Chain Length and Solvent Effects on Competitive Self-Assembly of

Elizabeth C. Landis and Robert J. Hamers ... Christian Clausen, Junzhong Li, Jonathan S. Lindsey, Werner G. Kuhr, and David F. Bocian .... Reed , and ...
4 downloads 0 Views 857KB Size
Langmuir 1994,10, 1186-1192

1186

Chain Length and Solvent Effects on Competitive Self-Assembly of Ferrocenylhexanethiol and 1-Alkanethiols onto Gold Gary K. Rowe and Stephen E. Creager' Department of Chemistry, Indiana University, Bloomington, Indiana 47405 Received December 13,1993. I n Final Form: February 11,1994" Competitiveself-assemblyof ferrocenylhexanethiol(Fc-Cs-SH)and 1-alkanethiolsonto gold from ethanol and 1-hexanol solution is interpreted using a model based on coupled Frumkin isotherms. Alkanethiola with long alkyl chains are alwayspreferentiallyadsorbed relative to Fc-Ce-SH. In the limit of low ferrocene coverage,that preference is described quantitatively in terms of a parameter A(AG2 - AGd' that describes the incremental change per methyleneunit in the characteristic free energy differencecontrollingcompetitive adsorption. That incrementalchange is found to be -1.9 kJ mol-' per methylene unit increase in alkanethiol chain length for adsorption from ethanol and -0.8 k J mol-' per methylene unit increase for adsorption from 1-hexanol. These values are smaller than would be anticipated from independent estimates of the incremental change in cohesive energy of n-alkanesper methylene unit derived from vapor pressure,fusion enthalpy, vaporization enthalpy, and sublimation enthalpy data. Finally, data on ferrocene coverage as a function of exposure time to a coating solution indicate that monolayer composition can change over time during self-assembly, reaching a limiting composition only after exposure times of between several hours and several days. It is postulated that the preference for adsorption of long-chain alkanethiols is manifest primarily in the slow relaxation of an initially-formed disordered monolayer to a more ordered structure.

Introduction The ability to prepare multicomponent surface layers is an important aspect of the use of self-assemblychemistry to modify ~urfaces.l-~3 One example of this is the use of self-assembly to create surfaces of controlled hydrophobicity; to first order, any desired hydrophobicity can be achieved by coimmobilization of hydrophilic and hydrophobic adsorbates from mixed solutions.14 Another example is the immobilizationof biological macromolecules onto surfaces to prepare chemical sensors; a sensor might require sucha biological agent for activity but might suffer because of denaturation of the biological agent, steric crowding around a binding site, and/or nonspecific binding of components from the medium onto the sensor surface. These problems can be controlled by coimmobilizing the biological agent with a molecule designed to create a local microenvironmentfavorable to the operation of the device.7 A third example is the immobilization of redox-active molecules onto electrode surfaces; attempts to immobilize close-packed arrays of redox-active groups frequently produce disordered layers with poorly-behaved redox a Abstract published

in Advance ACS Abstracts, April 1, 1994.

(1) Laibinis, P. E.; Nuzzo, R. G.; Whitesides, G. M. J. Phys. Chem. 1992,96, 5097. (2) Folkers.. J. P.:. Laibinis. P. E.: Whitesides. G. M. Lanamuir 1992. 8, 1330. (3) Bain, C. D.; Evall, J.; Whitesides, G. M. J. Am. Chem. SOC.1989, 111,7155. (4) Bain, C. D.; Whitesides, G. M. J . Am. Chem. SOC.1989,111,7164. (5) Sanaesy, P.; Evans, S. D. Langmuir 1993,9, 1024. (6) Ulman, A.; Evans, 5. D.; Shnidman, Y.; Sharma, R.; Eilers, J. E.; Chang, J. C. J. Am. Chem. SOC.1991,113, 1499. (7) Spinke, J.; Liley, M.; Guder, H. J.; Angermaier, L.; Knoll, W. Langmuir 1993,9, 1821.

(8)Chidsey, C. E. D.; Bertozzi, C. R.; Putvinski, T. M.; Mujsce, A. M. J. Am. Chem. SOC.1990,112,4301. (9) Collard, D. M.; Fox, M. A. Langmuir 1991, 7, 1192. (10) Finklea, H. 0.;Hanshew, D. D. J. Electroanal Chem. 1993,347, 327. (11) Zhang,L.T.;Lu,T. B.;Gokel, G. W.;Kaifer,A.E.Langmuir 1993, 9,786. (12) Rowe, G. K.; Creager, S. E. Langmuir 1991, 7,2307. (13) Creager, S. E.; Rowe, G. K. J . Electroanal. Chem., in press.

chemistry, whereas mixed layers with redox-active molecules present at low coverage and diluted with a second, redox-inactive component, are well behaved and serve as excellent model systems for studies on interfacial redox chemistry.Sl3 Finally, mixed monolayers that incorporate molecular "gate" sites are being developed as a way of adjusting the selectivity of electrode reactions for possible applications in ele~troanalysis.1~17 A particularly attractive system for such studies is that of alkanethiols self-assembled onto gold.1a20 Whitesides and co-workers have studied the competitive adsorption of simple alkanethiols with differing chain lengths and terminal functional groups onto gold'" and silver21using ellipsometry, X-ray photoelectron spectroscopy, and contact angle measurements. They make the following points: (i) long-chain alkanethiols are always preferentially adsorbed relative to short-chain alkanethiols; (ii) alkanethiols with polar functional groups can be either preferentially adsorbed or preferentially discriminated against, depending on solvent; (iii) competitive adsorption is generally not well described by simple equilibrium expressions; and (iv) despite this, monolayer composition appears to be dictated by thermodynamic rather than kinetic factors. The preference for long-chain adsorbates has been explained in terms of greater cohesive interactions among long alkyl chains in the monolayer, as reflected (for example) in solubility differences between long-chain and short-chain adsorbates. The inadequacy of simple adsorption equilibrium treatments has been attributed to a positive excess free energy of mixing of the two (14) Bilewicz, R.; Majda, M. J. Am. Chem. SOC.1991,113,5464. (15)Chailapakul, 0.;Crooks, R. M. Langmuir 1993,9, 884. (16) Fujihira, M.; Araki, T. J. EZectroanaZ. Chem. 1986,206, 329. (17) Steinberg, S.; Rubinstein, I. Langmuir 1992,8, 1183.

(18)Ulman, A. An Introduction to Ultrathin Organic Films. From Langmuir-Blodgett to Self-Assembly; Academic Press, Inc.: San Diego, CA 1991. (19) Bain, C. D.;Whitesides, G. M. Angew. Chem.,Int.Ed. Engl. 1989, 28,506. (20) Dubois, L. H.; Nuzzo, R. G. Annu.Reu. Phys. Chem. 1992,43,437. (21) Laibinis, P. E.; Fox, M. A.; Folkers, J. P.; Whitesides, G. M. Langmuir 1991, 7, 3167.

0743-7463/94/2410-1186$04.50/00 1994 American Chemical Society

Competitive Self- Assembly

components in the monolayer. Put another way, a preference for self-interactions (e.g., long-chain with longchain and short-chain with short-chain) over non-selfinteractions (eg., long-chain with short-chain) exists in the monolayers.2 Thermodynamic control of monolayer composition is inferred from the sensitivity of composition to solvent, chain length, and terminal functionality. The detailed mechanism of alkanethiol competitive adsorption on gold is complex, however, and involves intermediate states of variable composition and order that relax on a relatively long time scale.8?9*2&26 We report here the results of a study on competitive adsorption of ferrocenylhexanethiol (FC-C~-SH) with a series of n-alkanethiols of variable chain length. The ferrocenylalkanethiol/alkanethiolsystem is of particular interest as a model system for fundamental studies on the nature of long-range interfacial electron transfer.eJ2J3 It also has the advantage that incorporation of ferrocene into mixed monolayers can be quantitatively assessed with good precision using standard electrochemical methods such as cyclic voltammetry. This system differs from those studied by Whitesides and co-workers in that one of the monolayer components (Fc-Cs-SH) has a terminal substituent that is considerably larger than the projected area of an alkyl chain. Monolayers that incorporate significant quantities of this component are expected to be relatively disordered, since the ordering and tight packing of alkyl chains that typify alkane-based self-assembled systems will be disrupted. This can complicate interpretation of the competitive adsorption behavior. We will apply a physical model based on competitive (i.e., coupled) Frumkin adsorption isotherms to the interpretation of competitive adsorption data acquired in ethanol solvent as part of an earlier study12 and to new data acquired in 1-hexanol solvent as part of the present study. This model accounts explicitly for interaction energies of the monolayer components with themselves and with each ~ t h e r .In~ the ~ ~limit ~ ~ of high overall coverage but low Fc-Cs-SH coverage, the competitive adsorption process can be described in terms of a difference in characteristic free energies of adsorption for each pair of coadsorbates. That difference is found to vary in a regular way with the chain length of the alkanethiol coadsorbate. Solvent effects on the self-assembly process demonstrate that the energetics of solvation can moderate the dispersive forces in the monolayer that drive preferential adsorption of long-chain adsorbates. Finally, new data on the kinetics of formation of mixed monolayers by competitive self-assembly reveal that a relatively disordered monolayer with a composition close to that of the coating solution forms initially but then relaxes to yield a more ordered monolayer with a composition that reflects more strongly the forces that drive the preferential adsorption of long-chain adsorbates.

Langmuir, Vol. 10, No. 4, 1994 1187 was prepared from reagent grade HNOS(EM Science) and HCl (Baker). ”Piranha” solution used to clean glassware was made from regent grade HzSO4 (Baker) and 30% HzOz (Fisher). CAUTION Piranha solution reacts violently with most organic materials and must be handled with extreme care. 1-Butanethiol, 1-hexanethiol, 1-octanethiol, 1-decanethiol, and 1-dodecanethiol (Aldrich) were reagent grade and used without further purification except for 1-hexanethiol and 1-octanethiol, which were passed through activated alumina prior to use. Ferrocenylhexanethiol (Fc-Ce-SH) was prepared as described previously.12 Data on competitive adsorption from ethanol were taken from our earlier publication.12 New data on competitive adsorption from 1-hexanol and on the kinetics of competitive adsorption were acquired using gold disk electrodes prepared by sealing gold wires in epoxy.% The geometricelectrode area was estimated to be 0.011 cm2by manually measuring the electrode radius from a scanning electron photomicrograph (CambridgeStereoscan 250 Mk 11) of a sealed electrode. Electrodes were polished and then etched with dilute aqua regia (31:6 HCkHNOs:H20) in a quartz test tube at room temperature for 1min prior to immersion in the coating solution.26 Coating solutions were prepared in volumetric glassware cleaned by heating for 1 h in “piranha” solution ( 7 3 concentrated Has04 to 30% HzO2 at 90“C)followed by exhaustiverinsing with distilled water then ethanol, and drying in a glassware oven. Solutions containing two thiols were prepared in glass volumetric flasksbydilutingstocksolutions (5-50mM). Theaccuracy of the concentrations of stock solutions was limited by the analytical balance (Scientific Products Model SP 180) used to weigh the thiols; the error in concentration is f5%. The transfers were carried out in glass syringes in air. The transfer procedure may introduce a further error of up to *l%in the final composition of the coating solution. The total concentration of thiol in coating solutions was always 1.0 mM. Estimates of ferrocene and alkanethiolate coverage were obtained by integration of slow-scan cyclic voltammograms (0.1 V s-l) acquired in nitrogen-purged aqueous solutions of methanesulfonicacid or perchloric acid (0.1 M, for ferrocene oxidation) or potassium hydroxide (0.5 M, for alkanethiolate reductive desorption) using a conventional three-electrode cell with a platinum wire auxiliary electrode and a Ag/AgCl/sat’d KCl reference electrode. Electrochemical data were acquired with an EG&G PAR Model 362 scanning potentiostat and a Yokogawa Model 3023 x-y recorder.

Results and Discussion Adsorption of Ferrocenylhexanethiol and 1-Alkanethiols from Ethanol. Figure 1A presents a series of four adsorption isotherms (Le., plots of monolayer composition us coating solution composition) for competitive adsorption of Fc-Cs-SH and 1-butanethiol, 1-hexanethiol, 1-octanethiol, and 1-decanethiol from ethanol onto gold. These data were first presented in ref 12 and are replotted here for convenience since we will be undertaking a more detailed data analysis in the present work than was attempted in the earlier work. The dotted line indicates the behavior expected if the adsorption was completely ideal, i.e. with no preferential adsorption at Experimental Section any coverage. It is clear that alkanethiols that are more than six carbons long are preferentially adsorbed (the Electrolyte solutions were prepared from distilled deionized curves fall below the dotted line) relative to Fc-Cs-SH, water (Barnstead Nanopure) and either perchloric acid (Mallinckwhereas alkanethiols that are less than six carbons long rodt), methanesulfonic acid (Aldrich), or potassium hydroxide are not (the curve falls above the line). The general trend (Baker). Coating solutions were prepared from either reagent grade ethanol (MidwestGrain) or 1-hexanol(Baker). Aqua regia in these data that the adsorbate with the longest alkyl chain is always preferentially adsorbed is in keeping with (22) Bain,C.D.;Troughton,E.B.;Tao,Y.-T.;Whitesides,G.M.;Nuzzo, the observations of Chidsey and co-workerse for related R. G. J. J.Am. Chem. Soe. 1989,111, 321. ferrocene-containing mixed monolayers and with those of (23) Biebuyck, H. A.; Whitesides, G. M. Langmuir 1993,9,1766. Whitesides and co-workers1-4for electroinactive systems. (24) Hahner, G.; Woll, C.; Buck, M.; Grunze, M. Langmuir 1993, 9, 1955. Preferential adsorption of long-chain species has been (26) Creager, S. E.; Hockett, L. A.; Rowe, G. K. Langmuir 1992,8,854. attributed to dispersive forces in the monolayers, which (26) Soriaga, M. P.; Song,D.; Zapien, D. C.; Hubbard, A. T. Langmuir 1985,1,123. (27) Parsons,R. J. Electroanal. Chem. 1964,8, 93.

(28) Creager, S. E.; Rowe, G. K. Langmuir 1993,9,2330.

Rowe and Creuger

1188 Langmuir, Vol. 10, No. 4,1994 5 ,

,

I

0.0

‘FcCBSH

,

0.20 -

I

0.6

0.4

0.2

0 25

!



(‘FcC8SH

0.8 +

I

1.0

‘RW)

,

,

1.0

15

* B

00

0.5

2.0

C w w 1 CRsH

Figure 1. (A) Adsorption isotherms (Le., ferrocene coverage on the electrode, rFe, us fraction of Fc-Cs-SH relative to total thiol in the coating solution, C F ~ H , / ( C F+~ H CWH))for mixed monolayers of Fc-Cs-SH and 1-alkanethiols on gold prepared from ethanol solution: ( 0 )1-butanethiol; ).( 1-hexanethiol;(A) 1-octanethiol;(v) 1-decanethiol. Data from ref 12. (B)Plots of rFJ(rT- rpc) us C F ~ ~ C using W H data from Figure 1A and rT = 1.1 x 1O-g mol cm-2. See text for details.

should be greatest for long-chain adsorbates, and to a decrease in solubility of 1-alkanethiols in ethanol with increased chain length. The data presentation in Figure 1A provides qualitative insight into the relative affinities of Fc-Ce-SH and 1-alkanethiols for the surface; however it is not readily amenable to quantitative analysis. A more quantitative analysis can be obtained by invoking the formalism of competitive Frumkin isotherms.26 The relevant expressions are developed in Appendix 1;however for the specific case in which the sum of the surface coverages of components 1 and 2 approaches a specific limiting combined surface coverage (Le, conditions of high total surface coverage that would obtain for mixed monolayers of alkanethiols on gold), eq 1 applies

In this expression, Xlg& and X S , ~represent ~S the mole fractions of components 1and 2 on the surface, C1 and C2 the concentrations of components 1and 2 in the medium from which adsorption takes place, AGoa&,l and AGoaa,2 the characteristic free energies that describe adsorption of components 1 and 2 from solution in the absence of lateral intermolecular interactions, and an, a22, and a12 the self- and mixed-interaction energies among components in the monolayer. Component 1 is Fc-Cs-SH and component 2 the 1-alkanethiol in the present treatment.

Of particular interest is the situation when the Fc-Cs-SH coverage is very low; then, the final term in the exponential in eq 1becomes small and the magnitudeof the exponential term becomes approximately independent of Fc-Cs-SH coverage. This is the regime in which we have worked. We have not attempted to distinguish further between the adsorption free-energyterms A G o d , land AGoa&,z and the interaction terms a12 and a22 in the first term of the exponential in eq 1;rather, we have combined them into one global, characteristic adsorption free-energy difference (AG2 - AG1)’ that includes all the interactions that determine surface composition at low ferrocene coverage. At high total surface coverage, the ratio ( X I ~ ~ J X ~ ~ ~ S ) is simply given by I’l/(I’T - Fl), where rl is the coverage of Fc-Cs-SH and r T the total combined coverage of FcC&H and alkanethiol. Inspection of eq 1then reveals that a plot of Fl/(rT - I’l) vs C1/C2 will yield from its slope the term exp[(AGz - AGl)’/RTI for a given pair of vs adsorbates. (Note that a linear plot of rl/(I’T C1/C2 is diagnostic of the final term in the exponential in eq 1being small.) When (AG2 - AG1)’ is zero, the plot will have a slope of 1; a slope that differs substantially from 1can be quantitatively interpreted in terms of a difference in characteristic adsorption free energies for a given pair of coadsorbates. Figure l b presents four such plots prepared using the data in Figure l a and a maximum mol combined limiting surface coverage rT of 1.1 X cm-2 obtained via the reductive desorption method of Porter and co-workersmfor a monolayer of 1-hexanethiol. This value for rT is the most appropriate indicator of the maximum combined limiting surface coverage of thiolate under conditions of low ferrocene content in the monolayers. All the plots of Figure l b are approximatelylinear, indicatingthat eq 1applies over the range of concentrations and coverages included in the figure. Slopes for different Fc-C&H/alkanethiol pairs range from 2.0 to 0.016 as one or the other of the two components is preferentially incorporated into the monolayer. The voltammograms in Figure 2 illustrate the discrepancy between the maximum combined limiting surface coverages derived from ferrocene and thiolate electrochemistry. Ferrocene oxidation and alkanethiolate reductive desorption are both thought to be ,one-electron processes, yet integration of the charge for ferrocene oxidation in a Fc-Cs-SH monolayer yields a surface coverage (4.5 X 10-l0 mol cm-2; Figure 2, top) that is considerably smaller than that obtained by integration of the charge for reductive desorption of a hexanethiolate ~ ; 2, bottom). Porter monolayer (1.1X 10-Bmol~ m -Figure and co-workersattributed this discrepancy to the ferrocene moieties being too large to adopt the tight packing geometry characteristic of alkanethiolate monolayers without bulky substituents.m~30We have focused on the regime where Fc-Cs-SH is present only in small quantities; monolayers in this regime resemble pure alkanethiolate layers much more than they resemble pure Fc-Cs-SH monolayers, therefore use of the larger maximum limiting surface coverage obtained from alkanethiolate reductive desportion data is appropriate. Recognizing that each slope in Figure l b yields a term containing the difference in characteristic adsorption free energies (AG2 - AG1)’ for a different pair of coadsorbates, it is evident that a plot of RT ln(s1ope from Figure 1B) us chain length for the series of 1-alkanethiol coadsorbates (29) WalczaL, M.M.; Popenoe, D. D.; Deinhammer, R. S.;Lamp,B. D.; Chung, C.; Porter, M. D. Langmuir 1991, 7,2687. (30) Widrig, C. A.; Chung, C.; Porter, M. D. J. Electrooml. Chem. 1991,310,335.

Competitive Self-Assembly

I

20000

0.10 pA

2 3

18000 16000 -

$ 14000 p!

p 12000 I

I

I

0.0

I

I

0.2

I

0.4

I

0

I

0.6

$

10000 -

C iJ

8000 -

.5

T 0.25PA

I6000 4000 2000

J

I

-1.4

I

I

I

-1 2

I

-1.0

I

I

I

I

I

I

1

I

-0.8

E (V vs. Ag/AgCI)

Figure 2. (A) Cyclic voltammogram (0.1 V s-1,O.l M perchloric acid) showing ferrocene oxidation in a monolayer of Fc-C&H. Integration of the forward peak gives r F c = 4.5 x 10-10 mol cm-2. (B) Cyclic voltammogram (0.1 V s-l,0.5 M potassium hydroxide) showing reductive desorptionfor a monolayer of hexanethiolate. Integration of the forward peak gives rheunabiol.te = 1.1 X 1 V mol cm-2. 4000 2000

I -

0h

a

-g 5

-2000 -

v)

Y

-4000-

4

$

-6000-

\

-8000-

0

-10000-12000

The value of -1.9 kJ mol-' per methylene group corresponds essentially to an incremental increase in the cohesive energy in the monolayer relative to that of the individual monolayer components in solution with increasing chain length of one of the coadsorbates. The significanceof this value is most readily appreciated when it is compared with similar estimates of the incremental change in cohesive energy per methylene group for some related physical systems. One such system is the vaporization of liquid n-alkanes. The vapor pressure of a liquid at a given temperature is given by exp[-PG,dRTl where AGvap is the free energy of vaporization of the liquid. A plot of RT ln(vapor pressure) vs chain length for a series of n-alkanes will therefore reveal from ita slope the incremental change in the free energy of vaporization per methylene group increase in alkane chain length. The free energies of vaporization are determined primarily by intermolecular forces within the liquid; therefore such a plot will reflect the incremental increase in cohesive forces within the bulk liquid as the alkane chain length is increased. Figure 4 presents such a plot using vapor pressure data at 298 K for the series of n-alkanes from butane to n ~ n a n e . ~ The l * ~linearity ~ of the plot is excellent, and ita slope reveals an incremental effect on the vaporization free energy of -3.06 kJ mol-' per methylene group increase in alkane chain length. We consider the agreement of this value with our value of -1.9 kJ mol-' obtained from alkanethiol self-assembly data from ethanol to be quite good considering the disparate nature of the two physical processes involved. The direction in which the two values disagree is also easily rationalized; chain length should have a greater effect on vaporization than on competitive adsorption from solution since cohesive forces in the vapor phase are minimal whereas cohesive forces in solution (Le., solvation forces) could moderate the dispersive forces in the monolayer that drive the preference for adsorption of long-chain adsorbates. Further support for this rationalization is presented below in the form of a solvent dependence of the competitive adsorption process.

I 3

I

4

I

5

I

6

I

7

8

I

9

I

I

1

0

1

1

Carbon number Figure 3. Plot of RT ln(s1opes from Figure 1B) us l-akanethiol carbon number for competitive adsorption of Fc-Cs-SH with l-alkanethiols from ethanol. The slopes from Figure 1B were determined from a least-squares fit to the data. Temperature was 298 K. Slope = -1.9 kJ mol-' per carbon. will reveal from ita slope the incremental effect of changing the l-alkanethiol chain length on the relative adsorption free energies, i.e. the A(AG2 - AGd' per methylene unit change in l-alkanethiol chain length. Figure 3 presents such a plot, prepared using the slopes from Figure 1B and taking the temperature as 298 K. The slope of Figure 3 corresponds to a A(AG2 - AG1)' value of -1.9 kJ mol-' per methylene group increase in alkanethiol chain length. The negative sign is consistent with the expectation that adsorption of alkanethiols should become more favorable as chain length increases.

(31)Timmermans, J. Physico-Chemical Constant8 of Organic Compounds; Elsevier: New York, 1960;Vol. 1. (32) Smith, D. B. Thermodynamic Data for Pure Compounds. Part A: Hydrocarbow and Ketones; Elsevier: New York, 1986.

Rowe and Creager

1190 Langmuir, Vol. 10, No. 4, 1994 Other estimates of the incremental effect of increasing chain length on interaction energies among alkyl chains have been derived from the heats of fusion, vaporization, and sublimation of n-alkanes. Billmeyer cited values for an incremental change in fusion enthalpy of -3.86 kJ mol-’ per methylene group, a change in vaporization enthalpy of -4.09 kJ mol-’ per methylene group, and a change in sublimation enthalpy of -7.68 kJ mol-’ per methylene group increase in chain length.33 The latter value is in good agreement with an estimate from theory of the magnitude of the dispersion energy, Le. the van der Waals interaction energy, anticipated for a single methylene group in a paraffin crystalsMThese values are considerably larger than those estimated from Figure 3 using data for competitive adsorption of alkanethiols onto gold and from Figure 4 using vapor pressure data for n-alkanes. One probablereason of this is that the self-assembly and vaporpressure data include the effects of entropy on going from a dispersed phase to an ordered phase, whereas the heats of alkane fusion, vaporization, and sublimation do not. This highlights the fact that both enthalpic and entropic factors must be considered to fully describe the competitive adsorption process. Immobilizing a long-chain adsorbate from solution into a monolayer film involves a loss of entropy which is only partially counterbalanced by the stabilizing dispersive forces in the monolayer. The loss of entropy on immobilization may be greater for longer-chain adsorbates than for short-chain adsorbates, both because they have more degrees of freedom to lose and because they tend to form more ordered layers. Another factor contributing to the discrepancy between the self-assembly data and the n-alkane data may be that the self-assembled monolayers are not truly in equilibrium with the coating solution. We will show below that monolayer composition changes over relatively long time scales during the self-assembly process, so there is no guarantee that the data in Figures 1and 3 fully reflect the tendency for long-chain adsorbates to become preferentially adsorbed. Adsorption of Ferrocenylhexanethiol and 1-Alkanethiols from 1-Hexanol. The treatment above provides a quantitative, physically well-grounded interpretation of the preferential adsorption of long-chain alkanethiols. One insight that this interpretation provides is that the preference for long-chain adsorbates is partly aresult of dispersive forces in the monolayer being greater than those in the solution. We therefore reasoned that it should be possible to moderate those forces by adjusting the nature of the solvent from which the monolayers were formed. There is precedent for this in the work of Whitesides and co-workers,4 who noted that preferential adsorption of docosanethiol (C22H6SH) over dodecanethiol (ClzH26SH) is greater when self-assembly was performed from ethanol than when it was performed from isooctane solvent. It was thought that ismtanemoderates the preference for long-chain adsorbates since the chemical potentials of the alkanethiols should be more similar to one another in isooctane than in ethanol. Another way of thinking about this is that in an alkane-like solvent the solvation forces are similar to the forces in the monolayer that drive preferential adsorption of long-chainadsorbates; the energetic advantage of immobilization is moderated, so there is less of a preference for adsorption of longchain molecules. Experimental confirmation of this prediction in our system would serve to further validate the physical model that we have used to analyze our data. It would (33)Billmeyer, F. W.J. Appl. Phys. 1957,28, 1114. (34)Salem, L.J. Chem. Phys. 1962,37, 2100.

5 ,

I

0.0

I

I

I

I

0.2

0.4

‘FC€SSH

0.6

I (‘FeC6S-I

+

0.8

1.0

‘RSH)

1

B

0.00

I

i

A

. .

L

0.0

0.5

‘FcCBSH

1 .o

1.5

I ‘RSH

Figure 5. (A) Adsorption isotherms for mixed monolayers of Fc-C&H and 1-alkanethiolsprepared from 1-hexanolsolution: ( 0 )1-hexanethiol;(m) 1-octanethiol;(A)I-decanethiol. (b)Plota of rFJ(rT - rF3 us C F ~ ~ using C ~dataHfrom Figure SA and rT = 1.1 x W mol cm-2.

furthermore demonstrate how an adjustable system parameter, namely the nature of the coating solvent, could be used to control the composition of a redox-active organized assembly. Figure 5 presents data for competitive adsorption of Fc-C&H and 1-hexanethiol, 1-octanethiol,and 1-decanethiol from hexanol solvent, plotted in both of the formats used in Figure 1to present the data for adsorption from ethanol. Preferential adsorption of long-chainalkanethiols over Fc-C&H is evident in Figure 5 4 however, the deviation from the “ideal” dotted line is not nearly as great as it was for adsorption from ethanol. Plots of rl/(rTrl) vs C1/C2 are linear, and a plot of R T ln(s1opes from Figure 5B)vs alkanethiol chain length (Figure 6) reveals from its slope an incremental effect on the difference in adsorption free energies (Le., a A(AG2 - AG1)’ value) of -0.8 kJ mol-l per methylene group increase in alkanethiol chain length. The small magnitude of this value is consistent with the postulate that solvation forces in hexanol are sufficiently similar to those among the various components in the monolayers to help moderate the preference for adsorption of long-chain alkanethiols. Kinetics of Forming Mixed Self-Assembled Monolayers. The present model has implicitly assumed that the monolayer is in equilibrium with the solution and that the free-energy and interaction energy terms in eq 1are those associated with the thermodynamics of bringing absorbates from the bulk of solution to the electrode surface. The kinetics of adsorption and/or desorption are implicitly assumed to play only a minor role in dictating monolayer composition. The microscopic mechanism by which mixed self-assembled monolayers form is more

Competitive Self-Assembly

Langmuir, Vol. 10, No. 4, 1994 1191

-2000 -2500

0

-3000

n

2

-3500

-4000

W

C

$

4500

k

-5000 -5500

8000 8500

5

6

7

8

10

9

11

Carbon number Figure 6. Plot of RTln(s1opes from Figure 5B) us l-alkanethiol carbon number for competitive adsorption of Fc-C&H with l-alkanethiols from ethanol. Temperature was 298 K. Slope = -0.8 kJ mol-' per carbon.

5 -al I

1.5

-

1.0 -

A

t

0

0.0 0

,

2.0

1000 2000 3000 4000 5000

I

I

I

B 1.5 al

1.0

6 UU 0.5

0.0

0

loo0

2000

3000 4000 5000

Time (min.)

Figure 7. Ferrocene coverage us immersion time in the coating solution for a gold electrodes coated in (A) an ethanol solution containing0.50mM Fc-C&H and 0.50 mM l-dodecanethiol and (B)an ethanol solution containing 0.25 mM Fc-Ce-SH and 0.75 mM l-hexanethiol.

complex than this simple assumption would suggest, however, and kinetic factors almost certainly play a role in dictating the nature of the monolayers that are ultimately f ~ r m e d .We ~ ~therefore ~ ~ ~ sought to examine in more detail the question of thermodynamic us kinetic factors in determining monolayer composition for the FcCe-SH/alkanethiol system. Figure 7 presents data on ferrocene coverage us timeof-immersionfor two mixed monolayers,one preparedfrom Fc-Cs-SH and l-dodecanethiol and another from Fc-Ca-

SH and l-hexanethiol from ethanol solution. Each curve was acquired by immersing a suitably pretreated electrode in a coating solution, then periodically removing it to measure the ferrocene coverage electrochemically (after rinsing with ethanol), then reimmersing it in the coating solution (again after rinsing) to continue the experiment. A control experiment in which voltammograms were acquired at a series of electrodes that had been immersed for the desired time without periodic removal to determine ferrocene coverage revealed similar behavior, indicating that electrochemical interrogation of the monolayer does not disrupt the monolayer formation or relaxation processes. The composition of both monolayers changes with time as the electrodes are continuously exposed to the coating solution. Relaxation of the monolayer formed from Fc-Cs-SH and dodecanethiol is particularly strong, involving a decrease in ferrocenecoverage to a limitingvalue that is less than one-third of the initial value. Relaxation occurs in a different direction for each monolayer but occurs over approximately the same time scale in both cases, with monolayer compositions becoming independent of time after approximately 2 days of exposure to the coating solution. We postulate that these data reflect initial formation of a relatively disordered monolayer, the composition of which is determined primarily by the composition of the coating solution. Similar proposals have been put forth by several workers for single-component monolayers.22*u The direction of the change in ferrocene coverage is always such that the composition of the initially-formed monolayer is closer to that of the coating solution than is that of the monolayer obtained after a long relaxation time. Intermolecular forces that drive preferential adsorption of long-chain adsorbates have relatively little effect on the composition of the initially-formed monolayer; this suggests that the composition of the initially-formed monolayer is dictated by the kinetics of adsorption and/ or of transport to the electrode surface rather than by conventional adsorption free energy considerations. Relaxation to a final monolayer composition occurs as adsorbates undergo slow exchange with thiols in solution and the monolayer becomes more ordered.**9~~*~ It follows that the forces that drive preferential adsorption of longchain adsorbates are manifest primarily during the relaxation phase of monolayer formation and not in the initial stages of monolayer formation. In support of this postulate, Figure 8 presents two voltammograms for mixed monolayers of Fc-Cs-SH with l-hexanethiol recorded immediately after immersion in the coating solution (top) and after 67 h in the coating solution (bottom). The interfacial capacitance is much larger in the first voltammogram than in the second, as indicated by the larger double-layer charging current near 0 V. This indicates that the initially-formed monolayer is more easily penetrated by solvent and/or electrolyte ions than is the final monolayer. This trend is followed for all the monolayers that we have studied, at both high and low ferrocene coverage. (The multiple peaks in the voltammogramsin Figure 8 are thought to reflect structural disorder in the monolayers caused by steric crowding of the relatively large ferrocene g r 0 u p s , 8 , ~different ~ * ~ ~ total microenvironments for ferrocenes in different regions of the monolayers,%and double-layer effects.% Voltammograms at lower ferrocene coverage show no such multiple peaks, even though the trend with respect to interfacial capacitance after very short immersion times is preserved.) ~~

~

~

(36)Rowe, G.K.;Creager, S.E.J. Phys. Chem., in press. (36)Smith, C. P.;White, H.S.Anal. Chem. 1992,64,2398.

1192 Langmuir, Vol. 10, No. 4, 1994

Rowe and Creager

reflects the tendency for long-chain adsorbates to become incorporated into the monolayer.

Acknowledgment. The authors are grateful to the National Science Foundation (CHE-9216361)for financial support of this work. Appendix 1. Competitive Two-Component Adsorption with Interactions Included Competitive adsorption of two components from solution onto a solid surface can be described in terms of two independent adsorption free energies, suitably modified to consider interaction energies among the Mathematical expressions for these adsorption free energies are as follows:

0.0

0.2

0.4

0.6 0.8 E (V vs. Ag/AgCI)

Figure 8. Cyclic voltammograms (0.1 V s-I, 0.1 M perchloric acid) of mixed monolayers of Fc-C&H and hexanethiolon gold after 1 min (top) and 67 h (bottom) immersion in an ethanol solutioncontaining0.25mM Fc-Ce-SHand 0.75 mM hexanethiol.

This observation supports the postulate that the monolayers that are formed initially are disordered and that subsequent relaxation processes produce ordered monolayers that are more able to exclude ions and solvent from the electrode surface.

Summary This study has revealed the following aspects of the competitive self-assemblyof Fc-Ce-SHand 1-alkanethiols onto gold (i) alkanethiols with chain lengths longer than six carbons are preferentially adsorbed relative to Fc-CeSH; (ii) the preference for long-chain adsorbates can be described quantitatively in terms of an incrementalchange per unit increase in alkanethiol chain length in the difference in characteristic adsorption free energies for alkanethiols and Fc-Ce-SH; (iii) this incremental change is smaller than would be predicted from estimates of the incremental change in cohesive energy per unit increase in chain length for liquid and solid n-alkanes derived from vapor pressure, fusion enthalpy, vaporization enthalpy, and sublimation enthalpydata; (iv)the preference for longchain adsorbates is greater for adsorption from ethanol than for adsorption from hexanol solvent; (v) monolayer composition changes on a time scale of hours to days as the monolayer relaxes from a relatively disordered structure that forms initially to a more ordered structure that

The terms AGoah,l and AGoad,2 represent adsorption free energies of components 1 and 2 in the absence of interactions among adsorbates, all and a22 represent selfinteraction energies for components 1and 2, a12 represents the interaction energy between components 1and 2, and XI& and X2& represent the mole fractions of components 1and 2 adsorbed onto the solid surface. These expressions may then be inserted into equilibrium expressions that describe adsorption of 1 and 2:

(A-2a)

(A-2b) The terms Xl,l and XzpOl represent the mole fractions of components 1 and 2 in the solution from which the monolayer is formed. Equation A-2 is essentially identical to the Fnunkin adsorption isotherm for two adsorbates.mJ' At high total coverage, the term X2,hmay be replaced by (1 - XI#&);making this substitution, replacing mole fractions in solution with concentrations, and dividing eq A-2a by eq A-2b then yields eq A-3

X14h(all+ a22- 2u12)l/RZ"j(A-3)