Langmuir- Blodgett layers of amphiphilic molecules ... · CHAPTER 3 Structure and dynamics of...

Langmuir- Blodgett layers of amphiphilic molecules investigated by Atomic Force Microscopy

Langmuir- Blodgett lagen van amfifilische moleculen onderzocht met Atomic Force Microscopy

(met een samenvatting in het Nederlands)

Proefschrift

ter verkrijging van de graad van doctor aan de

Universiteit Utrecht op gezag van de rector

magnificus, prof.dr. W. H. Gispen, ingevolge het

besluit van het college voor promoties in het

openbaar te verdedigen op woensdag 23 mei 2007

des middags te 12.45 uur.

door

Aneliya Nikolova Zdravkova geboren op 10 december 1973, te Silistra, Bulgarije

Promotor: Prof.dr. J.P.J.M. van der Eerden

On the cover: Phase separation in binary mixed system of nuts (almonds and hazelnuts).

CONTENTS

CHAPTER 1 Introduction 1

1.1. Atomic force microscopy 2

1.2. Langmuir – Blodgett technique 4

1.3. Stability of Langmuir monolayer 8

1.4. Crystal structure of triglycerides 9

1.5. Outline of the thesis 12

CHAPTER 2 Phase behaviour in supported mixed monolayers of alkanols,

investigated by Atomic Force Microscopy 15

2.1. Introduction 16

2.2. Materials and methods 16

2.2.1. Chemicals 16

2.2.2. Aπ − isotherms 17

2.2.3. Langmuir - Blodgett film transfer 17

2.2.4. AFM measurement 17

2.3. AFM Observations 17

2.3.1. C16:C22 17

2.3.2. C18:C22 19

2.3.3. C18:C24 19

2.3.4. C16:C24 19

2.4. Thermodynamics 20

2.5. Conclusions 23

CHAPTER 3 Structure and dynamics of Langmuir – Blodgett Tristearin films:

Atomic Force Microscopy and theoretical analysis 25



3.2.1. Chemicals 27

3.2.2 Langmuir method 27


3.2.4. AFM measurements 28

3.3. Langmuir observations 29

3.3.1. Forced compression 29

3.3.2. Isobaric compression 30

3.4. AFM observation 32

3.4.1 Monolayer thickness 32

3.4.2. Initial structure, obtained by forced compression 36

3.4.3. Structural changes during isobaric compression 37

3.4.4. Stability of the transferred LB – film 41

3.4.5. Consistency of Langmuir and AFM data 41

3.5. Theory for nucleation, growth and coalescence of crystals 42

3.5.1. Qualitative interpretation of film evolution observations 42

3.5.2. Parameters and measurable variables 45

3.5.3. Avrami – Kolmogorov theory for coverage 45

3.5.4. Approximate theory for average crystal size and density 47

3.5.5. Interpretation of AFM – images of nucleation and growth 49

3.6. Conclusions 50

CHAPTER 4 Structure and stability of Triglyceride monolayers on water and

mica surfaces 53



4.2.1. Chemicals 55





4.3.1. Forced compression 57

4.3.2. Isobaric compression 60

4.4. AFM observations 64

4.4.1 Monolayer thickness 64

4.4.2. Stability of the transferred LB – film 68

4.4.2.1. Initial structure and structural changes of PPP – monolayer 68

4.4.2.2. Initial structure and structural changes of SSS – monolayer 71

4.4.2.1. Initial structure and structural changes of AAA – monolayer 75

4.5. Discussion 76

4.6. Conclusions 79

CHAPTER 5 Phase behaviour in binary mixed Langmuir-Blodgett monolayers of

Triglycerides 83



5.2.1. Chemicals 86





5.4. AFM observations 92

5.4.1 PPP – SSS structure 92

5.4.2 SSS – AAA structure 97

5.4.3 PPP – AAA structure 99

5.5. Discussion 102

5.6. Conclusions 106

CHAPTER 6 Summary 109

Samenvatting 113

List of Publications 117

Acknowledgements 119

Curriculum vitae 121

CHAPTER 1

Introduction

“Today…I propose to tell you of a real two-

dimensional world in which phenomena occur that

are analogous to those described in “Flatland”. I plan

to tell you about the behavior of molecules and

atoms that are held at the surface of solids and

liquids.”

I. Langmuir, Science 1936, 84,379

Since Irving Langmuir published his frist work on the study of two-dimensional systems of

molecular films at the gas-liquid interface [1], the interest in this area increased a lot. Many

scientists were fascinated by the idea to assemble individual molecules into highly ordered

architectures. They termed this materials engineering. Even though this is still a dream, the

Langmuir-Blodgett (LB) technique and Self-assembly (SA) process opened a window to the

realization of this goal. Presently LB and SA are widely used in areas like non-linear optics,

nanoelectronics, biosensors and piezoelectric devices [2].

Many molecules can form Langmuir films. We can describe them with one word-

amphiphiles. They have a hydrophilic head group and hydrophobic tail(s).The simplest amphiphilic

molecules are the aliphatic long-chain alcohols (CnH2n+1OH with n = 13-31). They form a

monolayer at the air-water interface, whose stability increases with the chain length. Other materials

like these are fatty acids and their salts, polymers, glycerides, phospholipids, pigments and proteins

[2, 3].

Self- assembled (SA) monolayers are molecular assemblies that are formed spontaneously

by the immersion of an appropriate substrate into a solution of an active surfactant in an organic

solvent [4, 5]. To investigate the surface and bulk properties of thin films, scientists use several

analytical tools. Ellipsometry to measure the thickness and uniformity of freshly prepared films;

1

Fourier transform infrared (FTIR) spectroscopy, in both grazing-angle and attenuated total

reflection (ATR) modes to learn about the direction of transferred dipoles, and to evaluate dichronic

ratios, molecular orientation, packing, and coverage; surface potential measurements to get

information on the coherence of the film at the water-air interface and on metal surfaces. A lot of

surface imaging technologies like X-ray Photoelectron Spectroscopy; Optical, Fluorescence,

Electron and Scanning Microscopy are used to study the surface topography [6].

In this thesis the main analytical tool, which was used for investigation is Atomic Force

Microscopy.

1.1. Atomic force microscopy

Atomic force microscopy (AFM) is one of the scanning probe microcopies. Common to these

techniques is that a probe is moved laterally (in x- and y- direction) across a sample surface, while

the height (z) or other parameters (force) are recorded. The first realization of this kind of

microscopy was the Scanning tunneling microscopy (STM) in 1981 by Binning and Rohrer [7]. An

electric current is measured due to electrons tunneling from a metal tip to a conducting sample. The

disadvantage of STM, that it is useful only for conducting samples, inspired scientists to generalize

this technique. This led to the invention of the atomic force microscopy (AFM) in 1986 [8]. AFM is

capable of scanning non-conductive samples. In an atomic force microscope a small tip on the end

of a cantilever-type spring is used as a probe. As a raster-scan drags the tip over the sample, some

sort of detection apparatus measures the vertical deflection of the cantilever, which indicates the

local sample height. The simplest deflection monitoring system is the laser beam reflection system.

A scheme of atomic force microscope setup is shown in Fig.1.

2

DetectorElectronics

AB

SplitPhotodiodeDetector

X ,Y

Z

Sample

Cantilever & Tip

Scanner

Laser

ControllerElectronics

Feedback Loop MaintainsConstant Cantilever Deflection

Measures

A + Bof deflection signal

A - B

Fig.1. Schematic presentation of Atomic force microscope setup.

The sample is mounted on top of a piezo crystal, which is used to position the sample very

accurately relative to the tip. A few micrometers above the sample a cantilever with the integrated

pyramidal tip is placed. A horizontally split photodetector detects the reflection of the laser beam

from the back of the cantilever. With the signal from this detector the point of contact of the tip with

the sample can be detected when the tip is lowered. Once the tip is in contact with the sample the

surface can be scanned. The distance the scanner moves vertically at each (x, y) data point is stored

by a computer to form the topographic image of the sample surface.

The AFM mode where the AFM tip is continuously in contact with the sample surface is

called Contact mode. Thus, in contact mode the AFM measures the repulsion force between the tip

and sample. The tip attraction by the capillary force determines the minimal force that can be used

in the AFM measurements, which is a few nanonewtons.

When measuring in air, damage of a sample by the AFM tip can not always be prevented. In

some cases it is useful to remove a small part of the sample material to investigate the thickness of a

complete layer. This can be done by increasing the setpoint, which causes the cantilever to move

3

downwards. From the observed change in the tip deflection the force increase can be calculated by

multiplying the change in distance with the spring constant of the cantilever. The maximum force,

which can be applied, is 200 nN for a cantilever with a spring constant of 0.6 N/m. Because of the

softness of the organic layers, described in this thesis, we did not use scanning forces beyond 30 nN

to make a hole in the layers. To prevent sample damaging, a different way of scanning the sample

with the AFM tip was invented in 1993: Tapping mode AFM [9]. It is a modulated technique where

the tip or the sample is subjected to a periodic vertical oscillation [10]. The advantage of this

technique is that the samples are less damaged by the forces exerted by the tip on the sample. The

disadvantage is that the Tapping mode AFM has slightly slower scan rate than contact mode AFM.

In general AFM has a lot of advantages, like very high resolution (for instance in contact

mode ‘atomic resolution’ images can be obtained). AFM is suitable tool for in-situ measurements,

i.e. materials can be studied in their natural environment [11]. Recently AFM was used for force

measurements in biological systems, for instance the strength of interaction of a membrane protein

in its natural surroundings [12-14].

AFM has also disadvantages. One of them is the heating of the sample by the laser beam

light. Another is the artifacts in the images caused by the interaction of the tip and the sample.

Despite of the disadvantages, AFM is one of the best techniques for observation of surfaces made of

different materials.

1.2. Langmuir-Blodgett technique

It is known that the surface structure of some materials is different from the bulk structure, which

leads to different macroscopic properties as compared to the bulk structure. An example for such

materials is provided by the triglycerides, which in crystals and in bulk solutions adopt a chair or

tuning fork conformation [15], but on the air-water interface they rearrange in a trident

conformation (all hydrocarbon chains pointing toward the same direction) [16, 17]. A detailed

description of the properties of triglycerides at the air-water interface is given in this thesis.

AFM can be used to study surface properties of materials. For this goal thin films are

transformed onto solid substrates via various deposition techniques. The technique we used is

Langmuir-Blodgett technique. This is the commonly used technique for preparation of monolayers

at air-water (or liquid-gas interface in general) interface and their transfer onto solid substrate. It

4

was introduced first by Irving Langmuir [1] and applied extensively by Katharine Blodgett. It

involves the vertical movement of a solid substrate through the monolayer - air interface [18].

In a Langmuir experiment a solution of amphiphilic molecules in an organic solvent is

spread on a liquid-vapor interface. An amphiphile is a molecule that is insoluble in water. One end

is hydrophilic, and, therefore, is preferentially immersed in the water and the other end is

hydrophobic, and preferentially resides in the air. Note that triglycerides, which are the major

substance investigated in this thesis, are lipophilic molecules. However, as an important finding of

our investigations, triglycerides spread as a monolayer on an air-water interface. They adopt a

trident conformation in which glycerol groups are immersed in the water phase and the hydrophobic

tails point into air. Therefore triglycerides behave as amphiphiles in this respect.

In a typical experiment a droplet of triglyceride solution is dripped on a water surface. After

spreading the solvent evaporates and the amphiphiles arrange in monomolecular layer (monolayer).

The molecular layer at the air-water interface is called Langmuir film [6, 19, and 20]. A typical

setup for LB experiments is a Teflon (PTFE) trough with three rigid walls and one movable barrier

(fig.2).

SubstrateWilhelmyplate

Barrier (PTFE)

Trough(PTFE)Amphiphilic molecules

SubstrateWilhelmyplate

Barrier (PTFE)

Trough(PTFE)Amphiphilic molecules

Fig.2. Schematic presentation of Langmuir-Blodgett Trough

By moving the barrier the monolayer can be compressed from an expanded state to a close packing

of the molecules. The amphiphiles have very small interaction, when the distance between them is

large. In this case they have very little effect on the surface tension on the subphase (usual it is

water). When the barrier compresses the layer, the molecules start to interact, which can be

5

regarded as a two dimensional analog of pressure, called surface pressure π . It is defined as

follows:

0π γ γ= − (1)

where 0γ is the surface tension in the absence of a monolayer, and γ the value with the monolayer

present. When the barrier is moved, the area of the film ahead of the barrier changes with , and

the area of the film behind the barrier by

TdA

,0TdA dAT= − . If the compression is isothermal, the Gibbs

free energy, G , of the total surface changes by:

0 ,0 0( )T T TdG dA dA dA dATγ γ γ γ π= + = − ≡ − (2)

The surface tension is measured with a Wilhelmy plate. This is usually a small platinum plate,

which is wetted completely. The downward force on a plate with length l, width w, and thickness t,

with a density pρ , immersed to a depth h in a liquid of density lρ is given by:

2 ( ) cospF glwt t w gtwl hρ γ θ ρ= + + − (3)

Where θ is the contact angle of the liquid on the solid plate, usually taken to be 0, and g is the

gravitational constant. From Eq.(3) changes of the surface tension γ are reflected as changes of the

force . F

π is recorded at constant temperature as a function of the surface area per molecule A , resulting in

a Aπ − isotherm. The measurement of A is straightforward, because it is linearly dependent on the

position of the barrier. A typical Aπ − isotherm is shown in Fig.3.

6

Area per molecule, A

Surfa

ce p

ress

ure,

ΠC

E

G

Phase transition

Phase transition

Fig.3. Schematic presentation of an ideal Aπ − isotherm (G - gaseous phase, E - expanded phase,

C - condensed phase).

A few regions are distinguished, corresponding to several phase transitions. These are,

almost, analogous to the three-dimensional gases, liquids and solids.

In the “gaseous” phase (G in fig.3), the molecules are far enough apart on the water surface

that they exert little force on one another. When the surface area of the monolayer is reduced, the

hydrocarbon chains will begin to interact. The state which is formed is called “expanded “phase

(E).The hydrocarbon chains of the molecules in such a film are in random, rather than regular

orientation, with their polar groups in contact with the subphase. The closest packed state is a state

in which the molecules have a packing resembling the packing in a two dimensional crystal. This is

referred to as the “condensed” phase (C). The area per molecule in such a state will be similar to the

cross-sectional area of the hydrocarbon chain, i.e., ≈ 0.19 nm2 molecule -1. If the monolayer is

compressed even further it collapses, resulting in a sudden decrease in the surface pressure. This is

referred to as collapse.

At and beyond the collapse pressure molecules are forced out of the monolayer and form

other structures, depending of their nature. For example, fatty alcohols and acids form micelles

beyond the collapse pressure. In micelles the molecules are arranged in spheres, with the polar head

groups on the outside and the hydrocarbon chains towards the center. Another arrangement is

characteristic of phospholipids molecules, which is called vesicles. In this arrangement, the double

layers form a shell with water both outside and inside [20]. In some cases multilayers can be

formed, when the monolayer is compressed on interface. E.g. for long-chain esters, up to eight

7

layers on top of each other were obtained [21]. This structure of multilayers on top of the monolayer

is typical also for triglycerides and bile acids [16, 17, and 22]. Recently was found that a single-

chain fatty acid methyl ester forms an unconventional air-stable interdigitated bilayer at the air-

water interface [23].

To investigate these structures the monolayers have to be transferred on a solid substrate,

which is either hydrophilic of hydrophobic. To achieve this, the method developed by Blodgett is

most frequently used and is commonly referred to as the Langmuir-Blodgett technique. With this

technique layers of molecules are deposited on a solid substrate by vertically dipping through the

liquid-vapor interface. During the deposition the surface pressure is kept constant by moving the

barrier to compensate the loss of the material that is transferred on the substrate. The typical dipping

speed is a few mm/s. It must be slow enough to allow the water to drain from the monolayer –

substrate interface and also to let films with a high viscosity adjust in the neighborhood of the

moving substrate. The most commonly used materials as substrates are mica, glass slides, oxidized

silicon wafers and graphite. Before the transfer the substrates can be treated to make them

hydrophilic or hydrophobic. It is possible to create multilayers by repeated dipping of the substrate.

One of the most used techniques for characterization of LB-films is AFM [25-28].

1.3. Stability of Langmuir monolayer

By definition a Langmuir monolayer is thermodynamically stable if under isobaric conditions at air-

water interface it does not change its structure. Conditions for thermodynamic stability can in

principle be established by measuring the equilibrium spreading pressure eqπ , i.e. the pressure at

which the surface area of the film does not change with time [3]. At this point it is important to

clearly discriminate between collapse pressure colπ and equilibrium pressure eqπ . For eqπ we use

the definition of Roberts [3]. The thermodynamic equilibrium (spreading) pressure is the surface

pressure that is spontaneously generated when a sample of solid material in its thermodynamically

stable phase, i.e. in the crystalline phase, is brought in contact with the water surface. Provided that

sufficient time is allowed for equilibration, one can, in principle, be sure that the monolayer which

has been formed by molecules detaching themselves from the crystal surface and spreading over the

subphase is in equilibrium with the crystals themselves. At surface pressures higher than eqπ there

will be a tendency for the monolayer to aggregate into crystals [3].

8

If the monolayer is compressed at a constant rate, at certain pressure it will collapse,

resulting in a sudden decrease in the surface pressure. This pressure is called collapse pressure. The

only way to determine the thermodynamic stability of the monolayer is to investigate it under

isobaric conditions at spreading pressures colπ π< . Note that sometimes one refers to equilibrium

spreading pressure if actually collapse pressure is meant, see e.g. [30].

It was found that some Langmuir monolayers are unstable at air-water interface at surface

pressures below the collapse pressure ( colπ π< ). One of the factors causing the loss of molecules

from the monolayer - “relaxation phenomena” can be desorption in the subphase, e.g. for

monoglycerides [29, 30], evaporation, e.g. for fatty acids. Other mechanisms, such as surface

rheology, surface chemical reaction, polar group hydration, the simultaneous motion of the

monolayer and the liquid substrate as a result of the surface pressure gradient, or structural

relaxation processes in the monolayer itself - such as change in the conformation of the molecules –

are difficult to quantify [24]. By definition these processes occur at pressure eqπ π> .

One of the surprising results of this thesis is that triglycerides, which are the main objects in

this work, also showed a thermodynamic instability at the air-water interface at surface pressures

far below the collapse pressure ( colπ π ). Under isobaric conditions at surface pressures eqπ π> a

molecular rearrangement process takes place which effectively thickens the film. Using Atomic

Force Microscopy for triglycerides we have shown that this process involves the growth of 3D

crystals of triglycerides on top of the monolayer, which is precisely what one should expect for

eqπ π> . For colπ π> similar crystallization processes take place, but in a less controlled and less

reproducible manner.

1.4. Crystal structure of triglycerides

Triglycerides (TAGs) are esterifications of three long-chain fatty acids with glycerol. Many

different types of TAGs exist because the three acids can all differ in chain length and degree of

saturation. The general formula for TAGs is:

CH -O-CO-R2 1

CH -O-CO-R2

CH -O-CO-R2 3

9

TAG molecules are able to pack in different crystalline arrangements or polymorphs, which exhibit

significantly different melting temperatures [15, 31]. It is well known that TAGs may crystallize in

the α (hexagonal, less stable), 'β (orthorhombic), or β (triclinic, most stable) form. However,

some fats display more polymorphs than this [32].

TAG molecules are “three legged” molecules that can pack with the acyl chains(“legs”) in

one of two conformations, neither of which involves all three chains packing alongside each other.

They can pack in a “chair” conformation where the acyl chain in the 2 position is alongside the

chain on either the 1 or 3 positions. Alternatively, a “tuning fork” conformation can be adopted

where the acyl chain in the 2 position is alone and the chains in the 1 and 3 positions pack alongside

each other (Fig.4)

Fig.4. Schematic representation of a tuning fork conformation (a) and a chair conformation (b). Either conformation naturally packs in a chair-like manner. The stacking of these chairs can

be in either a double or triple chain length structure and these stack side by side in crystal planes

(Fig.5).

τ

LL

τ

Double Triple

Fig.5. Schematic arrangement of triglycerides in double and triple layers. Both patterns may lead

to α , 'β or β crystalline phase.

10

The difference between polymorphs is most apparent from a top view of these planes, which

shows the subcell structure (Fig.6). These structures can be identified by X-ray diffraction patterns

[32].

H O T

α β’ β

Fig.6 Schematic presentation of the subcell structure of the three most common polymorphs in

TAGs (viewed from above the crystal plane).

The layer thickness or long spacing (L) gives information on the repeat distance between

crystal planes and obviously depends on the length of the molecules and, furthermore on the tilt

angle (τ ) between the chain axes and the basal plane. In the α phase the chains are oriented

perpendicular to the end-group plane (i.e. ). The o90τ = 'β and β phases have tilted chains (Fig.5).

The short spacing gives information on subcell structure (interchain distances). These

interchain distances depend on how the chains pack together and this is complicated by the “zigzag”

arrangement of successive carbon atoms in aliphatic chains. Closer packing is achieved when the

zigzag of adjacent chains are in step with each other (“parallel”) as opposed to out of step

(“perpendicular”).

In α - phase the chains are arranged in a hexagonal structure (H). They are not tilted and are

far enough apart for the zigzag nature of the chains to not influence packing.

In 'β - phase the chain packing is orthorhombic and perpendicular (O┴). Adjacent chains are

out of step with each other and they do not pack closely. The chains are tilted at 50 - 70o.

In β - phase the chain packing is triclinic (T). Adjacent chains are in step (“parallel”), and

thus pack closely together. This is the densest polymorphic form. The chains are tilted at 50 - 70o

[32].

11

The CnCnCn-type (n = even) TAGs have double chain length structure and the most stable

phase is β . They have asymmetric “tuning-fork” conformation [33]. Because this is the type of

TAGs, which we investigated in this thesis, in the next chapters we will use “tuning-fork”

conformation to describe their crystal structure.

1.5. Outline of the thesis

Langmuir-Blodgett technique and Atomic force microscopy were used to study the phase behaviour

of organic molecules at air-water and air-solid interfaces. Chapter 2 reports the structure of binary

mixed LB monolayers of fatty alcohols. It describes the dependence of phase separation phenomena

on the difference between the chain lengths of the two components and the surface pressure.

Chapter 3 reports the structure and temporal evolution of tristearin (SSS) monolayers at air-water

interface. In order to study the thermodynamic stability of SSS monolayers, they were incubated at

air-water interface, withdrawn and imaged with AFM. During incubation a crystal growth process

took place. A new model was developed to quantitatively describe this process. The crystal growth

theory for tristearin, which we propose was checked by investigating and comparing two more

triglycerides –tripalmitin (PPP) and triarachidin (AAA). In Chapter 4 we show the influence of the

chain length of triglycerides molecules on their stability on water and mica surfaces. Chapter 5

describes the phase behaviour of binary mixed LB- monolayers of triglycerides. We investigated the

relation between phase separation and chain length. In Chapter 6 all results presented in this thesis

are summarized and discussed.

References:

[1] Langmuir, I., The mechanism of the surface phenomenon of floatation, Trans. Faraday Soc.,

15(1920)62-74

[2] Petty, M.C., Langmuir-Blodgett films an introduction, Cambridge University Press, (1996)

[3] Roberts,G., Langmuir-Blodgett film Plenum Press, New York, (1990)

[4] Bigerow, W.C., Pickett, D.L., Zisman, W.A., J. Colloid Interface Sci. 1(1946) 513

[5] Zisman, W.A., Adv. Chem. Ser. 1 (1964) 43

[6] Ulman, A., An introduction to ultrathin organic films, Academic Press, London, (1991)

[7] Binning, G. and Rohrer, H., Helv.Phys. Acta 55 (1982) 726

[8] Binning, G., Quate, C.F. and Gerber, C., Phys. Rev. Lett. 56 (1986) 930

12

[9] TappingModeTm is patented by Digital Instruments, Santa Barbara CA, USA

[10] Radmacher, M., Tillman, R.W., Fritz, M . and Gaub, H.E., , Sciences 257 (1992) 409

[11] Rinia, H.A., Atomic force microscopy on domains in biological model membranes, Ph.D.

Thesis (2001), Utrecht University, The Netherlands

[12] Merkel, R. Phys. Rep. 346 (2001) 343-385

[13] Maeda, N., Senden, T.J., and Di Meglio, J.M., Micromanipulation of phospholipids bilayers by

AFM, Biochem. Biophys. Acta 1564 (2002) 165-172

[14] Ganchev, Dragomir N., Rijkers, D.T.S., Snel, M.M.E., Killian, J.A. and de Kruijff,

B.,Biochemistry 43(2004) 14987-14993

[15] Garti, N., Sato, K., In Crystallization and polymorphism of fats and Fatty Acids; Dekker, M.

New York (USA) 1988

[16] Bursh, T., Larsson, K., Chem. Phys. Lipids 2 (1968) 102-113

[17] Hamilton, J. A., Small, D.M., In Proc. Nat. Acad. Sci. USA 78 (1981) 6878

[18] Blodgett, K. B., Monomolecular films of fatty acids on glass, J. Am. Chem. Soc.,56 (1934) 495

[19] Gaines G.L., Insoluble monolayers at liquid gas interfaces, Wiley, New York, (1966)

[20] Petty, M.C., Langmuir-Blodgett films: an introduction, Cambridge University Press, 1996

[21] Lundquist, M., In Surface chemistry, Copenhagen. Munksgaard (1966) p.294

[22] Ekwall, P., Ekholm, R. and Norman, A., Acta Chem. Scand. 11 (1957) 703

[23] Chen X., et al., J.Phys. Chem. B 109 (2005) 19866-19875

[24] Chi, L.F. et al, Langmuir 8 (1992) 2255-61

[25] Flörsheimer, M., et al, Thin Solid Films 244 (1994) 1078-82

[26] Zasadzinski, J.A. et al., Science 263 (1994) 1726

[27] Sparr. E., Langmuir 17 (2001) 164-172

[28] Porter, M.D., Bright, T. B., Allara, D.L., Chidsey, C.F.D., J. Am. Chem. Soc. 109 (1987) 3559

[29] Fuente, J.F. and Rodriguez Patino, J.M. Langmuir, 10 (1994) 2317-2324

[30] Sanchez, C.C., Rodriguez Nino, M., Rodriguez Patino, J.M., Colloids and Surfaces B:

Biointerfaces 12 (1999) 175-192

[31] Chapman, D., Chem Rev 62 (1962) 433

[32] Hamawan, C., Starov, V.M., Stapey, A., Advances in Colloid and Interface Science 122 (2006)

3-33

[33] De Jong, S., PhD thesis (1980) University of Utrecht, The Netherlands.

13

CHAPTER 2

Phase behaviour in supported mixed monolayers of alkanols, investigated by Atomic Force Microscopy

Abstract

The structure of several mixed Langmuir-Blodgett monolayers of fatty alcohols, CnH2n+1OH

with even n = 16-24, was investigated by AFM at 20-22°C. Phase separation was found for

compressed films, if the chain length of the two components differed at least with six carbon atoms.

A strong dependence of the domain shape on the surface pressure was observed. The excess Gibbs

energy vs. surface pressure and mole fraction was calculated from π-A isotherms. In line with

thermodynamic expectation, the tendency of phase separation increased with increasing . A

surprising and as yet unexplained result was that we sometimes observed clear phase separation

already in the range

exG∆exG∆

0.1exG R∆ ≅ T

15

2.1. Introduction

Aliphatic long-chain alcohols Cn (CnH2n+1OH with n = 16–31) can be adsorbed on water surface.

Interestingly, adsorbed Cn turns out to enhance ice nucleation. Grazing-incidence X-ray diffraction

(GID) studies of Cn monolayers on water at 5°C revealed two-dimensional structure formation.

Wang et al. [1] concluded that the molecules in the Cn monolayers adopt a herringbone pattern.

According to the GID data, monolayers with n = 16 and 20 contain less crystalline material than

monolayers with n = 23, 30, 31 [2]. IR spectra of the same alcohol monolayers at an area per

molecule of 20 Å2 have been measured at the air/water interface at 20°C. These measurements also

showed that the hydrocarbon chains become more ordered with increasing length. It was found that

only alcohols with molecular areas of 18.5-20 Å2 significantly enhance nucleation of ice [2].

Combining these two types of experiments we expect that 2D layers of long alcohols (n > 20)

crystallize, when the molecular area is about 20 Å2. Kulkarni et al. [3] investigated mixed

monolayers of C16 and C22 at 25°C, studying surface viscosity and the area per molecule. Isotherms

of the system at five different mole fractions showed that all mixtures were thermodynamically non-

ideal.

In order to better understand mixed monolayers and to study the effect of chain length on

mixing, we investigated six mixed monolayer films thermodynamically and with AFM: C16:C22

with stoichiometry 1:1, 1:3 and 3:1, C18:C22 (1:1), C18:C24 (1:1) and C16:C24 (1:1). We used

Langmuir-Blodgett technique to transfer at several surface pressures binary mixed monolayers from

the water/air interface onto a mica substrate. Equilibrium layers were obtained by using a very small

initial surface pressure (π = 0 mN/m) of the Langmuir layer, and compressing slowly to the final

pressure.

2.2. Materials and methods

2.2.1. Chemicals:

Film material: Fatty alcohols (CnH2n+1OH, with n = 16, 18, 22, 24) were obtained from Merck and

used without further purification. Separate stock solutions of each alcohol with concentration of

5 mM in distilled chloroform were prepared. Solutions containing 1 mM mixtures of the alcohols in

mole ratios 1:1, 1:3 and 3:1 were prepared by proper mixing and diluting of stock solutions.

16

Subphase: MiliQ water was used as a subphase in our Langmuir system for all experiments. The

resistivity of the water is 18 MOhm*cm.

Substrates: All monolayers were transferred onto freshly cleaved mica.

2.2.2. π - A isotherms

Compression isotherms were measured on a Teflon trough (17.2×35.7 cm). The spreading pressure

π was measured with a Wilhelmy type balance consisting of a platinum plate coupled to an

electrobalance (Cahn Ankersmit 2000), with an accuracy of about 0.1 mN/m. The film material was

spread on the water subphase, using a 100 µL Hamilton syringe. The area per molecule A was

controlled by a moving barrier, at an accuracy of 1-2 Å2 per molecule. Spreading took place at

Å100A ≈ 2. Film compression started almost immediately after spreading, at a rate of 1 cm/min.

2.2.3. Langmuir-Blodgett film transfer

In order to obtain LB films, first a substrate was immersed perpendicularly in the aqueous subphase.

Equilibrium layers were obtained by using a very small initial surface pressur 0e (π = mN/m) of

the monolayer, and compressing slowly (1 cm/min) to the final pressure. Film transfer was then

accomplished by vertically lifting the substrate through the air-water interface at a speed of

2 mm/min. After deposition the monolayers were dried in air and kept in close containers until use.

All experiments were done at 20-22°C.

2.2.4. AFM measurements

The samples were examined with AFM within about 5 hours after preparation. Imaging was done

with a Nanoscope III (Digital Instruments) in contact mode with oxide-sharpened silicon nitride tip

(k = 0.06 N/m). The AFM was equipped with E scanner.

2.3. AFM Observations

2.3.1. C16:C22

17

A B

C D

E F

µm

1.038 nm 0.983 nm

0 1.25 2.5

02.

0-2

.0

AA BB

CC DD

EE F

µm

1.038 nm 0.983 nm

0 1.25 2.5

02.

0-2

.0

F

µm

1.038 nm 0.983 nm

0 1.25 2.5

02.

0-2

.0

Fig.1. AFM height image showing C16:C22 (1:1) mixed monolayers transferred at surface pressure

(A) π = 10 mN/m, (B) π = 15 mN/m, (C) π = 20 mN/m and (D) π = 35 mN/m. In panel E, an

enlarged height image is given, showing the tetragonal shape of C22 domains with corresponding

cross section in (F). The height difference between both alcohols is given by the vertical distance

between the markers. The scale bar is 1µm and the vertical scale is 4 nm for all images.

18

The AFM images in Fig.1 clearly show phase separation at surface pressures π ≥ 10 mN/m.

The thicker domains presumably mainly consist of C22, the thinner mainly of C16. The thickness of

the C22 domains and of the surrounding C16 film was found to be less than the thickness calculated

from X-ray data of crystals with vertically extended alcohols. The measured values are 1.0 nm (1.87

nm in crystals) for C16 and 2.0 nm (2.51 nm in crystals) for C22. This effect was observed before and

explained as monolayer depression, caused by the AFM tip [4]. Fig.1 F shows the height difference

between C16 and C22, to be 0.9 - 1.0 nm.

Phase separation in domains with the same height difference was found for C16:C22 mixtures with

(1:3) and (3:1) stoichiometry (data not shown).

2.3.2. C18:C22

The AFM images of this system showed a homogeneous monolayer at all surface pressures at

which the monolayer was compressed (π = 10, 20 and 35 mN/m). The measured thickness of the

monolayer was ~ 2.1 nm at π = 35 mN/m, as for C22.

2.3.3. C18:C24

This mixture with 6 carbon atoms length difference behaved similar to C16:C22. The AFM images

showed phase separation with C24 domains embedded in C18. At π = 35 mN/m the domains have

tetragonal shapes and they are more ordered than in the C16:C22 mixture. The measured thickness

for C18 is ~1.6 nm (2.09 nm in crystals) and for C24 it is 2.2~2.3 nm (2.7 nm in crystals) (Fig.2 A,C).

2.3.4. C16:C24

In this mixture we observed different C24 domain shapes as in other mixtures, they were very

irregular, at all final surface pressures. The height difference between the C24 domains and the C16

film is 1.1~1.2 nm (Fig.2 B, D).

19

B

D

0 1.25 2.5

0

-2.0

2.0

µm

1.198 nm 1.144 nm

A

C0.600 nm

0 1.25 2.5µm

0

2.0

-2.0

0.600 nm

BB

D

0 1.25 2.5

0

-2.0

2.0

µm

1.198 nm 1.144 nm

D

0 1.25 2.5

0

-2.0

2.0

µm

1.198 nm 1.144 nm

0 1.25 2.5

0

-2.0

2.0

µm

1.198 nm 1.144 nm1.198 nm 1.144 nm

A

C0.600 nm

0 1.25 2.5µm

0

2.0

-2.0

0.600 nm

C0.600 nm

0 1.25 2.5µm

0

2.0

-2.0

0.600 nm0.600 nm

0 1.25 2.5µm

0

2.0

-2.0

0.600 nm

Fig.2. AFM height image showing mixed monolayers transferred at surface pressure π = 35 mN/m.

(A ) C18:C24 and (B) C16:C24 (A ) with the corresponding cross sections in (C and D). The scale bar

is 1µm and the vertical scale is 4 nm for all images. The black lines show the area in the image,

where the cross section was taken.

2.4. Thermodynamics

In order to interpret the observed structures of mixed alcohol films, we introduce

thermodynamic information. The films were formed at the same temperature T 294 K, hence we

drop the temperature from the formulation. At given spreading pressure

≈

π and mole fraction the

structure with the lowest possible Gibbs energy (in J/mol) will be formed. Let be the Gibbs

energy for a homogenous, uniform film. If

x

G homG

( ),homG x π is a concave function of then a

homogeneous film is thermodynamically stable and . If

x

homG G= ( ),homG x π has a convex part then

a homogeneous film is unstable for a composition interval ( )0 1,x x x∈ that includes the convex part.

A homogeneous film with ( )0 1,x x x∈ can decrease its Gibbs energy by phase separating in

20

fractions ( ) ( )0 1 0/x x x x x= − − and ( ) ( )1 11 / 0x x x x x− = − − with composition 1x and

0x respectively:

( ) ( ) ( ) ( ) ( ) ( )0 1 0 1, , 1 , , hom ,x x x G x x G x xG x G xπ π π∈ → = − + < π (1)

The points ( )0 0x x π= and ( )1 1x x π= are the common tangent points to . A completely

immiscible film separates in pure phases, i.e.

homG

0 0x = , 1 1x = and hence x x= .

The Gibbs energy per mol can be determined, using ( )/x

G Aπ∂ ∂ = from Aπ − diagrams

(2) ( ) ( ) ( ) ( ) ( )( )A

G G A d A RT A dπ

π π

π π π π π π π∞

∞

∞ ′ ′− = ≈ − −∫ A∫

where π ∞ is the reference spreading pressure, which is chosen small enough that the film is

thermodynamically ideal at π ∞ . The Gibbs energy is split into an ideal and an excess part:

( ) ( ) ( ), ,id exG x G x G xπ π= + ,π (3)

( ) ( ) ( ) ( ) ( ) ( ) ( )0 1,1 ln 1

idG x G Gln 1x x x x x x

RT RT RTπ π π

= − + − − − − (4)

The Gibbs energy of mixing is defined as

( ) ( ) ( ) ( ) ( )0 1, , 1mixG x G x x G xGπ π π= − − − π (5)

Since , we get ( ) 0exG π ∞ = ( ),exG x π of a mixture using the right hand side of Eq. (2) for the mixed

and pure films and substituting in Eq. (5). By definition a mixture is non-ideal if ( ), 0exG x π ≠ .

Phase separation only occurs in non-ideal mixtures. Using Eqs. (3)-(5) it is seen that for ideal

mixtures the so-called additivity rule

( ) ( ) ( ) ( )0 1, 1A x x A xAπ π= − + π (6)

holds [5]. The reverse is not true. Indeed, Eq. (6) holds for completely immiscible films as well. It is

often believed that ( ),exG x π > 0 is necessary for phase separation to occur. In the most common

case where phase separation is driven by energetically unfavourable mixing, i.e. for all

,

, 0hom exG ≥

x ( ),exG x π > 0 indeed. But phase separation may occur also if energetical or entropical reasons

favour incorporation of a small fraction of the other component in a pure phase. Then may

have negative minima near

,hom exG

0x and 1x , and a maximum in between. This can cause phase separation

with ( ), 0exG x π < .

21

From the discussion so far it is clear that we need to measure ( ),exG x π as accurately as

possible. At the low reference pressure 0π the mixed film is ideal. We assume that upon decreasing

the film area, it stays ideal down to the molar area ( )A A x∗= where π starts to increase. The

results are given in Figure 3.

-0.3

-0.2

-0.1

0

0.1

0.2

0.3

0.4

0 5 10 15 20 25 30 35 40 45

surface pressure π (mN/m)

Gex

/RT

C16/C22(1:1)C16/C24C18/C22C18/C24C16/22(1:3)C16/22(3:1)

Fig. 3. Excess Gibbs energy for mixed monolayers as a function of spreading pressure π. The

compositions of the mixture are given by the labels at the curves.

From the figure we see that, as in [3], ( ),exG x π is small as compared to RT for all

mixtures, and that the noise is relatively large. Due to noise the sign can not be determined

unambiguously for C mixtures. The fact that with AFM we clearly saw phase separation,

suggests a special interaction between the relatively flexible hydrophobic tails and

alcohols, favouring incorporation of a small amount of in and reverse. The

16 22: C

16C 22C

16C 22C ( )0.5,exG x π=

curve for the C mixture is similar to that for C , suggesting that the difference in chain

length is the main parameter for demixing trends. In line with this G x

16 22: C C

)18 24:

( 0.5,ex π= tends to be

negative for : , which favours homogeneous films and positive for :C , which favours

phase separation.

18C 22C 16C 24

22

2.5. Conclusions

In this study we have obtained AFM images that reveal the structure of mixed alkanol monolayers,

and we applied out thermodynamic measurements and theory to interpret our observations.

As the head groups are the same for all alcohols used in this study, the observed differences

in monolayer structure have to be explained with the methylene-methylene interactions of the tails.

The longer alcohols (C22 and C24) interact more strongly, hence in a condensed layer they adopt a

crystalline, herringbone crystal structure [1, 2, 6] than the shorter ones (C16 and C18), which can be

fluid like. This is in agreement with IR spectra for single alcohol monolayers at 20°C [2].

For surface pressures of π = 10 − 35 mN/m we found phase separation for all systems,

except for C18:C22, with domains of the longer alcohol, embedded in a shorter alcohol film. This

leads to the conclusion that in a condensed monolayer phase separation takes place when the chain

length difference is 6 or more carbon atoms. The greater the length difference is, the more

unfavorable is the mixing free energy, which is also shown from the thermodynamic data.

At high surface pressure, π = 20 − 35 mN/m, the domains get tetragonal shapes. This can be

understood as at higher pressures crystalline packing of molecules is favoured. The π - A isotherms

show an area per molecule 19-20 Å2 for these surface pressures. At lower pressures, π = 10 -

15 mN/m, the excess Gibbs energy is small. Then disordered packing is more favourable and

domains are rounded.

If the chain length difference is only 4 methylene units, both the AFM images and the

thermodynamic data of the C18:C22 mixture indicate no phase separation.

In the case of a chain length difference of 8 units the excess Gibbs energy is so large that the

driving force for phase separation might be beyond the limit where equilibrium structures are

formed. Hence we think that the irregular domain shapes in the C16:C24 mixture are growth shapes,

rather than thermodynamic equilibrium shapes.

The result that we observed phase separation already in the range where our thermodynamic

measurements indicated ∆Gex ≅ 0.1 RT is surprising, since one would expect spontaneous phase

separation only if ∆Gex≥1 RT. This can not be explained yet, but it might be due to a too high

compression rate around the spreading pressure where phase separation starts.

23

References:

[1] Wang, J.L., et al., J. Am. Chem. Soc. 116 (1994) 1192

[2] Popovitz-Biro, R., et al, J. Am. Chem. Soc. 116 (1994) 1179

[3] Kulkarni, V.S., et al., J. Colloid Interface Sci. 89 (1982) 40

[4] Ten Grotenhuis, E., et al., Colloids and Surfaces, A: Physicochemical and Engineering Aspects

105 (1995) 309-318

[5] Gains Jr., G.L., Insoluble Monolayers at Liquid-Gas Interfaces, Interscience, New York, 1966

[6] Gavish, M., et al., Science 250, Issue 4983, (1990) 973

24

CHAPTER 3

Structure and dynamics of Langmuir – Blodgett Tristearin films: Atomic Force Microscopy and theoretical analysis

Abstract

The structure and temporal evolution of tristearin (SSS) monolayers at the air-water interface at

20 ± 1°C are investigated with the Langmuir method. The deposited Langmuir- Blodgett (LB)

layers were investigated with Atomic Force Microscopy (AFM). The LB experiments showed that

adsorption isotherms obtained with commonly used compression rates do not correspond to

thermodynamic equilibrium. Under isobaric conditions at 10 mN/mπ ≥ the film area slowly

decreased ,which corresponded to the formation of crystals on top of the monolayer. The AFM

images reveal that SSS initially form trident monolayers at air-water interface. These layers are

thermodynamically stable at surface pressure 5 mN/mπ ≤ . The thickness of the trident monolayer

was found to be 1.6 to 1.8 nm, corresponding to tilt angles of the molecule chains varying from

at o43τ = 10 mN/mπ = to at o53τ = 40 mN/mπ = . For 10 mN/mπ ≥ growth takes place of

crystals with a tuning fork conformation of the SSS molecules on top of the trident monolayer. The

crystals grow with time, mainly in lateral directions. The growth rate increases with surface

pressure. A new model is developed to quantitatively describe the crystal growth process. A lateral

growth rate of 2.3 nm/min and a vertical growth rate of 0.005 nm/min were calculated for 1

individual crystal at 10 mN/mπ = .The same growth process that was observed on the air-water

interface was also observed in transferred monolayers at room temperature, though the growth was

much slower.

25

3.1. Introduction

Many efforts have been made in investigating the structure of triglycerides. Most of the published

work has been on homogeneous triglycerides (their 3 fatty acid residues are identical). In the solid

state, triglycerides adopt a polymorphic crystalline structure.

Depending on the crystallization procedure, especially the thermal treatment, they may

crystallize in the α (hexagonal, less stable), β’ (orthorhombic), or β (triclinic, most stable) form. In

each of these polymorphic forms the molecules have a tuning fork conformation [1, 2], but the

packing of these tuning forks is different.

However in monolayers at a hydrophilic-hydrophobic interface, triglyceride molecules adopt

a trident conformation (all hydrocarbon chains pointing toward the same direction). This

conformation has been proposed by Bursh and Larsson, based on their Aπ − diagrams for

triglycerides on water at different temperatures [3]. The trident conformation was also found by

Hamilton, using NMR measurements for tripalmitin and triolein at the oil-water interface in

phospholipids vesicles [4, 5] and by Claesson for triolein in contact with mica [6]. In the trident

conformation the hydrophilic glycerol group is in contact with the water or the mica surface, and

the hydrophobic chains point into the air or oil. In some cases multilayers can be formed, when on

an interface a monolayer is compressed laterally [7-9].

Bursh and Larsson investigated what happened when a monolayer of triglyceride at the air-

water interface is compressed beyond the so-called collapse pressure, where the steady increase of

the spreading pressure upon lateral compression is interrupted. They concluded that some molecules

leave the monolayer to form new molecular layers. They proposed a trident conformation for the

first triglyceride monolayer and a tuning fork conformation in the next layers, with a packing

similar to that in the crystalline state [3]. Triple layer formation was reported also for bile acids

[10]. Only a few studies of triglycerides with Atomic force microscopy (AFM) were performed [11,

12]. Michalski investigated Langmuir-Blodgett monolayers on glass of tripalmitin by AFM [12].

The monolayer was compressed and withdrawn at a surface pressure, corresponding to the middle

of the condensed phase in the Aπ − . She suggested that the trident monolayer generally

reorganizes after being transferred to the glass, forming two different structures. The first one

corresponds to bilayers in a regular tuning fork crystalline structure. The second one corresponds to

the triple layer structure, proposed by Bursh and Larsson [12].

26

The aim of this chapter is to better understand the molecular structure and processes in

triglyceride films at the air-water interface (Langmuir film) and on a solid surface like mica

(Langmuir-Blodgett (LB) film). Therefore we measured the Aπ − (spreading pressure π vs area

per molecule A ) diagram of Langmuir films and we investigated LB films with AFM. In this

chapter we focus on tristearin (SSS), in subsequent chapters we extend the investigations to other

triglycerides. Starting with a Langmuir film at very small π , where the film is in a low-density

“gas” phase, we compressed the film, at a constant rate, to the desired pressure π (forced

compression). To investigate whether the Langmuir film was in thermodynamic equilibrium at this

pressure π , we sometimes left the film for some time at pressure t π (isobaric compression). The

Langmuir film was transferred to mica directly after forced compression ( ) or after

or of incubation time at constant pressure

0t =

30 mint = 60 mint = π (isobaric compression).


3.2.1. Chemicals

Film material: Tristearin (1, 2, 3, -trioctadecanoylglycerol: SSS) was purchased from Larodan with

a stated purity of >99 mass %. A stock solution of SSS with concentration of 1 mM in distilled

chloroform was prepared.

Subphase: Distilled water was used as a subphase in our Langmuir system for all experiments. The

resistivity of the water was 15 MOhm cm.


3.2.2. Langmuir method

Compression isotherms were measured on a home made instrument, using available components.

The instrument was equipped with a Teflon trough (8.6 ×14.8 cm). The spreading pressure π was

measured with a Wilhelmy type balance consisting of a platinum plate coupled to an electrobalance

(Cahn 1000, Ankersmit), with an accuracy of about 0.1 mN/m. The film material was initially

spread on the water subphase, dropping 20 µL of 1 mM SSS dissolved in chloroform, using a 25 µL

Hamilton syringe. The conditions were chosen such that initially the average area A per molecule is

.We started (asymmetric) film compression 2 min after spreading. In our system two 2110 ÅA ≈

27

modes of operation were available. First forced compression, where the position of the barrier, and

hence the trough length ( )l t ahead of the barrier, is given. Then the resulting spreading pressure

( )tπ is registered. In this mode we chose barrier velocities of the order of 1 cm/min, which

according to the literature should be slow enough that the Langmuir film stays close to

thermodynamic equilibrium.

Second we used the isobaric compression mode, where a constant spreading pressure π is

applied and the resulting trough length ( )l t is monitored. Obviously if the film is in equilibrium at

the applied pressure, then is constant. In practice however we often found the barrier to move

with velocities of the order of 1 m

( )l t

/secµ . This barrier motion reflects rearranging processes in the

Langmuir film. We use AFM images to interpret and quantify this process.


In order to obtain LB films, first a substrate was immersed perpendicularly in the aqueous subphase.

We started with a very small initial surface pressure ( 0π = mN/m), and compressed the monolayer

slowly (1 cm/min) to the final pressure. To obtain a LB film that is characteristic for forced

compression, the film was then transferred immediately by vertical pulling of the substrate through

the air-water interface at a speed of 2 mm/min. During the transfer the surface pressure was kept

constant by appropriately moving the barrier. The transfer process takes a few minutes.

In order to study the structural changes of the Langmuir film during isobaric compression

the film was left at constant pressure for 30 or 60 min before it was transferred to the substrate.

After deposition the LB-films were dried in air and kept in close containers until use. All

experiments were done at . o20 1 C±


The samples were examined with AFM within about 5 hours after preparation. We checked that the

length of this delay time is not critical. Imaging was done with a Nanoscope(R) IIIa (Digital

Instruments, Santa Barbara, CA) in contact mode with oxide-sharpened silicon nitride tip (k = 0.06

N/m). The AFM was equipped with a J scanner (176 x176 µm; z limit = 5.349 µm). All images

were processed using procedures for flattening in Nanoscope III software version 5.12r5 without

28

any filtering. To check if the monolayer is successfully transferred to the mica surface we measured

at least five different spots (each 150 µm2 ) of every sample. In order to detect structural changes in

the adsorbed film in contact with air we studied LB films several days after preparation as well.

3.3. Langmuir observations

3.3.1 Forced compression

0

10

20

30

40

50

60

0 20 40 60 80 100 120

2Area / molecule A (Å )

Surf

ace

pres

sure

π (m

N/m

) Fig.1.Example of surface pressure vs

area isotherm of tristearin (SSS) at air-

water interface, at 20o C, obtained by

forced compression at a rate of

1cm/min (x - observed data; - fit using

Eq. (1) with ,

260.4ÅcondA =

258ÅcolA = and 40.5mN/mcolπ =

Fig. 1 shows a typical Aπ − isotherm of tristearin (SSS), recorded at a barrier velocity of

1 cm/min. Three different regimes can be recognized. Starting at a large area per molecule A the

pressure is low and increases only slowly with decreasing A. Upon decreasing A further the

condensation area is reached and the pressure starts to increase more rapidly. Compressing

further it is seen that for A below the collapse area the increase of the pressure is slow again.

The explanation of this characteristic dependence is that for

condA

colA

cond colA A A= = the SSS molecules are

close enough together to form a condensed monolayer, whereas for this monolayer

collapses to form multilayer structures. The measured

colA A<

Aπ − data showed that the transition from

one regime to another were not very sharp. It order to get reliable and unbiased estimations for , and the collapse pressure , we fitted the isotherms with: condA colA colπ

( ) ( ) ( , ) ( , )col colcol col cond

col cond col cond

A s h A A a h AA A A A

π ππ ≈ − − + −

− −A b (1)

29

where , , , , a and b are fitting parameters, representing the slope of the

isotherm during collapse, i.e. for A < a and b characterize the smoothness of the transitions

from one regime to the other. The function

condA colA colπ cols cols

colA and

( ) ( )2 21,2

h x a x x a≡ − + (2)

is a hyperbola interpolating between ( ),h x a x≈ for large negative and x ( ),h x a ≈ 0 for large

positive . This function has no direct physical interpretation and was introduced for practical

purposes only. As shown in Fig.1 satisfactory fits were obtained. Fitting a number of isotherms that

were obtained at compression velocities varying from 0.5 cm/min to 2 cm/min we found

, and

x

262 2ÅcondA = ± 57.8 0.3ÅcolA = ± 41 1mN/mcolπ = ± . These values did not vary significantly

within the range of the barrier velocities that we applied .The is consistent with

the trident conformation of the SSS molecules in a monolayer film at the air-water interface. The

cross-sectional area per hydrocarbon chain for tristearin at 20

262 2ÅcondA = ±

oC in the α phase (α phase has the

most mobile acyl chains) is [13].Our isotherms are consistent with earlier reports [3, 12]. 219.7Å

3.3.2 Isobaric compression

Even though we found that the forced compression isotherms did not change appreciably for barrier

velocities between 0.5 and 2 cm/min, under isobaric conditions we did observe further compression

though at velocities that were one or two orders of magnitude smaller. We stopped the forced

compression when a certain surface pressure π was reached. Next we kept the surface pressure

constant at that value, allowing the barrier to move. This is shown in Fig.2.

30

Barrier position vs time

13.50

14.00

14.50

15.00

15.50

16.00

16.50

0 1000 2000 3000 4000 5000

time t (sec)

l(t)

(cm

)

π = 10mN/mπ = 35mN/m

Fig.2. Two examples of the measured barrier position as a function of time during forced and

isobaric compression. The almost vertical parts of the curves correspond to forced compression at

a rate of 1 cm/min. The slowly decreasing parts correspond to small residual isobaric compression

rates at the spreading pressure given in the figure.

After several minutes a constant velocity was reached usually. The evolution of the trough

length was fitted to

l t (3) ( ) ( ) ( ) ( )0 0 ,f 0l v t t v v h t t a≈ − − − − −

Here the five fitting parameters are l , the trough length at the start of the isobaric compression, t ,

the starting time of the isobaric period,

0 0

fv and v , the forced and isobaric barrier velocity

respectively, and a , characterizing the transition from the forced to the isobaric regime. The

accuracy of the fits typically was 0.2%. In all cases the fitted forced velocity fv was very close to

the applied barrier velocity.

Isobaric compression

0

5

10

15

20

25

0 10 20 30 40 50

Spreading pressure π (mN/m)

Vel

ocity

v (

um/s

ec)

Fig.3. Isobaric velocity ν (µm/sec) as a

function of spreading pressure π as obtained

by fitting the measured barrier position to

Eq.(3). Note the sharp increase of ν for

spreading pressure close to the collapse

pressure colπ .

31

In Fig. 3 we show the dependence of the isobaric velocity v on the surface pressure π . It

can be noted that for and depends linearly on for

. For pressures

0v ≈ 5 mN/mπ ≤ π

5 mN/m 35 mN/mπ≤ ≤ 42 mN/mπ = (the collapse pressure) a much faster

compression is found. These results show that the isotherm shown in Fig. 1, can be considered as an

equilibrium isotherm only for . For larger pressures the equilibrium value of A is

smaller than the value displayed in Fig.1. At this point it is worth wale to clearly discriminate

between collapse pressure

5 mN/mπ ≤

colπ and equilibrium pressure eqπ .We use the definition of Roberts in his

book [14],whereas sometimes in the literature one manes equilibrium spreading pressure what we

call collapse pressure, see e.g.[15]. Equilibrium (spreading) pressure is the surface pressure that is

spontaneously generated when a crystalline sample of the solid material is placed in contact with

the water surface. Provided that sufficient time is allowed for equilibration to occur one can, in

principle, be sure that the monolayer which has been formed by molecules detaching themselves

from the crystal surface and spreading over the subphase is in equilibrium with the crystals

themselves. At any surface pressure higher than this there should be a tendency for the monolayer

to aggregate into crystals [14]. According to our results (Fig.3) for tristearin at air-water interface is

5mN/meqπ = .

In the isobaric conditions some rearrangement must take place which effectively thickens

the film. We assume that this process involves the growth of 3D crystals of SSS, and we investigate

this hypothesis using AFM-imaging. To this end we compare LB-films obtained by transfer at

with films transferred 30 or 60 min after . 0t t= 0t

3.4. AFM observations

3.4.1. Monolayer thickness

From the AFM images of LB-films, withdrawn at 5 mN/mπ = (data not shown) it is seen that the

mica is covered with a homogeneous monolayer. The monolayer can be successfully transferred to

a mica surface and it is quite stable in the course of time. When the Langmuir film was prepared at

higher pressures a monolayer was observed as well, but now with embedded higher domains. After

1 day storage at room temperature of the withdrawn LB- film the monolayer is still present, though

with slightly higher thickness (fig.4, C, F).

32

A B C

D1.68 nm 1.73 nm

µm0 2.50 5.00

-2.0

00

2.00 E

1.53 nm 1.44 nm

µm2.50

-2.0

02.

00

5.000

0F

0

1.86 nm 1.78 nm

µm

-2.0

02.

00

0.50 1.501.00

0

AAA BB CCC

D1.68 nm 1.73 nm

µm0 2.50 5.00

-2.0

00

2.00D

1.68 nm 1.73 nm

µm0 2.50 5.00

-2.0

00

2.00 E

1.53 nm 1.44 nm

µm2.50

-2.0

02.

00

5.000

0E

1.53 nm 1.44 nm

µm2.50

-2.0

02.

00

5.000

0

1.53 nm 1.44 nm

µm2.50

-2.0

02.

00

5.000

0F

0

1.86 nm 1.78 nm

µm

-2.0

02.

00

0.50 1.501.00

0

F

0

1.86 nm 1.78 nm

µm

-2.0

02.

00

0.50 1.501.00

0

Fig.4. AFM height image of an SSS monolayer transferred immediately after forced compression to

surface pressure π = 30mN/m. The black squares are holes in the monolayer produced by scanning

at a high force (~30 nN). (A) image scanned at AFM force F = 1nN with corresponding cross

section (D). (B) same area as in (A) scanned with AFM force F = 7.6nN and the corresponding

cross section (E). (C, F) same sample exposed to air at room temperature for 1 day at F = 1nN.

The scale bar is 2 µm (A, B) and 500nm (C) and the vertical scale is 5 nm for all images.

We estimated the monolayer thickness using the following procedure. We first

scratched a rectangular hole in the monolayer with the AFM tip by scanning with a relatively large

force

( )0 0d d π=

30 nNF ≈ . Then a larger image, including the hole was scanned with small forces

1 8 nNF = − (fig.4). The height difference between the hole and the surrounding gives an

apparent thickness d . The fact that d turned out to depend on the scanning force F, shows that

the real monolayer thickness

′ ′

0 ( )d π depends on d . ′

In Fig.5 we show data, together with an overall fit of the form

( ),d F a b cF d Fπ π′ ≈ + + + π (4)

From this fit we can estimate the real thickness ( ) ( )0 ,d d Fπ π′≈ = 0 , corresponding to scanning

force , which is presented in Fig. 6. 0F =

33

Apparent monolayer thickness

1.0

1.2

1.4

1.6

1.8

2.0

0 5 10AFM force F ( nN )

AFM

thic

knes

s d'

(n

m)

10 mN/m10 mN/m20 mN/m20 mN/m30 mN/m30 mN/m

Fig.5. Measured layer thickness d as a

function of applied AFM force F and surface

pressure π. The surface pressures (π) at

which the monolayer was compressed are

given by the labels at the curves. The symbols

correspond to the measured data and the lines

are the fit according to Eq. (4).

′

0 '( 0)d d F

Real monolayer thickness

1.5

1.6

1.7

1.8

1.9

0 10 20 30 40

Spreading pressure (mN/m)

Laye

r thi

ckne

ss

d o (

nm)

Fig.6. Variation of the real thickness

= =

'( , )F

of the monolayer with varying

spreading pressures. Line: from the combined

fit with Eq. (4), squares: from independent

linear fits of d π at fixed π .

Note that the monolayer thickness varies from about 1.6 to 1.8 nm over the pressure range

that we study here. We interpret this change in thickness as reflecting a change in the tilt angle τ

between the alkyl chains and the substrate surface. Such a change in the tilt angle of amphiphilic

molecules on air-water interface due to compression was reported before [16, 17].

To translate the thickness into a tilt angle we need to estimate the effective chain length. A

first estimation we get from crystal data on the hexagonal α-phase [18, 19]. In this phase the SSS

molecules, in tuning fork conformation, are parallel to the c-axis. Then the interplanar distance d

(001), which is often referred to as long spacing, is equal to the length of the SSS molecule in

tuning fork conformation. This length is built up from two times the chain length plus the length of

the glycerol group, plus a small contribution from the contact region between SSS layers. Since

in the hexagonal α-phase, the alkyl chain length must be about 2.5 nm. A more

precise analysis and interpretation of crystallographic data of SSS in the stable β′-phase [20], where

and

( )001 5.06 nmd =

( )001 4.48 nmd = 60.8τ = ° , allows us to estimate an effective length of 5.13 nm of an SSS

molecule in tuning fork conformation. Correcting this for the length of the glycerol and the

34

contribution from the contact region in that phase, the alkyl chain length can be estimated as 2.31

nm.

We have no detailed information on the molecular conformation of the triglyceride

molecules in the monolayer. In order to estimate the tilt angle in the monolayer, we assume that the

glycerol part of the molecule makes close contact with the (hydrophilic) substrate. The alkyl chains

are stretched similar as in the α , β and 'β phases, though in different orientation with respect to

the glycerol group. This leads to a structure where alkane chains of 2.31 nm extend from the

substrate to the monolayer surface at a height above the substrate. Thus in the monolayer SSS

molecules adopt a trident conformation we get a simple relation:

0d

( ) 0sin /(2.31 nm)dτ = (5)

Interpreting our monolayer thickness data with Eq. (5), we see that the tilt angle varies from

43τ = ° at 10 mN/mπ = to 53τ = ° at 40 mN/mπ = . It is known that in the crystalline β′ and β-

phases of triglycerides the chains adopt specific tilt angles, which are characteristic for the chain-

packing in the given triglyceride. In these phases tilt angles always are above about 50o. Smaller tilt

angles are energetically unfavourable [1, 19]. Since presumably in the trident monolayer the alkyl

chains are less densely packed than the crystalline phases, a smaller tilt angle seems acceptable.

35

3.4.2. Initial structure, obtained by forced compression

A B

C D

AA BB

CC DD

Fig.7. AFM height image showing monolayers of SSS transferred immediately after forced

compression to surface pressure (A) 10 mN/mπ = , (B) 20 mN/mπ = , (C) 30 mN/mπ = and (D)

42 mN/mπ = . The density of higher domains, embedded in the monolayer, increases with the

surface pressure. The scale bar is 2 and the vertical scale is 20 nm for all images. µm

Figure 7 shows AFM images of SSS-layers that we transferred from the water-air surface to

mica, immediately after the spreading pressure π was reached by forced compression. Domains are

found that extend 3.5 nm or more above the monolayer level. Their density increases with

increasing π as shown in fig.8. We suggest that they are small initial crystals, formed in the period

where the spreading pressure increases from the small values at which the film is in a two-

dimensional gas state, to the final pressure π at which the condensed phase has formed. In this

period SSS molecules undergo major orientation and packing changes. Since the molecular surface

density of the adsorbed film is already high, in the last part of this period such motions are hindered

36

considerably. As a result the formation process of the domains will not be strictly deterministic and

a metastable film structure may form. We suppose that the domains serve as crystal nuclei from

which bigger crystals can grow when the Langmuir film is further compressed isobarically at

constant pressure π .

Initial coverage and density

0.00

0.04

0.08

0.12

0.16

0 10 20 30 40 50

Spreading pressure (mN/m)

Cov

erag

e

0.0

0.2

0.4

0.6

0.8

1.0

Den

sity

( um

-2)

Fig.8. Fraction θ of the film area, covered with crystals (▲), formed during the forced

compression to spreading pressure (π ) and crystal density ρ (♦). The curves are results obtained

fitting all forced and isobaric compression image data to the model described in Section 3.5.

3.4.3. Structural changes during isobaric compression

To investigate the structural changes of the Langmuir film in time, we transferred the Langmuir

film to the mica surface 0, 30 and 60 min after isobaric compression started. At surface pressure

5 mN/mπ = we observed no significant differences between the monolayers withdrawn 0 or 30

min after the start of isobaric compression.

37

B CA

D

1.59 nm

0 5.0 10.0

0-5

.05.

0

µm

E

10.05.00

-5.0

5.0

0

µm

3.49 nm 3.52 nm F0.192 nm

µm5.00 1

-5.0

5.0

0tmm m m

0.0

B CA

D

1.59 nm

0 5.0 10.0

0-5

.05.

0

µm

E

10.05.00

-5.0

5.0

0

µm

3.49 nm 3.52 nm F0.192 nm

µm5.00 1

-5.0

5.0

0

0.0

BB CCAAA

D

1.59 nm

0 5.0 10.0

0-5

.05.

0

µm

D

1.59 nm

0 5.0 10.0

0-5

.05.

0

µm

1.59 nm

0 5.0 10.0

0-5

.05.

0

1.59 nm

0 5.0 10.0

0-5

.05.

0

µm

E

10.05.00

-5.0

5.0

0

µm

3.49 nm 3.52 nmE

10.05.00

-5.0

5.0

0

µm

3.49 nm 3.52 nm

10.05.00

-5.0

5.0

0

10.05.00

-5.0

5.0

0

µm

3.49 nm 3.52 nm F0.192 nm

µm5.00 1

-5.0

5.0

0

F

0.0

0.192 nm

µm5.00 1

-5.0

5.0

0

0.0

0.192 nm

µm5.00 1

-5.0

5.0

0tmm m m

0.0

Fig.9. AFM height image of SSS monolayers transferred at π = 10 mN/m. (A) immediately after

forced compression, (B) after 30 min isobaric compression at air-water interface. (C) the same area

as in (B) after several scans with AFM force ~2nN. The scale bar is 2 µm and the vertical scale is

10 nm for all images. The corresponding cross sections are given in (D, E and F). Length

differences are given by the numbers at the markers. The symbols below the lines give our proposed

structure of the crystals (m - trident conformation; t – top layer tuning fork conformation)

At a surface pressure π = 10 mN/m, the AFM images show a homogeneous monolayer with

small defects when the LB-film was transferred to mica immediately after forced compression, as

shown in Fig.9A and D. After 30 min isobaric compression we observed a few higher domains,

embedded in the monolayer (fig. 9B, E). These domains were soft and could be scratched away

with the AFM tip, even at the normal scanning forces F that are normally used for imaging. After

several scans with F = 1-2 nN the second layer disappeared, leaving a flat film with the same

thickness as the trident monolayer, Fig.9C and F. The thickness of the domains, measured from the

monolayer, was 3.5 – 3.6 nm.

38

A D

µm

5.1 nm 4.9 nm

3.750 7

-9.5

09.

5

mαm

.50

B

µm

4.9 nm8.2 nm

3.75 7.500

-30.

030

.00

m αm

tαm

E

C F

µm3.75 7.500

-30.

030

.00

8.2 nm

5.0 nm15 nm

m tαm

tααm

tαααm

mica

AA D

µm

5.1 nm 4.9 nm

3.750 7

-9.5

09.

5

mαm

D

µm.50

5.1 nm 4.9 nm

3.750 7

-9.5

09.

5

mαm

.50

BB

µm

4.9 nm8.2 nm

3.75 7.500

-30.

030

.00

m αm

tαm

E

µm

4.9 nm8.2 nm

3.75 7.500

-30.

030

.00

m αm

tαm

µm

4.9 nm8.2 nm

3.75 7.500

-30.

030

.00

m αm

tαm

E

CC F

µm3.75 7.500

-30.

030

.00

8.2 nm

5.0 nm15 nm

m tαm

tααm

tαααm

mica

F

µm3.75 7.500

-30.

030

.00

8.2 nm

5.0 nm15 nm

m tαm

tααm

tαααm

mica

8.2 nm

5.0 nm15 nm

m tαm

tααm

tαααm

mica

Fig.10. AFM height image of SSS monolayers transferred at 20 mN/mπ = . (A) immediately after

forced compression, (B) after 30 min isobaric compression at air-water interface and (C) after 60

min isobaric compression. The corresponding cross sections are given in (D, E and F). The scale

bar is 2 µm for all images and the vertical scale is 20 nm for (A) and 70 nm for (B, C). Length

differences are given by the numbers at the markers. The symbols below the lines give our proposed

structure of the crystals (m – trident conformation; α - crystal tuning fork conformation; t – top

layer tuning fork conformation).

39

At surface pressure 20 mN/mπ = we found that the directly transferred LB-film consisted

of an almost defect free trident monolayer, in which many small domains were embedded. The

thickness of the domains was found to be 4.8 - 5.1 nm, as shown in Fig.10A and D. On LB-films

transferred after 30 min isobaric compression, the domains within the trident monolayer were

higher and bigger. The maximum measured thickness from the monolayer was 8.2 ± 0.2 nm

(fig.10B and E). After 60 min incubation even higher domains were found with thickness up to 20 ±

0.2 nm measured from the mica (Fig.10 C, F). On the highest domains we found terraces separated

by steps of height 4.9 ± 0.1 nm. In all cases, the domains were surrounded by the trident monolayer.

The same growth process was observed for LB-layers obtained at surface pressure

30 mN/mπ = . A closer AFM observation showed us that on the bigger crystals formed at

20 mN/mπ ≥ two different terraces can be found, with height thicknesses 3.5 nm and 5.1 nm from

the monolayer (fig.11).

A

B5.2 nm

3.5 nm5.1 nm

µm2.50 5.000

-20.

020

.00

mαm

αm

tm

A

B5.2 nm

3.5 nm5.1 nm

µm2.50 5.000

-20.

020

.00

AA

B5.2 nm

3.5 nm5.1 nm

µm2.50 5.000

-20.

020

.00

B5.2 nm

3.5 nm5.1 nm

µm2.50 5.000

-20.

020

.00

5.2 nm3.5 nm

5.1 nm5.2 nm3.5 nm

5.1 nm

µm2.50 5.000

-20.

020

.00

mαm

αm

tm

Fig.11. AFM height image of SSS monolayer

transferred at 30 mN/mπ = after 30 min isobaric

compression at air-water interface (A). The

corresponding cross section is given in (B). The scale

bar is 2 µm and the vertical scale is 50 nm.

40

3.4.4. Stability of the transferred LB-film

To check the stability of SSS-layer in air, we transferred it immediately after forced compression to

30 mN/mπ = and left it for 1 day at room temperature (fig.12). The monolayer became grainy and

slightly higher. The crystals grew slightly and the newly grown parts of the crystals were 3.5 ± 0.1

nm above the monolayer level. In some crystals higher domains (5.0 ± 0.1 nm) were observed.

During the incubation in air of the transferred LB- film not only the present already crystals were

growing, but also new very small nuclei appeared. We suggest that the grainy character of the

monolayer is due to molecules, which leave the monolayer to form the new nuclei and the new parts

of the present crystals.

Fig.11. AFM height image of a monolayer of SSS that

was transferred to mica immediately after forced

compression to surface pressure 30 mN/mπ =

and that

was left for 1 day in air at room temperature. The cross

section is shown in (B). The scale bar is 2 µm and the

vertical scale is 20 nm.

A

B

1.86 nm3.3 nm

5.1 nm

3.75 7.500

-10.

010

.00

µm

mtm

αm

AA

B

1.86 nm3.3 nm

5.1 nm

3.75 7.500

-10.

010

.00

µm

mtm

αm

B

1.86 nm3.3 nm

5.1 nm

3.75 7.500

-10.

010

.00

µm

1.86 nm3.3 nm

5.1 nm

1.86 nm3.3 nm

5.1 nm

3.75 7.500

-10.

010

.00

µm

mtm

αm m

tm

αm

3.4.5. Consistency of Langmuir and AFM data

If the densities of the monolayer and the higher domains were exactly the same, then the total film

volume ( )V t should remain constant during the isobaric compression process (SSS is not volatile).

In the Langmuir system we measure the temporal change of the film area ( )A t . From the AFM

41

images we can estimate the average film thickness ( )d t . Ideally ( ) ( )A t d t V= is constant,

whence ( ) ( )0 /A A t , which can be obtained from the Langmuir experiment should be equal to

as measured by AFM. ( ) ( )/ 0d t d

Fig.13 shows that, within the experimental accuracy this is true. The relatively large

uncertainty of 5 - 10% of as estimated from AFM is due to the inherent inaccuracy of the

standard Nanoscope “bearing analysis” software for estimating crystal volumes. The uncertainty in

the Langmuir estimation of 2 - 4% is caused by the differences in the observed isobaric velocities

as obtained from the fits in section 3.3.2. As the individual fits are accurate, these velocity

differences reflect accidental differences in the structure of the film that was being compressed.

( )d t

v

Scaled film thickness

1

1.2

1.4

1.6

0 10 20 30 40 50Spreading pressure (mN/m)

d(t)

/ d(0

)

Langmuir 30 min

Langmuir 60 min

AFM f it 30 min

AFM f it 60 min

AFM data 30 min

AFM data 60 min

Fig.13. Scaled film thickness estimated by Langmuir machine and AFM

3.5. Theory for nucleation, growth and coalescence of crystals

3.5.1 Qualitative interpretation of film evolution observations

In the sequel we shall interpret the observed film structure on the basis of a model that is

schematically presented in figure 14.

42

~1.75 nm

4.8 - 5.1 nm~ 3.5 nm

~ 3.5 nm

4.8 - 5.1 nm

45 2± o

40 -45 o

40 -45 o

~ 90 o

~ 90 o

3.5 0.1nm±43 -45 o

43o

A B C D E

Fig.14. Schematic illustration of the structures proposed for thin layers of SSS molecules.

(A) Monolayer structure. At 5 mN/mπ ≤ this is the only structure found, at 10 mN/mπ ≥ it is the

structure around the higher domains, (B) Structure of stable thin crystals. At 10 mN/mπ = all

observed crystals have this structure. (C) Structure of metastable crystals. Such crystals are found

on films that are withdrawn immediately after forced compression to 20 mN/mπ = . (D, E)

structure of higher crystals. Such crystals are observed after 30 or 60 min isobaric compression at 20 mN/mπ ≥ .

Our observations suggest that an SSS trident monolayer is thermodynamically unstable for

spreading pressure 5 mN/mπ . Therefore during isobaric compression at 5 mN/mπ , some

SSS molecules move to the top of the monolayer. These molecules rearrange in higher domains

where they presumably adopt the more stable tuning fork conformation and pack similar as in the

crystalline α and β crystal forms. This film structure was first proposed by Bursh and Larsson to

interpret the triple chain LB-film thickness that they observed for LB-films that were compressed

beyond the collapse pressure [3, 21]. Based on a careful analysis of the observed domain height we

propose a new model for the structure and packing characteristics of the domains.

Using, as above the estimated effective length of 5.13 nm for an SSS molecule in the tuning

fork conformation, the observed domain thickness of 3.5 nm at 10 mN/mπ = , corresponds to a tilt

angle 43 44.5τ = − ° , i.e. somewhere between the estimated tilt angle in the trident monolayer and

43

the tilt angle in the stable β′ phase, Fig.9B, E. We may suppose that the structure of these layers can

be described as a slightly deformed β or β′ phase. As we observed this layer thickness always at the

upper film layer, we refer to this structure as the top layer structure (‘t’ in the figures). The fact that

we have observed this structure to be common for top layers shows that the increased tilt is a form

of surface relaxation, caused by the different interaction with other crystals layers.

At surface pressures 20 mN/mπ ≥ some domains extended as much as 5.0 ± 0.1 nm above

the surrounding monolayer. This suggests that in these domains the molecules are fully stretched

(5.13 nm) and oriented perpendicular to the monolayer, i.e. the structure of these domains is similar

to the crystalline α phase (fig.14, C). During forced compression of the Langmuir film at

20 mN/mπ ≥ , the crystal growth process is so fast that this metastable, α -like , polymorph with

layer thicknesses d(001) ~ 5 nm is formed. Domains that are grown in the α - phase will not

spontaneously transform to the β or 'β phase because this involves a very slow solid-solid

transformation process.

For the slow growth process at 10 mN/mπ = the stable β - phase is grown immediately,

though with a top layer slightly thinner than the interplanar distance d(001) = 4.48 nm of the real

β - phase (fig.9).

The higher domains (8.2 ± 0.2 nm) found after isobaric compression can be explained with

the formation of a second layer on top of the first one (fig.14, D). This measured thickness does not

correspond to two fully stretched layers (~ 10 nm). We assume that the first layer is in α - phase

(5.1 nm) and the second layer, which in this case is the top layer, has the ‘t’- structure, having a tilt

of 40-45o and a thickness of 3.4 ± 0.2 nm .The reproducibility of the step height in the next layers,

which is 4.8-5.1 nm, supports this interpretation.

Combining our observations we concluded that for 5 mN/mπ crystals of SSS in tuning

fork conformation are growing on top of a trident monolayer at the air-water interface. If the growth

is slow enough (e.g. at 10 mN/mπ = or in the last stages of growth at 15 mN/mπ > ) the crystals

grow in the β or 'β phase. The top layer is tilted at 43-45o, i.e. the molecules are somewhat more

flat than in the β or 'β phase. At larger growth rates (e.g. the initial stages of growth at

15 mN/mπ > ) crystals are formed with a metastable α - like structure. The transformation of α -

like crystals to a β or 'β -like crystal structure is too slow to be observed. Only the top layer may

relax to an inclined molecular orientation. The same crystal growth processes occur at the mica-air

44

interface of the transferred LB-film as at the air-water interface of the isobarricaly compressed

Langmuir film, though much slower.

The type of crystal growth process proposed here, where the structure of the first layer is

very different from that of the subsequent layers, is known as the Stranski - Krastanov growth

mode.

3.5.2 Parameters and measurable variables

At this point we know that the size of the crystals increases with time, the growth being mainly in

lateral directions, and that the growth rate increases with surface pressure. The number of crystals

too, increases initially with time and with increasing surface pressure. In later stages, when a

significant fraction of the monolayer film is covered by crystals, crystals start to coalesce, and their

number decreases again. To interpret these observations more quantitatively we develop a simple

model.

Our model provides us with the dependence on time and on surface pressure of the crystal

density the average crystal area and the fraction (dimensionless) of the film

that is covered by crystals (area of the crystals divided to the total area of the image). At

random nucleation of crystals starts at a rate I . We assume that the crystals have a roughly

cylindrical form, with initial radius and height , and that they grow with rates and in the

lateral and vertical direction respectively. We consider , , , and as time independent

physical parameters that may depend on the spreading pressure of the film. Inspection of the data

suggests that some nucleation of crystals takes place before the sample is removed from the

Langmuir through. This means that a non-zero initial substrate coverage

2( mρ µ − ) )2(a mµ θ

0t =

0R 0h lv vv

I 0R 0h lv vv

( )0tθ = and crystal

density have to be considered. ( 0tρ = )

3.5.3 Avrami - Kolmogorov theory for coverage

In the beginning of the nucleation and growth process the crystals are far enough away from each

other to grow independently. We shall refer to the crystal density and the covered fraction in the

initial stages as “free” values fρ and fθ

45

(7) ( ) ( ) ( ) ( )0

0t

f f ft I t dtρ ρ ρ′ ′= + = +∫ 0 It

( ) ( ) ( ) ( )( ) ( )2 2 2

0 0 0 0 00

13

t

f f l f l lt t I t R v t t dt t I R t v R t vθ θ π θ π ⎛ ⎞⎟′ ′ ′ ⎜≡ + ⋅ + − = + + + ⎟⎜ ⎟⎜⎝ ⎠∫ 2 3t (8)

The first terms in Eq(7) and Eq.(8) describe the effect of crystals that were already present at

the beginning, , of the isobaric compression. Their density does not change in time, but their

coverage grows according to

0t =

( ) ( )( )( )

2

0

00

0f

f ff

tθ

θ ρ ππρ

⎛ ⎞⎟⎜ ⎟⎜= ⎟⎜ ⎟⎜ ⎟⎜⎝ ⎠lv t+

)

(9)

This expression can be found by using that the average area of the crystals at

equals

( 20R tπ = 0t =

( ) ( )0 / 0f fϑ ρ .

In the later stages of film growth, the actual values of and will be smaller than

the free values,

( )tρ ( )tθ

( ) ( )ft tρ ρ< and . The fact that crystals can only grow and nucleate in

the uncovered film area between the already existing crystals, is captured by the Avrami -

Kolmogorov theory, leading to

( ) ( )ftθ θ< t

)tθ (10) ( ) ( )(1 exp ftθ = − −

This expression gives us the actual coverage of the film by crystals. It depends however, on

too many physically important parameters to hope that all these parameters can be derived from

observed curves alone. Therefore we want to use observed crystal sizes and crystal densities

as well. What we need is an expression for the number

( )tθ

( )c cN N t= of free crystals that have

merged to form one actual crystal. Then the experimental crystal density and crystal size are found

from

/f cNρ ρ= (11)

(12) / /c fa Nθ ρ θ ρ= =

Unfortunately, no general theory to obtain is available. In the next section we develop an

approach to the problem.

cN

46

3.5.4 Approximate theory for average crystal size and density

In the spirit of the Avrami-Kolmogorov theory the first step is to consider the growing film as if

circular crystals nucleate and grow independently. For the free crystal coverage and density we use

Eqs.(7) and (8). The average area fa of freely growing crystals is not simply the area ( )20 lR v tπ +

of crystals that nucleate at time . Crystals that nucleate later have a smaller area at time .

Taking this into account we obtain

0t = t

( ) ( ) ( ) 20 0

1/3f f f l la t t t R R v t v tθ ρ π⎛ ⎟⎜= = + + ⎟⎜ ⎟⎜⎝

2 2⎞⎠ (13)

( ) 20 0

13

ff

aR t R R v t v t

π≡ = + + 2 2

l l (14)

for the average area fa and radius fR of freely growing crystals.

In the next step we take the merging and overlapping of these free crystals into account.

Two crystals merge to form one new crystal if they are located close enough together. Two circular

crystals with radius fR touch each other, and will probably coalesce, if their (centre-to-centre)

distance is 2 fR or less. Generally, we assume that two crystals merge if one is located within the

merging region A of the other. The area of is written as + A+ 2fa η+ = a , with . 2η ≈

The key idea is to define as the density of original crystals that are, directly or

indirectly, connected to a original crystal in the origin. This density satisfies

( )cρ r

( ) ( )( )01c f Pρ ρ= −r r (15)

Here ( is the probability that an original crystal at r is not connected to the crystal in

the origin. Let be the merging region of a crystal at . All original crystals in the merging

region

)0P r

( )A+ r r

( )0A+ of the central original crystal are connected to this crystal, hence ( )c fρ ρ=r for

inside . Further away the probability of a given crystal at r to be consider to the central crystal

is equal to the probability to find at least one connected crystal in its merging region .

Therefore

r

(0)A+

( )A+ r

( )cρ r satisfies the implicit equation

(16) ( )

( )

( )( )

( )2

0

1 exp 0

f

cf c

A r

for r A

rr d r for r A

ρ

ρρ ρ

+

+

+

⎧ ∈⎪⎪⎪⎪⎪ ⎡ ⎤⎛ ⎞= ⎨ ⎟⎜⎢ ⎥⎪ ⎟′ ′⎜− − ∉⎟⎪ ⎢ ⎥⎜ ⎟⎪ ⎜ ⎟⎢ ⎥⎝ ⎠⎪ ⎣ ⎦⎪⎩∫

47

The latter expression derives from to the case of the probability 0N =

1 ( ) exp( )!

NNP a

Naρ ρ= − (17)

to find N objects within an area a ,when the average density of these objects is ρ .

Eq.(16) is an integral equation for , which can not be solved exactly. We make the

following observations. If the free crystal density

( )cρ r

fρ is so high that the summed merging area is

larger than the total film area, , then far enough outside 2 1f fa ρ η θ+ = > ( )0A+ a constant solution

(18) ( ) ((1 expc c f aρ ρ ρ ρ+= = − −r ))cexists. In this case merging of the original crystals leads to an infinite crystal.

For smaller free crystal densities fρ , the connected crystal density decays to 0 with

increasing distance from the central crystal. Assuming that varies slowly over the circular

region of radius

( )cρ r

( )cρ r

( )A+ r fRη and area 2 2fa πη+ = R , we have for the integrand in Eq.(16)

( )( )

( ) ( ) ( )22 21 .......

8c c cA

d r a aρ ρ ρ+

+ +′ ′ ≈ + ∇ +∫r

r r r (19)

where is the two-dimensional Laplace operator. Here the first two terms of a Taylor expansion

of have been used.

2∇

( )cρ r

Substituting Eq.(19) in Eq.(16) for , we get for a linear differential

equation

( )0A+∉r 1ca ρ+

( ) ( ) ( ) ( )2 21

8c f c ca aρ ρ ρ ρ+ +⎛ ⎟⎜= + ∇ ⎟⎜ ⎟⎜⎝r r ⎞⎠r (20)

The general solution that vanishes at infinity is

( ) ( )( )( )( )

0

0

K /

Kf

c c f

r Rr c

β ηρ ρ ρ

β= =r (21)

where is a Bessel function. The scaling factor can be expressed in terms of the free coverage 0K β

fθ as given by Eqs.(8)-(13)

( )2

2

8 1 f

f

η θβ

πη θ−

≡ (22)

The constant c is obtained from a matching criterion at the edge of the merging region

of the central crystal, i.e. at (0)A+fr Rη= .Of the circular region with the same area ( )A+ r a+ at

48

the edge of , a fraction (0)A+ ( )2/3 3 / 2γ π= − overlaps with .In that part all crystals

are connected to the central crystal, whence

(0)A+

( )c fρ ρ=r . On the remaining fraction 1 γ− of

we approximate the density of connected crystals by the density at the edge:

( )A+ r

( ) ( )c c fRρ ρ η≈r . Thus

we approximate the surface integral in Eq.(16) by

(23) ( )( )

( ) (2 1c fA

d r a Rρ γρ γ ρ+

+ ⎡′ ′ ≈ + −⎢⎣∫r

r )c fη ⎤⎥⎦

Substituting Eq. (20) with fr Rη= into Eq. (15) leads to an implicit equation for c :

(24) ( )[( 21 exp 1fc η θ γ γ= − − + − ])cfrom which c can be solved by numerical iteration.

Upon integrating ( over the whole surface we obtain the average number of original

crystals that are, directly or indirectly connected to a given cluster, i.e. for the average number

of original crystals that merge to form a new crystal:

)cρ r

cN

( ) ( )( )( )

22 2 1

200 0

K1 1 2 2 1 1

K

f fR R

c c f c fcN d r rdr r rdr

η ηβηρ π ρ π ρ η θ

β β⎛ ⎞⎟⎜ ⎟≡ + ≈ + + = + +⎜ ⎟⎜ ⎟⎜⎝ ⎠∫ ∫ ∫r (25)

Note that this average crystal size diverges if the free coverage fθ approaches the limiting

value where we found that Eq.(16) had an infinite cluster solution Eq.(18). 2 1fη θ →

3.5.5 Interpretation of AFM-images of nucleation and growth

Combining Eq.(25) with Eqs.(7)-(12), we can express the observed film coverage, crystal density

and crystal size in terms of the nucleation rate and the lateral growth rate. To interpret data on the

individual crystal volumes as well, we assume that the nuclei have a monolayer height and that

vertical growth takes place on top of these with a constant velocity . We have simultaneously

fitted all our AFM data to Eqs.(7)-(12), (25). The results of the fit are given in the table below.

vv

49

Symbol in text Value Unit

Equilibrium pressure eqπ 5.0 0.5± mN/m

Initial density ( )0f tρ = ( )( )0.020 0.002 eqπ π± − µm-2

Initial radius 0R ( )( )4.5 0.5 eqπ π± − nm

Nucleation rate I ( )0.0002 eqπ π< − µm-2 min-1

Lateral growth rate lv ( )( )0.45 0.05 eqπ π± − nm/min

Vertical growth rate vv ( )( )0.0009 0.0001 eqπ π± − nm/min

The first fit parameter is the equilibrium value eqπ of the spreading pressure. Nucleation and

growth of crystals is expected to take place in the film only for pressures eqπ π> . For the other

model parameters we compared the quality of the fit for the two cases that the parameter was taken

independent of π or proportional to eqπ π− . In all cases the second choice performed better. We

also found that the fitting model did not produce a significantly positive value for the nucleation

rate (after withdrawal of the film from the through). As a matter of fact, direct inspection of the

AFM images confirmed that only rarely the number of crystal increased with increasing with time

after withdrawal. The decrease in the number of crystals due to coalescence and merging was a

more dominant process. Therefore we took

t

0I = to estimate the other model parameters (thus

leaving us with a 5-parameter fit to 4 independent observables. We used 10 different combinations

of time and spreading pressure (altogether 78 AFM images were taken into account). For the

nucleation rate we give an upper limit.

In view of the uncertainty and experimental spread of the AFM images the fits are quiet

satisfactory. Using more elaborate models does not seem justified or useful.

3.6. Conclusions

In this study we have obtained AFM images that reveal the structure of thin tristearin (SSS) films,

formed at air-water interface. Based on Langmuir and AFM experiments we estimated crystal

growth rates in the lateral and vertical direction. Our investigations lead to the following

conclusions.

50

Compressing an extended SSS film at the air-water interface slowly, starting at very low

surface pressure, monolayers of SSS molecules in trident conformation are formed. These

monolayers are thermodynamically stable at surface pressure .This is the equilibrium

spreading pressure for this system. At higher surface pressures a monolayer with a trident

molecular conformation is still formed, but it is metastable and transforms slowly to form

crystalline domains. The monolayer can be successfully transferred onto a mica surface and the

resulting Langmuir-Blodgett film is quite stable in the course of time. Using AFM-imaging the

monolayer thickness of the trident monolayer can be measured. We find that the thickness depends

on the surface pressure, which we interpret as a change in the tilt angle between the average chain

direction and the crystal surface. From thicknesses between 1.6 and 1.8 nm, we conclude that the tilt

angle gradually increases from

5 mN/mπ ≤

43τ = ° at 10 mN/mπ = to 53τ = ° at 40 mN/mπ = .

At pressures the trident monolayer is thermodynamically metastable.

Macroscopically it is immediately determined by the behaviour of Langmuir films during isobaric

compression. The film area decreases slowly but steadily when the film is subject to a constant

surface pressure. We investigated this process on the molecular scale by comparison of AFM

images of LB-films that were produced after different periods of isobaric compression. We

concluded that crystal growth takes place under these conditions. Within the crystals the molecules

adopt a tuning fork conformation, except for the bottom layer where the molecules stay in the

trident conformation. The crystal sizes increase with time, the growth being mainly in lateral

directions. The growth rate increases with surface pressure.

10 mN/mπ ≥

We developed a new model to describe the growth and coalescence of crystals in the film.

Fitting the AFM-data to this model we estimate a lateral growth rate of 2.3 nm/min and a vertical

growth rate of 0.005 nm/min at 10 mN/mπ = . The data were insufficient to get a reliable estimate

of the nucleation rate of new crystals, we merely estimate an upper limit of 0.001 new nuclei per

µm2 and per minute.

At and above the collapse pressure the crystal growth process is the same, though faster than

in the intermediate regime. The isobaric compression velocity increases rapidly from 3 µm/sec at

to 18 µm/sec at . mN/m35π = mN/m42π =

The transformation processes that took place in the Langmuir film (floating on water), were

observed in the transferred LB-film (on mica) as well. In the latter case however, the processes were

orders of magnitude slower.

51

References:

[1] Garti, N. and Sato, K., In Crystallization and polymorphism of fats and Fatty Acids; Dekker, M.

New York (USA) 1988

[2] Ollivon, M., Triglycerides. In Manuel des Corps Gras. Ed.A.Karieskind, Lavoisier, Paris

(France) 1992; p. 469

[3] Bursh, T., Larsson, K. and Lundquist, M., Chem. Phys. Lipids 2 (1968) 102-113

[4] Hamilton, J.A., Small, D.M., In Proc. Nat. Acad. Sci. USA 78 (1981) 6878

[5] Hamilton, J.A., Biochem. 28 (1989) 2514

[6] Claesson, P.M., Dedinaite, A., Bergenstahl, B., Campbell B. and Christenson, H., Langmuir, 13

(1997) 1682

[7] Lundquist, M., In Surface chemistry, Copenhagen. Munksgaard, 1966 p.294

[8] Kuzmenko, I., Buller, R., Bouwman, W.G., Kjaer, K., AlsNielsen, J., Lahav, M. and

Leiserowitz, L., Science 274 (1996) 20046-20049

[9] Xue, J.Z., Jung, C.S., Kim, M.W., Phys. Rev. Lett. 69 (1992) 474-477

[10] Ekwall, P., Ekholm, R., and Norman, A., Acta Chem. Scand. 11 (1957) 703

[11] Birker, P.J. and Blonk, J.C., J. Am. Oil Chem. Soc. 70 (1993) 319-321

[12] Michalski, M., Brogueira, P., Goncalves da Silva, A. and Saramago, B., Eur. J. Lipid Sci.

Technol. 103 (2001) 677-682

[13] Akita, C., Kawaguchi, T., Kaneko, F., Yamamuro, O., Akita, H., Ono, M. and Suzuki, M.,

Journal of Crystal Growth 275 (2005) 2187-2193

[14] Roberts, G., Langmuir-Blodgett Films Plenum Press, New York (1990) p.21

[15] Gaines, G.L., Insoluble Monolayers at Liquid-Gas Interface, Wiley, New York, 1966

[16] Karaborni S. and Toxvaerd, S., J.Chem.Phys. 97 (8) (1992) 5876-5883

[17] Lin, B., Shih, M.C., Bohanon, T.M., Ice, G.E. and Dutta, P., Physical Review Letters Vol.65,

No.2 (1990) 191-194

[18] Yase, K., Ogihara, S., Sano, M., Okada, M., Journal of Crystal Growth 116 (1992) 333-339

[19] Takeuchi, M., Ueno, S. and Sato, K., Crystal Growth & Design Vol.3, NO.3 (2003) 369-374

[20] De Jong, S., Triacylglycerol crystal structures and fatty acid conformations, a theoretical

approach- PhD thesis (1980).University of Utrecht, The Netherlands

[21] Larsson, K., Physical Properties - Structural and Physical Characteristics. In: The Lipid

Handbook. Eds. Gunstone, F.D.; Harwood, J.L.; Padley, F.B. Chapman & Hall, London (UK) 1986;

p.321

52

CHAPTER 4

Structure and stability of Triglyceride monolayers on water and mica surfaces

Abstract

The structure and the stability of tripalmitin (PPP), tristearin (SSS) and triarachidin (AAA)

monolayers at the air-water are investigated with the Langmuir method. The Langmuir-

Blodgett (LB) layers obtained by deposition on mica were investigated with Atomic force

microscopy (AFM). Our experiments show that the three triglycerides can form monolayers

with molecules in trident conformation at the air-water interface. We determined the

equilibrium spreading pressure eqπ below which such monolayers are thermodynamically

stable. Under isobaric conditions at 0π π> the film area decreased slowly for PPP and SSS,

corresponding to crystal formation with molecules in tuning fork conformation on top of the

monolayer. Here 0π may be significantly larger than eqπ . In the intermediate range

0eqπ π π< < film area decrease was not measured. The isobaric compression rate was highest

for PPP and almost zero for AAA. Using carefully AFM the thickness of the trident

monolayers was measured. It is 1.49 nm for PPP, 1.75 nm for SSS and 2.2 nm for AAA,

corresponding to tilt angles of the molecules of 46.4o, 49.2o and 59.0o respectively. The LB-

monolayers of PPP and SSS, which were transferred at 0π π≥ are thermodynamically

unstable in air. Small crystals form on top of the monolayer, presumably in β -phase for SSS.

Contrary to SSS, domains with α -like and β -like structure coexist in the LB film of PPP.

The nucleation rate increases with increasing surface pressure π and with decreasing chain

length of the triglyceride. For AAA no well - defined crystals were found on top of the LB-

monolayer during the investigated period of days. The trident monolayer is the less mobile

and the crystal phase is the more stable the longer the alkyl chains are.

53

4.1. Introduction

The interfacial behaviour of surfactants and their mixtures is of importance in a wide range of

applications. The most commonly used emulsifiers in the food industry are the

monoglycerides [1]. Spread monolayers at air-water interface can show relaxation phenomena

mainly because of instability due to desorption or collapse [2].

For monoglyceride monolayers, it was observed that the main causes of instability are

desorption in subphase competing with collapse followed by nuclei formation. It was found

that the stability of the monolayers depends on the film structure, subphase composition, the

temperature, the surface pressure [3, 4] and aqueous phase pH [5]. Some of the investigated

monoglycerides were unstable at surface pressures π below the so-called collapse pressure

colπ . The rate of monolayer molecular loss due to desorption increased with surface pressure.

Molecular loss at the interface depended also on the hydrocarbon chain length. The longer

monoglycerides were more stable than the shorter [3, 5].

The triglycerides are another class of molecules of great importance in the food

industry. They are isolated from plant seeds or animal tissues and processed into edible-fat

products, of which they are the main constituents. The crystallization of triglycerides is a key

step both during manufacturing fat products and fractionating fats and oils. In all cases, the

crystallization behaviour is very complex due to the intricate composition of fat blends and

the tendency of triglycerides to crystallize in a variety of morphological forms. Depending on

the crystallization procedure, especially the thermal treatment, they may crystallize in the α

(hexagonal, less stable), β’ (orthorhombic), or β (triclinic, most stable) form. Each of these

polymorphic forms consist of layers in which the molecules have a tuning fork conformation

but the orientation of the tuning forks within the layers, as well as the packing of the layers

is different [6, 7].

On the other hand, in monolayers at a hydrophilic-hydrophobic interface, triglyceride

molecules adopt a trident conformation (all hydrocarbon chains pointing toward the same

direction). In the trident conformation the hydrophilic glycerol group is in contact with the

water or the mica surface, and the hydrophobic chains point into the air or oil [8-12].

In previous work [Chapter 3] we investigated monolayers of tristearin (SSS, chain

length 18 C atoms) floating on water in a Langmuir system and deposited on mica with AFM.

The Langmuir experiments showed that adsorption isotherms obtained with commonly used

compression rates do not correspond to thermodynamic equilibrium. Under isobaric

54

conditions the film area decreased slowly, corresponding to the formation of crystals on top of

the monolayer. The AFM images revealed that SSS initially forms trident monolayers at air-

water interface. These layers are thermodynamically stable only at surface pressure

5mN/mπ ≤ . For 10mN/mπ ≥ growth takes place of crystals with a tuning fork conformation

of the SSS molecules on top of the trident monolayer. The crystals grow with time, mainly in

lateral directions. The growth rate increases with surface pressure. A new model was

developed to quantitatively describe the crystal growth process. A lateral growth rate of

2.3 nm/min and a vertical growth rate of 0.005 nm/min were calculated for one individual

crystal at 10mN/mπ = . The same growth process that was observed on the air-water interface

was also observed in transferred monolayers at room temperature, though the growth was

much slower.

The aim of this work is to understand the behavior of different triglycerides

(tripalmitin-PPP chain length 16 C atoms, tristearin-SSS and triarachidin- AAA chain length

20 C atoms) at air-water interface (Langmuir film) and on solid surface like mica (Langmuir-

Blodgett film) and to establish the relation between their molecular structure and their

monolayer stability. Two kinds of experiments have been done. First, we measured the Aπ −

(spreading pressure π vs area per molecule A ) diagram of Langmuir films. Starting with a

Langmuir film at very small π , where the film is in a low-density “gas” phase, we

compressed the film, at a constant rate, to the desired pressure π (forced compression). To

investigate whether the Langmuir film was in thermodynamic equilibrium at this pressure π ,

we sometimes left the film for some time t at pressure π (isobaric compression). In the

second type of experiment the Langmuir film was transferred to mica directly after forced

compression ( ). We investigated LB films with AFM immediately and a few days after

incubation in air at room temperature.

0t =


4.2.1. Chemicals

Film material: In our experiments we used saturated monoacid triglycerides (their three acyl

chains are the same). Tripalmitin (1, 2, 3-Propanetriyl trihexadecanoate: PPP, chain length 16

C atoms), Tristearin (1, 2, 3, -trioctadecanoylglycerol: SSS, chain length 18 C atoms) and

55

Triarachidin (trieicosonoin: AAA, chain length 20 C atoms) were purchased from Larodan

(Sweden) with a stated purity of >99 mass %. Stock solutions of PPP, SSS and AAA with

concentration of 1 mM in distilled chloroform were prepared.

Subphase: Distilled water was used as a subphase in our Langmuir system for all

experiments. The resistivity of the water was 15 MOhm cm.



Compression isotherms were measured on a commercial, fully automated Langmuir Blodgett

Trough (model: 311D, Nima Technology Ltd., England). The instrument was equipped with a

Teflon trough (283.0 cm2) and one Delrin barrier. The spreading pressure π was measured

with an accuracy of about 0.1 mN/m. The film material was initially spread on the water

subphase, dropping 30 µL of 1 mM stock solution dissolved in chloroform, using a 100 µL

Hamilton syringe. The conditions were chosen such that initially the average area A per

molecule is .We started (asymmetric) film compression 2 min after spreading. In

our system two modes of operation were available. First forced compression, where the

position of the barrier, and hence the trough length

2110 ÅA ∼

( )l t ahead of the barrier, is given. Then

the resulting spreading pressure ( )tπ is registered. In this mode we chose barrier velocities of

the order of 1 cm/min, which according to the literature should be slow enough that the

Langmuir film stays close to thermodynamic equilibrium.

Second we used the isobaric compression mode, where a constant spreading pressure

π is applied and the resulting trough length ( )l t is monitored. Obviously if the film is in

equilibrium at the applied pressure, then ( )l t is constant. In practice however we often found

the barrier to move with velocities of the order of 1 m/secµ .


In order to obtain LB films, first a substrate was immersed perpendicularly in the aqueous

subphase. We started with a very small initial surface pressure ( 0π = mN/m), and

compressed the monolayer slowly (1 cm/min) to the final pressure. To obtain a LB film that is

56

characteristic for forced compression, the film was transferred immediately by vertical pulling

of the substrate through the air-water interface at a speed of 2 mm/min. During the transfer the

surface pressure was kept constant by appropriately moving the barrier. The transfer process

takes a few minutes. After deposition the LB-films were dried in air and kept in close

containers until use. All experiments were done at 20 ± 1°C.


The samples were examined with AFM immediately after preparation. Imaging was done with

a Nanoscope (R) IIIa (Digital Instruments, Santa Barbara, CA) in contact mode with oxide-

sharpened silicon nitride tip (k = 0.06 N/m). The AFM was equipped with a J scanner

(176 x176µm; z limit = 5.349 µm). All images were processed using procedures for flattening

in Nanoscope III software version 5.12r5 without any filtering. To check if the monolayer is

successfully transferred to the mica surface we measured at least five different spots (each

150 µm 2) of every sample. In order to detect structural changes in the adsorbed film in

contact with air we studied LB films several days after incubation at room temperature (20 ±

1°C) as well.


4.3.1. Forced compression

Fig.1. Surface pressure vs area

isotherms of tripalmitin (PPP),

tristearin (SSS) and triarachidin

(AAA) at air-water interface, at

20oC, obtained by forced

compression at a rate of 1cm/min

(the thermodynamic equilibrium

isotherms are obtained, in

principle, at very slow

compression rates).

02468

101214161820

40 50 60 70 80 90 100 110Area/molecule A ( A2)

Surfa

ce p

ress

ure

(mN

/m)

PPP

SSS

AAA

57

Fig. 1 shows typical Aπ − isotherms of tripalmitin (PPP), tristearin (SSS) and

triarachidin (AAA), recorded at a barrier velocity of 1 cm/min. Two different regimes can be

recognized for the three triglycerides. Starting at a large area per molecule A the pressure is

low and increases only slowly with decreasing A. Upon decreasing A further the condensation

area is reached and the pressure starts to increase more rapidly. The pressure at which

was reached we will call condensation pressure

condA

condA condπ .We interpret the low pressure

phase as “gaseous”, and the high pressure phase as “condensed”. On the basis of Langmuir

experiment alone we can not conclude whether the condensed phase is liquid-like or solid-

like. For condπ π≥ the molecules are close enough together to form a condensed monolayer.

The collapse pressure colπ is the surface pressure at which the monolayer collapses to form

multilayer structures. For the studied triglycerides it was in the range of 40 48mN/mπ = −

and it increases in order: (AAA) (SSS) (PPP)col col colπ π π< < . With our LB instrument the

collapse pressure was difficult to reproduce because of technical problems. colπ

The measured Aπ − data showed that the transition from one regime to another was

not very sharp and it was different for the investigated triglycerides. In order to get reliable

and unbiased estimations for and condA condπ , we fitted the isotherms with:

( ) ( , )condA ch A A aπ ≈ − (1)

here , , c and are fitting parameters. The function condA a h

( )2 21( , )2

h x a x x a≡ − + (2)

is a hyperbola interpolating between for large negative and for

large positive x . This function has no direct physical interpretation and was introduced for

practical purposes only, i.e. to arrive at an unambiguous definition and evaluation of

( ),h x a x≈ x ( ), 0h x a ≈

/ 2cond caπ = and . Fitting a number of isotherms that were obtained at compression

velocity 1cm/min we found and

condA

262 1ÅcondA = ± 6 2mN/mcondπ = ± for PPP;

and

262 1ÅcondA = ±

9 2 for SSS and and 265 1ÅcondA = ± 8 1mN/mcondπ = ± for AAA (fig.2). mN/mcondπ = ±

58

B Condensation pressure πcond

0

5

10

15

20

25

14 16 18 20 22

Number carbon atoms

πco

nd (m

N/m

)

Fig.2. Condensation area (A) and condensation pressure condA condπ (B) as a function of

number of carbon atoms in the alkyl chains of the triglycerides: 16 (PPP), 18 (SSS) and

A Condensation area A cond

60

61

62

63

64

65

66

67

14 16 18 20 22

Number carbon atoms

A con

d (Ag

stro

m s

quar

ed)

20 (AAA).

The fact that is around for all studied triglycerides is consistent with a

trident conformation of triglyceride molecules in a monolayer film at the air-water interface.

Indeed, the cross-sectional area per hydrocarbon chain for tristearin at 20

condA 263 Å

oC in the α phase

(the α phase has the most mobile acyl chains) is [13]. Our isotherms are consistent

with earlier reports [8, 12] and the results in Chapter 3.

219.7Å

The fact that condπ is almost the same for the investigated triglycerides as well (Fig.2

B), is consistent with the idea that the packing properties of the hydrocarbon chains is mainly

determined by short range repulsive interactions. The effective repulsion is quite independent

of the chain length.

The value of the fitting parameter (Eq.1) describes the sharpness of the gas -

condense transition and depends strongly on the chain length. This is also seen in Fig.1 where

the

a

Aπ − isotherm for PPP is sharper than those for SSS and AAA. This effect is of a kinetic

nature, i.e. PPP realizes the transition from gas to condensed phase faster than the longer SSS

and AAA. The longer chains need more time to transform and rearrange in perpendicular

position. If this was solely due to the flexibility and mobility of individual triglycerides, the

sharpness of the adsorption isotherms would be strongly dependent on the compression rate.

As we did not observe such dependence we conclude that the sharpness of the PPP isotherm

also is influenced by interactions between the molecules. The isotherms in Fig.1 suggest that

in a moderately dense packed monolayer at the air-water interface the longer triglycerides

59

(SSS and AAA) repel each other significantly at larger intermolecular distances than the

shorter PPP. From this perspective we shall discuss the experimental results in the next

section.

4.3.2. Isobaric compression

To investigate the stability of the triglyceride films at an air-water interface we made the

following experiment. We stopped the forced compression at a constant compression rate fν

when a certain surface pressure was reached. Next we kept the surface pressure constant at

that value, allowing the barrier to move. After several minutes a constant isobaric velocity

π

ν

was reached usually. Due to molecular loss from the monolayer the barrier moved forward to

keep the surface pressure constant and the through length l(t) decreased. This is shown in

Fig.3.

Barrier position vs time for speading pressure π = 20mN/m

9.50

10.00

10.50

11.00

11.50

0 1000 2000 3000 4000 5000time t (sec)

l(t)

(cm

)

AAASSSPPP

Fig .3. Examples of the measured barrier position as a function of time during forced isobaric

compression for ( )PPP, (□) SSS and (○) AAA. The almost vertical parts of the curves

correspond to forced compression rate of

∆

fν ≈ 1 cm/min. The slowly decreasing parts

correspond to small residual isobaric compression rates at velocity ν at the spreading

pressure 20mN/mπ = . Note that the molecular loss, presented by l(t) is higher for PPP.

60

The evolution of the trough length was fitted to

(3) ( ) ( ) ( ) ( )0 0 ,fl t l v t t v v h t t a≈ − − − − − 0

Here the five fitting parameters are l , the trough length at the start of the isobaric

compression, , the starting time of the isobaric period,

0

0t fv and v , the forced and isobaric

barrier velocity respectively, and a , characterizing the transition from the forced to the

isobaric regime. The accuracy of the fits typically was 0.2%. In all cases the fitted forced

velocity fv was very close to the applied barrier velocity.

Fig.4. Isobaric velocity

ν (µm/sec) as a function

of the spreading

pressure π, as obtained

by fitting the measured

barrier position to Eq.

(3) for PPP, SSS and

AAA. Note the sharp

increase of ν for PPP for

spreading pressure close

to the collapse pressure

48mN/mcolπ ≈

In previous work [Chapter 3] we found that for SSS the adsorption isotherms that were

Isobaric compression

0

5

10

15

20

25

30

35

40

45

0 5 10 15 20 25 30 35 40 45

Spreading pressure ( mN/m)

Vel

ocity

v (u

m/s

ec)

PPP

SSS

AAA

obtained by forced compression did not change appreciably when increasing the barrier

velocities from 0.5 to 2 cm/min. Nevertheless under isobaric conditions we observed further

compression. We noted that the isobaric velocities were one or two orders of magnitude

smaller than commonly used forced velocities of ~ 1cm/min. The same behavior we observe

here for PPP and AAA. In Fig. 4 we show the dependence of the isobaric velocity v on the

surface pressure π for the three triglycerides. It can be noted that for PPP 0v for

10mN/m

≈

π ≤ and ν depends linearly on π for 10mN/m 25mN/mπ≤ ≤ . For pressures

61

30mN/mπ ≥ , i.e. close to the collapse pressure, a m is found. These

that the PPP isotherm in Fig.1 cannot be considered as an equilibrium isotherm

for 10mN/m

uch faster compression

results show

π > . For such pressures the equilibrium value of A is smaller than the value

give

At this p

n in Fig.1.

oint it is important to clearly discriminate between collapse pressure colπ and

equilibrium pressure eqπ . In isobaric conditions a molecular rearrangement process takes

place which effectively thickens the film. Using Atomic Force Microscopy for SSS we have

shown that this process involves the growth of 3D crystals of SSS on top of the monolayer,

which is precisely what one should expect for eqπ π> . For colπ π> the same crystallization

process takes place, but in a less controlled and less reproduc nner. This interpretation

is in line with the definition of Roberts [14]. Note however that sometimes one names

equilibrium spreading pressure what we call collapse pressure, see e.g. [5]. According to

thermodynamics equilibrium (spreading) pressure is the surface pressure that is

spontaneously generated when a sample of triglyceride in its thermodynamically stable phase,

i.e. in the crystalline

ible ma

β phase, is brought in contact with the water surface. Provided that

sufficient time is allowed for equilibration, one can, in principle, be sure that the monolayer

which has been formed by molecules detaching themselves from the crystal surface and

spreading over the subphase is in equilibrium with the crystals themselves. At surface

pressures higher than eqπ there will be a tendency for the monolayer to aggregate into crystals

[14]. Another way to express this, is that the chemical potential of the triglycerides in the β

phase, βµ , is equal to the chemical potential Lµ of triglycerides in the Langmuir layer on th

water surface:

e

( )L eq βµ π π µ= ≡

Therefore we can, in principle estimate eqπ from the dependence of the isobaric

velocity ν on π . In Fig.4 we suggest that there exist a surface pressure 0π , such that ν is

proportional to 0π π− for 0π π> . If we assume that 0 eqπ π≈ we get an estimation 0π of

eqπ . On the other hand, we know from our AFM observation that the process taking place for

0π π is crystal growth and therefore a non-linear dependence of > ν on eqπ π− is expected.

stal growth theory for faceted crystals 0 eqIn cry π π− would be called “nucleation gap”,

62

meaning that for π in the regime 0eqπ π π≤ ≤ no observable growth takes place. Thus we

should realize that the real equilibri equm pressure π may be well below the linear estimate

0π . According to our results (Fig.4) for tripalmitin (PPP) at air-water interface 0 10mN/mπ = .

e instability of the monolayer of PPP at 0Th π π> increases with the surface

same dependence was reported before for m almitin at air-water interface [3, 5]. The

isobaric velocity for SSS given here is slightly smaller than what we measured before

[Chapter 3] and 0 15mN/m

pressure. The

onop

π = is higher (fig.4). The difference in the results probably is

caused by the bett trough that we used in this work. It is known that the use of

Delrin barrier in our present system drastically improves the quality and reproducibility of the

experiments [15]. It is also been observed that simple Langmuir systems (like the one we used

before) tend to underestimate the spreading pressure in condensed films. From our

experiments it was difficult to estimate 0

er Langmuir

π for AAA. The measured isobaric velocity was too

close to zero for all investigated surface pressures (fig.4). From our experiments it can be

concluded that both the mobility, characterized by ν and the effective stability, characterized

by 0π of triglycerides monolayers at an air-water interface depend on the film composition,

notably on the length of the triglycerides. For PPP and SSS we can determine an upper bound

0π for the equilibrium pressure eqπ and we see that 0π is lower for PPP than for SSS. This

e mobility and flexibility will decrease with increasing of

the chain length.

The increa

dependence can be understood as th

sing interactio th will also influence the thermodynamic stability of n streng

the crystal phase, as is signified both by an increasing melting point and by decreasing

equilibrium vapor pressure. Therefore one would expect ( ) ( ) ( )eq eq eqPPP SSS AAAπ π π≥ ≥ .

This seems contradictory to the inequality 0 0( ) (PPP )SSSπ π< that we found above. It i

lar i on gen

s

well-known however, that a stronger molecu erally leads to stronger non-

linearity of the dependence of the growth rate and the nucleation rate on eq

nteracti

π π− . As a

consequence the nucleation gap 0 eqπ π− will be larger for SSS than for PPP. Only if we could

make full non-linear ( )ν π fits to data like those in Fig.4, we might find the real dependence

of eqπ on the chain length.

One should also expect that the stronger interaction reduces the mobility of

triglyceride molecules in a condensed phase. This effect is seen by the decrease of the isobaric

63

velocity when changing from PPP to SSS and to AAA. The isobaric velocity for PPP and

SSS, and presumably for AAA as well, depends roughly linearly on the surface pressure when

π is far enough above eqπ , i.e. for 0π π> . The slope /d dν π is positive and increases with

decreasing chain length (Fig.4).


ess

In es form a trident monolayer at an air-water

interface, which can be transferred to mica. We found that the monolayer thickness varied

4.4.1. Monolayer thickn

Chapter 3 we showed that SSS molecul

from 1.6 to 1.8 nm over the pressure range we studied. The molecules were tilted and the tilt

angle varied from o43τ = at 10mN/mπ = to o53τ = at 40mN/mπ = .

From the AFM images of LB-films of PPP, SSS and AAA, withdrawn at

20 mN/mπ = (fig. is s e m5) it een that th ica substrate is covered by a homogeneous

water surface

monolayer. Apparently the Langmuir monolayer can be successfully transferred from the

in the Langmuir trough to a mica surface to form a Langmuir-Blodgett film

there. When the LB-films were prepared at lower pressures (data not shown) a monolayer was

observed as well, but with a lot of holes in it.

64

A B C

0

0

2.5-5.0

5.0

5.0

1.35 nm 1.39 nm

µm

D

µm2.50-5

.0

5.0

5.0

0

1.57 nm 1.61 nm

E

0µm

0

2.5 5.0

5.0

-5.0

2.12 nm 2.12 nm

F

AA BB CC

0

0

2.5-5.0

5.0

5.0

1.35 nm 1.39 nm

µm

D

0

0

2.5-5.0

5.0

5.0

1.35 nm 1.39 nm

0

0

2.5-5.0

5.0

5.0

1.35 nm 1.39 nm

µm

D

µm2.50-5

.0

5.0

5.0

0

1.57 nm 1.61 nm

E

µm2.50-5

.0

5.0

5.0

0

1.57 nm 1.61 nm

2.50-5.0

5.0

5.0

0

1.57 nm 1.61 nm

E

0µm

0

2.5 5.0

5.0

-5.0

2.12 nm 2.12 nm

F

0µm

0

2.5 5.0

5.0

-5.0

2.12 nm 2.12 nm

0

2.5 5.0

5.0

-5.0

2.12 nm 2.12 nm

F

Fig.5. AFM height image of monolayers transferred immediately after forced compression to

surface pressure π = 20mN/m. The black squares are holes in the monolayer produced by

scanning at a high AFM force (F~30 nN). The monolayers are scanned at AFM force F =

1nN. (A) PPP monolayer with corresponding cross section (D). (B) SSS monolayer and the

corresponding cross section (E). (C, F) AAA monolayer. The scale bar is 2 µm and the

vertical scale is 10 nm for all images.

We estimated the monolayer thicknesses using the following procedure. By

scanning with a relatively large force

0d

30 nNF ≈ we scratched a rectangular hole in the

monolayer with the AFM tip. Then a larger area, including the hole was scanned with small

forces 1 8 nNF = − (fig.5). The height difference between the hole and the surrounding film

gives an apparent thickness d . Analyzing our data carefully we found that d turned out to

depend on the scanning force F for all investigated triglycerides. Therefore the real monolayer

thickness differs from d . In Fig.6 we present the observed dependence of d on . It is

seen that d is larger for larger chain length, as expected. It is also seen that the vertical

compressibility of the monolayer, given by the slope of curves, is the same for PPP,

SSS and AAA. The real thickness

′ ′

0d ′ ′ F

′

'( )d F

0 '( )d d F= , corresponding to scanning force is

presented in Fig. 7. It is linear dependent on the length of the chains in the triglyceride

molecules.

0F =

65

Apparent monolayer thickness at spreading pressure π = 20 mN/m

0

0.5

1

1.5

2

2.5

3

0 2 4 6 8 10

AFM force F (nN)

AFM

thic

knes

s d'

(nm

)

PPP

SSS

AAA

Fig.6. Measured layer thickness d for PPP, SSS and AAA as a function of applied AFM

force F at surface pressure

′

20mN/mπ = .

Real monolayer thickness at spreading pressure π =20 mN/m

1

1.2

1.4

1.6

1.8

2

2.2

2.4

12 14 16 18 20 22 24

number of carbon atoms in the chain

Laye

r thi

ckne

ss d

0 (n

m)

Fig.7. Real monolayer thickness for PPP (16), SSS (18) and AAA (20) at surface pressure 0d

20mN/mπ = . The values are found by extrapolation of the apparent monolayer thickness

in Fig.6 to AFM force F = 0 nN.

'd

To translate the thickness into a tilt angle τ of the acyl chains with the mica surface

we need to estimate the effective chain length. A precise analysis and interpretation of

crystallographic data for the long spacing of PPP, SSS and AAA in the stable β-phase

[16], allows us to estimate an effective length of PPP, SSS and AAA-molecules in tuning

fork conformation. Correcting this for the length of the glycerol group and the contribution

(001)d

effd

66

from the contact region of (001) layers in the β phase, the alkyl chain length can be

estimated. When carrying out an analogous procedure for the α and β’ phase the same lengths

are found.

chaind

chaind

We have no detailed information on the molecular conformation of the triglyceride

molecules in the monolayer. In order to estimate the tilt angle in the monolayer, we assume

that the glycerol part of the molecule makes close contact with the (hydrophilic) substrate.

The alkyl chains are stretched similar as in α , β and 'β phases, though in different

orientation with respect to the glycerol group. This leads to a structure where alkyl chains

extend from the substrate to the monolayer surface at a height above the substrate.

Thus in the monolayer the triglyceride molecules adopt a trident conformation we get a

simple relation:

chaind 0d

0sin( ) / chaind dτ = (4)

The estimated chain lengths and the tilt angles are given in the table below.

triglyceride (001)d

β - phase (nm) effd (nm) chaind (nm) 0d (nm) of

monolayer

Tilt angle τ

PPP 4.03 4.62 1.80 1.49 46.4o

SSS 4.48 5.13 2.31 1.75 49.2o

AAA 4.92 5.62 2.82 2.20 59.0o

In the stable β-phase the triglyceride molecules are tilted at tilt angle [16].

Since presumably in the trident monolayer the alkyl chains are less densely packed than the

crystalline phases, a smaller tilt angle seems acceptable. Interpreting our monolayer thickness

data with Eq.4, we see that the tilt angle increases with increasing the length of the alkyl

chains, i.e. the longer chains in the monolayer are more perpendicular to the substrate than the

shorter. In the aliphatic alcohol monolayers on water the same dependence of the tilt angle on

chain length was found. IR spectra of these alcohol monolayers showed that the hydrocarbon

chains become more ordered with increasing length [17].

o60.8τ =

67

4.4.2. Stability of the transferred LB-film

In our previous investigation we found that during incubation in air transferred SSS - films

become slightly thicker and grainy. Crystals that were present directly after the transfer were

growing and new very small nuclei appeared [Chapter 3]. We suggested that the grainy

character of the monolayer is due to small clusters of molecules, which leave the monolayer

to form new crystals or contribute to the growth of existing crystals. To investigate this

mechanism further we transferred PPP, SSS and AAA Langmuir layers to mica, immediately

after forced compression to spreading pressure π, and left them in closed containers for a few

days at room temperature ( C). o20 1±

4.4.2.1. Initial structure and structural changes of PPP-monolayer

After 30 min incubation on water surface at surface pressure 10mN/mπ = (this is 0π for

PPP), which we expect to be close to eqπ for PPP, the AFM images show a homogeneous

monolayer with small holes when the LB-film was transferred to mica (data not shown). At

surface pressure 20mN/mπ = ( 0π π> ) we found that the directly transferred PPP-film

consists of a closed monolayer, onto which many small domains were positioned. The

thickness of the domains ranged from 3.3 nm to 4.6 nm above the monolayer level (Fig.8A,

D). These domains were soft and could be easily scratched away with the AFM tip, even at

the low AFM forces that are normally used for imaging. Indeed, after a few scans with

these domains usually had disappeared, leaving a flat film with the thickness of

the trident monolayer. The maximum measured height of the domains 4.6 ± 0.1nm

corresponds to fully extended PPP molecules in tuning fork conformation perpendicular to the

substrate, i.e. to one crystalline layer in the

1 2 nNF = −

α phase. This result is consistent with the growth

of crystals in tuning fork conformation on top of the trident monolayer, reported for SSS

[Chapter 3]. We therefore may assume that the observed domains are small crystals, formed

in the period where the spreading pressure increased from the small values at which the film

is in a two-dimensional gas state, to the final pressure 0π π− at which the condensed phase

has formed. In this period PPP - molecules undergo major orientation and packing changes.

As a result the formation process of the domains will not be strictly deterministic and a

68

metastable film structure may form. The domains serve as crystal seeds which can grow into

bigger crystals when the LB - film is incubated longer time in air (Fig.8).

B

C

3.3 nm4.5 nm

tm

αm

m

10.0

-10.

00

10.05.00µm

D

0 5.0 10.0

10.0

-10.

0

0

3.3 nm 4.6 nm

m tm

αm

µm

E

10.0-10.

0

0

10.0

5.00

tm

αm

αm

m

4.5 nm4.6 nm3.3 nm

µm

F

A

BB

CC

3.3 nm4.5 nm

tm

αm

m

10.0

-10.

00

10.05.00µm

D3.3 nm4.5 nm

tm

αm

m

10.0

-10.

00

10.05.00µm

D

0 5.0 10.0

10.0

-10.

0

0

3.3 nm 4.6 nm

m tm

αm

µm

E

0 5.0 10.0

10.0

-10.

0

0

3.3 nm 4.6 nm

m tm

αm

µm0 5.0 10.0

10.0

-10.

0

0

3.3 nm 4.6 nm

m tm

αm

µm

E

10.0-10.

0

0

10.0

5.00

tm

αm

αm

m

4.5 nm4.6 nm3.3 nm

µm

F

10.0-10.

0

0

10.0

5.00

tm

αm

αm

m

4.5 nm4.6 nm3.3 nm

µm10.0-1

0.0

0

10.0

5.00

tm

αm

αm

m

4.5 nm4.6 nm3.3 nm

µm

F

AAA

Fig.8. AFM height image of PPP monolayers transferred at 20 mN/mπ = . (A) immediately

after forced compression, (B) the same sample after 1 day incubation in air at room

temperature, (C) the same sample after 2 days incubation in air. The corresponding cross

sections are given in (D, E and F). The scale bar is 2 µm and the vertical scale is 20 nm for

all images. Height differences are given by the numbers at the markers. The symbols below

the lines give our proposed structure of the crystals (m – monolayer in trident conformation;

t – top layer in tuning fork conformation; α - crystal in tuning fork conformation).

69

The AFM images of LB-films of PPP incubated for 1 day in air showed that the

monolayer is still present and the density of the crystals is higher. Most of the new crystals

were small and with thickness 4.6 ± 0.1nm. Some bigger domains were found with thickness

3.3 ± 0.1 nm (Fig.8B, E). After 2 days incubation the domains were bigger, mostly with

thickness 4.6 ± 0.1 nm and still surrounded by the trident monolayer (Fig.8C, F). Apparently

immediately after the transfer a few crystals are present, some in α phase, other in β phase. In

air these crystals grow and new crystals, mostly in α phase, nucleate. The same processes

were observed for PPP monolayer transferred at 10 mN/mπ = after a few days incubation in

air, but the speed of nucleation is lower (Fig.9).

A Crystal density for PPP

0

0.5

1

1.5

2

2.5

3

3.5

0 1 2 3days

Cry

stal

s pe

r µm

2


B Volume per total area for PPP

0

0.00002

0.00004

0.00006

0.00008

0.0001

0.00012

0 1 2 3days

Volu

me

per t

otal

are

a ( µ

m)


Fig.9. Crystals per area (A) and average crystal layer thickness (B), formed during

incubation of PPP LB-monolayer in air at room temperature.

70

From these observations we conclude that the LB monolayer is unstable both with

respect to the α -phase and to the β -phase:

( ) ( ) (LB PPP PPP PPPα β )µ µ µ> > (5)

where ( )LB PPPµ is the chemical potential of PPP molecules in a monolayer on the mica

surface. Note that from the LB layer mainly the less stable α -phase grows. This is a

manifestation of Ostwalds rule, which states that if two (or more) phases can grow in

principle, then the least stable of these phases usually will dominate because it is less ordered

and hence its growth kinetics are faster.

4.4.2.2. Initial structure and structural changes of SSS-monolayer

When the SSS-film was withdrawn after 30 min incubation at air-water interface at

15mN/mπ = (this is 0π for SSS) we observed only a homogeneous (trident) monolayer with

a lot of holes. After 2 days incubation in air the LB - monolayer was covered with small

domains with thickness 3.6 ± 0.1nm (data not shown). At surface pressure

20mN/mπ = ( 0π π> ) we found that the directly transferred SSS film consisted of an almost

defect free monolayer, in which a few domains were imbedded. The thickness of the domains

was 3.6 ± 0.1nm (fig.10A, D). After 1 day incubation in air the LB-monolayer was covered

with many new nuclei with thickness 3.6 ± 0.1nm (fig10B, E). After 2 days incubation in air

the observed nuclei were bigger, all with the same thickness (fig.10C, F).

71

B

C

5.0µm

tm

mtm

tm

3.6 nm10.0

10.0-10.

00

0

3.7 nm 3.7 nmE

5.00

10.0

-10.

0

10.0

0

µm

3.6 nm

mtm

D

3.6 nm

1.7 nmtmm

mica

µm5.0 10.00-1

0.0

010

.0F

1.7 nm3.6 nm

micam

tm

A

BB

CC

5.0µm

tm

mtm

tm

3.6 nm10.0

10.0-10.

00

0

3.7 nm 3.7 nmE

5.0µm

tm

mtm

tm

3.6 nm10.0

10.0-10.

00

0

3.7 nm 3.7 nm

5.0µm

tm

mtm

tm

3.6 nm10.0

10.0-10.

00

0

3.7 nm 3.7 nmE

5.00

10.0

-10.

0

10.0

0

µm

3.6 nm

mtm

D

5.00

10.0

-10.

0

10.0

0

µm

3.6 nm

mtm

5.00

10.0

-10.

0

10.0

0

µm

3.6 nm

mtm

D

3.6 nm

1.7 nmtmm

mica

µm5.0 10.00-1

0.0

010

.0F

1.7 nm3.6 nm

micam

tm

3.6 nm

1.7 nmtmm

mica

µm5.0 10.00-1

0.0

010

.0F3.6 nm

1.7 nmtmm

mica

µm5.0 10.00-1

0.0

010

.0

3.6 nm

1.7 nmtmm

mica

µm5.0 10.00-1

0.0

010

.0F

1.7 nm3.6 nm

micam

tm

AA

Fig.10. AFM height image of SSS monolayers transferred at 20mN/mπ = . (A) immediately




all images. Length differences are given by the numbers at the markers. The symbols below


α - crystal in tuning fork conformation; t – top layer in tuning fork conformation).

72

The structure of SSS-films immediately transferred after forced compression to

30mN/mπ = is reported in Chapter 3.

A

10.0

µm10.0

D4.8 nm

mαm

αm

5.00

0-1

0.0

4.7 nm

B

3.5 nm 4.7 nm

0 5.0µm

-10.

010

.00

αm

tm

m

E

10.0

3.8 nm 3.6 nm 3.6 nm

tm

tm

tmm

10.05.0µm

0-10.

00

10.0F

C

AA

10.0

µm10.0

D4.8 nm

mαm

αm

5.00

0-1

0.0

4.7 nm10.0

µm10.0

D4.8 nm

mαm

αm

5.00

0-1

0.0

4.7 nm

BB

3.5 nm 4.7 nm

0 5.0µm

-10.

010

.00

αm

tm

m

E

10.0

3.5 nm 4.7 nm

0 5.0µm

-10.

010

.00

αm

tm

m

10.0

3.5 nm 4.7 nm

0 5.0µm

-10.

010

.00

αm

tm

m

E

10.0

3.8 nm 3.6 nm 3.6 nm

tm

tm

tmm

10.05.0µm

0-10.

00

10.0F

3.8 nm 3.6 nm 3.6 nm

tm

tm

tmm

10.05.0µm

0-10.

00

10.0

3.8 nm 3.6 nm 3.6 nm

tm

tm

tmm

10.05.0µm

0-10.

00

10.0F

CC

Fig.11. AFM height image of SSS monolayers transferred at π = 30 mN/m. (A) immediately




all images. Length differences are given by the numbers at the markers. The symbols below


α - crystal in tuning fork conformation; t – top layer in tuning fork conformation).

73

Most of the observed crystals had a thickness 4.9 ± 0.1 nm above the monolayer level. This

thickness corresponds to fully extended alkyl chains of SSS (fig.11A, D). When the LB-

monolayer was incubated in air for 1 day the existing crystals were growing and new very

small nuclei appeared. The thickness of the new nuclei and the newly grown parts of the

crystals was 3.6 ± 0.1(fig.11B, E). After 2 days incubation in air the new nuclei became

bigger (fig.11C, F).

The density of the crystals of SSS formed during incubation in air is initially growing

with time (fig.12A).

A Crystal density for SSS

00.20.40.60.8

11.21.41.6

0 1 2 3days

Cry

stal

s pe

r µm

2

π = 15mN/m

π = 20mN/m

π = 30mN/m

B Volume per total area for SSS

0

0.00001

0.00002

0.00003

0.00004

0.00005

0.00006

0.00007

0 1 2 3days

Vol

ume

per a

rea

( µm

)

π = 15mN/m

π = 20mN/m

π = 30mN/m

Fig.12. Crystals per area (A) and average crystal layer thickness (B), formed during

incubation of SSS LB-monolayer in air at room temperature.

Note however that the density decreased at 30mN/mπ = after 2 days of incubation.

This is due to coalescence of the large number of relatively large crystals at this pressure. The

74

average crystal layer thickness during the incubation increases with the surface pressure and

the time (Fig.12B).

Summarizing, we found that for SSS immediately after transfer, several crystals are

observed which are mainly in the β - phase for 20mN/mπ ≤ and mainly in α - phase for

30mN/mπ ≥ . This suggests that 0 (SSS, )π β is 15mN/m or smaller, and

0 (SSS, ) 25mN/mπ α ∼ . The fact that 0 0( ) ( )π α π β> could have been explained

since ( ) ( )eq eqπ α π β> , because the β -phase is more stable than the α -phase, and the

nucleation gap 0 eqπ π− probably is not much different for α and β . After transfer to mica

surface the existing crystals grow further and new crystals appear, all mainly in the β -

phase, but not with respect to the α - phase. In terms of chemical potentials this means that

( ) ( ) (LBSSS SSS SSSα )βµ µ µ> > (6)

Note that the location of the LB chemical potential LBµ with respect to αµ and βµ is

different for SSS and for PPP.

4.4.2.3. Initial structure and structural changes of AAA-monolayer

Fig.13 presents a LB-film of AAA transferred immediately after forced compression at

20mN/mπ = and incubated 2 days in air. We did not observe any crystals, but the monolayer

was somewhat coarse. We suppose that this coarsening was due to very small nuclei, which

were difficult to detect directly with the AFM. If we would have incubated the monolayer for

a longer time, then probably the existing nuclei would grow into crystals. The absence of well

developed crystals after two days incubation is due to very slow kinetics of AAA. In term of

chemical potentials we hypothesize the same relative positions as for SSS:

( ) ( ) (LB )AAA AAA AAAα βµ µ µ> > (7)

The slower kinetics of the AAA monolayer (as compared to SSS and PPP) is expected in

view of the stronger interaction between the longer alkyl chains of AAA.

75

A

10.05.00

10.0

-10.

00

µm

2.1 nm

m

BAA

10.05.00

10.0

-10.

00

µm

2.1 nm

m

B

10.05.00

10.0

-10.

00

µm

2.1 nm

m

10.05.00

10.0

-10.

00

µm10.05.00

10.0

-10.

00

µm

2.1 nm

m

B

Fig.13. AFM height image of AAA monolayer transferred at 20mN/mπ = immediately after

forced compression and incubated 2 days in air (A) with the corresponding cross section (B).

The scale bar is 2 µm and the vertical scale is 20 nm (m – monolayer in trident

conformation).

4.5. Discussion

By definition trident monolayers formed of PPP, SSS and AAA are thermodynamically stable

for eqπ π≤ at the air-water interface. We discussed several methods to estimate eqπ . First

from the collapse pressure colπ . This leads to a large overestimation, col eqπ π . A practical

estimate is condπ , obtained from fitting experimental isotherms. We found 8mN/mcondπ ∼ for

all three triglycerides. The reliability of the assumption cond eqπ π≈ however, is unclear, both

theoretically and experimentally as condπ may depend on the forced compression rate, eqπ

not. From isobaric compression we obtained the pressure 0π below which compression was

absent or too slow to be measured. It is clear that 0 eqπ π≥ , but unfortunately the amount of

overestimation, i.e. the “nucleation gap” 0 eqπ π− , can not be deduced from our data. We

found that 0 10mN/mπ = for PPP, 0 15mN/mπ = for SSS and 0 20mN/mπ = for AAA. We did

not observe any changes in the structure of the trident monolayers on the air-water interface in

the regime 0π π≤ . Combining all our observations with a physically reasonable picture we

conclude that 10mN/meqπ ≤ for all three triglycerides, thus supporting the idea eq condπ π≈ .

For all eqπ π> the Langmuir monolayers are thermodynamically unstable though this

becomes evident in the compression data only for 0π π> .

76

Under isobaric conditions at 0π π> slow, but observable compression rates were

found for PPP and SSS. The isobaric velocity was highest for PPP and almost zero for AAA.

We demonstrated that the stability and the kinetics of Langmuir monolayer depend strongly

on the length of the triglycerides alkyl chains. The trident monolayer is the less mobile, and

the crystal phase is the more stable, the longer the alkyl chains are.

The AFM images of LB-films transferred immediately after forced compression at

0π π> for PPP and SSS showed some domains on top of the monolayer. The molecules in

these domains presumably adopt the tuning fork conformation and pack similar as in the

crystalline α and β crystal forms. This film structure for SSS was explained with the model

we proposed in Chapter 3. The observed domain thickness of 3.6 ± 0.1 nm at 20mN/mπ =

for SSS, corresponds to a tilt angle , i.e. somewhere between the estimated tilt

angle in the trident monolayer and the tilt angle in the stable

o43 44.5τ = −

β phase (fig.10A, D). We

suppose that the structure of these layers can be described as a slightly deformed β or 'β

phase. As we observed this layer thickness always at the upper crystal layer, also for

multilayer crystals, we refer to this structure as the top layer structure (‘t’ in the figures).

At surface pressure 30mN/mπ = some domains extended as much as 5.0 ± 0.1 nm

above the surrounding SSS monolayer (fig.11A, D). This suggests that in these domains the

molecules are fully stretched (5.13 nm) and oriented perpendicular to the monolayer, i.e. the

structure of these domains is similar to the crystalline α phase. The fact that we did not

observe such α -domains for 20mN/mπ ≤ shows that 0 (SSS, ) 20mN/mπ α > , which is

consistent with the higher stability of the β -phase, since 0 (SSS, ) 15mN/mπ β ≈ .

Even though the SSS trident monolayer was stable at eqπ π≤ ( 15mN/mπ = ) at the

air-water interface, a LB-monolayer that was transferred changed its structure during

incubation in air. Small nuclei with thickness 3.6 ± 0.1nm ( β - phase structure) appeared on

top of the monolayer (data not shown). The same crystal growth process takes place during

incubation in air of LB-monolayer of SSS that was transferred at 20mN/mπ = (Fig.10 B, E).

Again the thickness of the newly formed nuclei was 3.6 ± 0.1nm. After 2 days incubation the

density of the nuclei was higher, but their thickness remained the same (fig.10 C, F). On a

LB-film of SSS transferred at surface pressure 30mN/mπ = initially some α - like structures

were found, but they did not change in time. The newly formed parts around them, and all

new nuclei as well, have a thickness of 3.6 ± 0.1nm, which corresponds to the β or 'β phase

77

(fig.11). Domains that were grown in the metastable, α - like, polymorph phase on the water

surface, do not spontaneously transform to the β or 'β phase because this would involve a

very slow solid - solid transformation process. We conclude that, on mica, β is more stable

than a trident monolayer, and α probably not. Stated in terms of chemical potentials this is

expressed in Eq.6 in section 4.4.2.2.

PPP behaves similar to SSS. The Langmuir monolayer of PPP does not change at the

air-water interface at 0 10mN/mπ ≤ and it is thermodynamically unstable at 0 10mN/mπ ≥ .

Domains with different thickness were found in LB monolayer, that was transferred

immediately after forced compression at 20mN/mπ = (fig.8A, D). Judging from their

thickness most of them were in α - phase, some in the β - phase. Using the estimated

effective length of 4.62 nm for a PPP molecule in tuning fork conformation, the observed

domain thickness of 3.3 ± 0.1 nm corresponds to a tilt angle , which is

between the estimated tilt angle in the trident monolayer and the tilt angle in the stable

o43.8 47.3τ = −

β

phase (‘t’ in the figures). The height of 4.5 ± 0.1 nm corresponds to fully extended PPP

molecules (4.62 nm) almost perpendicular to the substrate, similar to the crystalline α - phase.

Contrary to SSS, domains with α - like and β - like structure coexist in the LB film of PPP.

All transferred LB monolayers at 0 10mN/mπ ≥ change during incubation in air. The AFM

images showed that after 1 day in air small nuclei in α - phase and the β - phase appeared on

top of the PPP monolayer and the density of the crystals extremely increased. After 2 days the

crystals become bigger and their density decreased due to the crystals coalescence (fig.9 A).

Comparing figures 9 (in 4.4.2.1.) and 12 (in 4.4.2.2.) we see that the growth and

nucleation rates depend on the surface pressure and the nature of the triglyceride. The LB-

monolayers, which were transferred at eqπ π≥ are thermodynamically unstable in air. The

different mobility of the molecules in the trident monolayer of SSS and PPP is the main

reason for the different rate of nucleation. The increase of the nucleation rate with increasing

π reflects that the initial monolayer on mica is denser, and hence more unstable, when the

Langmuir layer is transferred at higher π . Indeed we may expect that the chemical potential

of monolayer on mica is close to the chemical potential of the monolayer on water, i.e.

( 0) (LB Lt )µ µ π= ≈ . The monolayer molecules thus can reduce their free energy from about

( )Lµ π to αµ or βµ by moving to the top of the monolayer to form new crystals in tuning

fork conformation.

78

Based on our results for isobaric velocity at the air-water interface (section 4.3.2.) we

concluded that PPP has a smaller 0π (corresponding to faster kinetics) than SSS even though

eqπ is larger. The same will be true for layers on mica. For PPP both the α and β phase are

more stable than the LB monolayer. The driving force for β - formation is larger than for α -

formation, but the kinetics are faster for α , therefore we observe both domains with α - like

and domains with β - like structure. For SSS only the β - phase seems more stable. Also for

the longest triglyceride AAA, the monolayer is thermodynamically unstable at the air-mica

interface, but the crystallization kinetics are so slow that on the time scale of days, or even

weeks, they behaves as if they were stable.

4.6. Conclusions

In this study, we investigated the behavior of three triglycerides: PPP, SSS and AAA at the

air-water interface and on a solid substrate. Based on Langmuir and AFM experiments, we

established the relation between the molecular structure and the stability of the monolayers.

Our investigations lead to the following conclusions.

At the air-water interface all investigated triglycerides form monolayers of molecules

in trident conformation. These monolayers are kinetically stable at air-water interface at

surface pressure 0π π≤ . 0π is the surface pressure below which we did not observed any

changes in the Langmuir monolayer under isobaric cnditions. We know that 0 eqπ π≥ , where

eqπ is the thermodynamic equilibrium pressure, but the amount of overestimation, i.e. the

“nucleation gap” 0 eqπ π− , can not be deduced from our data. We found that 0 10mN/mπ =

for PPP, 0 15mN/mπ = for SSS and 0 20mN/mπ = for AAA. From dynamic adsorption

isotherms, obtained at a compression rate of 1 cm/min we find a condensation pressure

8mN/mcondπ ∼ for all three triglycerides. Since eqπ must be smaller for AAA and larger for

PPP we conclude that 10mN/meqπ ≤ for all three triglycerides. Thus we arrive at the

conclusion that eq condπ π≈ .

For eqπ π> the Langmuir monolayers are thermodynamically unstable at air-water

interface. Under isobaric conditions at 0π π> slow compression was found for PPP and SSS.

The isobaric compression rate is highest for PPP and almost zero for AAA.

79

LB-monolayers can be successfully transferred onto a mica surface. Using the AFM

imaging, the thickness of the trident monolayers can be measured. We demonstrated that the

apparent thickness depends strongly on the AFM scanning force and we showed that the

compressibility of the investigated triglycerides is the same. The thickness of the monolayers,

obtained by extrapolation to zero scanning force is 1.49 nm for PPP, 1.75 nm for SSS and 2.2

nm for AAA. These monolayer thicknesses correspond to tilt angles of the molecules of 46.4o,

49.2o and 59.0o respectively. We conclude that the tilt angle increases with increasing the

length of the alkyl chains, i.e. the longer chains in the monolayer are more perpendicular to

the substrate.

The LB-monolayers transferred immediately at surface pressure 0π π> for PPP and

SSS contain domains on top of the monolayer. The molecules in these domains adopt the

tuning fork conformation and pack similar as in the crystalline α and β crystal phase.

The LB-monolayers of PPP and SSS, which were transferred at 0π π≥ are

thermodynamically unstable in air. Small nuclei in tuning fork conformation form on top of

the monolayer. For SSS they are all in β -phase, for PPP, domains with α - like and β - like

structures coexist in the LB film. The density of the crystals increases in time. We conclude

that the different mobility of the molecules in the trident monolayer of SSS and PPP is the

main reason for the different rate of nucleation and growth. For AAA, the monolayer is

thermodynamically unstable at the air-mica interface, but the crystallization kinetics are so

slow that on the time scale of days they behaves as if they were stable.

We conclude that the stability and the kinetics of Langmuir-Blodgett monolayer depend

strongly on the length of the triglycerides alkyl chains and also on the surface pressure at

which the deposition took place. The trident monolayer is the less mobile and the crystal

phase is more stable the longer the alkyl chains are. The nucleation rate increases with

increasing π , due to the fact that the LB-monolayer is denser , and hence more unstable,

when the Langmuir layer is transferred at higher π .

80

References:

[1] Charalambous, G., Doxatakis, G., In Food Emulsifiers: Chemistry, Technology,

Functional Properties and Applications; Elsvier, Amsterdam, 1989

[2] Smith, R. and Berg, J. J. Colloid Interface Sci. 74 (1980)273-286

[3] Fuente, J.F. and Rodriguez Patino, J.M. Langmuir 10 (1994) 2317-2324

[4] Fuente, J.F. and Rodriguez Patino, J.M. Langmuir 11 (1995) 2090-2097

[5] Sanchez, C.C., Rodriguez Nino, M., Rodriguez Patino, J.M., Colloids and Surfaces B:

Biointerfaces 12 (1999)175-192

[6] Garti, N. and Sato, K., In Crystallization and polymorphism of fats and Fatty Acids;

Dekker, M. New York (USA) (1988)

[7] Ollivon, M., Triglycerides. In Manuel des Corps Gras. Ed.A.Karieskind, Lavoisier, Paris

(France) (1992) p. 469



[10] Hamilton, J.A., Biochem. 28 (1989) 2514-2520

[11] Claesson, P.M., Dedinaite, A., Bergenstahl, B., Campbell B. and Christenson, H.,

Langmuir 13 (1997)1682-1688


Technol. 103 (2001) 677-682

[13] Akita, C., Kawaguchi, T., Kaneko, F., Yamamuro, O., Akita, H., Ono, M. and Suzuki,

M., Journal of Crystal Growth 275 (2005) 2187-2193

[14] Roberts, G., Langmuir-Blodgett Films Plenum Press, New York (1990) p.21

[15] Hardy, N.J., Richardson, T.H. and Grunfeld, F., Colloids and Surfaces A:

Physicochemical and Engineering Aspects 284-285 (2006) 202-206


approach- PhD thesis (1980) University of Utrecht, The Netherlands

[17] Popovitz-Biro, R., Wang, J.L., Majewski, J., Shavit, E., Leiserowitz, L. and Lahav, M.,

J. Am. Chem. Soc. 116 (1994) 1179

81

CHAPTER 5

Phase behaviour in binary mixed Langmuir-Blodgett monolayers of Triglycerides

Abstract

The structure of binary mixed monolayers of triglycerides: tripalmitin (PPP), tristearin (SSS)

and triarachidin (AAA) at air-water interface are investigated with the Langmuir method. The

Langmuir-Blodgett (LB) layers obtained by deposition on mica were investigated by Atomic

Force Microscopy. Based on Langmuir and AFM results the relation between the phase

behavior of binary mixed TAGs and the difference in the chain length is established. Our

experiments show that TAGs mixtures form monolayers with molecules in trident

conformation at the air-water interface, like pure TAGs. The condensation area

and the condensation pressure

263 ÅcondA =

8 10 mN/mcondπ = − are found to be the same for all mixtures

and pure systems. The sharpness of the transition from “gas” to “condensed” phase in the

Aπ − isotherms decreases linearly with the average chain length for all systems. Using AFM

carefully the monolayers thicknesses for the mixtures were measured and compared to

those of the pure systems. We found that is linear dependent on the average chain length

of the TAGs molecules. We determined the relative film compressibility and found that it

is higher for mixed monolayers ( ) than for pure systems

( ). The AFM results show phase separation in the systems PPP-SSS and

PPP-AAA, which is not complete. The solubility of the shorter PPP molecules in the “long”

(SSS-rich and AAA-rich) phase is significant. For the mixture SSS-AAA, phase separation

was not observed. In that mixture the monolayer thickness varies linearly with

composition, supporting the conclusion that SSS and AAA mix almost ideally. In general the

0d

0d

K-10.08 0.01 nNK = ±

-10.07 0.01 nNK = ±

0d

83

main driving force for phase separation is the difference in the alkyl chain length. Indeed

PPP-AAA (length difference 4 C atoms) shows the most clear phase separation. The relatively

weak phase separation in PPP-SSS and the absence of phase separation in SSS-AAA shows

that the influence of chain length difference decreases with increasing average chain length.

In air the PPP-SSS and PPP-AAA mixtures monolayer are unstable and crystals with

α - like and β - like structure are formed on top of the monolayer as in pure PPP and SSS

systems.

84

5.1. Introduction

Because triglycerides (TAGs) are the main components in the natural fats, they have been

studied for many years. Most of the research focuses on investigating the melting and

crystallization properties of TAGs. An overview of the thermodynamic and kinetic aspects of

fat crystallization was published recently [1]. It is well known that TAGs may crystallize in

the α (hexagonal, less stable), 'β (orthorhombic), or β (triclinic, most stable) form; the

nomenclature scheme following Larsson [2] as reviewed in Hagemann [3], Hernqvist [4],

Wesdorp [5], Sato [6], and Ghotra [7]. Each of these polymorphic forms consists of layers in

which the molecules have a tuning fork conformation but the orientation of the tuning forks

within the layers, as well as the packing of the layers is different. This polymorphic behavior

of TAGs strongly determines the physical properties of the fats. The monoacid saturated

TAGs (the three acyl chains are identical) are the simplest in the TAGs family. Because of

the simplicity of their structure this group has been examined in more detail than other

groups. The three basic α , 'β and β polymorphic forms have been formed [3]. Generally,

the polymorphic behaviour of TAGs with an even carbon number n are well represented by

the behaviour of PPP (n = 16) and SSS (n = 18) [3, 8-13]. The crystallization and the phase

transformation properties of these two triglycerides were found to be very similar. They only

differ with respect to the rate at which these processes occur. Both TAGs exhibit a

preferential tendency for β - crystallization. The 'β - form can only be crystallized from the

isotropic melt within a narrow temperature range and shows the typical 'β - wide-angle

diffraction pattern [13].

Binary mixtures of TAGs show far more complex polymorphic behaviour as

compared to pure TAGs. For binary TAG mixtures, the primary factors determining phase

behaviour are differences between the TAGs in chain length, the degree of saturation and

position of the fatty acid moieties, and which polymorphs are involved. Different phase

behaviour is frequently observed for different polymorphs. For example in the mixture PPP-

SSS the triglycerides are completely miscible in the less stable phases (α and 'β ) but they

form a eutectic system in the stable β - form [14-16]. The same behaviour was observed by

Takeuchi et al. [17] for the mixture LLL-MMM, where the carbon numbers in the fatty acid

chains differ by 2. When the difference in the carbon chain length differs by 4 or 6 like in the

mixtures LLL-PPP and LLL-SSS respectively, the metastable phases (α and 'β ) turn out to

85

be immiscible. Eutectic and monotectic behaviour is observed in the β - form for the LLL-

PPP and LLL-SSS systems, respectively, with the α form of SSS co-existing with the β

form of LLL under certain conditions [17].

In monolayers at a hydrophilic-hydrophobic interface, e.g. water/oil, water/air or

mica/air triglyceride molecules adopt a trident conformation (all hydrocarbon chains pointing

into the same direction). In the trident conformation the hydrophilic glycerol group is in

contact with the water or the mica surface, and the hydrophobic chains point into the air or the

oil [18-22].

In previous work [Chapter 4] we investigated monolayers of tristearin (SSS, chain

length 18 C atoms), tripalmitin (PPP chain length 16 C atoms) and triarachidin (AAA chain

length 20 C atoms), at air-water interface (Langmuir film) and on solid surface like mica

(Langmuir- Blodgett film). We established the relation between their molecular structure and

their monolayer stability. We found that the trident monolayer is the less mobile and the

crystal phase is the more stable the longer the acyl chains are. Using AFM carefully the

thickness of the trident monolayers was measured. It is 1.49 nm for PPP, 1.75 nm for SSS and

2.2 nm for AAA, corresponding to tilt angles of the molecules of 46o, 49o and 59o

respectively.

The aim of the work, presented in this chapter is to understand the phase behavior of

binary mixed TAGs: PPP-SSS, PPP-AAA and SSS-AAA at air-water interface (Langmuir

film) and on solid surface like mica (Langmuir-Blodgett film). We measured the Aπ −

(spreading pressure π vs area per molecule A ) diagram of Langmuir films. Starting with a

Langmuir film at very small π , where the film is in a low-density “gas” phase, we

compressed the film, at a constant rate, to the desired pressure π (forced compression). The

Langmuir film was transferred to mica directly after forced compression ( ) and

investigated with AFM immediately.

0t =


5.2.1. Chemicals

Film material: In our experiments we used saturated monoacid triglycerides (their three acyl

chains are the same). Tripalmitin (1, 2, 3-Propanetriyl trihexadecanoate: PPP, chain length 16

86

C atoms), Tristearin (1, 2, 3, -trioctadecanoylglycerol: SSS, chain length 18 C atoms) and

Triarachidin (trieicosonoin: AAA, chain length 20 C atoms) were purchased from Larodan

(Sweden) with a stated purity of >99 mass %. Stock solutions of PPP, SSS and AAA with

concentration of 1 mM in distilled chloroform were prepared. The stock solutions were mixed

in ratios 1:1, 1:3 and 3:1.

Subphase: Distilled water was used as a subphase in our Langmuir system for all

experiments. The resistivity of the water was 15 MOhm cm.



Compression isotherms were measured on a commercial, fully automated Langmuir Blodgett

Trough (model: 311D, Nima Technology Ltd., England). The instrument was equipped with a

Teflon trough (283.0 cm2) and one Delrin barrier. The spreading pressure π was measured

with an accuracy of about 0.1 mN/m. The film material was initially spread on the water

subphase, dropping 30 µL of 1 mM stock solution dissolved in chloroform, using a 100 µL

Hamilton syringe. The conditions were chosen such that initially the average area A per

molecule is . We started (asymmetric) film compression 2 min after spreading. In

our system we used the forced compression operation mode, where the position of the barrier,

and hence the trough length ahead of the barrier, is given. Then the resulting spreading

pressure

2110 ÅA ∼

( )l t

( )tπ is registered. In this mode we chose barrier velocities of the order of 1 cm/min,

which according to the literature should be slow enough that the Langmuir film stays close to

thermodynamic equilibrium.


In order to obtain LB films, first a substrate was immersed perpendicularly in the aqueous

subphase. We started with a very small initial surface pressure ( 0π = mN/m), and

compressed the monolayer slowly (1 cm/min) to the final pressure. To obtain a LB film that is

characteristic for forced compression, the film was transferred immediately by vertical pulling

of the substrate through the air-water interface at a speed of 2 mm/min. During the transfer the

surface pressure was kept constant by appropriately moving the barrier. The transfer process

87

takes a few minutes. After deposition the LB-films were dried in air and kept in closed

containers until use. All experiments were done at 20 ± 1°C.


The samples were examined with AFM immediately after preparation. Imaging was done with

a Nanoscope (R) IIIa (Digital Instruments, Santa Barbara, CA) in contact mode with oxide-

sharpened silicon nitride tip (k = 0.06 N/m). The AFM was equipped with a J scanner

(176 x176 µm; z limit = 5.349 µm). All images were processed using procedures for

flattening in Nanoscope III software version 5.12r5 without any filtering. To check if the

monolayer is successfully transferred to the mica surface we measured at least five different

spots (each 150 µm 2) of every sample.


A

0

5

10

15

20

25

30

35

40

50 60 70 80 90

Area/molecule A (A2)

Sur

face

pre

ssur

e (m

N/m

)

PPPSSSPPP-SSS (1:1)

88

B

0

5

10

15

20

25

30

35

40

50 60 70 80 90Area/molecule A (A2)

Sur

face

pre

ssur

e (m

N/m

)

PPPAAAPPP-AAA (1:1)

C

0

5

10

15

20

25

30

35

40

50 60 70 80 90

Area/molecule A ( A2)

Sur

face

pre

ssur

e (m

N/m

)

SSSAAASSS-AAA (1:1)

Fig.1. Surface pressure vs area isotherms of tripalmitin (PPP), tristearin (SSS) and

triarachidin (AAA) and their mixtures at air-water interface, at 20oC, obtained by forced

compression at a rate of 1cm/min.

In the previous Chapter 4 we already discussed the shape of the typical Aπ −

isotherms of PPP, SSS and AAA, where two different regimes can be recognized for the three

triglycerides. The condensation area and condensation pressure condA condπ have been

described as values at which the transfer from “gaseous” to “condensed” phase occur. The

collapse pressure colπ is the surface pressure at which the monolayer collapses to form

multilayer structures. For the studied triglycerides it was in the range of 40 48 mN/mπ = −

and it increased in order: (AAA) (SSS) (PPP)col col colπ π π< < . With our LB instrument the

collapse pressure was difficult to reproduce because of details in its construction. For the

mixtures we measured similar

colπ

Aπ − isotherms as for the single components (Fig.1). The

89

measured Aπ − data for the mixtures showed that the pressure range, where the transition

from one regime to another takes place, was rather wide. For the mixtures the adsorption

isotherm was always between the isotherms of the single components. An exception was the

mixture SSS-AAA, for which the Aπ − isotherm sometimes almost coincided with the

isotherm of AAA (Fig.1C). In order to get reliable and unbiased estimations for and condA

condπ , we fitted the isotherms with:

( ) ( , )condA ch A A aπ ≈ − (1)

where , a , and h are fitting parameters. The function condA c

( )2 21( , )2

h x a x x a≡ − + (2)

is a hyperbola interpolating between for large negative x and for

large positive x . This function has no direct physical interpretation and was introduced for

practical purposes only, i.e. to arrive at an unambiguous definition and evaluation of

( ),h x a x≈ ( ), 0h x a ≈

/ 2cond caπ = and . Fitting a number of isotherms (15) that were obtained at compression

velocity 1cm/min we found and

condA

264 1 ÅcondA = ± 9 3 mN/mcondπ = ± for SSS-AAA;

and 263 1 ÅcondA = ± 10 3 mN/mcondπ = ± for PPP-SSS and and 263 3 ÅcondA = ±

11 2 mN/mcondπ = ± for PPP-AAA. Together with the corresponding data for the pure PPP,

SSS and AAA systems, these results are presented in Figure 2.

A Condensation area A cond

56

58

60

62

64

66

68

70

14 16 18 20 22

Average chain lenght (C atoms)

Aco

nd (Ǻ

2 )

90

B Condensation pressure πcond

2

4

6

8

10

12

14

14 16 18 20 22


Con

dens

atio

n pr

essu

re (m

N/m

)

C Fitting parameter a

0

2

4

6

8

14 16 18 20 22


a

Fig.2. Condensation area (A), condensation pressure condA condπ (B) and fitting parameter

(C) for triglycerides (▲) and their mixtures (■). X axis presents the number of the carbon

atoms in the triglyceride chains: PPP (16), SSS (18) and AAA (20). For the mixtures it was

calculated as follow: 17 = PPP-SSS (1:1), 18 = PPP-AAA (1:1) and 19 = SSS-AAA (1:1).

a

The fact that is around for all studied triglycerides and their mixtures is

consistent with a trident conformation of triglyceride molecules in a monolayer film at the air-

water interface. Indeed, the cross-sectional area per hydrocarbon chain for tristearin at 20

condA 263 Å

oC in

the α phase (the α phase has the most mobile acyl chains) is [23]. 219.7Å

The fact that condπ is almost the same for the investigated pure triglycerides and their

mixtures as well (8 1 ), is consistent with the idea that the packing properties of the

hydrocarbon chains is mainly determined by short range repulsive interactions. The effective

repulsion is quite independent of the chain length and compositions, which shows that mixing

of triglycerides does not change their packing properties drastically. The tendency of

0 mN/m−

condA

91

and condπ to increase slightly with increasing chain length reflects a slightly enhanced

repulsion of longer chains.

The value of the fitting parameter (Eq.1) describes the sharpness of the gas -

condensed transition and is found to depend strongly on the chain length ( the smaller , the

sharper is the transition). This is also seen in Fig.1 where the

a

a

Aπ − isotherm for PPP is

sharper than those for SSS and AAA. This observation can be understood if one realizes that

the shorter PPP molecules are stiffer than the longer SSS and AAA. The longer chains will

spread somewhat more in lateral direction. The isotherms in Fig.1 suggest that in a

moderately dense packed monolayer at the air-water interface the longer triglycerides interact

already at significantly larger intermolecular distances than the shorter ones. The fitting

parameter is rapidly increasing with increasing chain length. Apparently the presence of

PPP in a mixture reduces the hindering of the motion of the longer molecules and thus

sharpens the transition from the gas to the condensed phase.

a

In general the Aπ − isotherms of the mixtures interpolate linearly between the

isotherms of the pure components. E.g. the isotherm of the PPP-AAA mixture (average chain

length 18) is very similar to the isotherm of pure SSS (chain length 18). Thus from the Aπ −

isotherms alone one would be tempted to conclude that the triglycerides mix (almost) ideally.

In the remains of this chapter, we show that this conclusion is incorrect. We investigated

Langmuir – Blodgett monolayers of the mixtures with AFM. Our experimental results clearly

show non-ideal behaviour, and even phase separation.


To investigate the structure of the three mixtures we withdrew Langmuir monolayers

immediately after forced compression to 20 mN/mπ = .We chose this surface pressure

because is in the middle of the condensed region of the Aπ − isotherms. We know from the

Aπ − isotherms that it is well above the condensation pressure condπ , but still below the

collapse pressure colπ .

5.4.1. PPP-SSS structure

92

0 2.5

0

5.0

5.0

-5.0

µm

1.7 nm 1.6nm

DA

C

0 1.0-1.5

01.

5

0.21 nm 0.25 nm

µm

F

2.0

B

0 2.5

2.5

-2.5

0

µm

1.69 nm3.52 nm

E

5.0

0 2.5

0

5.0

5.0

-5.0

µm

1.7 nm 1.6nm

D

0 2.5

0

5.0

5.0

-5.0

µm

1.7 nm 1.6nm

DAA

CC

0 1.0-1.5

01.

5

0.21 nm 0.25 nm

µm

F

2.00 1.0-1.5

01.

5

0.21 nm 0.25 nm

µm

F

2.0

BBB

0 2.5

2.5

-2.5

0

µm

1.69 nm3.52 nm

E

5.00 2.5

2.5

-2.5

0

µm

1.69 nm3.52 nm

E

5.0

Fig.3. (A) AFM height image of PPP-SSS monolayer transferred immediately after forced

compression to π = 20 mN/m. The black square is a hole in the monolayer produced by

scanning at a high AFM force (F~30 nN). The monolayer is scanned at AFM force F~1 nN.

(B) another area of the same sample, where the onset of phase separation was observed. (C)

zoomed image of (B). The corresponding cross sections are given in (D, E and F). The scale

bar is 2 µm for (A and B) and 1 µm for (C). Height differences are given by the numbers at

the markers.

93

The AFM images of PPP-SSS (1:1) showed a homogeneous monolayer with thickness

1.6 ± 0.1 nm, i.e. somewhere between the measured thicknesses of PPP and SSS (Fig.3A, D).

The monolayer contains more holes than monolayers of the pure systems, which are almost

defect free. This is the first indication that due to the difference in the chain length of the two

components in the mixture, random packing of longer and shorter molecules is

thermodynamically not optimal. In some regions of the samples the onset of phase separation

was observed (Fig.3B, C). It was difficult to measure directly the height difference by the

AFM. We estimated a height difference of 0.2 ± 0.1 nm (Fig.3C, F). The carbon chain length

of PPP and SSS differs by 2 carbon atoms, which is ~ 0.5 nm ( sc = 0.254) [24]. The thickness

of the monolayers, which we obtained by extrapolation to zero scanning force, is = 1.49

nm

0d for PPP, = 1.75 nm for SSS and = 2.19 nm for AAA. These monolayer thicknesses

correspond to tilt angles of the molecules of 46

0d 0do, 49o and 59o respectively [Chapter 4]. The

measured height difference of 0.2 ± 0.1 nm in the mixture PPP-SSS (1:1) is slightly below the

expected height difference of ~ 0.3 nm between tilted PPP and SSS monolayers. A reasonable

explanation of the small height difference is that PPP and SSS are not completely separated.

The fact that most of the AFM images of PPP-SSS (1:1) showed a homogeneous monolayer

supports the idea that PPP and SSS have only a weak tendency to phase separate.

In Chapter 4 we demonstrated that the apparent thickness depends strongly on the

AFM scanning force. Even relatively small scanning forces may compress triglyceride

monolayer. We showed that the compressibility varies little between the investigated pure

triglycerides. To measure the real monolayer thickness of the mixture PPP-SSS (1:1) we

used the same procedure as in Chapter 4. By scanning with a relatively large force

0d

30 nNF ≈ we scratched a rectangular hole in the monolayer with the AFM tip. Then a

larger area, including the hole, was scanned with small forces 1 8 nNF = − (fig.4). The

height difference between the hole and the surrounding film gives an apparent thickness d for each strength of the scanning force . We investigated three different holes in one sample

(Fig.4).

′

F

94

Apparent monolayer thickness of PPP- SSS

0.8

1

1.2

1.4

1.6

1.8

2

0 2 4 6 8 10AFM force F (nN)

AFM

thic

knes

s d'

(nm

)

hole 1hole 2hole 3SSSPPP

Fig.4. Measured layer thickness d for PPP, SSS and PPP-SSS (1:1) as a function of applied

AFM force F at surface pressure

′

20 mN/mπ = .

Real monolayer thickness for PPP-SSS

1.4

1.5

1.6

1.7

1.8

0 50SSS in the mixture (%)

Laye

r thi

ckne

ss d

0 (n

m)

100

Fig.5. Real monolayer thickness for PPP, SSS and PPP-SSS (1:1) for three different holes

at surface pressure

0d

20 mN/mπ = . The values are found by extrapolation of the apparent

monolayer thickness in Fig.4 to AFM force F = 0 nN.

0d

'd

In Fig.4 we see that the dependence of d on the scanning force F for the mixtures of

triglycerides and for the pure phases is very similar. By definition the isothermal

compressibility of 3 dimensional materials is:

′

95

1T

T

VKV P

∂⎛ ⎞= − ⎜ ⎟∂⎝ ⎠ (3)

Analogously the isothermal film compressibility can be defined as

0

0

131

film TT

dKd P

∂⎛ ⎞= − ≈⎜ ⎟∂⎝ ⎠K (4)

where the last approximation is valid if the material properties of the film are the same as of

the bulk material. In our system we measure the AFM force . The pressure in this case

would be

F

/P F contact area= (5)

but unfortunately we cannot accurately estimate the contact area. For practical reasons we

define the quantity as K

0

0

1 dKd F

∂≡ −

∂ (6)

and, somewhat loosely, we shall refer to as film compressibility from now on. K

The film compressibility of the monolayer, given by the slope of curves, is

slightly higher for PPP-SSS (1:1) ( ) than for pure PPP and SSS

( ). The real thickness , corresponding to scanning force is

presented in Fig. 5. As shown in Fig.5 for PPP-SSS (1:1) we found two distinct results,

and . These values are close to

and

'( )d F

-10.08 0.01 nNK = ±

-10.07 0.01 nNK = ± 0d 0F =

0 1.50 0.02 nmd = ± 0 1.69 0.01 nmd = ±

0 (PPP) 1.49 0.02 nmd = ± 0 (SSS) 1.75 0.02 nmd = ± respectively, and we suppose that

they are the thickness of PPP-rich and SSS-rich areas in the monolayer respectively.

We conclude that in the mixture PPP-SSS (1:1) phase separation takes place, which is

not complete. The PPP-rich regions contain dissolved SSS molecules and SSS-rich regions

contain dissolved PPP molecules. As the dissolved molecules will influence the average

thickness it is now clear why the height difference of the domains in Fig.3, as well as the

difference in monolayer thickness does not correspond to the length of 2 carbon atoms but 0d

96

is slightly smaller. Note that we can not completely exclude the possibility that PPP and SSS

are well separated in very small domains, which cannot be detected by the AFM.

Like in our previous investigations for the single components [Chapter 3 and 4] we

found higher domains on top of the PPP-SSS monolayer. Most of them had a thickness of 3.5

± 0.1 nm (Fig.3 B and E). This corresponds to molecules of PPP or SSS in tuning fork

conformation. Similar domains were formed when a Langmuir monolayer of the single

component was transferred immediately at 20 mN/mπ = (3.3 ± 0.1 nm for PPP and 3.5 ± 0.1

nm for SSS) [Chapter 4]. The composition of the crystals on top of the PPP-SSS monolayer is

not clear. They could contain either PPP or SSS or both types of molecules.

5.4.2. SSS-AAA structure

A

0 2.5

5.0

-5.0

0

µm

1.93 nm 1.86 nm

B

5.0

AA

0 2.5

5.0

-5.0

0

µm

1.93 nm 1.86 nm

B

5.00 2.5

5.0

-5.0

0

µm

1.93 nm 1.86 nm

5.00 2.5

5.0

-5.0

0

µm

1.93 nm 1.86 nm

B

5.0

Fig.6. (A) AFM height image of SSS-AAA (1:1) monolayer transferred immediately after

forced compression to 20 mN/mπ = and the corresponding cross section in (B). The black

square is a hole in the monolayer produced by scanning at a high AFM force (F~30 nN). The

monolayer is scanned at AFM force F~1 nN. The scale bar is 2 µm and the vertical scale is

10 nm. Height differences are given by the numbers at the markers.

The AFM images of SSS-AAA (1:1) showed a homogeneous monolayer (Fig.6). To

measure the thickness of the monolayer we used the same procedure as described in Section

5.4.1. In order to get reliable and unbiased estimations for this procedure was repeated for

3 independent holes in one sample (Fig.7). For SSS-AAA (1:1) we found

by extrapolation of the apparent monolayer thickness d in Fig.7 to AFM force .

0d

0 1.95 0.02 nmd = ±

′ 0 nNF =

97

The film compressibility of the mixture monolayer was very close to

that of the pure monolayers ( ). We did not observe any crystals on top of

the monolayer. Contrary to PPP-AAA we saw no indications for phase separation. No

domains were observed. This suggests the absence of phase separation in PPP-AAA. To

investigate this further we also studied (1:3) and (3:1) mixtures. The same absence of domains

was observed for 1:3 and 3:1 mixtures (data not shown). The monolayer thickness

depended linear on composition (Fig.8). The film compressibility was the same for all SSS-

AAA mixtures. All these observations support that PPP-AAA forms (almost) ideal mixtures.

-10.08 0.01 nNK = ±

-10.07 0.01 nNK = ±

0d

Apparent monolayer thickness for SSS - AAA (1:1)

1

1.2

1.4

1.6

1.8

2

2.2

2.4

0 2 4 6 8 10AFM force F (nN)

AFM

thic

knes

s d'

(nm

)

hole1

hole 2

hole 3

SSS

AAA

Fig.7. Measured layer thickness ' for SSS, AAA and SSS-AAA (1:1) as a function of applied

AFM force at surface pressure

d

F 20 mN/mπ = .

Real monolayer thickness of SSS-AAA

1

1.2

1.4

1.6

1.8

2

2.2

2.4

0 25 50 75 100

AAA in the mixture (%)

Laye

r thi

ckne

ss d

0 (nm

)

98

Fig.8. Real monolayer thickness for SSS-AAA at surface pressure 0d 20 mN/mπ = as a

function of the mole fraction of AAA in the mixture. The values are found by extrapolation of

the apparent monolayer thickness to AFM force F = 0 nN. 'd

5.4.3. PPP-AAA structure

In some cases AFM images of LB-films of PPP-AAA (1:1) show areas where phase

separation is hardly visible (Fig.9A, B and C) as in the SSS-PPP (1:1) mixture. There are

other areas however, with very well separated domains (Fig.9D, E and F). To measure

monolayer thicknesses for different domains, we used the same procedure as described in

section 5.4.1. We measure the monolayer thickness in different areas independently. As

before we corrected for the compression of the AFM at low scanning force (Fig.10).

0 2.5

5.0

-5.0

0

µm

1.72 nm1.65 nm

C

5.0

BA

D E

0 2.5-5.0

5.0

0

µm

1.98 nm1.49 nm

F

5.0

0 2.5

5.0

-5.0

0

µm

1.72 nm1.65 nm

C

5.00 2.5

5.0

-5.0

0

µm

1.72 nm1.65 nm

C

5.0

BBAA

DD EE

0 2.5-5.0

5.0

0

µm

1.98 nm1.49 nm

F

5.00 2.5-5.0

5.0

0

µm

1.98 nm1.49 nm

5.00 2.5-5.0

5.0

0

µm

1.98 nm1.49 nm

F

5.0

Fig.9. (A and B) AFM height images of PPP-AAA (1:1) monolayer transferred immediately

after forced compression to 20 mN/mπ = with little indication of phase separation. The

corresponding cross section is given in (C). (D and E) present areas of the same sample,

99

where phase separation is clearly visible. The corresponding cross section is given in (F).

Black squares are holes in the monolayer produced by scanning at a high AFM force (F~30

nN). The monolayers are imaged at AFM force F~1nN. The scale bar is 2 µm and the vertical

scale is 10 nm for all images. Height differences are given by the numbers at the markers.

The thickness of the PPP-AAA (1:1) monolayer, at positions where the phase

separation is not obvious, is , i.e. between the monolayer thicknesses of

the single components. The film compressibility at such positions is

larger than for PPP and AAA separately ( ).

0 1.86 0.05 nmd = ±

-10.11 0.02 nNK = ±

-10.07 0.01 nNK = ±

Apparent monolayer thickness for PPP-AAA

0.5

1

1.5

2

2.5

0 2 4 6 8AFM Force F (nN)

AFM

thic

knes

s d'

(nm

)

AAA-rich domain

AAA-pure

PPP-rich domain

PPP pure

PPP-AAA (1:1)

Fig.10. Measured layer thickness for PPP, AAA from the monolayers of the pure

components and in the mixture PPP-AAA (1:1) as a function of applied AFM force F at

surface pressure

'd

20 mN/mπ = .

At positions with clear phase separation the two different domains had thicknesses

0 1.42 0.01 nmd = ± and (Fig.9E, F). These heights are close to the height

of PPP ( ) and AAA (

0 1.96 0.03 nmd = ±

0 1.49 0.02 nmd = ± 0 2.19 0.04 nmd = ± ) monolayers, respectively

[Chapter 4]. The thickness of the higher domains is somewhat less than the thickness of pure

AAA. We assume that in regions like those in Fig. 9 D and E the higher domains are AAA-

rich. The film compressibility of AAA-rich areas is higher than the one

of the pure AAA ( ). This is consistent with the hypothesis that the AAA-

rich domains contain some PPP molecules, which lower the packing density of AAA

-10.10 0.01 nNK = ±

-10.07 0.01 nNK = ±

100

molecules and thus make the monolayer more compressible. The lower domains are similar in

thickness and have the same film compressibility as pure PPP monolayers

( ).We assume that they are PPP-rich. The fact that in the phase separated

regions the AAA-rich domains occupy a significantly larger fraction of the surface area than

the PPP-rich domains (Fig.9) indicates that PPP is more soluble in AAA than AAA in PPP.

Moreover, the fact that the lower, PPP-rich, domains are similar in thickness to pure PPP

monolayers suggests that AAA is hardly soluble in PPP. Such a tendency is reasonable, since

it will be energetically more unfavourable to dissolve long molecules in a thin layer than

reverse. In some of the AFM images of PPP-AAA (1:1) monolayer we observed three

different levels (Fig.11). The highest and lowest levels corresponded to the thickness of

AAA-rich and PPP-rich layers respectively. The area at the middle level was very irregular,

showing the onset of phase separation as for PPP-SSS in Fig.3. Its thickness of about 1.65 nm

corresponds to the thickness of PPP-AAA monolayers before demixing, see Fig.9A-C.

-10.07 0.01 nNK = ±

A B

0 2.5

0

2.5

-2.5

µm

1.65 nm 1.48 nm1.92 nm

mica level

5.0

AAA B

0 2.5

0

2.5

-2.5

µm

1.65 nm 1.48 nm1.92 nm

mica level

B

5.00 2.5

0

2.5

-2.5

µm

1.65 nm 1.48 nm1.92 nm

mica level

5.00 2.5

0

2.5

-2.5

µm

1.65 nm 1.48 nm1.92 nm

mica level

5.0

Fig.11. (A) AFM height image of PPP-AAA (1:1) monolayer transferred immediately after

forced compression to 20 mN/mπ = . The corresponding cross section in (B) presents three

different domains in the mixture. The black square is a hole in the monolayer produced by

scanning at a high AFM force (F~30 nN). The monolayer is scanned at AFM force F ~1nN.

The scale bar is 2 µm. Height differences are given by the numbers at the markers.

On top of the PPP-AAA (1:1) monolayer we observed crystals with thickness vary

from 3.5 to 4.7 nm (Fig.12). These values coincide with the measured values for PPP crystals

on top of monolayer in α and β - like phases. We also know that PPP crystallizes faster from

101

a monolayer than the longer AAA [Chapter 4]. In view of these facts we suppose that the

crystals, formed on top of PPP-AAA monolayer, are (almost) pure PPP. This conclusion

however, can not be drawn with certainty from the present AFM observation alone.

0 2.50

-5.0

5.0

µm

4.75 nm 3.5 nm

PPP level

B

5.0

A

0 2.50

-5.0

5.0

µm

4.75 nm 3.5 nm

PPP level

B

5.00 2.50

-5.0

5.0

µm

4.75 nm 3.5 nm

PPP level

5.00 2.50

-5.0

5.0

µm

4.75 nm 3.5 nm

PPP level

B

5.0

AAA

Fig.12. AFM height image of PPP-AAA (1:1) monolayer transferred immediately after forced

compression to 20 mN/mπ = (A) and the corresponding cross section in (B). The image

presents crystals on top of the monolayer. Their height is measured from the PPP level. The

scale bar is 2 µm. Height differences are given by the numbers at the markers.

5.5. Discussion

It is known from the literature that the phase behaviour in binary mixed TAGs in 3D crystal

structure strongly depends on the difference between the TAGs chain length [14-17]. When

the carbon numbers in the alkyl chains differ by 2 the mixtures are complete miscible in the

less stable forms (α and 'β ) and they form a eutectic system for the β - form. When the

difference in the carbon chain length is 4 or 6 also the metastable phases (α and 'β ) are

immiscible. Eutectic and monotectic behaviour is observed in the β - form for the LLL-PPP

and LLL-SSS systems, respectively. The α - form of SSS may co-exist with the β - form of

LLL under certain conditions [17].

In our 2D systems of binary mixed TAGs we have found a similar dependence on the

difference between the TAGs chain length. In order to compare quantitatively the

condensation area , the condensation pressure condA condπ and the isotherm sharpness , a

102

Langmuir isotherms were fitted for all mixtures and single components. At an air-water

interface we expect the binary mixed TAGs to form a trident monolayer like the single

components. This is confirmed with the Langmuir adsorption isotherms, which show that

for all investigated mixtures. The condensed pressure 63 2 ÅcondA 2= ± condπ also turned out

to be very similar for all systems (see Section 5.3.).

The value of the fitting parameter , which characterizes the sharpness of the

transition from “gas” to “condensed” phase in the

a

Aπ − isotherm, varied considerably for

different systems. In the mixtures was larger for SSS-AAA, than for PPP-SSS and PPP-

AAA. In the single components was smaller for PPP, than for SSS and AAA. We found a

linear increase of with average chain length for all systems (Fig.2C).

a

a

a

Based on our results we conclude that the smaller molecules increase the mobility in

the film and make the transition from one phase to another sharper. This was clear for the

single component films and for all mixtures. Judging on Langmuir data alone, one could be

tempted to conclude that PPP, SSS and AAA are well miscible in the monolayer regime. Our

AFM analysis however, shows that this conclusion would be incorrect. AFM thus is shown to

be essential for a sound interpretation of monolayer data.

From the AFM images of LB-films of PPP-SSS, SSS-AAA and PPP-AAA, withdrawn

at 20 mN/mπ = it is seen that the mica substrate is covered by a monolayer. Apparently the

Langmuir monolayer can be successfully transferred from the water surface in the Langmuir

trough to a mica surface to form a Langmuir-Blodgett film there.

The AFM images of LB-monolayers transferred immediately after forced compression

at 20 mN/mπ = for PPP-SSS (1:1) showed a monolayer, which in some areas was quite

homogeneous. In other areas of the same sample the onset of phase separation was observed.

In the areas with the onset of phase separation two different layer thicknesses were found. The

estimated height difference (0.2 ± 0.1nm) was somewhat smaller than the length of a two

carbon atom chain at a tilt angle of about 50o (0.3 nm) and considerably smaller than the

length of a perpendicular two carbon atom chain (~ 0.5 nm). The thickness of the monolayer

, measured from the homogeneous areas of the sample varied from 1.5 to 1.7 nm (Fig.5). 0d

We concluded that in the mixture PPP-SSS (1:1) phase separation takes place, but it is

not complete, i.e. the difference in composition of the PPP-rich and SSS-rich phases is small.

The alternative hypothesis, that the phase separation is complete, but PPP and SSS form very

103

small domains, which cannot be detected by the AFM, seems unlikely but cannot be excluded

with certainty.

The other mixture SSS-AAA, where the difference in the carbon chain length is also

two atoms behaved different from PPP-SSS (1:1). All AFM images of SSS-AAA (1:1, 1:3

and 3:1), transferred immediately after forced compression to surface pressure 20 mN/mπ = ,

showed homogeneous monolayers. We never observed the onset of phase separation. The

monolayer thickness varied linearly with composition between the monolayer thicknesses

of the single components (Fig.8). Based on these observations we concluded that SSS and

AAA mix (almost) ideally. We suppose that, due to stronger interactions between longer alkyl

chains, the sensitivity for differences in the chain length decreases with increasing alkyl chain

length.

0d

In the mixture PPP-AAA (1:1) (4 carbon atoms difference) phase separation was

observed. From the AFM images it is seen that there were areas, where only the onset of

phase separation was clearly visible and areas, where it was well defined (Fig.9). The

measured height difference between the higher and the lower domains was 0.6 ± 0.1 nm. This

difference is close to the difference in monolayer thickness of the single components PPP and

AAA with and respectively. One could assume that the highest

domains in the mixture correspond to AAA and the lowest to PPP. Precise measurement of

the thickness of the different domains in the mixture as a function of the AFM force showed

that the thickness of the higher domains is somewhat smaller than the thickness of pure AAA

monolayers. The film compressibility of the higher domains is slightly larger than for pure

AAA monolayers. The thickness of the lower domains was close to the thickness of a PPP

monolayer (Fig.10). The area occupied by higher domains in PPP-AAA (1:1) monolayer was

more than 50%. Based on these results we concluded that the higher domains in the mixture

are not “pure” AAA but are rich in AAA, with a significant fraction of dissolved PPP

molecules. Probably the lower domains are PPP with little dissolved AAA. Some AFM

images of PPP-AAA (1:1) (Fig.11) showed even other domains with thicknesses between the

highest and lowest measured thickness. These regions may correspond to PPP-AAA systems

that are in initial state of phase separation.

0 1.49 nmd = 0 2.19 nmd =

Our results for the film compressibility (Fig.13) showed that the monolayers of the

mixtures have higher compressibility than those of the pure components. We suppose that in

104

the pure systems due to the identity in the chain length the molecules pack better and thus

make the monolayer less compressible.

Film compressibility K

0

0.02

0.04

0.06

0.08

0.1

0.12

0.14

15 16 17 18 19 20 21

Average chain length (C atoms)

Film

com

pres

sibi

lity

K (n

N-1

)

Fig.13. Film compressibility as a function of the chain length for the pure components (■)

and their mixtures (○).

K

From Chapter 4 we know that at surface pressure 20 mN/mπ = PPP and SSS form

crystals in fork conformation on top of the monolayer. For SSS they are all in β - phase, for

PPP, crystals with α -like and β -like structures coexist in the LB-film. We also know that

the rate of crystallization on top of the monolayer is highest for PPP, lower for SSS and

almost zero for AAA.

The AFM images of LB-films, withdrawn at surface pressure 20 mN/mπ = of PPP-

SSS and PPP-AAA mixtures often showed domains on top of the monolayer as well. On SSS-

AAA, like on AAA, such domains were not observed. The thickness of the domains in PPP-

SSS and PPP-AAA vary from 3.5 to 4.7 nm (Fig.3 B, E). For PPP-AAA the thickness of the

crystals was measured from the PPP level (Fig.12). These thicknesses correspond to those of

PPP and SSS crystals in fork conformation in α -like and β -like structures, found in the LB-

films of the pure components. From these data we cannot deduce the composition of the

crystals that are formed in the second layer.

105

5.6. Conclusions

In this study we have obtained AFM images that reveal the structure of binary mixed TAGs

films (PPP-SSS, PPP-AAA and SSS-AAA), formed at air-water interfaces. Based on

Langmuir and AFM experiments we established the relation between their phase behavior and

the chain length of the two components. We compared the results with those for the pure

components. Our investigations lead to the following conclusions.

Compressing an extended binary mixed TAGs film at the air-water interface slowly,

starting at very low surface pressure, monolayers of TAGs molecules in trident conformation

are formed. We obtained reproducible Aπ − isotherms for all mixtures. After comparing with

the Aπ − isotherms of the single components we find that the condensation area

and the condensation pressure263 ÅcondA = 8 10 mN/mcondπ = − are the same for all

investigated systems. We find that the value of the fitting parameter , which characterizes

the sharpness of the transition from “gas” to “condensed” phase in the

a

Aπ − isotherms

increases linearly with the average chain length for all systems. For the mixtures it is larger

for SSS-AAA than for PPP-SSS and PPP-AAA. In the single components a is smaller for

PPP, than for SSS and AAA. We conclude that the smaller molecules increase the mobility in

the film and make the transition from one phase to another sharper. From the Langmuir

results we see that the thermodynamics (i.e. Aπ − isotherms) is insufficient to determine the

monolayer phase behaviour. For better understanding of the phase behavior of TAGs we used

our AFM results for mixtures and single components. AFM is shown to be essential for a

correct interpretation of monolayer data.

We obtained the AFM monolayer thickness for the pure systems and mixtures and

we find that it is linear dependent on the average chain length of the TAGs molecules. We

estimated the film compressibility and found that mixed monolayers are slightly easier to

compress ( ) than pure monolayers ( ).

0d

K-10.08 0.01 nNK = ± -10.07 0.01 nNK = ±

Based on the AFM images of LB-monolayers of PPP-SSS and PPP-AAA we conclude

that in these systems incomplete phase separation takes place. The solubility of the shorter

PPP molecules in the “long” (SSS-rich and AAA-rich) phase is significant.

For the mixture SSS-AAA, where the difference in the chain length is two carbon

atoms as for PPP-SSS we did not observe phase separation. The linear dependence of the

monolayer thickness of the composition supports the conclusion that SSS and AAA 0( )d x

106

mix almost ideally. We conclude that, due to the stronger interaction between the longer alkyl

chains, the sensibility for differences in the chain length decreases.

In air the mixtures have the same instability and form crystals on top of the monolayer

as the pure PPP and SSS systems. The thickness of the crystals on top of PPP-SSS and PPP-

AAA corresponds to PPP and SSS crystals in α - like and β - like structures, as in the LB-

films of pure PPP and SSS layers. From our data we cannot deduce the composition of the

crystals that are formed in the second layer. In the mixture SSS-AAA, like on pure AAA,

such crystals were not observed.

References:

[1] Himawan, C., Starov, V.M., Stapley A.G.F., Advances in Colloid and Interface Science

122 (2006) 3-33

[2] Larsson, K., Acta Chem Scand 20 (1966) 2255

[3] Hagemann, J.W. In: Garti, N. and Sato, K., Editors, Crystallization and polymorphism of

fats and fatty acids, Marcel Dekker, New York (1988), p. 9.

[4] Hernqvist L., In: Garti, N. and Sato, K., Editors, Crystallization and polymorhism of fats

and fatty acids, Marcel Dekker, New York (1988), p. 97.

[5] Wesdorp LH, Liquid-Multiple Solid Phase Equilibria in Fats, PhD dissertation, Delft

University of Technology; (1990)

[6] Sato, K., Fett-Lipid 101 (1999) 467

[7] Ghotra, B.S., Dyal, S.D. and Narine, S.S., Food Res Int 35 (2002) 1015

[8] Sato, K. and Kuroda, T., J Am Oil Chem Soc 64 (1987) 124

[9] Kellens, M., Meeussen, W., Riekel, C. and Reynaers, H., Chem Phys Lipids 52 (1990), 79-

98

[10] Kellens, M., Meeussen, W., Gehrke, R. and Reynaers, H., Chem Phys Lipids 58 (1991),

131-144

[11] Kellens, M., Meeussen ,W. and Reynaers, H., J Am Oil Chem Soc 69 (1992) 906.

[12] Desmedt, A., Culot, C., Deroanne, C., Durant, F. and Gibon, V., J Am Oil Chem Soc 67

(1990). 653

[13] Kellens, M., Meeussen, W. and Reynaers, H., Chem Phys Lipids 55 (1990) 163-178

[14] Lutton, E.S., J Am Chem Soc 77 (1955) 2646

107

[15] Kellens, M., Meeussen, W., Hammersley, A. and Reynaers, H., Chem Phys Lipids 58

(1991) 145-158

[16] MacNaughtan, W., Farhat, I.A., Himawan, C., Starov V.M. and Stapley, A.G.F., J Am

Chem Soc 83 (2006) 1-9

[17] Takeuchi, M., Ueno S. and Sato, K., Cryst Growth Des 3 (2003). 369-374



[20] Hamilton, J.A., Biochem. 28 (1989), 2514-2520

[21] Claesson, P.M., Dedinaite, A., Bergenstahl, B., Campbell B. and Christenson, H.,

Langmuir 13 (1997) 1682-1688


Technol. 103 (2001) 677-682

[23] Akita, C., Kawaguchi, T., Kaneko, F., Yamamuro, O., Akita, H., Ono, M. and Suzuki,

M., Journal of Crystal Growth 275 (2005)2187-2193


approach- PhD thesis (1980) University of Utrecht, The Netherlands

108

Summary

Amphiphiles are molecules, which have a hydrophilic head group and a hydrophobic tail.

They form monolayers at air-water interfaces (Langmuir monolayer). These monolayers can

be transferred on solid surfaces by the Langmuir - Blodgett technique (LB layers) and

investigated further by Atomic Force Microscopy. In this thesis we used these techniques to

study the phase behaviour of organic molecules at air-water and air-solid interfaces.

We started our investigation with the simplest amphiphilic molecules - fatty alcohols

(CnH2n+1OH, with even n = 16-24). Their behaviour can be used as a starting point for

understanding other amphiphiles with more complex structure. In Chapter 2 we discuss the

structure of binary mixed LB monolayers of fatty alcohols. The dependence of phase

separation on the difference between the chain lengths of the two components and the surface

pressure is described. Based on our results we conclude that phase separation occurs in

compressed films, if the chain length of the two components differs at least six carbon atoms.

A strong dependence of the domain shape on the surface pressure is observed. At high surface

pressure, 20 35 mN/mπ = − , the domains have tetragonal shapes. This is explained by the fact

that at higher pressures crystalline packing of molecules is favored as compared to disordered

or liquid like packing. The excess Gibbs energy ∆Gex vs. surface pressure and mole fraction is

estimated from Aπ − isotherms. In line with thermodynamic expectation, the tendency of

phase separation increases with increasing ∆Gex. The result that we observe phase separation

already in the range where our thermodynamic measurements indicated is

surprising, since the equilibrium thermodynamics predict phase separation only if

. The stability against phase separation of monolayers of fatty alcohols in non -

equilibrium isobaric conditions apparently is smaller than in equilibrium.

0.1exG R∆ ≅ T

T1exG R∆ ≥

Another interesting research in this area would be to image binary mixed monolayers

of odd fatty alcohols and investigate the conditions (the difference between their chain

lengths) for phase separation.

In the next two Chapters, 3 and 4 we describe the structure and stability of triglyceride

monolayers at air-water and air-solid (mica) interfaces. The triglycerides are typical examples

for materials, whose surface structure is different from the bulk structure. For such materials

different macroscopic properties are expected for surfaces and films as for the bulk. They

adopt a chair or tuning fork conformation in crystals and in bulk solutions, but at air-water

109

interface they rearrange in trident conformation (all hydrocarbon chains pointing in the same

direction). In the trident conformation they behave as amphiphilic molecules at the air-water

interface, though in general they are lipophilic. The idea to study these molecules in 2D

system was based on the previous study done on triglycerides at air-water interface by Bursh

and Larsson (1968). They first proposed the trident conformation for the triglyceride

molecules at an air-water interface. They found that if such a monolayer is compressed

beyond the collapse pressure, some of the molecules leave the monolayer to form new

molecular layers. Bursh and Larsson proposed a trident conformation for the first triglyceride

monolayer and a tuning fork conformation with a packing similar to that in the crystalline

state in the next layers.

By definition a Langmuir monolayer at a given spreading pressure π is

thermodynamically stable if under isobaric conditions at air-water interface it does not change

its structure, i.e. the area of the film remains constant. The pressure at which this happens is

called equilibrium spreading pressure eqπ . At spreading pressures eqπ π> one would expect

that the film area decreases, resulting in the formation of a new structure. The new structure,

depending of the film material, could be e.g. micelles in the subphase or multilayers on top of

the monolayer. Such structures can be formed at a certain pressure if the monolayer is

compressed at a constant rate. This pressure is called collapse pressure. The only way to

determine the thermodynamic stability of the monolayer is to investigate it under isobaric

conditions at spreading pressures colπ π< .

It is well known that some Langmuir monolayers are unstable at air-water interface at

surface pressures below the collapse pressure. Indeed, one of the surprising results, described

in Chapters 3 and 4, is that triglycerides turned out to be thermodynamic instable at air-water

interface at surface pressures far below the collapse pressure. Under isobaric conditions a

molecular rearrangement process takes place which effectively thickens the film. Using

Atomic Force Microscopy for triglycerides we have shown that this process is growth of 3D

crystals of triglycerides on top of the monolayer for surface pressures eqπ π> . In Chapter 3

we present a new crystal growth model to quantitatively describe this process.

In Chapter 4 this crystal growth model for tristearin is also applied two more

triglycerides – tripalmitin (PPP) and triarachidin (AAA). It was found that the three

investigated triglycerides behave similar. We investigated the influence of the chain length of

triglyceride molecules on the stability of their films on water and mica surfaces. The trident

110

Langmuir – Blodgett monolayers were the less mobile and the crystal phase was the more

stable, the longer the alkyl chains were. The nucleation rate increased with increasing surface

pressure π .The monolayer was compressed and transferred at the sameπ .

In Chapter 5 we describe the phase behaviour of binary mixed LB - monolayers of

triglycerides. We discuss the relation between phase separation and chain length. Incomplete

phase separation was observed for systems with two or more carbon atoms difference of the

chain length of the two components. The solubility of the shorter PPP molecules in the “long”

(SSS-rich and AAA-rich) phase was significant. An interesting result is that we did not

observe phase separation in the mixture SSS-AAA, although the difference in the chain length

is two carbon atoms as for the PPP-SSS mixture. The linear dependence of the monolayer

thickness of the composition supports the conclusion that SSS and AAA mix almost

ideally. The conclusion was that, due to the stronger interaction between the longer alkyl

chains, the sensibility for differences in the chain length decreases.

0( )d x

The study in this thesis illustrates that the Langmuir – Blodgett technique and Atomic

force microscopy are useful tools in the study of phase behavior of organic molecules on

different interfaces. The results could be used as a template for investigation of the phase

behaviour of other kinds of triglycerides (mixed-acid saturated/unsaturated) and their

mixtures in 2D dimension.

I would like to finish this work with a Japanese proverb:

“Beauty is only one layer.”

111

Samenvatting

Amphiphielen zijn moleculen met een hydrofiele kop en een hydrofobische staart. Op het

grensvlak tussen lucht en water vormen zij zogenaamde Langmuir monolagen. Deze

monolagen kunnen door middel van de Langmuir-Blodgett (LB) techniek worden

overgebracht op vaste oppervlakken en verder worden onderzocht met atomic force

microscopy (AFM). In dit proefschrift wordt het fase gedrag van organische moleculen aan

het lucht-water en lucht-vaste stof grensvlak onderzocht met behulp van deze technieken.

Dit onderzoek startte met de eenvoudigste amphifilische moleculen - vetachtige

alcoholen (CnH2n+1OH, met even n = 16-24). Hun gedrag kan gebruikt worden als startpunt

voor het begrijpen van het gedrag van andere amphifilische moleculen met complexere

structuur.

In Hoofdstuk 2 wordt de structuur van binaire vermengde LB monolagen van vetachtige

alcoholen besproken. De afhankelijkheid van fasescheiding van het verschil in ketenlengte

van de twee componenten en de oppervlaktedruk is onderzocht. Uit deze resultaten

concluderen we dat fasescheiding optreedt in samengedrukte films indien de ketenlengte van

de twee componenten tenminste zes koolstof atomen verschilt. Een sterke afhankelijkheid van

de vorm van de verschillende domeinen op de oppervlaktedruk is waargenomen. Bij een hoge

oppervlaktedruk, 20 35 mN/mπ = − , hebben de domeinen een tetragonale vorm. Dit wordt

verklaard door het feit dat bij een hogere druk een kristallijne pakking van de moleculen

gunstiger is dat een vloeistofachtige of wanordelijke pakking. De exces Gibbs energie ∆Gex

uitgezet tegen de oppervlaktedruk en molfractie is met behulp van Aπ − isothermen geschat.

In overeenstemming met de verwachting volgens de thermodynamica neemt de neiging tot

fasescheiding toe met toenemende ∆Gex. Het resultaat dat wij fasescheiding waarnemen waar

thermodynamische metingen aangeven dat is nogal verrassend, omdat er

volgens de evenwichts thermodynamica alleen fasescheiding optreedt als . De

stabiliteit ten opzichte van fasescheiding van monolagen van vetachtige alcoholen in niet-

evenwicht isobare omstandigheden is kleiner dan in evenwichtsomstandigheden. Het zou

interessant zijn om te onderzoeken of de condities voor fasescheiding (verschil in ketenlengte)

hetzelfde zijn voor vetachtige alcoholen met een oneven ketenlengte.

0.1exG∆ ≅ RT

T1exG R∆ ≥

113

In de volgende twee hoofdstukken, 3 en 4, i de structuur en stabiliteit van triglyceride

monolagen aan lucht-water en lucht-vaste stof (mica) grensvlakken onderzocht. Triglyceriden

zijn typische voorbeelden voor materialen met een oppervlaktestructuur die verschillend is

van de bulkstructuur. Voor zulke materialen worden verschillende macroscopisch

eigenschappen voor oppervlakten en films ten opzichte van de bulk verwacht. Deze

moleculen nemen een stoel of stemvork conformatie aan in kristallen en in oplossingen, maar

aan het lucht-water grensvlak herschikken deze moleculen zich in een drietand conformatie

(alle koolwaterstof ketens wijzen in dezelfde richting). In de drietand conformatie gedragen

zij zich als amphifhiele moleculen aan het lucht-water grensvlak, hoewel deze moleculen in

het algemeen lipofiel zijn. Het idee om deze moleculen in een 2D systeem te onderzoeken is

gebaseerd op een eerdere studie van triglyceriden aan het lucht-water grensvlak door Bursh en

Larsson (1968). Zij stelden de drietand conformatie voor de triglyceride moleculen aan een

lucht-water grensvlak voor en vonden dat indien zo'n monolaag voorbij de zogenaamde

'collapse pressure' wordt samengedrukt, sommige moleculen de monolaag verlaten om nieuwe

lagen te vormen. Bursh en Larsson stelden een drietand conformatie voor voor de eerste

triglyceride monolaag en een stemvork conformatie met een gelijkaardige structuur als in de

kristallijne vorm voor de volgende lagen.

Per definitie is een Langmuir monolaag voor een gegeven oppervlaktedruk π

thermodynamisch stabiel indien onder isobare condities aan het lucht-water grensvlak de

structuur niet verandert, d.w.z. de oppervlakte van de film is constant. De druk waarvoor dit

geldt wordt de evenwicht oppervlaktedruk eqπ genoemd. Bij een lagere druk verwacht men

dat de oppervlakte van de film afneemt, resulterend in een nieuwe structuur. De nieuwe

structuur (die afhangt van het materiaal), zou b.v. micellen in de subphase of multilagen

boven op de monolaag kunnen zijn. Zulke structuren kunnen bij een zekere druk gevormd

worden indien de monolaag bij een constante snelheid wordt samengedrukt. Deze druk wordt

de 'collapse pressure' genoemd. De enige manier om de thermodynamisch stabiliteit te

bepalen is door de monolaag te onderzoeken bij isobare condities voor een oppervlaktedruk

kleiner dan de ‘collapse pressure’, colπ π< .

Het is bekend dat sommige Langmuir monolagen instabiel zijn aan het water-lucht

grensvlak voor oppervlaktedrukken die kleiner zijn dan de ‘collapse pressure’. Een van de

verrassende resultaten (beschreven in hoofdstukken 3 en 4) is dat triglyceriden

thermodynamisch instabiel zijn aan het lucht-water grensvlak voor oppervlaktedrukken die

114

veel kleiner zijn dan de ‘collapse pressure’. Bij isobare condities vindt een moleculair

hershikkingsproces plaats dat resulteert in een dikkere film. Met behulp van AFM metingen

aan triglyceriden hebben we laten zien dat dit proces de groei van 3D kristallen bovenop de

monolaag voorstelt voor oppervlaktedrukken eqπ π> . In hoofstuk 3 wordt een nieuw model

geïntroduceerd voor de kwantitatieve voorspelling van deze kristalgroei.

In Hoofdstuk 4 is dit groeimodel voor tristearin (SSS) ook toegepast op twee andere

triglycerides: tripalmitin (PPP) en triarachidin (AAA). De drie onderzochte triglycerides

vertoonden een gelijkaardig gedrag. De invloed van de ketenlengte van triglyceride moleculen

op de stabiliteit van hun films op water en mica oppervlakten is onderzocht. Voor langere

ketenlengtes waren de drietand Langmuir-Blodgett monolagen het minste mobiel en de

kristalfase was de meest stabiele. De nucleatiesnelheid nam toe met toenemende

oppervlaktedruk. The monolaag werd samengedrukt en overgebracht bij dezelfde druk π .

In Hoofdstuk 5 wordt het fase gedrag van binaire LB monolagen van triglycerides

onderzocht, waaronder de relatie tussen fasescheiding en ketenlengte. Onvolledige

fasescheiding werd waargenomen voor moleculen met twee of meer koolstof atomen verschil.

De oplosbaarheid van de kortere PPP moleculen in de fase met de 'lange' moleculen (rijk aan

SSS en AAA) was aanzienlijk. Een interessant resultaat is dat wij geen fasescheiding in het

mengsel SSS-AAA hebben waargenomen, hoewel het verschil in de ketenlengte twee koolstof

atomen is (net zoals voor het PPP-SSS mengsel). De lineaire afhankelijkheid van de dikte van

de monolaag van het mengsel onderbouwt de conclusie dat SSS en AAA een vrijwel ideaal

mengsel vormen. De conclusie was dat, tengevolge van de sterkere wisselwerking tussen de

langere alkyl ketens, de gevoeligheid voor verschillen in de ketenlengte afnam.

Dit proefschrift illustreert dat de Langmuir-Blodgett en AFM technieken nuttig zijn

voor het bestuderen van het fasegedrag van organische moleculen aan verschillende

grensvlakken. De resultaten zouden als een leidraad voor onderzoek van het fasegedrag van

andere soorten van triglyceriden (vermengde-zuur verzadigd/onverzadigd) en hun mengsels in

2D gebruikt kunnen worden.

Graag beëindig ik dit proefschrift met een Japans gezegde:

'Schoonheid is een enkele laag'.

115

List of publications

• A.N. Zdravkova, J.P.J.M. van der Eerden and M.M.E. Snel

Phase behaviour in supported mixed monolayers of alkanols, investigated by AFM

Journal of Crystal Growth, 275 (1-2) (2005) 1029-1033

• Rick van Beek, Leonardus W. Jenneskens, Aneliya N. Zdravkova, Jan P. J. M. van der

Eerden, Cornelis A. van Walree.

Polythiophenes Containing Oligo (oxyethylene) Side Chains as a Thin Film on a ZnSe Single

Crystal Surface, Macromolecular Chemistry and Physics, 206 (10)(2005)1006-1014

• Reza Dabirian, Aneliya N. Zdravkova, Peter Liljeroth, Cornelis A. van Walree, and

Leonardus W.Jenneskens

Mixed Self-Assembled Monolayers of Semirigid Tetrahydro-4H-thiopyran End-Capped

Oligo(cyclohexylidenes), Langmuir, 21 (23) (2005)10497-10503

• Aneliya N. Zdravkova and J.P.J.M. van der Eerden

Structure and dynamics of Langmuir – Blodgett Tristearin films: Atomic Force Microscopy

and theoretical analysis, Journal of Crystal Growth, 293(2) (2006) 528-540


Structure and stability of Triglyceride monolayers on water and mica surfaces,

submitted to Crystal Growth & Design 10/2006


Phase behaviour in binary mixed Langmuir-Blodgett monolayers of Triglycerides

submitted to Crystal Growth 01/2007

117

Acknowledgements

If someone wants to make a dream come true they need two people, one to give him the

opportunity to develop and prove themselves and another to tell them “You can do it!”

This is why first of all I would like to express my sincere gratitude to my promoter

and supervisor Prof. Dr. Jan van der Eerden for offering me the opportunity and privilege to

work in his group. Jan, thank you very much for your guidance and for everything you did to

make me a scientist. You showed me the beauty of the science and I learned so much from

you. Thank you for being understanding, helpful and patient with me.

The second person I would like to thank is my best friend Katya Ivanova, who told me

“You can do it!” and showed me the way to do it. Katya, thank you for your non - stop

support, in whatever the decisions I have made.

Next I would like to thank to my colleague and friend Thijs Vlugt who was

encouraging me all these four years. Thijs, thank you very much for your help, professional

and personal advices. You made my life in Holland much easier.

My deepest acknowledgements go now to my family, my mother Pavlina, my father

Nikola, my sister Mariana and my niece Niya, whose endless support and love encouraged me

most.

Майко, от цялото си сърце искам да ти благодаря за жертвите и лишенията,

които трябваше да понесеш, за да мога да продължа образованието си и да го завърша с

докторска степен. Мери и Ния, благодаря ви за любовта и подкрепата, без които не бих

могла да се справям с проблемите в живота. Татко, съжалявам че не си между нас, за да

можеш да споделиш с мен радостта от успеха ми. Благодаря ти, че ме научи да бъда

силна!

I would like to thank to my very good friend and first housemate in Holland Marija

Matovic for the support and the fruitful discussions about our work and the life in general.

I would like also to thank to my Bulgarian friends Nikoleta, Ivan, Petar, Boryana, and

Veselka for holding my hand and listening me complaining, when I had problems. Special

thanks to Nikoleta, who was always on line, when I needed her. Petar and Veselka, thank you

for the unforgettable time we spent together in Utrecht. Boryana, thank you for your useful

advices. Ivan, thank you for your warm hospitality and delicious meals.

And last, but not least I would like to thank to my colleagues, especially to Dennis,

Arjan, Floris, Peter Vergeer, Linda and Philipp for the good working atmosphere.

119

Curriculum Vitae

Aneliya Zdravkova was born on 10th of December 1973 in Silistra, Bulgaria. In 1991 she

finished the Chemistry class in the Grammar School for Mathematics and Life Science in

Silistra, Bulgaria. She continued her education at Sofia University “St. Kliment Ohridski”,

Sofia, Bulgaria in the Faculty of Chemistry. In 1998 she obtained her Master of Science degree

in Chemistry and Physics. From 1998 until 2001 she was working as logistic controller in a

trade company for pharmaceutical products “Sanita Trading ltd.”, Sofia, Bulgaria. From 2001

until 2002 she worked in the Laboratory of Ultrastructure Research, National Institute for

Physiological Sciences, Okazaki, Japan as research assistant. In 2003 she started her PhD

research in the Condensed Matter and Interfaces group, Utrecht University, The Netherlands

under the supervision of Prof. Dr. Jan van der Eerden. The results of this research are

presented in this thesis.

121

Langmuir- Blodgett layers of amphiphilic molecules ... · CHAPTER 3 Structure and dynamics of...

Documents

Transcript of Langmuir- Blodgett layers of amphiphilic molecules ... · CHAPTER 3 Structure and dynamics of...