J . Phys. Chem. 1992, 96, 5220-5222
5220
Quantitative Analysis of Thermodesorption of Langmuir-Blodgett Monolayers P. Tippmann-Krayer and H. Mohwald* Universitat Mainz. Inst. f. Physikalische Chemie, Welder Weg 1 1 , 06500 Mainz, Germany (Received: March 16, 1992)
Thermal desorption of Langmuir-Blodgett (LB) monolayers is studied for fatty acid monolayers with different chain lengths (magnesium palmitate, C16;magnesium arachidate, C20)and for phospholipid (DMPE) monolayers with two hydrocarbon chains. In all cases thermodesorption follows first-order kinetics. The difference of binding energiesof C16 and C20 corresponds to calculated differences of van der Waals interaction energies. The value of lOI3 s-' for the frequency factor k in an Eyring ansatz is typical for molecules which desorb out of the fluid phase of the monolayer (e&, DMPE monolayer). The very high frequency factor k = loz3s-I for fatty acid monolayers can be understood by Eyring's theory for activated complexes. In this case desorption occurs from a solid state via an intermediate "fluid" state of high entropy.
Introduction Studies of thermal desorption of Langmuir-Blodgett (LB) monolayers give information about lateral interaction energies between amphiphilic molecules and about binding energies between lipids and substrate. These parameters are important for the thermal stability of LB films. In order to understand these interactions, it is important to perform systematic experiments varying specific molecular groups and phases from which desorption occurs. Thermodmrption was observed by measuring thickness changes of the monolayer via optical reflection (interference-enhanced reflection (IeR)') when heating the sample with a constant rate j3. We studied magnesium palmitate (C16)and magnesium arachidate (C20)monolayers to analyze the influence of the length of hydrocarbon chains on lateral interactions. Additionally, we inspected phospholipid (DMPE) monolayers to learn about the effects of two hydrocarbon chains on thermal stability. Experimental Details To measure thickness changes, we used the simple optical technique IeR:'v2 the intensity change of specularly reflected light (A = 633 nm, incident angle = exit angle = 70') with perpendicular polarization depends linearly on the change of monolayer thickness when using a Si/Si02 substrate with 1700-Aoxide thickness. The Si/Si02 substrate was heated from the backside via a gold film serving as heater electrode. The temperature of the wafer was measured by a thermocouple and calibrated by substances with known melting point. The coverage is determined by IeR as described before. All samples had been heated twice. This was necessary because for temperatures above 150 OC there were changes for the reflected intensity due to twisting of the pure substrate. The coverage was evaluated by the difference of reflected intensities between the first and second heating cycle. All monolayers were prepared by the conventional dipping technique.3 The lipids (palmitic or arachidic acid, L-a-dimyristoylphosphatidylethanolamine (DMPE); Sigma, Munich) were spread without further purification from a chloroformmethanol (3:1,Merck, Darmstadt, p.a.) mixture onto a Lauda film balance (Lauda FW 1, Lauda-Konigshofen). For fatty acid monolayers the subphase consists of a solution of M MgC12 and 10 mM Na2HP04/NaH2P04buffer (pH = 8.3) in Millipore filtered water (conditions for maximal binding of Mg2+to COOof lipids). For DMPE monolayers we used only Millipore filtered water as subphase. Monolayer transfer was performed at a constant surface pressure of 30 mN/m for C16 and C20 and 35 mN/m for DMPE at a dipping speed of about 6 mm/min at room temperature. The substrates were pieces of thermally oxidized silicon wafers (Si02 thickness 1700 A; Wacker-Chemie, Burghausen). They were cleaned for several hours in concentrated chromosulfuric acid, methanol, and Millipore water before being dried at elevated temperatures.
-
-
0022-3654/92/2096-5220$03.00/0
TABLE I C16 c20 DMPE
E , kJ/mol 161 185 151
k , s-I 4 x 1022 4 x 1023 3 x 1014
Results Figure 1 shows the coverage of the substrate versus temperature when linearly increasing the temperature with the same rate 0 (j3 = 0.7K/s). The desorption spectra of Figure 1 give T, = 81 OC ( T , is the temperature at which desorption rate is maximal) for C16 and T , = 116 OC for C20. This means that thermal stability is increased by going to longer hydrocarbon chains for identical head groups of the lipids. Two hydrocarbon chains per lipid increase T , considerably: T , = 228 OC for DMPE. These observations are in good agreement with former e~periments.~ It has to be mentioned that within the same preparation (identical subphase) thermodesorption spectra are identical (AT, I A2 "C). Different subphases at nominally identical preparation conditions lead to AT, 5 f5 OC. Within these uncertainties no differences in thermodesorption spectra had been obtained when cleaning the Si/Si02 wafer with different methods: with Hellmanex, with chromosulfuric acid, plasma cleaning, or UV illumination. First-order kinetics is described by5 -de = k8 exp(-E/RT)
dt where 8 is the surface coverage, E is the binding energy of the lipids or the activation energy of desorption, k is the frequency factor or the rate constant, and R is the universal gas constant. For a linear change of sample temperature with time ( T = To + Pt) and for E independent of 8, eq 1 is solved to find the temperature T,,, at which the desorption rate is maximal: = -exp ART, P
(),: --
This means that for first-order kinetics E and k can be determined independently by varying j3 and plotting 1 / T , versus In T,2)5 (Arrhenius plot): the slope ( = - E / R ) of the straight line gives the binding energy E; the intersect with they axis (=ln (kR/E)) yields the frequency factor k . The Arrhenius plots for C16,C20,and DMPE monolayers are shown in Figure 2. The linear dependence of the measuring points demonstrates the correct assumption of first-order kinetics. From the linear plot we obtained the values of Table I. The accuracy for E amounts to f4 kJ/mol, for k f 1 order of magnitude. Table I clearly shows that as in the case of the DMPE monolayer a higher temperature T , does not necessarily result in a higher binding energy E. A conclusion from T , on E is only valid for the same frequency factor k. Interpretation For fatty acid monolayers the frequency factors are of the same order of magnitude. The differences of binding energies are due
(a/
0 1992 American Chemical Society
The Journal of Physical Chemistry, Vol. 96, No. 13, 1992 5221
Letters
is about unity), kB is the Boltzmann constant, h is the Flanck constant, and AS is the molar entropy of activation. Assuming K = 1, Table I gives AS = 206 J/(K mol) for magnesium arachidate. This value agrees very well with the increase of entropy for the melting of cadmium arachidate.I0
p=0,7K/s
Temperature [ “ C ]
Figure 1. Coverage as a function of temperature at a constant heating rate j3 (j3 = 0.7 K/s) for magnesium palmitate (C16), magnesium arachidate (C20), and DMPE monolayers on a Si wafer with about 1700-A oxide thickness. -10.01,.
,
I
,
,
,
I
,
,
,
I
,
,
1
I
,
-11,ol -12.0
-15.0 -16.0
,
\
\
\
,
I
.
,
,
j
-
1 \+
1j, \
-17.0 0.0018
0.002
,
0.0022
,
,
I
,
0.0024
,
,
,
\I
i,
0.0026
,
,
0.0028
,
0.003
1 ITm
Figure 2. I/T, ( T , = temperature at which the desorption rate is a maximum) versus In (j3/Tm2)for magnesium palmitate (C16), magnesium arachidate (CZO), and DMPE monolayers on Si/Si02 wafers.
to different chain lengths. The van der Waals interaction energy Wfor two saturated hydrocarbon chains can be calculated with6 W = -1.24 X 103(N/D5) [kcal/mol]
(3) where N is the number of CH2 groups per chain and D [A] the distance between two chains. The fatty acid chains are laterally packed in a hexa on with a lattice spacing of d = 4.1 1 A.’ This leads to D = 4.7 and therefore to an increase of van der Waals interaction energy of 3 W ( N = 1) = 6.4 kJ/mol per additional CH2 group. Thus, the experimentally determined increase of binding energy going from C 16 to C20 is in very good agreement with the theoretical value of 25.6 kJ/mol. The increase of E is mainly due to the increase of van der Waals interaction energy for lipids with different chain lengths and identical head groups. The decrease of the binding energy of the DMPE monolayer in comparison to the fatty acid monolayers cannot be explained by van der Waals interaction of the chains because the head groups are not at all comparable. The frequency factor k is very much different for fatty acid monolayers and for the DMPE monolayer. This can be explained by desorption of the molecules out of different phases of the monolayer: For temperatures above 60 OC the hydrocarbon chains of DMPE are melted; a fluid monolayer remains on the solid support.8 A frequency factor on the order of 10” s-’ is as well expected if the transition state is assumed to be a two-dimensional gas and the adsorbate layer is fully delocalized (mobile in the statistical ~ e n s e ) . The ~ fatty acid molecules desorb out of the solid state of the chains. Calorimetric measurementsIO and experiments on cadmium arachidate multilayen” prove that the chains of cadmium arachidate do not melt before 110 OC. As shown in Figure 1, the fatty acid monolayers desorb far below 110 OC. The very high frequency factors can be explained by Eyring’s theory for activated complexes.12 For a first-order reaction proceeding through an activated transition state k can be written as kbT k = Kexp(AS/R) (4) h where K is called the transmission coefficient (in many cases it
x
Discussion These experiments have shown that it is necessary to determine the binding energy E and the frequency factor k by analyzing the thermodesorption spectra for different heating rates to obtain quantitative information on thermal stability of LB monolayers. The frequency factor k is very much different for molecules desorbing out of the solid or fluid phase of the monolayer. Our interpretation for the dependence of the magnitude of k as well holds for the thermodesorption experiments of LB multilayers studied by mass spectrometry.” For 20 layers of pure arachidic acid extremly high values (k SKI)have been determined, whereas for 6 monolayers of cadmium arachidate k = 7 X 10l2 s-I is reported. Desorption starts at very low temperatures (-55 “C) for the pure arachidic acid multilayer where the multilayer is in the solid state. The cadmium arachidate multilayer only desorbs at about 200 OC where the lipids are in the fluid state. (The melting point for cadmium arachidate is known to be at 110 OC.10) In ref 14 isothermal desorption experiments using ellipsometry are reported. The binding energy of stearic acid evaporated on a nickel foil amounts to 185 kJ/mol, in good agreement with our observations (see Table I). This value is interpreted as the sublimation energy, as the sum of the vaporization energy (106.8 kJ/mol) and the fusion energy (68.7 kJ/mol), so that these authors as well come to the conclusion that the lipids desorb out of the solid phase. This argumentation as well holds for the binding energies we observed for fatty acid monolayers. Only the values for the vaporization energies (100 kJ/mol for palmitic acid and 108 kJ/mol for arachidic acid”) are too small to explain the binding energies of Table I. But it has to be mentioned that in ref 14 the values for E of the lowest stearic acid monolayer are much smaller than ours: 152 and 128 kJ/mol for a nickel and a platinum surface. On the other hand, the value for a stearic acid monolayer on silver, reported in ref 15, is very similar to the one we observed: E = 181 kJ/mol, but k 1030 S-I. In ref 17 a quartz crystal microbalance was used to study desorption of LB layers. The differences of the frequency factors between fatty acid and phospholipid monolayers have as well been noticed, but absolute values of E and k are smaller than the ones in Table I: E = 129 f 8 kJ/mol, k = lOI9 s-l for a stearic acid monolayer; E = 42 f 7 kJ/mol, k = lo9 s-l for a DPPA monolayer. It is not possible to compare E and k we determined for magnesium arachidate with the values for cadmium arachidate (E = 81 kJ/mol, k = 6 X lo9 SKI) and magnesium arachidate (E = 76 kJ/mol, k = 6 X lo9 s-l) in ref 1. These discrepancies can be explained by the difference of preparation: pH 5.5 instead of pH 8.3. At pH 5.5 a huge amount of the lipids is not dissociated; these lipids exist as pure acid molecules. The melting point for the pure acid is much lower than for the fatty acid salt, so that the low values for the binding energy and the frequency factor can be explained by a desorption out of the fluid phase of the monolayer. Altogether these results show that binding energies as well as frequency factors are to a large extent preparation dependent, but there are general trends and features which can be understood within classical theories and which now have to be elaborated in more detail to understand the stability of organic films.
-
References and Notes ( I ) Laxhuber, L. A.; Rothenhausler, B; Schneider, G.; Mohwald, H. Appl. Phys. A 1986, 39, 173. (2) Tippmann-Krayer, P.; Mohwald, H.; Schreck, M.; Gopel, W. Thin Solid Films, in press.. (3) Kuhn, H.; Mobius, D. Angew. Chem. 1971.83, 672. (4) Laxhuber, L. A.; Mohwald, H. Langmuir 1987, 3, 837. (5) Redhead, P. A. Vakuum 1962, 2, 203.
J . Phys. Chem. 1992, 96, 5222-5224
5222
(6) Salem, L. J . Chem. Phys. 1962, 37, 2100. (7) Kjaer, K.; Als-Nielsen, J.; Helm, C. A,; Tippmann-Krayer, P.; Mohwald, H. Thin Solid Films 1988, 159, 17. (8) Kenn, R.; Tippmann-Krayer, P.; Mohwald, H.; Kjaer, K.; Als-Nielsen, J . Hasylab Report, 1988, p 251. (9) Menzel, D. Interactions on Mefal Surfaces; Gomer, R.. Ed.; Topics in Applied Physics, Vol. 4; Springer: Berlin, 1975; p 101. (10) Naselli, C.; Rabolt, J. F.; Swalen, J . D. J . Chem. Phys. 1985, 82, 2136.
(1 1) Tippmann-Krayer, P.; Mohwald, H.; L’vov, Yu.M. Langmuir 1991, 7, 2298. (1 2 ) Atkins, P. W. Physical Chemistry, 3rd ed.;Oxford University Press: London, 1986; p 747. (13) Schreck, M.; Schier, H.; Gopel, W. Lnngmuir 1991, 7 , 2287. (14) Pimbley, W. T.; MacQueen, H. R. J . Phys. Chem. 1964,68, 1101. (1 5) Lederer, E. L. Seifensieder-Ztg. 1930,57, 67. (16) Spink, J . A. J . Colloid Interface Sci. 1967, 24, 61. ( 1 7 ) Jones, J. P.; Webby, G. M. Thin Solid Films 1989, 178, 21 1.
Substrate Induced Ordering of Molecular Adsorbates on Au( 111) Joachim Hossick Schott,*vt Claudia R. Arana,* Hector D. Abruiia,*Helen Hurrell Petach,s C. M. Elliott,§ and Henry S. White*,+ Department of Chemical Engineering and Materials Science, University of Minnesota, Minneapolis, Minnesota 55455, Department of Chemistry, Baker Laboratory, Cornell University, Ithaca, New York 14853, and Department of Chemistry, Colorado State University, Fort Collins, Colorado 80523 (Received: March 5, 1992; In Final Form: May 7, 1992)
Adsorbed monolayers of [R~(bpy)~(bpy-(CH,);bpy)]~+ (x = 4 and 5 ) on the unreconstructed 4 3 X 22 reconstructed surfaces of Au( 1 1 1) were imaged by scanning tunneling microscopy in dimethylformamide. On the reconstructed surface, the adsorbed films display highly ordered patterns that are commensurate with the measured corrugation patterns of the substrate. In contrast. when adsorution is on the atomicallv smooth unreconstructed surface, STM images reveal a random spatial distribution
We report scanning tunneling microscopy (STM) in dimethylformamide (DMF) of monolayers of [Ru(bpy)’(bpy(CH2)x-bpy)]2f (I) (bpy-(CH,),-bpy is an a,o-bis(4’-methyl2,2’-bipyridyl-4-yl)alkane,and x, the number of methylene groups in the alkane chain, is either 4 or 5) adsorbed on the unreconstructed and 4 3 X 22 reconstructed surfaces of Au( 11 1). The I = [ Ru(bpy)zbpY.(CHz)x-bpy l z +
key finding of our study is that the observed supramolecular structures of electroactive films of I (for x = 4, 5 ) on the 4 3 X 22 reconstructed surface are commensurate with the long-range corrugation of the underlying substrate. In contrast, long-range ordering is absent in films of I adsorbed on the unreconstructed and atomically flat Au( 11 1) surface. The results suggest that the adsorption of I, through interactions of the pendant bipyridyl group with the Au atoms, is modulated in real space with a periodicity of -30 A. Several previous STM studies have focused on the specific orientation and/or location of adsorbates with respect to the substrate’s atomic ~ t r u c t u r e . ~The - ~ present work demonstrates that variations in the electron density at a surface, which occur as a result of a reconstruction, can dictate the structural features of an adsorbate layer. The study of Au( 11 1) has received a great deal of attention from STM researchers in recent years. Atomically flat, clean surfaces can be prepared in ultrahigh vacuum (UHV) and in ambient environments, and atomic corrugation can be measured in UHV,5 air,5 and liquids6 A particularly interesting feature
’Universit)
of Minnesota
I Cornell University f
Colorado State Lniversity
of the Au( 11 1) surface, which will play a major role throughout this report, is its reconstruction. Au( 111) is the only faceantered cubic (fcc) (1 11) metal surface known to reconstruct, as established by a number of surface sensitive However, the exact atomic positions within the unit cell of the reconstructed surface and the new long-range features of this reconstruction were revealed only recently in an ultrahigh vacuum STM study by Bart et a1.I0 Haiss et aL6 subsequently demonstrated that this information could also be obtained with the sample being exposed to air or organic solvents. Generally, reconstructions result in surface layer atomic and electronic distributions which differ, sometimes markedly, from an ideally bulk terminated lattice. Thus, it is expected that the adsorption and nucleation behavior of monolayer quantities of adsorbed species may reflect these variations. Indeed, Chambliss et al.” demonstrated in a recent STM study that submonolayer quantities of Ni adatoms nucleate on the reconstructed Au( 111) surface in a pattern which reflects topographical features of the underlying reconstruction. The 4 3 X 22 phase can be obtained by thermal annealing of the Au( 11 1) surface or by a tip-induced electronic transition.I2 Figure 1 shows an STM image of the 4 3 X 22 reconstruction obtained in our l a b o r a t ~ r y . ~The ~ . ~reconstructed ~ surface is imaged in STM as parallel pairs of corrugation lines with a horizontal pair-to-pair separation of -65 A and a vertical corrugation amplitude of -0.2 A. The spacing between corrugation lines within a pair is 20-25 A. Barth et al. have shown that each corrugation line on the bare, reconstructed surface marks a transition between hexagonal close packed and face centered cubic stacking regions. Parallel line pairs can either extend over several hundreds of angstroms, as shown in Fi ure 1, or they can be organized in domains, separated by -250 and rotated by f120° with respect to each other, yielding a zigzag “herringbone” pattern (not shown here). Similar surface features have been observed by many groups in various sample environments: UHV,l0 air>’* polar organic solvents,6 and e1ectrolytes.I6 In contrast, images of the unreconstructed Au(ll1) surface show the normal hexagonal arrangement of Au atoms (2.9-A spacing), but none of the features described above. Figure 2 shows images of Au( 11 1) surfaces after adsorption of I from DMF solutions. Unreconstructed and reconstructed Au surfaces were prepared by thermal treatment as described in
0022-3654/92/2096-5222$03.00/00 1992 American Chemical Society
1
