Temporal Stability of a Bilayer in a Langmuir−Blodgett Multilayer and

Department of Chemical Engineering, Rose-Hulman Institute of Technology,. Terre Haute, Indiana 47803. Received September 28, 1995. In Final Form: Marc...
1 downloads 0 Views 213KB Size
Langmuir 1996, 12, 3015-3023

3015

Temporal Stability of a Bilayer in a Langmuir-Blodgett Multilayer and Its Dependence on Multilayer Structure H. Wu,†,‡ M. D. Foster,*,† S. A. Ross,§ W. L. Mattice,† and M. A. Matties†,| Institute of Polymer Science, The University of Akron, Akron, Ohio 44325-3909, and Department of Chemical Engineering, Rose-Hulman Institute of Technology, Terre Haute, Indiana 47803 Received September 28, 1995. In Final Form: March 15, 1996X The stability of a bilayer at or near the surface of a Langmuir-Blodgett multilayer of cadmium arachidate is investigated by fluorescence spectroscopy. Changes in the fluorescence spectrum with time, monitored over more than 200 h, indicate that rearrangement and loss of labeled molecules occurs even at room temperature in air. The nature of this change is perturbed by dipping the multilayer into the subphase an additional time, even when no further material is deposited atop the multilayer. Placing an additional bilayer of either cadmium arachidate or a polypeptide copolymer with alkyl side chains atop the labeled bilayer sharply reduces its mobility and suppresses somewhat the rate of change of structure in that bilayer. Adding a bilayer of polyglutamate affects the labeled bilayer’s lateral mobility more than does adding a bilayer of cadmium arachidate. This is due to the interdigitation of the polyglutamate’s side chains with the layer upon which it is deposited. However, addition of the cadmium arachidate bilayer is more effective in reducing sublimation.

* Author to whom correspondence should be addressed. Phone: (330) 972-5323. Fax: (330) 972-5290. E-mail: foster@ polymer.uakron.edu. † University of Akron. ‡ Present address: Tesa Tape, Inc., Charlotte, North Carolina 28209. § Rose-Hulman Institute of Technology. | Present address: Department of Chemical Engineering, University of Cincinnati, Cincinnati, Ohio 45221. X Abstract published in Advance ACS Abstracts, May 1, 1996.

optical microscopy,8 and electron diffraction2b as well as by fluorescence9 and neutron reflectometry,1,5,10-12 Many of these studies have focused on the sample behavior at temperatures above ambient, but some work on mobility at room temperature has been done. Schwartz et al.3 have directly imaged the restacking of molecules of a single monolayer and three-layer multilayers to form thicker multilayer regions using AFM. In their case, CdA is deposited originally on a hydrophilic substrate and maximization of the very favorable headgroup-headgroup interactions drives the rearrangement.3 This rearrangement is much faster in water than in air. Reflectometry measurements1,10,11,13 also reveal rearrangements during deposition, though the exact character of the rearrangement is not as readily ascertained in those results. Fujihira and co-workers9 attempted to characterize the degradation of a molecular device on the basis of photoinduced electron transfer between donor and acceptor molecules in an LB film using fluorescence. They investigated the interdiffusion of an excimer-forming probe, pyrenedecanoate anion, in calcium arachidate (CaA) layers. They showed that mixed monolayers of CaA and the pyrene-labeled fatty acid contain a nonuniform twodimensional distribution of chromophores and that the ratio of excimer to monomer fluorescence intensities could be used to follow changes in the pyrene concentration on a microscopic level. From the decay of this ratio they concluded qualitatively that interlayer diffusion did occur over a period of tens of hours after deposition. Two limitations in their published discussion of the results present opportunities for a closer look using fluorescence, while at the same time attempting to reproduce their general results. First, the calibration procedure they used

(1) Vierheller, T. R.; Foster, M. D.; Wu, H.; Schmidt, A.; Knoll, W.; Wegner, G.; Satija, S.; Majkrzak, C. F. In preparation. Vierheller, T. R. Ph.D. Thesis, The University of Akron, 1994. (2) Schreck, M.; Schier, H.; Go¨pel, W. Langmuir 1991, 7, 2287. Riegler, J. E. J. Phys. Chem. 1989, 93, 6475. Tippmann-Krayer, P.; Laxhuber, L. A.; Mo¨hwald, H. Thin Solid Films 1988, 159, 387. Laxhuber, L. A.; Mo¨hwald, H. Langmuir 1987, 3, 837. (3) Schwartz, D. K.; Garnaes, J.; Viswanathan, R.; Zasadzinski, J. A. N. Science 1992, 257 , 508. Schwartz, D. K.; Viswanathan, R.; Zasadzinski, J. A. N. J. Phys. Chem. 1992, 96, 10444. (4) Naselli, C.; Rabolt, J. F.; Swalen, J. D. J. Chem. Phys. 1985, 82, 2136. (5) Shimomura, M.; Song, K.; Rabolt, J. F. Langmuir 1992, 8, 887. (6) Rabe, J. P.; Swalen, J. D.; Rabolt, J. F. J. Chem. Phys. 1987, 86, 1601.

(7) Barbaczy, E.; Dodge, F.; Rabolt, J. F. Appl. Spectrosc. 1987, 41, 176. (8) Rabe, J. P.; Novotny, V.; Swalen, J. D.; Rabolt, J. F. Thin Solid Films 1988, 159, 359. (9) Fujihira, M.; Nishiyama, K.; Hamaguchi, Y. J. Chem. Soc., Chem. Commun. 1986, 823. (10) Musgrove, R. J.; Grundy, M. J.; Roser, S. J.; Richardson, R. M. ISIS Annu. Report, 1990, A113. (11) Grundy, M. J.; Musgrove, R. J.; Richardson, R. M.; Roser, S. J.; Penfold, J. Langmuir 1990, 6, 519. (12) Stroeve, P.; Rabolt, J. F.; Hilleke, R. O.; Felcher, G. P.; Chen, S. H. Mater. Res. Soc. Symp. Proc. 1990, 166, 103. (13) Foster, M. D.; Vierheller, T. R.; Schmidt, A.; Knoll, W.; Satija, S.; Majkrzak, C. F. Mater. Res. Soc. Symp. Proc. 1992, 41, 248.

Introduction Molecular movement in Langmuir-Blodgett (LB) films has important implications both for the potential exploitation of LB multilayers in nanoscale devices and for our basic understanding of the dynamics and stability of supramolecular structures. Since the structure of cadmium arachidate (CdA) LB films has been widely studied, much of the work done to date on the stability of LB systems has focused on this molecule. In a reflectometry study of multilayer stability to interdiffusion1 it was found that sublimation plays an important role in rearrangement of the films, even at temperatures considerably below the bulk order-disorder transition temperature for CdA. Studies by others,2 particularly Mo¨hwald and co-workers, have also indicated the important role of sublimation. One method which suggests itself as a means of providing an independent measure of molecular movement is fluorescence. This method is employed here to probe this phenomena as it manifests itself in the changes which occur in room temperature structure on the time scales of minutes to tens of hours after deposition. Rearrangement, or molecular movement, in multilayers has been observed by using AFM,3 IR,4 FTIR,5 Raman,6,7

S0743-7463(95)00807-9 CCC: $12.00

© 1996 American Chemical Society

3016 Langmuir, Vol. 12, No. 12, 1996

Wu et al.

could be improved upon by using a sample more similar in structure to the actual test multilayer. Second, they made no attempt to independently analyze the individual monomer and excimer intensity signals. Of particular interest in the present work is the behavior of those layers closest to the air interface. Arndt et al.14 have showed using grazing incidence reflection IR that the layer adjoining the CdA/air interface generally is more tilted or more disordered than layers in the “bulk” of the multilayer. X-ray scattering work by Skita et al.15 has led them to a similar conclusion. In addition, they have reported that the ordering of this layer may be improved by the subsequent deposition of an additional monolayer atop it. These results raise the question of the relationship of the top layer’s disorder to the stability of the multilayer as a whole. In research done by the author and coworkers,1 it has been observed with both X-ray and neutron reflectometry that the mass density of the top bilayer is consistently lower than what may be rationalized on the basis of transfer coefficients when the samples are measured within a few days time after deposition. The objectives of this work are to gain insight into the change taking place in a labeled top bilayer at room temperature during the first hours and days after deposition and to study the possibility of reducing the change that is not favorable to future applications. Sublimation has been observed in a sample with pyrene-labeled molecules right next to air. An additional bilayer of either cadmium arachidate or a polypeptide copolymer with alkyl side chains is found to act efficiently as a sublimation barrier. The mobility of the pyrene molecules is restricted by the capping layers as well. Experimental Section 1. Materials. Arachidic acid of purity greater than 99% (C20) was purchased from Fluka Chemika. The pyrene end-tagged 10-(1-pyrene)decanoic acid (Py-C10) was purchased from Molecular Probes. Structures of the probe and the matrix molecules are shown in Figure 1. These materials were used as received. The polyglutamate statistical copolymer (PG60) used was a gift from Prof. G. Wegner of the Max-Planck-Institut fu¨r Polymerforschung. This molecule has octadecyl side chains on 60% of its repeat units, as suggested by the schematic structure shown in Figure 1 as well. The polymer's molecular weight is 113000 g/mol, and the melting temperature of the crystallized side chains in a bulk sample is 42 °C.16 Cadmium chloride was purchased from Aldrich and used without further treatment for preparing the subphase. Chloroform of 99.9% purity was used for preparing deposition solutions and also for trough cleaning. Milli-Q water was used as the subphase, while distilled water was used in the substrate cleaning procedure. 2. Sample Preparation. Before deposition, quartz substrates were sonicated in chloroform for 20 min, then in distilled water for 5 min, and afterward in nitric acid for 20 min. After another 5 min of sonication in distilled water, the substrates were sonicated in saturated sodium bicarbonate solution for 10 min. Three rinses with water of 5 min each followed, and the substrates were dried with nitrogen.17 The resulting hydrophilic substrates had contact angles of 100.0 ( 0.5°. Multilayers were deposited on the substrates within 3 h after cleaning. The subphase for the fatty acid monolayers was a 8 × 10-4 molar solution of CdCl2 in Milli-Q water adjusted to pH 6.8 ( 0.2 with the help of TRIZMA buffer (Sigma). At this pH there was nearly 100% dissociation of the arachidic acid upon contact with the subphase. The subphase for PG60 was Milli-Q water. Arachidic acid (AA) solution was spread on the subphase from a solution with a concentration of about 1 mg/mL, while the (14) Arndt, T., Bubeck, C. Thin Solid Films 1988, 159, 443. (15) Skita, V.; Richardson, W.; Filipkowski, M.; Garito, A.; Blasie, J. J. Phys. (Paris) 1986, 47, 1849. (16) Duda, G. Ph.D. Thesis, University of Mainz, 1988. (17) Aoki, K.; Seki, T.; Suzuki, Y.; Tamaki, T.; Hosoki, A.; Ichimura, K. Langmuir 1992, 8, 1007.

Figure 1. Chemical structure schematic for (a) arachidic acid, C20, (b) 10-(1-pyrene)decanoic acid, Py-C10, and (c) poly[(γmethyl-L-glutamate)-co-(γ-octadecyl-L-glutamate)], PG60. Note that the labeled molecule is somewhat shorter, thicker, and heavier than the arachidic acid molecule. Table 1. Solutions Used in Spreading Langmuir Monolayers solution notation AA AP PG60

molecule(s) used arachidic acid (C20) arachidic acid (C20) and 10-(1-pyrene) decanoic acid (Py-C10) PG copolymer, PG60

concentration (mg/mL) 1.00 0.97 0.1

mixture of C20 and Py-C10 (AP) was spread from a solution of concentration 0.97 mg/mL (the concentration of Py-C10 in the solution was around 0.1 M). In this mixture, the number ratio of C20 to Py-C10 molecules was 27:1. The PG60 solution (PG60) had a concentration of 0.1 mg/mL. Solutions were stored in the freezer when not in use. Table 1 shows the details of the solutions used in this study. The trough (Nima 611, 29 × 18 × 0.25 cm3) was cleaned by wiping the TEFLON portions with chloroform and waiting for 15 min to allow all the chloroform to evaporate. Any remaining contamination was removed by multiple rinses with water. Each monolayer was left on the trough for 15 min before being compressed. The AA and AP monolayers were compressed under feedback pressure control to 30 mN/m and were annealed at this surface pressure for 5 min. The PG60 monolayer was compressed to 20 mN/m and deposited after it reached the surface pressure. All monolayers were transferred at 21 ( 1 °C using dipping speeds of 5 mm/min for the fatty acid and 20 mm/min for PG60. The transfer ratios for both up and down strokes were 1.00 ( 0.07. Samples with fluorescently labeled layers were prepared for three different multilayer arrangements shown in Table 2. For

Temporal Stability of a Bilayer in LB Films

Langmuir, Vol. 12, No. 12, 1996 3017

Table 2. Detailed Sample Structures of Each System system I II III

notation

design structure

q/5/2 q/5/2/0 q/7 q/5/2/2-CdA q/9 q/5/2/2-PG60 q/7/2-PG60

quartz/5CdA/2CdAP quartz/5CdA/2CdAPa quartz/7CdA (background sample) quartz/5CdA/2CdAP/2CdA quartz/9CdA (background sample) quartz/5CdA/2CdAP/2PG60 quartz/7CdA/2-PG60 (background sample)

a Sample q/5/2/0 was dipped through the subphase without any monolayer on the subphase surface after the first fluorescence measurement was done for the deposited 5CdA and 2CdAP layers.

each arrangement a corresponding reference sample containing the same number of layers, but having no pyrene labeled molecules, was made as well. System I had seven layers of fatty acid salt, with the top bilayer being labeled. System II samples looked like system I samples with an additional CdA bilayer atop the labeled bilayer. In system III samples, the outermost bilayer contained PG60 rather than CdA. One limitation of the present study is the disruption of order in the mixed monolayers which results from the mismatch in molecular length between C20 and Py-C10 as well as the mismatch in molecular diameter due to the bulkiness of the pyrene label. This disruption may mean that layers containing labeled molecules are somewhat less stable than pure layers. The concentration of labeled molecules has been kept small in order to minimize their impact on the stability. A sample having the same layer configuration as system I but deposited on a silicon substrate (obtained from Semiconductor Processing, Inc.) was characterized by X-ray reflectivity. The reflectometer, mounted on a 12 kW rotation anode (Rigaku) with Cu target, had a pyrolytic graphite primary monochromator and adjustable collimating slits which provided a nominal resolution of order 0.004 Å-1 in scattering vector, q ()4π sin θ/λ). 3. Fluorescence Measurements and Sample Characterization. Fluorescence emission spectra of the LB multilayer samples were obtained with a SLM 8000 spectrofluorometer with a 450 W Xenon arc lamp excitation source. The excitation wavelength was set at 333 nm, and the spectra were taken in the front-face mode. Due to the low signal intensity from these LB samples, the photon-counting mode was employed with a 4 nm bandwidth for the emission and excitation slits. All the spectra shown have been corrected for background measured using the appropriate reference sample, and the signal from the labeled sample was much higher than that from the reference sample. For example, for sample q/5/2/2-PG60 the background sample was q/7/2-PG60. The only difference between the two samples was the presence of Py-C10 in the q/5/2/2-PG60 sample. Five layers were placed between the substrate and the layer with pyrene labels in every case, in order to eliminate substrate effects on the fluorescence emission. The overall thickness of these five CdA layers was about 125 Å. The amount of time required to accumulate a spectrum was typically around 6 min. The time zero event was completion of dipping the CdAP bilayer.

Results and Discussion 1. As Deposited State. We consider first the structure of the various samples at time zero, remarking on the significance of a spectrum’s general features and then comparing spectra from different samples and different times. The emission spectra taken as soon as possible after deposition of our four different samples are shown in Figures 2, 3a, 4, and 5. These spectra were taken at a few minutes after deposition of the labeled layers when these layers were the last ones deposited on the sample. The time lag for the first spectrum increased to 1-2 h in samples where additional unlabeled layers were deposited. The quantities of interest are the monomer emission intensity, Im, the excimer emission intensity, Ie, and the ratio Ie/Im. Since fluorescence from any depth in the sample will be registered in our measurement, the

Figure 2. Five selected emission spectra from sample q/5/2 taken at 0.033, 2.9, 20.5, 84.8, and 133.8 h after deposition of the second labeled layer. Very strong excimer emission is evident at 470 nm, and both monomer emission and excimer emission decrease with time.

monomer intensity is proportional to the concentration of isolated, labeled molecules in the entire sample. Likewise, Ie is proportional to the concentration of aggregates of labeled molecules in the sample. Since the proportionality constants for relating Im and Ie to concentrations may vary from sample to sample, comparison of absolute levels of Im and Ie among samples is impractical. However, the ratio Ie/Im, which is indicative of the relative concentration of aggregates vs isolated molecules, may be readily compared. Each spectrum has three pronounced peaks. The two at ca. 378 and 398 nm are associated with monomer emission, while that at 470 nm corresponds to excimer emission. In each case the two monomer peaks are seen to behave very similarly. Therefore, the peak at 378 nm is singled out as representative of the monomer behavior and all the references to “monomer emission” below correspond to that peak. For sample q/5/2, the first fluorescence spectrum was collected 4 minutes after completing the deposition of the pyrene-labeled bilayer. Figure 2 is a collection of background-corrected spectra measured at different times after the deposition with the first measurement designated as t ) 0.033 h. The presence of both monomer emission and excimer emission indicates the existence of both isolated pyrene molecules as well as pyrene aggregates, which suggests the nonuniform distribution of the pyrene-tagged molecules. The same kind of nonuniform distribution of labeled pyrene molecule was observed by Fujihira et al. as well.9 Changing over to a CdAP monolayer from CdA and then depositing the CdAP bilayer required a total of about 50 min. Therefore, addition of this bilayer plays an important role in determining the time from initial immersion of the sample into the subphase to measurement of the first spectrum. The most important differences noted among the first spectra of different systems measured after deposition reveal that adding another bilayer atop the labeled bilayer suppresses the mobility of the labeled molecules. Two differences between the spectra for the uncapped sample and the sample capped by CdA are evident in a comparison of Figures 2 and 4. The formation of pyrene excimer in solution is a diffusion controlled process, in which the excimer formation rate (kAB) is inversely proportional to the viscosity of the system (η), as pointed out by Einstein-Smoluchowski diffusion

3018 Langmuir, Vol. 12, No. 12, 1996

Wu et al.

Figure 4. Five selected emission spectra from sample q/5/2/ 2-CdA taken 1.53, 2.1, 22.3, 94.0, and 129.0 h after deposition of the second labeled layer. The ratio of excimer to monomer emission is considerably suppressed in this system in comparison to the sample q/5/2.

Figure 3. (a) Emission spectra from sample q/5/2/0 taken before (s) and after (- - -) completing an extra dipping cycle in the subphase. The changes in monomer emission intensity at λ ) 378 nm and excimer emission intensity at 470 nm indicate a reduction in the fraction of pyrene-labeled molecules present as aggregates. (b) Five selected emission spectra from sample q/5/2/0 taken at 0.43, 2.7, 19.4, 90.8, and 129 h after the extra dipping cycle. Monomer emission intensity increases in time while excimer emission intensity decreases.

theory.18,19 The same viscosity dependence will operate in the LB system, with perhaps an additional effect coming from excimer formation arising from static sites that depend on structure and defects. In sample q/5/2, the lower CdAP layer is facing a neighboring CdA layer. The upper CdAP layer, however, is facing air, which has much higher mobility and lower viscosity. Adding the extra capping layer, in the q/5/2/2-CdA case, restricts movement in the top CdAP layer. That is, the viscosity seen by a pyrene-labeled molecule moving laterally in the layer is increased. This increase in viscosity will cause a decrease in the excimer formation rate, and less excimer emission will be observed, as confirmed in these measurements. Supposing that the mobility has dropped is also consistent with the observations of Arndt14 that the ordering of a (18) Birks, J. B. Photophysics of Aromatic Molecules; WileyInterscience: New York, 1970. (19) Debye, P. Trans. Electrochem. Soc. 1942, 82, 205. Umberger, J. Q.; La Mer, V. K. J. Am. Chem. Soc. 1945, 69, 1089. Noyes, R. M. Prog. React. Kinet. 1961, 1, 131.

Figure 5. Five emission spectra from q/5/2/2-PG60 taken at 1.16, 2.9, 24.2, 87.0, and 112.0 h after deposition of the second labeled layer. Ie/Im is the smallest for this system.

layer next to air may be improved by subsequently adding an additional layer. More highly ordered CdA layers are less mobile than less ordered ones. One may question whether the difference between the capped and uncapped samples is indeed brought about only by the change in the neighborhood of the labeled bilayer or rather is brought about simply by submerging the sample longer in the subphase. To address this issue, a measurement was performed using a second sample having the same design structure as that of q/5/2 but a slightly different deposition history. This sample, designated q/5/2/0, was measured immediately (3 min) after deposition and was then measured again after being subjected to an additional dipping cycle in a subphase having no monolayer atop it. In this way, the amount of time sample q/5/2/0 spent in the subphase is the same as that of sample q/5/2/2-CdA, but the amount of time used between the dipping of layers may be different. The spectrum for sample q/5/2/0 before the extra dipping, shown in Figure 3a, has features similar to those for sample q/5/2. The Ie/Im ratio has a value of 1.10, as opposed to a value of 1.35 seen for sample q/5/2. While this difference appears significant when comparing just these two samples, measurements of Ie/Im for four other samples

Temporal Stability of a Bilayer in LB Films

(deposited from fresh solutions at different times over a 5 month period) of the same structure and preparation history as that of q/5/2 suggest an average value for Ie/Im of 1.21, with the variations being (0.12. This slight variability of initial structure before dipping notwithstanding, comparison of the two spectra in Figure 3a suggests that changes in the labeled bilayer are caused by additional time in the subphase. Specifically, the ratio Ie/Im now drops from 1.10 to 0.72 after the extra dipping cycle during which no material is deposited. This change is similar in character to the variation in Ie/Im seen between q/5/2 and q/5/2/2-CdA, although it is not as large. The drop in Py-C10 aggregate density caused by dipping could be due either to layer rearrangement or to loss of labeled molecules out of the sample directly from the aggregate domains. Neutron reflectometry measurements of CdA multilayers containing both layers of normal CdA and layers of deuterated CdA give evidence of rearrangement under water.1,10 In situ AFM work by Zasadzinski and co-workers3 has shown that when, left in the aqueous subphase, a single CdA monolayer anchored by tailsubstrate interactions or tail-tail interactions rearranges in a matter of minutes to create domains with three or even five layers. The driving force for this rearrangement is the maximization of favorable headgroup-headgroup interactions. No doubt we have some rearrangement here as well, and rearrangement of the pyrene-labeled layers would disturb some aggregates as well as creating a terraced surface for the multilayer. The results of Claesson and Berg20 suggest that escape of molecules into the subphase is unimportant when depositing under the conditions used here. Addition of a polyglutamate bilayer suppresses the mobility of the labeled bilayer even more than does the addition of a CdA bilayer. The initial value of the ratio Ie/Im is even lower for q/5/2/2-PG than for q/5/2/2-CdA, as shown by comparison of Figures 4 and 5. The initial monomer signal for q/5/2/2-PG (4.0) is the largest of all the samples. In fact, in the spectra for both q/5/2/2-CdA and q/5/2/2-PG a third monomer peak at about 420 nm becomes evident due to the lower relative strength of the excimer peak intensity in these samples. While, for multilayers of CdA only, one generally pictures the interfaces between aliphatic tails as welldefined, research in our group21 has demonstrated that there is some interdigitation of PG side chains with the tails of an underlying CdA layer. This effect is seen with both reflectometry and investigations of layer mechanical stability done with AFM.21 Merkel et al.22 have shown by way of lateral diffusion coefficient measurements in lipid bilayers supported on a solid substrates or monolayers that coupling to a neighboring monolayer can have a strong effect on lateral mobility. In particular, they find decreasing lateral mobility with increasing interdigitation between the lipid layer and a neighboring monolayer. The interdigitation reduces the mobility of the molecules, and thus the possibility for the molecules to form excimers during the excitation lifetime is limited. To summarize, as one compares the initial state of the three fundamentally different types of samples, the results suggest that the mobility of the labeled bilayer decreases from the uncapped sample to the CdA-capped and then PG-capped samples. Some of the change in Ie/Im between the uncapped and capped samples may, however, be due simply to an increased residence time in the subphase (20) Claesson, P. M.; Berg, J. M. Thin Solid Films 1989, 176, 157. (21) Tsukruk, V.; Foster, M. D.; Reneker, D. H.; Schmidt, A.; Wu, H.; Knoll, W. Macromolecules 1994, 27, 1274. (22) Merkel, R.; Sackmann, E.; Evans, E. J. Phys. (Paris) 1989, 50, 1535.

Langmuir, Vol. 12, No. 12, 1996 3019

Figure 6. Schematic of all the potential fates of monomer and aggregate species originally in the labeled bilayer: (a) sublimation of monomeric species; (b) sublimation as dimer; (c) exchange among various states of aggregation; (d) interdiffusion to another layer.

which provides the opportunity for layer rearrangement to reduce the frequency of pyrene-pyrene interaction. 2. Time Dependent Behavior. It is worth noting again that the change in fluorescence emission signal with time reflects the change in the pyrene-labeled molecules only; the matrix molecules, on the other hand, do not necessarily change in the same fashion. The room temperature aging of all four samples was followed by successively measuring the emission spectra at progressively greater time intervals. The first several spectra were collected at time intervals of several minutes (6-30 min), and the last ones were spaced apart by several hours (ca. 12 h). It is important to make four general observations about the significance of the changes in Ie and Im seen with time. First, since a fairly low concentration of pyrene-labeled molecules (0.1 M) is used for preparation of the sample, one may assume that the quantum yield for a given sample changes little in time. Thus, observed changes in Ie and Im, as well as changes in the ratio Ie/Im, reflect structural changes in the originally labeled bilayer and other layers into which the labeled molecules may have come through rearrangement or interdiffusion. Second, connections between the signals and multilayer structure must take into account all potential fates of isolated labeled chains and aggregates, as summarized in Figure 6. Isolated chains may move about within the multilayer or sublime. In the first case they continue to contribute to Im, whereas if some chains are lost, Im should be reduced, since the quantum yield for either the monomer emission or the excimer emission is assumed to be constant. In contrast, the movement of chains out of aggregates will cause a reduction in Ie, even if those chains do not leave the system. The word “aggregate” here includes both cases where two pyrene moieties are in direct contact and cases where they may not be in contact but are separated by such small distances that they can diffuse together during the lifetime of the excited state. Thirdly, it is imperative to recognize that it is the behavior of the pyrene-labeled molecules which is directly probed. The behavior of the CdA molecules themselves may be somewhat different, due to the differences in chemical structure and length between the two types of molecules. In fact, the locations of the labeled molecules actually represent defects in the CdA layer structure. Therefore, in seeking to minimize its free energy, the system may eliminate the labeled molecules at a rate faster than that at which it loses arachidic acid molecules. Finally, all changes in fluorescence emission during aging in air are assumed to be dictated by changes in the amount and distribution of the labeled molecules. Aging is seen, in the case of the q/5/2 sample, to cause a decrease in both Ie and Im. The decrease in Im may be

3020 Langmuir, Vol. 12, No. 12, 1996

Wu et al.

Figure 7. Graphical summaries of the time variations in monomer intensity, Im (solid circle), excimer intensity, Ie (open circles), and the ratio Ie/Im (open triangles) for all four samples. Shown with the data in the inset are lines corresponding to power law expressions which come closest to capturing the overall time dependence: (a) sample q/5/2; (b) sample q/5/2/0; (c) sample q/5/2/ 2-CdA; (d) sample q/5/2/2-PG.

seen qualitatively in Figure 2 and is summarized quantitatively in Figure 7a, which plots the decrease with time using both linear and logarithmic coordinates. The data for sample q/5/2 are unique in that they include information from times right after the deposition of the labeled bilayer (t ) 0), whereas the others begin sometime later due to further deposition. A reduction in Im of roughly a factor of three occurs in the first 20 h for sample q/5/2, and change continues to take place at an ever slower rate over the subsequent 200 h during which the sample is monitored. This description of the intensity change which suggests two separate regions of behavior is not so helpful as the linear plot may suggest, for on a log-log representation (inset in Figure 7a) it is difficult to clearly identify a single distinct shift in slope. In order to see if the data could be readily represented by a time dependence characteristic of, for example, a first-order process, an attempt was made to fit them using a simple exponential with a single characteristic time. Even when fitting only the data at “short” times or only at “long” times, exponential representations proved to be poor. Much better agreement was found with power law

Table 3. Values of Exponent (B)a Characterizing Overall Rate of Change Over Time Examined Im Ie Ie/Im

q/5/2

q/5/2/0

q/5/2/2CdA

q/5/2/2PG

-0.18 -0.39 (-0.38 -0.18)b -0.21 (-0.23 -0.14)b

+0.02 -0.14

-0.009 -0.14

-0.07 -0.21

-0.16

-0.13

-0.14

a Y ) AtB. b The exponents found for two different regimes, t < 20 h and t > 20 h.

expressions of the form Y ) AtB, where Y was Im, Ie, or Ie/Im. Fitting the curves in separate regions (e.g. t < 20 h and t > 20 h) appeared to have some merit in the case of q/5/2. For the other samples, such a distinction was difficult to defend. Values of B for power laws fit over the entire range of time available varied from +0.02 to -0.39 and are listed in Table 3 for purposes of comparison. No attempt was made to assign a mechanistic interpretation to these exponent values. The reduction in Im with time in sample q/5/2 may be clearly ascribed to pyrene-labeled molecules subliming

Temporal Stability of a Bilayer in LB Films

Figure 8. X-ray reflectivity curves taken from sample s/5/2. The first measurement (open circles) was done right after the last CdAP layer was deposited. The second measurement (solid circles) was done 50 h after the deposition. The slight difference in the reflectivity curve at high q is insignificant if the counting statistics are considered.

from the sample. However, there is more than one possible explanation for the drop in Ie seen in Figure 7a. One cause is movement of chains from aggregates into isolated states. Such a change by itself would increase Im while decreasing Ie/Im. The reduction in Ie may also be due in part to sublimation of chains directly from aggregates. While the loss from aggregates to the surroundings is a rare occurrence when the multilayer is under water, it will happen more readily in air. The rate at which Ie decreases is always larger than that for Im, and the ratio Ie/Im diminishes with time roughly as t-0.21, the fastest rate seen for any of the samples. Therefore, both the total concentration of labeled molecules and the fraction of molecules present as aggregates drop with aging of the sample. We are unable to distinguish whether molecules are escaping directly from aggregates or if they all become isolated first and then sublime. The overall structure of the first system was checked by X-ray reflectivity using a sample (s/5/2) that had the same layer structure as sample q/5/2 but was deposited on top of a silicon substrate. X-ray reflectometry measurements made at 12 h intervals during the first 50 h of aging for the sample s/5/2 showed no evidence of significant alteration of the overall multilayer structure (e.g. disordering) with time, as suggested by the selected reflectivity curves presented in Figure 8. Fitting of these X-ray reflectivity curves indicated that the scattering length density of the top bilayer could be estimated with a precision of about (3%, meaning that if there were a 3% mass loss from the top bilayer, the effects on the reflectivity curves would not be resolved in the range of q studied. In both the s/5/2 and the q/5/2 samples, the pyrene-labeled molecules constituted 3.7% of the total mass of molecules in the top bilayer. On the basis of the sensitivity of the XR measurement and a crude calculation, one may argue that perhaps at least one in four or five of the molecules leaving the sample is labeled. That is, there is a strong bias in favor of losing the labeled molecules. To do this estimation, one must assume that quantum yield remains constant with varying concentration (that is, Im varies linearly with isolated monomer concentration) and that the distribution among the excimer and monomer states does not vary markedly in time. (The last condition is apparently not quite met but is convenient.) If these assumptions are fulfilled, then the fractional loss in monomer signal is the same as the fractional loss in the mass of molecules participating in monomer emission. At one extreme is the case that all

Langmuir, Vol. 12, No. 12, 1996 3021

molecules lost are pyrene-labeled molecules. The ca. 60% drop in monomer emission over 50 h seen for sample q/5/2 would correspond to a loss of 60% of the molecules yielding monomer emission. In order to relate that fraction to a mass loss, one must assume some fraction of molecules fluorescing as monomers, which for the present purposes may be taken to be of order 0.5. In this case the loss of 1% of the mass of the top bilayer would be indicated, since the labeled molecules giving monomer fluorescence initially accounted for roughly 1.8% of that bilayer’s total mass. A 1% change in mass would not be discernible in the X-ray measurements done here. However, if three CdA tails leave with every labeled molecule lost, the estimated percentage of bilayer mass lost over 50 h becomes ca. 4%, just within the range observable with XR. Of course, if only a very small fraction of all labeled molecules are yielding monomer fluorescence, then these XR measurements would not be able to resolve the corresponding mass loss, even if 27 unlabeled tails were left with every labeled molecule. Study of sample q/5/2/0 indicates the extra time in the subphase affects its aging behavior as well as the initial structure. Im changes very little over 130 h, as shown in Figure 7b. Since excimer emission intensity does drop with time, changing approximately as t-0.14, one is inclined to suppose that the sublimation of chains from the monomer state is offset by the conversion of chains from the aggregate to monomer states. Ie decreases much less rapidly than it does for q/5/2, and the fraction of molecules in aggregates therefore is also dropping more slowly in this case than for q/5/2. Overall, the extra dipping cycle lowers the initial aggregation density, leads to a nearly constant monomer concentration, and results in a more slowly decreasing aggregation density. Capping with an additional bilayer of CdA or PG also alters the time dependent behavior of the labeled molecules in the multilayer, as may be seen by comparing Figure 7c and d with Figure 7a for q/5/2. The monomer signal changes very little in time for q/5/2/2-CdA, though it drops slightly overall and drops somewhat faster for q/5/2/2PG. However, as can be seen in Figure 8a, the rates of decay for Im for both capped samples are much lower than that for the uncapped sample q/5/2. When the CdA cap is in place, it appears that a nearly steady-state concentration of isolated monomers is maintained. The fact that Ie does drop with long times while Im remains nearly constant in the case of the CdA capping bilayer demonstrates that labeled molecules are lost with time despite the cap. These molecules are most likely escaping through the top CdA bilayer by exchange with molecules in that bilayer or by diffusion through defects in that bilayer. The overall rate of loss of molecules cannot be accounted for solely by loss from the perimeter of the sample. The rate at which molecules may find their way to the perimeter or to a given defect by lateral diffusion may be bounded on the high side using results from Merkel et al.22 They measured lateral diffusion coefficients of order 10-8 cm2/s in lipid bilayers supported on a solid substrate or monolayer. Since a CdA layer is less fluid than the lipid bilayers they considered, a Py-C10 molecule moves less than 0.038 cm in a 10 h period at room temperature. It will take a molecule in the middle of the sample thousands of hours to make it to the edge. The experiments suggest that labeled molecules sublime from all the samples, since there is always an overall loss of labeled species with time. The experiments are unable to distinguish whether the molecules sublime directly from aggregates or move from aggregates to the isolated state and then escape as isolated molecules. However, it is possible to conjecture about the mechanism on the basis

3022 Langmuir, Vol. 12, No. 12, 1996

Wu et al.

Figure 10. Comparison of the decrease in time in fraction of labeled molecules present as aggregates among the three samples q/5/2 (open circles), q/5/2/2-CdA (solid circles), and q/5/2/2-PG (solid triangles).

Figure 9. Comparison of the decrease in time in relative monomer emission (a) and excimer emission (b) among the three samples q/5/2 (open circles), q/5/2/2-CdA (solid circles), and q/5/2/2-PG (solid triangles).

of energetic considerations. There are two kinds of interaction in these LB films, the van der Waals interaction between the hydrocarbon tails and the electrostatic interaction between the headgroups. Depending on the relative strength of these two interactions, the molecules may sublime as monomers or dimers. If the van der Waals interaction is stronger, the molecules will continue to associate even as they leave the system. If the electrostatic interaction is stronger, the molecules will leave one by one. Surface force measurements by Claesson20 have shown that the energy for the Cd2+ head interaction is in the range 1.5-3 kT per molecule. Molecular simulations23 have suggested that the hydrocarbon interaction is around 3 kT as well. So, it is possible that, in a 100% CdA LB sample, there is both monomer and dimer sublimation. Differences in the molar cohesive energies of the arachidic tails and labeled molecule tails may be considered in the mixed layers. Molar cohesive energies can be measured experimentally and can be calculated using a group contribution theory.24 There is good agreement between the experimental value and theoretical calculation of 131 kJ/mol for arachidic acid without taking into consideration the COOH group. A group contribution calculation for Py-C10 results in a lower energy value of 88.9 kJ/mol without the COOH. This energy difference, along with the difference in geometry for the two types of chains, suggests that the mixed layer may be subject to more (23) Ulman, A. An Introduction to Ultrathin Organic Films: from Langmuir-Blodgett to Self-Assembly; Academic Press: San Diego, CA, 1991. (24) Israelachvili, J. N. Intermolecular and Surfaces Forces; Academic Press: New York, 1992.

sublimation and less order than a 100% CdA layer. It would also be expected that in a layer containing only CdA that dimer sublimation is more likely than in the case of the mixed layer due to the larger interaction strength for arachidic acid. The capping layers do not just hinder the sublimation process, they also continue to restrict the molecular mobility of the pyrene-tagged molecules. Figure 9 summarizes the change in Ie/Im ratio with time for the three systems. The pyrene molecules in sample q/5/2 are the least restricted, as indicated by the highest Ie/Im ratio for that sample. After the placement of the capping layer, the mobility of the molecules has been restricted, resulting in lower Ie/Im values. For sample q/5/2/2-PG, the restriction is more severe due to the fact that there is interdigitation of the PG side chains and the CdA tails. The relationships among these mobilities do not change markedly with time, as indicated by the parallel manner in which the Ie/Im curves decrease in time for the three samples. It is clear that a power law with a constant exponent cannot fit the change of Ie/Im with time for sample q/5/2, which suggests that there is more than one process going on at the same time. It may be that two characteristic times are present, a shorter one for the more mobile CdAP molecules in the layer next to air and a longer one for CdAP molecules in the next layer down. After molecules leave the top layer, escape of molecules from the second layer becomes more important in determining the rate of loss. The parallel behavior of the Ie/Im ratio curves seen in the log-log plot in Figure 10 indicates that the relative mobilities of the samples do not change in time. That is, the ratios (Ie/Im)A/(Ie/Im)B, where A and B are sample names, do not change. For example, sample q/5/2 always has the highest mobility. Conclusions Measurements using pyrene-labeled short chain amphiphiles have identified interesting differences among the “as-deposited” states as well as among the room temperature aging behaviors of multilayers containing a labeled bilayer atop five CdA layers. These variations result from the treatment of the multilayer subsequent to deposition of the labeled bilayer. The as-deposited labeled bilayer contains both isolated chromophores and chromophores sufficiently close to one another to fluoresce

Temporal Stability of a Bilayer in LB Films

as aggregates. The proportion of aggregates is reduced by subsequent dipping into a CdCl2 solution subphase or by deposition of a CdA or PG bilayer. When capping layers are added, at least some portion of the reduction in aggregate density may be ascribed to a reduction in mobility within the bilayer. The strongest perturbation occurs when the PG bilayer is added, due to interdigitation. Labeled chains escape from all the samples over time due to sublimation, although the sublimation rate is slowed when a capping layer is added. The successful hindrance of the molecular sublimation and the change in the

Langmuir, Vol. 12, No. 12, 1996 3023

molecular mobility are of importance for efforts at incorporating LB multilayers in real devices. Acknowledgment. H.W. would like to thank Dr. T. Vierheller and Dr. A. Schmidt for their helpful advice and encouragement. Research support from the National Science Foundation (Grant CTS 91-10110), The University of Akron, and a Sikka Fellowship (S.A.R.) is gratefully acknowledged. LA950807I