PHYSICAL REViEW E VOLUME 53, NUMBER 3 MARCH 1996

Pattern formations in lipid monolayers under illumination

Mu Wang, Cheng Sun, Willem J. P. van Enckevort, Jan van Esch, Gerald Wildburg, , Ru-Wen Peng,Nai-Ben Ming, Piet Bennema, Helmut Ringsdorf, and Roeland 3. M. Nolte

National Laboratory of Solid State Microstructures, Nanjing University, Nanjing 9l0099, Chinaand Center for Advanced Studies in Science and Technology of Microstructures, Nanjing 210098, China

Department of Solid State Chemistry, Faculty of Science, University of Nijmegen, 6M5 ED Nijmegen, The NetherlandsInstitute of Organic Chemistry, University of Mainz, D 5509-9 Mainz, Germany

Department of Organic Chemistry, Faculty of Science, University of ¹jmegen, GM5 ED ¹jmegen, The Netherlands(Received 19 December 1994; revised manuscript received 1 September 1996)

In lipid monolayer film we observed that the domains of the liquid-condensed phase may growunder continuous illumination of microscope light. This phenomenon occurred in the coexistenceregion of liquid-condensed (LC) and liquid-expanded phases. The average area per molecule in themonolayer film remained constant during the domain growth. By analyzing the growth behaviors ofthe I C domains carefully, we found that the observed domain growth arises from an illumination-related local mass transfer, which can be attributed to the light-induced damage to the Quorescencemolecules; our previous explanations [Wang et al. , Phys. Rev. Lett. 71, 4003 (1993)] should bemodified. Our results demonstrate the significant eKect of the decomposed Quorescence moleculeson the growth of the domains in the liquid-condensed phase. This efFect may also be used to studythe domain growth dynamics.

PACS number(s): 68.10.—m, 61.30.—v, 81.10.Aj, 64.60.—i

I. INTRODUCTION

Monolayers on the air-water interface provide an in-

triguing model system in which fundamental propertiesof two-dimensional pattern formation can be studied [1—3]. In this system the amphiphilic molecules that formthe monolayers have hydrophilic and hydrophobic por-tions. The balance between these opposing natures keepsthe molecules on the air-water interface. Conventionally,when a monolayer is isothermally compressed, it under-goes a series of phase transitions. By introducing fluores-cence microscopy and Brewster angle microscopy [1—4],transition process from a liquid-expanded (LE) phase toa liquid-condensed (LC) phase can be visualized directly.Previous experimental studies have shown that the do-mains of LC phase may vary from compact faceted pat-terns [5,6] to dendrites [7] and fractals [3,6,8], and some-times to even more complicated morphology [9]. Mor-phology transitions from fractal to dendrite [6], and fromfractal to nonfractal [10], etc. , have also been found.Therefore, monolayer film is an ideal two-dimensional(2D) system to study pattern formation and pattern se-lection problems. Normally, monolayers can be charac-terized by an isotherm of surface pressure (II) versus areaper molecule (A). When the area of the monolayer filmis large, the monolayer is in gaseous phase, where thearea per molecule is much greater than the molecule di-mension. In this case, the hydrocarbon portions of themolecules make a significant contact with water surface.As the area of the monolayer film decreases, a transi-tion from gaseous phase to a liquid-expanded phase (LE)occurs. In LE phase, the hydrocarbon chains lift from

the surface but remain largely disordered. Further com-pression of the monolayer film leads to the appearance ofthe liquid-condensed. (LC) phase, in which hydrocarbonchains lift from the surface and the degree of chain align-ment is much higher than in the LE phase. There areevidences of long range orientational order in LC phase.To some extent, we may regard the domains in LC phaseas 2D crystals. It is very interesting that II-A curveof some lipid compounds shows a long, flat plateau inthe LC-LE coexistence region. This implies that theLE-LC transition may be a first-order one. Figure 1shows the measured isotherms of monolayer film of N, %-dihexadecyl-3-(1-imidazolyl)-propyamine as a function oftemperature. When temperature increases, the isothermshifts upwards, and the flat two phase coexistence regionshrinks. Further increase of temperature may lead to thedisappearance of the coexistence region, i.e. , a criticalpoint is expected. These features resemble very muchthe phase transition in van der Waals gas, which is athoroughly studied model system in statistical physics.In Fig. 1, however, we did not observe the critical point,because the monolayer film collapsed when the film wascompressed to a certain extent at higher temperature.When collapse happened, the surface pressure did notincrease any more even when the film was compressed toa very condensed situation, as indicated by the arrow inFig. 1. Strong thermofluctuation at high temperature issuggested to be responsible for the collapse of a mono-layer film. The first-order-transition-like feature shownin a monolayer system suggests that the concepts andthe techniques used in the first-order phase transitionsmay be applied to the studies of LC domain growth in

1063-651X/96/53(3)/2580(8)/$10. 00 53 2580 1996 The American Physical Society

53 PATTERN FORMATIONS IN LIPID MONOLAYERS UNDER. . . 2581

50

40—

the surface-tension gradient, but also depends on the de-composition of Huorescence molecules upon continuousillumination, which changes the chemical potential in theilluminated region.

30-

P 20-II. EXPERIMENT

10-

0.2I

0.4I I I

0.6 0.8 1.0

Area (nm /molecule)

I

1.2

FIG. 1. II-A isotherms of the monolayer film of N, N-dihexadecyl-3-(1-imidazolyl)-propyamine as a function oftemperature. The LE-LC coexistence region is characterizedby the long, 8at plateau. We are not able to see the criti-cal point in this figure because the monolayer film collapsedwhen the area per molecule was decreased to a certain extentat high temperature, as indicated by the arrow.

the LE-LC coexistence region.Morphology of LC domains in lipid monolayers is rich.

Previous experiments have shown that domain morphol-ogy may transit from one to another during growth [6].In situ observation is proven to be a useful tool to investi-gate the mechanism behind these phenomena. However,it was not easy to apply in situ observation in a langmuirmonolayer system. In conventional langmuir monolayerexperiments, LC phase is nucleated by moving a teflonbarrier to decrease the area of monolayer film. Duringthe compression, strong convection always occurs in thewater subphase. Hence it was normally difIicult to tracea growing domain over a long period. By introducingan inhomogeneous electric field, it was possible to collectdomains under the microscope and study their evolvingprocess [11].However, the applied electric field may theninHuence molecular aggregation dynamics, because theelectric dipole-dipole interaction is believed to play animportant role in LC domain growth.

In this paper, we report a phenomenon that the LCdomains may grow under continuous illumination of mi-croscope light in the LE-LC coexistence region of lipidmonolayers. A device is designed to suppress the convec-tion in the water subphase, so we are able to trace thegrowing process of a specific domain for a period of about20—30 min and even longer. Technically, the phenomenonreported here implies a system with a controllable driv-ing force to study the development of LC domains. Thisphenomenon is also enlightening to the domain growthmechanism. Actually, the illumination-related domaingrowth phenomenon was first reported in 1993 [6]. How-ever, in that paper the role of surface-tension gradientwas misinterpreted. Our further investigation indicatesthat the observed. domain growth not only depends on

Imidazole surfactants R (-) N-, N--dihexadecyl-[2-(1-imidazolyl)propyl]amine (A) and N, N-dihexadecyl-3-(1-imidazolyl)-propylamine (B) were synthesized by one ofthe authors (van Esch) [12]. The synthesized imida-zole surfactants were purified by column chromatogra-phy. Before doing the experiments, Imidazole surfac-tant was dissolved in chloroform (P.A. grade, Merck).0.5 mol%%uo fiuorescent probe was added into the solu-tion for fluorescence microscopy. Two diff'erent fluo-rescent probes, DPPE-sulforhodamine and fluorescein-PE (Molecular Probes, USA) were introduced in theexperiments separately. Filters were used to generatequasimonochromatic light for the excitation of fluores-cent molecules. Green light (510—560 nm) was applied toexcite DPPE-sulforhodamine. For fluorescein-PE we uti-lized blue light (450—480 nm). Monolayer film was formedby carefully spreading the mixed solutions onto the watersurface in a tefion-coated trough (350 cm2), which wasthermostatted to 20.0+0.1 'C and placed on the stage ofa fluorescence microscope. The water for the monolayerexperiments was purified by a Millepore system. The pHof the water was 5.5. The trough was equipped with amotorized teflon barrier, allowing compressing monolay-ers with a controllable rate. In our experiments, this ratewas of the order of several A molecule i min . Surfacepressure-area (II-A) isotherm of the monolayers was mea-sured simultaneously. Surface pressures were detectedby Wilhelmy balance and were accurate to within 0.1mN m . The trough, Wilhelmy balance, and tempera-ture measuring system were all interfaced to a personalcomputer for data acquisition and system control. Inorder to monitor the behavior of a selected domain, thesurface flow in the plane of the monolayer was suppressedgreatly by an open circular mask of teflon (diameter 20mm), which was placed in the subphase under the objec-tive of the microscope. The opening on the mask guar-anteed that the surface pressure inside and outside of themask were identical at equilibrium situation.

The experimental procedure was as follows. First wecompressed the lipid monolayer films to the LE-LC co-existence region. In our system, the coexistence regionwas characterized by the formation of dark, fractal-likeLC domains. When the LC domains became sufFicientlylarge, the barrier was stopped and fixed thereafter. Thedomains ceased growing when an equilibrium surfacepressure was reached. Then we chose a LC domain forcontinuous observation. The water flow in the troughwas weakened when compression had been finished fora long time. Also the mask suppressed the convectiongreatly. So it was easy to keep the domain under inves-tigation at the center of the view field of the microscope.The optical field diaphragm (OFD) of the microscope was

MU WANG et aI. S3

adjusted so that a suitable size of the shade of OFD wasachieved. The domains in the shade of OFD were pro-tected from illumination. In the illuminated region thedomain growing process was recorded by a microscope-matched video system. Quantitative data about the do-main growth, such as the interfacial (interline) growthrate and increasing rate of the fraction of LC domains,etc. , were obtained by image analysis.

a much longer time to nucleate new I C domains. It hadalso been observed that the LC domains grew faster whenthe intensity of the illuminating light became stronger.

As the domains grew in the illuminated region, thosebranches in the shade of OFD, which were not illumi-

III. B.ESUITS AND DISCUSSION

For compound A, typical morphology of LC domaingenerated by compressing the monolayer film is shown in

Fig. 2(a). Because a molecule of compound A is a chiralone, some long branches of the fractal-like domain rotateclockwise O. n. the right corners of Fig. 2(a), part of theshade of optical field diaphragm can be seen. Continuousillumination of the monolayer 6.lm locally by microscopelight may induce the growth of the LC domain, as shownin Figs. 2(b) —2(d). It is noteworthy that in the initialperiod of illumination, some of the outer arms of the LCdomain of A melted and the whole pattern shrank [Figs.2(a) and 2(b)]. Thereafter the cluster started to grow

[Figs. 2(b) —(d)]. In the early stage of the illumination-related growth, facets could be formed on the tips of thefractal branches [Fig. 2(b)]. Yet these facets were not sta-ble. As illustrated in Figs. 2(c) and 2(d), dendritic pat-terns finally emerged after continuous growth. It shouldbe pointed out that although the branches trapped inbetween the others were growing, the most outward tipsgrew much faster, as that indicated by the arrows in Figs.2(b) and 2(c). This phenomena, known. as screening ef-fect, was a typical feature in diffusion-li. mited growth.

Similar growth phenomena happened in the monolayerof compound B, as shown in Figs. 2(a') —2(d'). In thisprocess, the initial melting of the LC domain upon illumi-nation was not evident. For the domains of compound B,anisotropy during the development of the interface (in-terline) between LC phase and LE phase is much weakerbecause B is a nonchiral molecule. According to the mi-croscopic solvability theory [13,14], anisotropy is requiredin the interfacial dynamics for a stable dendritic growth.If anisotropy is not sufFicient, repeated tip splitting oc-curs and dense branched pattern instead of dendrite mayfi.nally be formed. This idea is supported by Fig. 2. FromFigs. 2(a') —2(d') one may also find that LC domains be-came more "fat" during the growth. This eKect may re-late to the positional relaxation of amphiphilic moleculeswhen they incorporated into the LC domain [10].

In the LE-LC coexistence region, we carefully chosean area about 100 micrometers in radius under. the mi-croscope, which was free of any LC domains. When welooked under the Buorescence microscope, this area had ahomogeneous contrast. By illuminating this area contin-uously, dendritic LC domains were nucleated, as shownin Fig. 3. Our experiments showed that the illumination-related nucleation process depended on the intensity ofthe Inicroscope light. If the light intensity was attenuatedby adding a polaroid into the optical system, then it took

I I I!

'P

i

FIC. 2. The successive micrographs to show the illumi-nation-related LC domain growing process. During the do-main growth the average area per molecule on the air-waterinterface remained a constant. (a)—(d) show the growthof LC domain of compound A upon illumination. Thefractal-like pattern developed into dendrite after continuousgrowth [(a)—(d)]. (a'd') illustrate the illumination-relatedgrowth of LC domains of compound R. The initial fractalpattern was thickened during its growth and 6nally evolvedto a fingering pattern (d'). The difference between compoundA and R is that compound A has stronger anisotropy. Thebars represent 50 pm. On the upper right corners of (a') —(d'),the codes in the timer represent hours, minutes, and secondsfrom left to right, respectively.

53 PATTERN FORMATIONS IN LIPID MONOLAYERS UNDER. . . 2583

FIG. 3. The successive micrographs toshow the illumination-related nucleation pro-cess of LC domains of compound A. Initiallyan area free of any LC domains was selectedunder the microscope. By illuminating thisarea continuously, a LC domain was nucle-ated (a). Continuous illumination made thedomain grow. After a long time illumination,new generation of LC domains was induced,as shown in (f). The bar represents 100 pm.

nated, melted gradually in most cases, as indicated bythe arrows in Figs. 4(b) and 4(c). The trough on thestage of the microscope was slightly shifted in order toobserve these branches in the shade of OFD. It shouldbe pointed out that the process shown in Fig. 4 occurredwhen the average molecular density in monolayer filmremained a constant. The local growth of domains inthe illuminated region and melt of domain branches inthe surrounding region, as indicated in Fig. 4, suggestthat there exists a mass flux moving into the illuminatedregion from the surrounding areas. The melting of do-mains in the shade of OFD provides the nutrient neededfor the domain growth in the illuminated region. Mean-while, one may ask what will happen if the mass transferis blocked by a 2D barrier. Figure 5 gives an example,which demonstrates the domain morphologies near theboundary separating the illuminated area (lower-rightpart of Fig. 5) and the region protected by the shade ofOFD (upper-left corner). There happened to be a longdomain branch near the boundary of illumination, whichacted as a 2D barrier. On the upper-left corner of thepicture, as indicated by the arrow, the initial fractal-likedomain morphology has been broken, and a pattern ofBlaments was formed on one side of the branch. The sidebranches on the other side, however, have been melted.Faceted patterns have developed in the lower-right partof Fig. 5, which was continuously illuminated. Figure 5further supports the idea that there exists a local masstransport across the boundary of illumination. If thereis not much resistance to the molecular transport, theamphiphilic molecules will move into the illuminated re-gion and support the local domain growth there. If themolecular diQ'usion is blocked locally, as that shown onthe corner of Fig. 5, the molecules will be compressednear the obstacle. Consequently the side of the branchthat faces the coming nutrient may grow and form a pat-tern of filaments. Figure 5 also indicates that the masstransport in this system is really two dimensional.

Concerning the growth phenomena presented above,one may argue that the observed domain growth may re-sult from an inhomogeneous distribution of surface pres-sure during the compression, for example, local excesssurface pressure over the monolayer film after a veryquick compression [15]. To avoid the possible inhomo-geneous distribution of surface pressure, we left the LCdomains on water surface for some time after the com-pression. During the period waiting for the equilibrium,some minor side branches of the LC domains shrank dueto the effect of line tension. We selected a shrinking LCdomain for continuous observation. It turned out thatthe shrinking LC domain also grew upon illumination,as shown in Figs. 6(a)—6(d). If the LC domains wereleft on the air-water interface for even longer time, thoseinitially fractal-like domains could evolute to a compactcircular pattern. The illumination-related growth alsooccurred to these compact circular domains [16]. Wetherefore conclude that the observed domain growth doesnot relate to the possible compression-induced inhomo-geneous distribution of surface pressure. One may findthat screening e8'ect is clearly demonstrated in Fig. 6.Those branches marked by the arrows grew very slowlycomparing to other branches. This means that the do-main growth shown here should not be described by alocal growth model. Otherwise those marked branchesmay grow as fast as others.

From a molecular structure point of view, the photo-chemical reactions of compounds A and B are unlikely.The UV-vis spectra of these two compounds indicatedthat the absorption band of these two compounds is be-low 250 nm [17]. On the other hand, however, the fluores-cent molecules may decompose upon illumination. Whenthis happens, the illuminated region will become darker.This eÃect is remarkably illustrated in Fig. 7. The lowerpart of Fig. 7 has been illuminated for a long period,where well developed compact faceted and dendritic do-mains can be seen. The upper part of the figure was

2584 MU WANG et al. 53

protected from illumination by the shade of OFD. Inthe unilluminated region, the initial fractal-like brancheshave been broken and many side branches melted. Figure7 indicates that the chemical environment in the illumi-nated region was diferent from that in the unilluminatedregion due to the decomposition of fluorescent probes.

We found that the growth rate of the illuminated LCdomains depends on the concentration of fluorescent dye.We measured the increasing rate of the fraction of thearea occupied by the LC domains of compound B overthe whole illuminated region. Concentration of fluores-cent dye varied from 0.18 mo1% to 1.07 mol%. The ex-perimental result is shown in Fig. 8, which indicates thatthe illumination-related domain growth rate increaseswhen the fluorescence concentration raises.

We also changed the fluorescent probe from DPPE- 25',m

FIG. 5. The luminescence micrograph to show the situ-ations near the boundary of the illuminated region. Thelower-right part of the micrograph was continuously illumi-nated. Faceted LC domains of compound A. have developedin this region. The upper-left corner was protected by theshade of OFD. A long domain branch happened to be nearthe boundary of illumination. As indicated by the arrow, thebranches in the previously unilluminated region were broken,and some filaments were formed on one side of the long do-main branch.

(b)

(c)

FIG. 4. Corresponding to the growth of LC domains ofcompound B in the illuminated area, the domain branches inthe shade of OFD melted gradually, as indicated by the arrowsin (b) and (c). The trough on the stage of the microscope wasslightly shifted when the pictures (b) and (c) were taken. Thetime interval between each micrograph was about 2 min. Thebar represents 50 p,m.

sulforhodamine to fluorescein-PE to study the behav-ior of LC domains upon illumination. Meanwhile thewavelength of the illumination light was correspondinglychanged to 450—490 nm (blue light). It turned out thatthe LC domains still grew under continuous illumination.It seems that the illumination-related domain growthdoes not depend on the wavelength of light. Our previ-ous experiments have shown that the illumination-relateddomain growth do depend on the amphiphilic moleculeswhich form the monolayer film. As we reported ear-lier [16], LC domains of L-o.-dimyristoyl phosphatidyl-ethanolamine (DMPE) dissolved when they were contin-uously illuminated.

Rice and McConnell once stud. ied a kind of patterntransition of lipid domains [18], which was attributedto the photochemical effects on the monolayer films.Mobius et at. also suggested that photochemical effectmay change the local physical property over the mono-layer film, for example, local surface pressure [19]. Inour case, as we stated above, photochemical reactions onboth A and B were unlikely. Moreover, the screeningeKect in the domain growth suggests that the domaingrowth should not be the direct result of photochemicalreactions. If the observed domain growth were directlyresulted from photochemical reaction, all the LC-LE in-terfaces would grow simultaneously, because the photo-chemical product should. be homogeneously distributed.over the illuminated region. Yet what we observed wasthat the growth usually started from the most outwardtips (screening efFect).

So far what we learned from the experimental datais as follows. The domain growth upon illumination isindependent of the type of fluorescent dye, independentof the illuminating wavelength and the status of the do-

2585

'INNING]N

iI

ii

tee eR IBKIIB eee

8( .-mm a

3

FIG. 6. When the LC domains of com-pound B' w'ere left on air-water interface forsome time, the whole pattern shrank underthe inHuence of surface tension. Some minorbranches disappeared and only backboneswere left (a). It is shown in these micro-graphs that shrinking LC domain may alsogrow under continuous illumination. Thosebranches marked by the arrows grew veryslowly comparing to the other branches.This efFect is the typical feature of diKu-sion-limited growth. On the upper right cor-ners of (a)—(d), the first three codes in thetimer represent hours, minutes, and seconds,respectively. The bar represents 50 p,m.

Ie' ~ I tli

Ie II i'

i( )[Ql Rl Rk2R. B

' 'iNSN!IISml BR w i ii

%$IL!I48!lIII I~ ll

imam mi

llil'le i'm/

i[

g. x

ii e

31;

i It

main. Yet it depends on the amphiphilic compound thatforms the monolayer film. In the illuminated region, thechemical environment may be diff'erent from the otherregions due to the decomposition of fluorescent probes.The evidences of local mass transfer into the illuminatedregion from surrounding areas have also been observed.

To understand the observed phenomenon, it will behelpful to investigate both the physical and chemical en-vironment on the monolayer film across the illuminated

region. The temperature diff'erence between the illu-minated area and the surrounding unilluminated regionwas measured by a pair of identical diode thermometers,which were calibrated before experiments. The sensi-tivity of the measuring system was about 0.1 'C jmV.Figure 9 shows the result of measurement, which indi-cates that the temperature in the illuminated area wasabout 0.2 'C higher [20]. A stable temperature distribu-tion can be reached when the area has been illuminatedfor about one minute. It is known that surface tensionof a monolayer 61m declines when the temperature in-creases. Therefore, a centripetal surface-tension gradientis established around the microscope light spot on the

20

16-

12-

FIG. 7. The luminescence micrograph to illustrate the de-composition of Huorescent probes upon continuous illumina-tion. The up region of the picture was protected by the shadeof OFD, where the initial fractal-like domain branches havebeen melted. The lower part of the picture was continuouslyilluminated, where LC domains have grown significantly. Thedarkening of the illuminated region was due to the decompo-sition of the fluorescent probes. The amphiphile involved inthis picture was compound A. The bar represents 25 pm.

I

0.9I

0.300.0 1.20.6

Concentration ( mo1% )FIG. 8. The plot to show the increasing rate of the frac-

tion of LC domains in the illuminated region as a function ofconcentration of Buorescent probe. The illumination-relateddomain growth rate increased with the increase of Huorescentconcentration.

PATTERN FORMATIONS IN LIPID MONOLAYERS UNDER. . .

2586 MU WANG et al. 53

15.0

12.5-

10.0

light on

0.2 C

light off

bQ

7.5-CS

0

5.00.0 10.0 15.0

I

20.0

monolayer 61m. This surface-tension gradient, however,does not transport amphiphilic molecules into the illumi-nated region as what we stated in our previous paper [6].Contrary, it drives molecules out of the illuminated area.So the effect the surface-tension gradient is to dissolvethe LC domains. This effect was experimentally provenby slightly heating a single LC domain with a speciallydesigned device in the water subphase, w hereupon weobserved the gradual melting of a fractal-like LC domain[21]. On the other hand, as we demonstrated in Fig.7, continuous illumination may decompose the fluores-cent probes in the illuminated region. The decomposedfluorescent molecules in the illuminated area may createa different chemical environment than the surroundingregion. This chemical potential difference is responsiblefor the transport of amphiphilic molecules. The directionand speed of the local. molecular transfer, which is deter-mined by molecular interactions, governs the behavior ofLC domains in the illuminated region.

Actually, the local temperature build up upon illu-mination shown in Fig. 9 has two effects. One is lo-cal thermoexpansion of the monolayer film, which movesthe amphiphile out of the illuminated region. However,thermoexpansion stops as soon as equilibrium is reached.The thickness of a monolayer film is only about severaltens of angstrom, so the time required to establish ther-moequilibrium in the illuminated region should be veryshort. Therefore the effect of thermoexpansion is neg-ligible. Another effect comes from the surface-tensiongradient around the illuminated region, which also drivesmolecules out of the illuminated region. So the efFects re-lated to the local temperature increase always make theLC domains melt. On the other hand, the decomposedHuorescent probes in the illuminated region may induce acentripetal chemical potential difference. If the chemicalpotential in the illuminated region becomes lower thanthat in the surrounding region, a mass Hux moving into

Time (min)FIG. 9. The plot to show the temperature difference be-

tween the area illuminated by the microscope light and thearea protected by the shade of OFD, which was measuredby two identical diode thermometers placed in these two re-gions, respectively. One may 6nd that a stable temperaturedifference can be established after continuous illumination forabout 1 min.

the illuminated region will be generated. When the Hux

moving into the illuminated area is stronger than the out-ward moving Hux induced by the surface-tension gradi-ent, the molecular density in the illuminated region willincrease continuously. In this case, LC domains in theilluminated region will grow. The growth rate of LC do-main depends on the competition between the inward andoutward Hux. If, however, the chemical potential inside ofthe illuminated region becomes even higher than that inthe surrounding region because of the role of the decom-posed Huorescent probes, the amphiphilic molecules will

definitely move outward. Meanwhile, LC domain will dis-solve upon continuous illumination [16]. Therefore, thebehavior of I C domains upon continuous illuminationdepends on the chemical potential difference across theboundary of the illuminated area, while this potential dif-ference depends on the interactions between amphiphilicmolecules and decomposed Huorescent probes. Accordingto solution theory, one may easily write down the relationbetween molecular interaction and chemical potential. Itturns out that a stronger attractive interaction betweenimpurity and compound molecules favors the growth ofLC domain [16].

The above picture of domain growth mechanism is sup-ported by our observation of the domain growth rate asa function of Huorescent concentration, as that shownin Fig. 8. %Then the fluorescent concentration becomeshigher, the concentration of the light-decomposed fluores-cent molecules will also be high. The effect of impuritywill then be more evident. Hence LC domains grow morerapidly at higher Huorescent concentration.

The slight melt of LC domain before the overwhelminggrowth as that shown in Figs. 2(a) and 2(b) can be un-

derstood in the framework of crystal growth theory andsolution theory. In Huorescence microscopy, most Huores-cence probes are kept out of the LC domain during thedomain growth. Only a very small part remains in theLC domain. In LE-LC coexistence region, when equi-librium is reached, chemical potential p should be equaleverywhere, i.e. ,

p (z) = p, (i) =y, (o)

where "i" and "o" in parentheses stand for "inside" and"outside" of the area that will be illuminated. , "LE" and"LC" in superscript stand for the LE phase and LCphase, respectively. Upon illumination decomposed flu-

orescent probes (impurities) are introduced, which maychange the chemical potential. For both the compoundswe studied in this paper, chemical potential declines dueto the decomposed fluorescence probes. So p (i) de-

Z.E'creases to p, (i). At the same time, p," (i) decreases to

/p~c (i) because the fluorescent probes within the LC do-main may also decompose. It is reasonable to expect thatthe degree of the decline of chemical potential within LCdomains is much less than that in the LE phase, becausethe impurity concentration in LC phase is much lower.So, upon illumination, the following situation holds:

Therefore the LC domains in the illuminated region melt

PATTERN FORMATIONS IN LIPID MONOLAYERS UNDER. . .

because p (i) ( p," (i). At the same time, am-

phiphilic molecules flow into the illuminated region fromI

the surrounding areas because p" (i) ( p (o). Con-sequently the amphiphilic molecule concentration in LEphase in the illuminated region increases. According to

LE'the solution theory, p (i) will then increase. As soonas pL (i) = p"c (i) is reached, melting of LC domains inthe illuminated region stops. Thereafter, because the re-lation p, (i) ( p~E(o) still holds, amphiphilic moleculesare continuously transported into the illuminated areaand the concentration of amphiphilic molecules in LEphase in the illuminated region increases continuously.

I

Therefore p" (i) is further increased. Eventually, weLC' LE'may have p," (i) ( p (i) in the illuminated region,

which means that a positive driving force for the growthof LC domain is established. Hence LC domains start togrow.

Whether the initial melting of LC domains upon illu-mination as that shown in Figs. 2(a) and 2(b) can be ob-served depends on the initial difference between p (i)and p + (i). The degree of this difference relates to themolecular interactions. If p,

" (i) and p" (i) do not dif-fer very much, they may become equivalent quickly uponillumination. Meanwhile we are not able to observe ob-vious melting of LC domains in the initial period of illu-mination. We suggest this is the situation of compoundB [Figs. 2(a') —2(d')].

In conventional langmuir monolayer experiments, inorder to reach the LE-LC coexistence region, p(r, t) isincreased by continuous decreasing of the area of mono-layer 61m. In the illumination-related domain growth,however, p(r, t) is increased by the local mass transport.Whatever the way that p(r, t) is increased, as soon asp(r, t) exceeds the critical value p„LCdomains grow.Corresponding to p, the surface pressure is IIo(Ao, 7'),where Az is the area per molecules at the critical densityp, T is temperature. When p(r", t) increases, the localsurface pressure II(A"E, T) also increases. The locally

increased surface pressure is dissipated by the growth ofLC domains. Consequently the average surface pressureover the whole monolayer 6.lm remains constant. Analogto crystal growth, we suggest that the dimensionless driv-ing force for the domain growth, Ap/kT, be expressed as

Ap ( II(A"E, T) 5 . t'p(r, t) )kT ~ II, (ALE, T) y q p.

ocln[ — 'i

='Ini

In the case of illumination-related domain growth, thisdriving force may be adjusted by changing the concen-tration of the fluorescent probe, or varying the intensityof the illuminating light.

Briefly, we found in lipid monolayers a growth phe-nomenon of LC domains, which can be attributed tothe light-induced damage of' fluorescence molecules. ThisefFect depends on the molecular interactions betweenamphiphile and decomposed fluorescence molecules, andthe illuminating light intensity as well. When studyingthe domain morphology by fluorescence microscopy, oneshould be careful of this illumination-related efFect. Onthe other hand, by carefully controlling the illuminatinglight of the microscope and the concentration of fluores-cence molecules, it is possible to create a local growthenvironment for LC domains with low driving force. Onemay then investigate the growth dynamics of the 2Dfaceted crystals and dendrites, as well as 2D morpho-logical instabilities.

ACKNOWLEDC MENTS

The authors thank Professor H. Mohwald for discus-sion. The experiments of fIuorescence microscopy werecarried out in the group of Professor H. Ringsdorf. M.W.wishes to acknowledge the support of the Institute of Or-ganic Chemistry in Mainz University. This work was alsopartially supported by a grant from the Natural ScienceFoundation of China (IVo. 19425007) and a grant fromthe State Science and Technology Commission of China.

[1] C. M. Knobler, Science 249, 870 (1990).[2] R. M. Weis and H. M. McConnell, Nature 810, 47 (1984).[3] A. Miller, W. Knoll, and H. Mohwald, Phys. Rev. Lett.

56, 2633 (1986).[4] D. Honig and D. Mobius, J. Phys. Chem. 95, 4590 (1991);

S. Henon and 3. Meunier, Rev. Sci. Instrum. 62, 936(1991).

[5] B. Berge, L. Faucheux, K. Schwab, and A. Libchaber,Nature 850, 322 (1991).

[6] Mu Wang, G. Wildburg, J. van Esch, P. Bennema, R. J.M. Nolte, and H. Ringsdorf, Phys. Rev. Lett. 7'1, 4003(1993).

[7] S. Akamatsu, O. Bouloussa, K. To, and F. Rondelez,Phys. Rev. A 46, R4504 (1992).

[8] K. A. Suresh, J. Nittmann, aud F. Rondelez, Europhys.Lett. 6, 437 (1988).

[9] R. M. Weis and H. M. McConnell, J. Phys. Chem. 89,4453 (1985); K. J. Stine, C. M. Knobler, and R. C. Desai,Phys. Rev. Lett. 65, 1004 (1990).

[10] Mu Wang et al. (unpublished).[11] A. Miller and H. Mohwald, Europhys. Lett. 2, 67 (1986).[12] Jan van Esch, Ph. D. thesis, Nijmegen University, The

Net her lands (1993) .

[13] E. Ben-Jacob and P. Garik, Nature 848, 523 (1990).[14] D. A. Kessler and H. Le~;ine, Phys. Rev. B 88, 7867

(1986).[15] A. Miller and H. Mohwald, J. Chem. Phys. 86, 4258

(1987).[16] Mu Wang, Ru-Wen Peng, Nai-Ben Ming, W. J. P. van

Enckevort, P. Bennema, 3. van Esch, H. Ringsdorf, andR. J. M. Nolte, Phys. Lett. A 208, 235 (1995).

[17] R. J. M. Nolte (private communication); see also Ref.[12].

[18] P. A. Rice and H. M. McConnell, Proc. Natl. Acad. Sci.86, 6445 (1989).

[19] D. Mobius, in Organic Phototransformations in Xonhomogeneous Media, American Chemical Society Sympo-sium Series, edited by M. A. Fox (ACS, Washington, DC,1985), Ser. No. 278, p. 113.

[20] The local temperature increase was most probably dueto the absorption of the energy of microscope light by the+rater subphase.

[21] Mu Wang et al. (unpublished).