Langmuir 1985, 1, 45-52 45

Spontaneously Organized Molecular Assemblies. 1.

Formation, Dynamics, and Physical Properties of
n -Alkanoic Acids Adsorbed from Solution on an Oxidized
Aluminum Surface
David L. Allara*
Bell Communications Research, Murray Hill, New Jersey 07974

Ralph G . Nuzzo*
AT&T Bell Laboratories, Murray Hill, New Jersey 07974

Received August 16,1984

This study shows that closest packed, oriented monolayers of n-alkanoic acids can be formed on oxidized
aluminum substrates by adsorption from dilute solution. The formation of these organic surface phases
is characterized by complicated kinetics in which surface and/or monolayer defects, as well as impurities,
seemingly play important roles. The structures so obtained are dynamic in nature in that they undergo
rapid exchange with ligands in solution. The data further suggest an important limitation to self-assembly
in this system which is directly related to the length of the n-alkanoic acid tail. The influences of experimental
variables on the formation of oriented monolayers is discussed.

Introduction of quasi-two-dimensionalorganic assemblies and the pro-

The adsorption and spontaneous organization of am-
phiphilic molecules, especially the n-alkanoic acids, on
smooth, ambient' metal surfaces, are the subject of a vast
l i t e r a t ~ r e . ~ - 'The
~ interest in these surface adsorbate
structures traditionally has been motivated by their rele-
vance to such important technological areas as lubrication,
corrosion, and catalysis. In recent years, speculative
technological applications, involving such areas as optics
and microelectronics,16 as well as the development of
surface-modified electrodes for electrochemical applica-
' ~ generated a renewed interest as to the nature
t i o n ~ have
(1)The term ambient is used to indicate that moat attention has
centered on metals having oxide overlayers which may or may not have
contained various levels of impurities and/or contaminants.
(2)A review of earlier work can be found in: Bowden, F. P.; Tabor,
D."The Friction and Lubrication of Solids", Oxford University Press:
London; 1968; Part 11, Chapter 19.
(3)(a) Chapman, J. A.; Tabor, D. R o c . R. SOC.1967,London, Ser. A
242, 96-107. (b) Brockway, L. 0.; Karle, J. J. Colloid Sci. 1947, 2,
277-287. (c) Menter, J. W.; Tabor, D.R o c . R. SOC.London, Ser. A 1961,
204,514-524. (d) Sagiv, J. J. Am. Chem. SOC.1980,102,92-98.
(4)Walker, D.C.; Ries, H. E. Adv. Chem. Ser. 1964,No. 43,295-301.
(5)Doyle, W. P.; Ellison, A. H. Adu. Chem. Ser. 1964,No. 43,268-274.
(6)Gaines, G. L. J. Colloid Sci. 1960,15, 321-339.
(7)Timmons, C.0.; Zisman, W. A. J. Phys. Chem. 1965,69,984-990.
(8)Timmons, C. 0.; Patterson, R. L.; Lockhart, L. B. J. Colloid In-
terface Sci. 1968,26, 120-127.
(9)Bornong, B. J. Surf. Sci. 1969,16,321-330.
(10)Allara, D.L.; Tompkins, H. G. J . Colloid Interface Sci. 1974,49,
410-421.
(11)Boerio, F. J.; Chen, S. L. J. Colloid Interface Sci. 1980, 73,
176-185.
(12)Brown, N. M. D.; Floyd, R. B.; Walmsley, D. G. J. Chem. SOC.,
Faraday Trans. 2 1979,75,261-270.
(13)Lewis, B. F.; Mosesman, M.; Weinberg, W. H. Surf. Sci. 1974.41,
142-164.
(14)Golden, W. G.; Snyder, C.; Smith, B. J. Phys. Chem. 1982,86,
4675-4678.
(15)Cass, D.A,; Strauss, H. L.; Hansma, P. K. Science (Washington,
D.C.)1976,192,1128-1130.
(16)For example, see: (a) Vincett, P. S.; Roberts, G. C. Thin Solid
film 1980,68,135-171.(b) Special issue on molecular electronics: Proc.
IEEE 1983,230,197-263.
(17)For example, see: Faulkner, L. R. Chem. Eng. News 1984,28-45
and references therein.
cesses by which they can be prepared. As regards the latter
point, spontaneous adsorption methodologies are to be
contrasted with the Langmuir-Blodgett transfer process'*
in which ordered, multilayer assemblies can be produced
for the longer chain acids &e., fatty acids), particularly
when the total number of carbon atoms is greater than 18.
In the case of spontaneous adsorption, no preassembling
of the adsorbate molecules is carried out; the substrate is
exposed directly to the adsorbate in solution or the gas
phase and any ordering that occurs is thus considered to
be spontaneo~s.'~The questions of and related to the
adsorption mechanism(s) (reflecting kinetics) and equi-
librium structures (reflecting thermodynamics)are of great
interest, but past studies have been intended to concen-
trate almost exclusively on the latter. It is not clear in all
cases, however, whether true equilibrium structures, com-
prised solely of the intended adsorbate(s), in fact have been
studied.
Most studies of the adsorption of carboxylic acids on
metals have involved surfaces exposed to air prior to ad-
sorption and thus largely refer to adsorption on native
oxide overlayers. These studies, which largely complement
the voluminous data on bulk oxide powders, can be divided
into three general functional classes: (1)acids chemisorbed
to form metal carboxylate salts, processes that may involve
substrate corrosion and multilayer formation; (2) acids
chemisorbed via proton transfer to a lattice oxygen atom
to yield monolayer structures; (3) acids chemisorbed
(perhaps physisorbed is a more appropriate description)
with no proton transfer. Representative examples of these
three are carboxylic acid chemisorption on copper,6,10J1
aluminum,'2-16 and siliconFOrespectively. Although the
cases of multilayer formation are quite important and of
significant interest, the less complicated cases of monolayer
chemisorption are more attractive as beginning points for
(18)For example, see: Gaines, G. L. 'Insoluble Monolayers at Liq-
uid-Gas Interfaces"; Interscience: New York, 1966.
(19)As we shall detail, spontaneous, in this context, cannot be used
interchangeably with the word instantaneous.
(20)Iler, R. K. "The Chemistry of Silica"; Wiley: New York, 1979;
Chapter 6.

0743-7463/85/2401-0045$01.50/0 0 1985 American Chemical Society

46 Langmuir, Vol. 1, No. 1, 1985
careful structural studies of the first formed layer.
A concern with regard to many of the studies cited
above, in particular those that provide contrasts with
Langmuir-Blodgett techniques, is whether the term self-
assembling is in fact appropriate. These concerns focus
on two issues. First, are monolayers with oriented (or
perhaps ordered), closest-packed structures of low surface
free energy indeed formed? Second, to what extent do the
various experimental variables-solvent, temperature,
carbon-chain length, substrate, and concentration as rep-
resentative examples-influence the nature of the surface
phases that are formed?
The classic study of Timmons and Zisman7concluded
that for chain lengths of C14 and longer, close-packed
monolayers can be prepared on Pt; on NiO substrates,
chain lengths as low as Clo were successfully employed. An
earlier study of Zisman21 presented the interesting ob-
servation that the coefficient of friction of a glass surface,
covered with a homologous series of fatty acids, drops to
a minimum (and nominally constant) value for chain
lengths of C14or greater. Borno~$ observed that coverages
of fatty acids (C6--c30)adsorbed from the melt on chro-
mium substrates were less than that expected for a
Langmuir-Blodgett-type film; no discontinuities of the
type described above were observed. Various studies of
specific longer chain acids (CI6and above) seem to offer
conflicting results as to whether submono-, mono-, and/or
multilayer coverages are obtained for ambient metal sur-
faces such as copper416J1and aluminum.6J4 For example,
Gaines reported that, for the latter, greater than monolayer
coverages were obtained for stearic acid (C1J in sharp
contrast to the results of Golden, Snyder, and Smith14
which show submonolayer coverages for arachidic acid
(C20). The literature is repleat with such discrepancies
despite an impressive array of experimental techniques
(electron diffraction, reflection infrared spectroscopy, ra-
dioisotopic labeling, etc.) which have been brought to bear
on this problem. Thus, while many additional comparisons
can be made, no general conclusions can be drawn. These
continuing significant and puzzling differences, in our
opinion, reflect the need for a detailed, self-consistent
reexamination of this classic and widely reported chemi-
sorption system.
It has been our contention22that spontaneous adsorp-
tion, under the appropriate conditions, out to be capable
of producing monolayers with structural features akin to
those of the Langmuir-Blodgett type. This contention is
not new to us, having been suggested by previous authors%3
and recently broadly demonstrated by Hubbard et al.23*24
for the adsorption of nonamphiphilic molecules (differing
structurally from the type reported herein) on metal
surfaces. In the present and the following paperz5 we
report the results of our detailed reexamination of the
adsorption of carboxylic acids on native aluminum oxide
surfaces. From reflection infrared spectroscopy,z6ellip-
sometry, isotopic labeling (both 2H and 3H), X-ray
photoelectron spectroscopy (XPS), and contact angle
~~~ ~ ~~ ~
(21)Zisman, W.A. In "Friction and Wear"; Davies, R., ed.; Elsevier:
New York, 1959;pp 110-148.
(22)Nuzzo, R. G.;Allara, D. L. J. Am. Chem. SOC.1983, 105,
4481-4483.
(23)Soriaga, M. P.;Hubbard, A. T. J. Am. Chem. SOC.1982,104,
3937-3945 and references cited therein.
(24)Hubbard, A. T.Acc. Chem. Res. 1980,13, 177-184.
(25)Allara, D.L.;Nuzzo, R G. Langmuir, following paper in this isaue.
(26)For discussions of the application of infrared spectroscopy to the
determination o f alkyl chain orientation and ordering in monolayer
structures, see: Allara, D. L.; Swalen, J. D. J . Phys. Chem. 1982,86,
2700-2704. Rabolt, J. F.; Burns, F.C.; Schlotler, N. E.; Swalen, J. D. J.
Chem. Phys. 1983,78,946-952.
Allara and Nuzzo
wetting data, an improved understanding of the structure
and dynamic character of the self-assembly of n-carboxylic
acid monolayers on aluminum has been obtained. Our
results suggest an extremely important role for kinetics
in the formation of equilibrium closest packed structures.
Through an examination of both a homologous series of
high-purity acids (C,-c24) and vinyl- and propargyl-ter-
minated C22acids, the influence of adsorbate structure on
monolayer structure has been detailed. Our results further
demonstrate that these monolayer structures are capable
of undergoing extremely facile exchange reactions with
materials in solution. Taken together, these results develop
a rich though complicated coordination chemistry for
carboxylic acids chemisorbed on this (and presumably
other) oxide surface.
Experimental Section
Materials. The n-alkanoic acids from C6 and above were
obtained from Sigma Chemical Co. and were used without further
purification. GC analyses were obtained for the liquid acids and
showed purities of >99.9%. The melting points of the higher acids
were checked and all showed melting ranges of less than 2 "C.
The terminally substituted acids CH2=CH(CH2)lgC02H and
CH=C(CH2),,C02H were synthetic samples supplied by Prof.
George Whitesides, Harvard University, and had melting points
of 70.0-70.5 and 78.5-79.5 "C, respectively. Hexadecane, obtained
from Aldrich Chemical Co., had a nominal purity of 99%. It was
further purified by slow elution through a column of activity 1
alumina and stored under nitrogen. Deuterated alkanoic acids
were obtained from Cambridge Isotope Laboratories (Cambridge,
MA) and were used without further purification. Acetic acid was
obtained from J. T. Baker Chemicals as the Ultrex grade (99.9%)
and was used as received. Aluminum used for the substrate
preparation was 99.999% purity. Tritiated acetic anhydride was
obtained from New England Nuclear Co. (Boston, MA).
Sample Preparation and Treatment. Aluminum substrates
were prepared by the evaporation of aluminum from resistively
heated tungsten boats onto 2-in.-diameter single-crystal silicon
wafers which had been polished to C40-A surface roughness.
Depositions were carried out in the lo4 to torr range in an
oil diffusion pumped system equipped with a liquid nitrogen trap
assembly. After melting the aluminum, a shield was removed,
thus exposing both the samples and a quartz crystal thickness
monitor which was placed nearby in order to determine the
thickness of the deposited film. Approximately 200 nm of A1 was
deposited and the chamber then backfilled with high-purity re-
search-grade oxygen to 1 atm. The samples were removed and
treated in one of two ways: (1) the samples were examined quickly
by ellipsometry and, within several minutes, placed into the
adsorbate solution; (2) the samples were placed directly into an
ethanol solution of acetic acid (0.05 M) for -15 min, removed,
rinsed with pure ethanol on a spinner (Headway photoresist type),
examined by ellipsometry, and then placed in the adsorbate
solution. The acid solutions were made by dissolving the weighed
portion of acid in hexadecane contained in a Pyrex flask which
had been precleaned in ethanolic KOH followed by repeated water
and methanol rinses. For the acids C18 and above, solution only
could be obtained by warming the contents to -35-40 "C. The
solutions were transferred to wide-mouth, screw-top polypropylene
containers, precleaned with chloroform and methanol washes, and
maintained a t 25.0 f 0.3 "C during sample immersion. The
solutions of acids CISand above tended to remain supersaturated
at 25 "C although occasionally one would form crystals and require
rewarming. Following immersion, the samples were removed and
the surface rinsed quickly on the spinner with HPLC grade
hexane.
Infrared Measurements. IR spectra of adsorbates were taken
by reflection using a Digilab 15B spectrometer with modified
optics. Details are given elsewhere.25
Ellipsometry Measurements. Ellipsometric measurements
were made using a Rudolph 423 null ellipsometer equipped with
a Babinet-Solet compensator set a t 7r/4 retardation and with the
-
optical axes at 45" to the instrument verical axis. Measurements
were made with an 1-mm beam at 623.8 nm (He-Ne laser) and
Spontaneously Organized Molecular Assemblies. 1 Langmuir, Vol. I , No. 1, 1985 47
at 442.0 nm (He-Cd laser) using a 70" angle of incidence at the
sample. Little difference was seen, however, in the results at the
two wavelengths, and for convenience,most measurements were
done at 632.8 nm. Spatial variations across the sample were - II
usually within the experimental errors of m-fl.5 A in terms of I
the final film thicknesses calculated from the readings. I
Contact Angle Measurements. Static contact angles were 30 - 0 I 0
measured with a Ram6-Hart goniometer. Drops were placed on
the surface in the ambient environment and read several times
on both sides of the drop. Angles were read to -&loand were
reproducible from sample to sample to <f3".
Radioisotopic Labeling. Tritiated acetic anhydride (5 mCi;
50.0 mCi/mmol) was dusolved in a sufficient quantity of absolute 20 -
ethanol under nitrogen to give a specific activity of acetic acid
(arising from esterolytic cleavage) of 25.0 mCi/mmol with a
concentration of 0.0050M. Sampleswere incubated by contacting
only the aluminum surface with the radiolabel. Calculations were 0 8.1
based on the measured activity of aliquots of known volume which 10 - 0

t
1
I
had been removed from the exchange or sample digestion media.
Aquasol-2 scintillation cocktail (NEN) and a Beckman 300
scintillation counter were used in all experiments.
Photoemission Measurements. A Kratos XSAMSOO pho-
- . a 'i
8
',e
I

toemission spectrometer utilizing a hemispherical analyzer and I

operating in a fixed analyzer transmission mode was employed 0 I , I I I I
for data acquisition. The instrumental resolution was 1 eV, using
1-mm mechanical apertures in conjunction with a Mg Ka X-ray
source. Digital data were acquired by using 0.05-V steps. The
base operating pressure was <5 X lo-" torr. CARBON NUMBER
Figure 1. Ellipsometry data showing the limiting f i i thicknesses
Results of n-alkanoic acids adsorbed on an oxidized aluminum substrate
Calculation of Film Thicknesses from Ellipsometry [RC02H]= 0.0050 M in n-hexadecane, 25 "C.
Data. In order to transform the raw data of the ellipso-
metry measurements into film thicknesses, the standard
parallel layer, classical electromagnetic model was cho-
sen.25p27 In this model one assumes that the system is
reasonably described as a flat sample consisting of an in-
finite substrate with a parallel overlayer of thickness d.
All phases are considered to be of homogeneous compo-
sition and isotropic, with the optical behavior of each phase
being accurately described by a single dielectric function.
A more complete description of this approach is given
elsewhere,25but specific relevant details are given here.
The measured analyzer and polarizer angles, A and P ,
respectively, are related to the complex electric fields of
incident and reflected light via the following equations:
6
a
/o:
/ 0
rp/r, = tan $eiA 0
0
.8

$ = A(10) (3) /: 0
a
a

+
A = 2P ( x / 2 ) (4) /
/
m a
a

/
The El's are the complex electric fields of the incident (+) ' a
and reflected (-) light beams of p and s polarization. The /
terms rp and rs are the standard Fresnel coefficients and
a
are functions of both the amplitude ratio tan $ and the
optical phase shift A. These latter parameters are calcu-
0 5 10 15 20 25
lated directly from the raw ellipsometry data. From the
theory mentioned above, r /rs can be related to the optical CARBON NUMBER
properties of the ample!^^^^^ With this approach it is Figure 2. Ellipsometry data showing the film thicknesses of
possible to determine a complex refractive index which n-alkanoic acids adsorbed on an oxidized aluminum substrate:
describes a multilayered sample as a pseudo-single-phase [RC02H]= O.oOo50 M in n-hexadecane,25 "C. Incubation times
are as follows: ( 0 )2, (m) 28, (0)
51, (V)73, (A)384 h. The dotted
material. In the case of a "clean" substrate, this provides line represents the best fit to the data in Figure 1for n-alkanoic
a reference value. For the same substrate, but with an acids of 12 carbons or greater.
overlayer of refractive index, iio = no (assuming no ab-
sorption term for the overlayer at the probe wavelength), calculating no for monolayer samples are such that it is not
the thickness and no in principle can be calculated by possible to routinely obtain reliable values of both no and
numerical iteration techniques. In practice, the errors in d. For this reason, and because the actual value of no for
a close-packed monolayer differs only slightly from the
(27) Aspnes, D. E. In 'Opical Properties of Solids";Seraphin, B. O., bulk material, we assign the bulk value to noand calculate
Ed.; North Holland Amsterdam, 1975. d. The errors inherent in this procedure are estimated to
48 Langmuir, Vol. I , No. I, 1985
be no greater than -4~0.5 A for reasonable errors of
f0.05 in no (typical value -1.50).
- samples or solutions (see Experimental Section).
Film Thickness Measurements. Monolayer films of
linear carboxylic acids (prepared as described in the ex-
perimental techniques section) were examined by ellipso-
metry in order to explore qualitatively the effects of ex-
perimental variables on the nature of the layers which
form. Film thicknesses, d, are presented in Figures 1 and
2 as plots of d vs. the number of carbons in the n-alkanoic
acid. The values of d were calculated assuming a refractive
index of 1.5 for the organic overlayer, a reasonable value
for a hydrocarbon material with a density close to that of
the bulk solid. For films of these thicknesses, the exper-
imental errors were too large to permit an accurate cal-
culation of n as an independent variable (see above).
These errors notwithstanding, the data could be fit rea-
sonably using a value of n = 1.50 f 0.05. The largest single
source of variability in the ellipsometry measurements was
found to result from spurious contamination of the alu-
minum-aluminum oxide substrate during the determina-
tion of the “clean” substrate constants. In order to provide
a uniform and displaceable initial overlayer, we have
preadsorbed acetic acid on most of the substrates exam-
ined. After determining the optical constants for these
coordinated surfaces, monolayers were prepared by
exchanging acetate with a longer n-alkanoic acid. The data
given in Figures 1 and 2 are not corrected for the contri-
bution of the displaced acetate monolayer (see below).
Control experiments indicate that this corection, in the
absence of other variables, would add an additional 2-3
A to the measured thicknesses. Control experiments also
showed that over the course of time required for sample
incubation, oxide thickening almost exactly compensated
the apparent lost thickness due to acetate displacement.
The effect of surface roughness on the thickness calcu-
lated from the ellipsometry measurements is expected to
be a small but systematic source of error in the data re-
ported. Scanning electron micrographs show that the
substrate oxide topography is comprised of shallow, rolling
-
hills of -50-200-A periodicity. Roughness of this order
( 1.2 to 1.3 times the geometric area) is expected to yield
a systematic variation of much less than 10% of the total
film thickness.
Taken together, the data presented in Figures 1and 2
demonstrate the importance of kinetics and competitive
adsorption in the formation of self-assembling monolayers.
At higher concentrations of fatty acids in hexadecane (-5
X M), the limiting film thicknesses described in Figure
1were obtained. Such values usually required incubation
times of the order of several days to be reached. I t is of
interest to note that the thickness of most films increased
significantly (220%) after they had become autophobic
(i.e., had nonzero contact angles with hexadecane). It is
clear from the data that there exists a continuity of gross
structure28and packing density for acids of chain length
C12or greater. Below C12,there was a pronounced tend-
ency for the films to form multilayers. The kinetics of
multilayer formation were extremely variable-hence the
range of values seen in Figure 1-and few samples ever
were found to develop nonzero contact angles with hexa-
decane. Careful control experiments and analysis of the
acids by GC-mass spectroscopy did not suggest that this
behavior was related to any atypical impurities in the
(28) As discussed above, the most significant sources of error are
systematic in nature. Thus, for organic overlayers of well-known bulk
refractive index, ellipsometry measurements, when carefully made, can
provide a simple but powerful method for determining the manner in
which the apparent thickness of an adsorbate layer varies with structure.
Allara and Nuzzo
Figure 2 shows data that further indicate the importance
of kinetics in the formation of self-assembling monolayers.
In this series of experiments (similar to that above), the
concentration of the fatty acids in hexadecane was de-
creased to 5 X M, a 10-fold dilution. The substrates
for this experiment were all evaporated and prepared si-
multaneously, and replicate runs of a given acid were in-
cubated in the same solutions. Several features of the
results obtained deserve specific comment. First, consid-
erable variation in formation kinetics is seen for duplicate
samples in the same solution. Second, the mean variation
between samples decreases as they approach their limiting
thickness. Third, limiting film thicknesses are found which
correspond exactly to those obtained at higher concen-
tration. Fourth, the time needed to reach a maximum
packing density is now of the order of 2 weeks. Fifth,
multilayers are still seen for acids within the chain lengths
below C12.
We note that while most of these data were obtained
using samples pretreated with acetic acid, these results are
generally predictive of adsorption behavior on this oxide
material. Acetate merely provides a uniform perturbation
of substrate constants, thus allowing the results of different
experiments to be correlated and compared. The obser-
vation of large variations in the kinetics of assembly in
both the presence and absence of acetate suggests im-
portant constraints on self-assembly other than the pres-
ence of contaminating overlayers on the oxide support.
Studies of Surface-Exchange Reactions. In the
chemistry described above, the importance of ligand-ex-
change reactions in the formation of self-assembling
monolayers was suggested. The high sorption activity of
the native aluminum oxide surface virtually guarantees
that some material from the laboratory ambient will be
adsorbed on the substrate before it is immersed in the
n-alkanoic acid solution. These spurious ligands-water,
carbonate, air-borne organics, etc.-or even the acetate
ligands described above, can and do greatly influence the
kinetics of closest packed monolayer formation. The sim-
ple notion of a surface formation constant, in analogy to
transition-metal coordination chemistry in solution, re-
quires that these ligands be displaceable by a long chain
carboxylic acid of a given concentration if a good mono-
layer is to be formed. To examine the dynamic charac-
teristics of such surface coordination reactions and further
develop our understanding of the mechanismb) involved,
we have conducted three independent labeling studies. In
the first, we have examined the displacement of tritiated
acetate ( [2-3H])from native aluminum oxide substrates
by Cg, CIS,and Czon-alkanoic acids. In the second, X-ray
photoelectron spectroscopy (XPS) was used to follow the
comparable displacement of trifluoroacetic acid by this
same series of acids. The third study, by reflection IR,
examined the competitive formation of and displacement
from closest packed monolayers using protio and per-
deuterated stearic acids. Taken together, these results
allow strong inferences to be made as to the nature of the
monolayers formed and point out several complicated,
though little appreciated, characteristics of the substrate
used. Each is discussed in turn.
(a) Radioisotopic Labeling Studies. When aluminum
oxide substrates (which had been previously incubated in
a 0.005 M solution of [2-3H]aceticacid in ethanol) were
immersed in hexadecane solutions of Cg, C18,and Cz0 n-
alkanoic acids (0.0050 M), significant quantities of the
tritiated label were displaced from the surface into the
solution. this process was slow; after several days, a lim-
Spontaneously Organized Molecular Assemblies. 1
BINDING ENERGY (ev)
Figure 3. X-ray photoelectron spectra showing (a) the Cls core
level spectra which result from treating an oxidized aluminum
substrate with a 0.050 M solution of CFBCOzHin ethanol at 25
"C for 15 min and (b)this same sample after immersion in a 0.0050
M solution of stearic acid in n-hexadecane.
iting activity was counted which corresponded to the av-
erage displacement by the higher acid of 6.7 X 1014acetic
acid molecules per square centimeter of substrate surface
area. This value is larger than that expected for a closest
packed monolayer (-5.1 X 1014molecules/cm2)6but is
remarkably close when this measured site density is cor-
rected for substrate surface roughness (the real surface area
is -1.3 times the geometric value). As had been observed
in the ellipsometry experiments, significant quantities of
radiolabel, 520% of the total, were displaced by the C18
and Czoacids after the samples had developed nonzero
contact angles with the hexadecane solution.
Digestion of these films and their substrates (after
washing with hexane) in concentrated aqueous sodium
hydroxide yielded a surprising result. The aqueous solu-
tion thus produced contained an activity that corresponded
formally to an additional 7.2 X 1014acetic acid molecules
per square centimeter of substrate surface area being left
on the support. We will defer our discussion of this result
to the following section and the companion paper to this
study.
(b) XPS Study of the Displacement of Trifluoro-
acetic Acid by Stearic Acid. Aluminum substrates
which had been incubated in 0.05 M solutions of tri-
fluoroacetic acid in ethanol showed complex Cls core level
spectra. At least four discrete peaks were observed (Figure
3), only two of which are assignable to the labeled marker
(BE = 293.7 and 290.6 eV). The large peak at 285.1 eV
was almost entirely absent on control samples (510% of
the intensity found here). Incubation in a 0.005 M hexa-
decane solution of stearic acid for 5 days resulted in the
loss of most of the trifluoracetate marker. A new, intense
peak at 286.4 eV was observed. The large peak observed
initially a t 285.1 eV is still seen and shows little change
in absolute integrated intensity. These results, and those
obtained above using tritiated acetic acid, argue strongly
for the absorption and incorporation of small molecules
into the oxide overlayer. We further suspect that reduction
of these small organics at or near the metal-metal oxide
Langmuir, Vol. 1, No. 1, 1985 49
m *
*\
0.10
. .
0 20 40 60 80 100 120 (40
TIME (hr)
Figure 4. First-order kinetic plot of the fractional coverage of
the substrate surface by stearic-do acid. In this experiment, a
closest packed monolayer of the doacid has been placed in 0.0050
M solution of the perdeuterated (d,) acid in n-hexadecane at 25
"C.The two symbols (m, 0 ) represent replicate experiments on
one and three samples, respectively.
interface may be an important process.29
(c) Kinetics of Surface-Exchange Reactions in
Closest Packed Monolayers. The data described above
suggest the importance of kinetics in determining the
structural nature of the monolayers obtained by adsorp-
tion. There further exists a strong indication that tradi-
tional macroscopic indicators of monolayer "perfection",
such as wetting, are not sufficiently sensitive to monitor
the approach to equilibrium, closest packed structures. To
further explore the dynamics of the formation of self-as-
sembling monolayers, we have followed the exchange of
stearic acid for its perdeuterated analogue by reflection
infrared spectroscopy. These data were determined from
the ratio of the integrated areas of the most intense bands
in the CH and CD stretching-mode regions, respectively.
The spectral characteristics in each of these regions were
found to be insensitive to the degree of coverage of each
isotopic component. These results provide an insight into
the reactive coordination chemistry relevant to this oxide
at high packing densities.
Figure 4 shows a first-order plot of the fractional cov-
erage of the surface by the protio (initial) acid as a function
of time. For the formal reaction described by eq 5, where
A + B*SE
kl
ASS+ B
k-1
(5)
A and B are the labeled and unlabeled exchange pair re-
spectively and S is the substrate, simple approach to
equilibrium kinetics is predicted. In the range of con-
centrations relevant to this experiment ([A] = M,
(29) For example, the intense peak seen in the XPS experiments at
a binding energy of 285.1 eV is shifted 1.3 eV to lower binding energy than
the Cls core level emission for the stearic acid hydrocarbon tail (no
correction for charging). Such relative shifts generally indicate a highly
reduced form of carbon. Angular-dependentXPS measurements are also
consistent with the suggestions given in the text.
50 Langmuir, Vol. 1, No. 1, 1985
CARBON NUMBER
Figure 5. Contact-angle data for n-alkanoic acid monolayers
adsorbed on an oxidized aluminum substrate [RC02H]= 0.0050
M in n-hexadecane,25 OC. The immersion times in the n-alkanoic
acid solutions were as follows: (a)15, (m) 72 h. The three wetting
liquids were from top to bottom water, methylene iodide, and
n-hexadecane.
kl[A] >> k-,[B]), this should reduce to pseudo-first-order
behavior. As shown in Figure 4, this clearly is not ob-
served. These data, representing two independent runs
with four separate samples, suggest both nonideality and
remarkable self-consistency. While the precision of the
data precludes a detailed analysis (the apparent two com-
pleting first-order processes evidenced in the plot may be
an artifact), the general conclusion is clear. Simply stated,
exchange in closest packed monolayers on native aluminum
oxide substrates is not kinetically simple.
(d) Formation of Mixed Monolayers of Stearic-do
Acid and Stearic-d36Acid. When an aluminum sub-
strate was incubated in a solution containing a mixture of
deuterated and nondeuterated stearic acids in n-hexade-
cane (2.5 X M in each), a distinct preference for the
adsorption of the protio acid was observed by reflection
infrared spectroscopy. We estimate, based on a compar-
ison of the spectra of pure do and d, monolayers to that
obtained above, that the observed composition reflects an
equilibrium ratio of do/d35 of the order of 1.4 f 20% at
25 OC.
Wetting Characteristics of Adsorbed Monolayer
Films. The wetting characteristics of our monolayer films,
as judged by contact angle data, strongly parallel the classic
data in the literature on the properties of oleophobic
monolayer film^.^ In our studies, we have examined three
liquids: water, methylene iodide, and n-hexadecane. The
data in Figure 5 illustrate the influence of two experi-
mental variables-immersion time and carbon chain
length-on the wetting properties of n-alkanoic acids ad-
sorbed from n-hexadecane solutions (0.005 M). The times
chosen (16- and 72-h incubations in the n-alkanoic acid
solutions) correspond roughly to the times when consid-
erable thickening was noted by ellipsometry under com-
parable experimental conditions. First, the contact angle
data does not seem to mirror this effect. Second, there
does appear to be a discontinuity of behavior below CI2,
a feature similar to that seen in the ellipsometry experi-
ments. Third, for carboxylic acids with chain lengths XlZ,
materials that consistently yield oleophobic films, some
variation is noted. In general, across this series the average
contact angles increase slightly and reach limiting values,
-
depending on the wetting liquid, at carbon chain lengths
of 18-19. We note, with some interest, that the Czzacids
(30) For leading references, see: "Advances in Chemistry Series 43";
Gould, R. F., Ed.; American Chemical Society: Washington, D.C., 1964.
Allara and Nuzzo
ADSORBATE
~~
2700 2800 2900 3000 3100
WAVENUMBERS
Figure 6. Infrared spectra of n-C19H&02H. The ordinate is
adsorption defined as -log (Illo)where Z is the intensity of de-
tected light for the sample and Io that of a reference. Lower:
transmission spectrum of the bulk acid dispersed in KBr. Upper:
glancing angle reflection spectrum of a monolayer of the acid
solution adsorbed onto an oxidized aluminum film. The features
at 2851 and 2919 cm-' are assigned to the C-H stretching modes
of the CH2groups. The features at 2879 cm-l and those in the
2938-2966-cm-' region are assigned to the C-H stretching modes
of the CH, groups.
terminated with an acetylene and vinyl functional group
(see Experimental S e ~ t i o n )both
~ ' have nonzero contact
angles with hexadecane. The observed value of 36' for the
vinyl-terminated derivative is remarkably close to that of
the methyl-terminated acid of the same chain length.
Finally, the wetting data do not suggest a significant
odd-even progression in surface tension for acids C12and
above. This is a likely consequence of the near-orthogonal
orientation of the adsorbate to the surface (see below).
Discussion
The ellipsometry data in Figure 1, given the known
optical constants of the fatty acid series, argue strongly
that the acids of chain length CI2 or longer are at or near
bulk packing density. This suggestion is fully supported
by independent radioisotopic labeling and
our own results on the displacement of [2-3H]CH3C02H
from this substrate. Taken together, these data strongly
suggest the formation of highly oriented, closest packed
monolayer structures. In Figure 6, we compare the infrared
spectra, in the C-H stretching region, of arachidic acid
C
(), taken under two sets of experimentalconditions. The
lower portion of the figure, taken in transmission, shows
the isotropic spectrum of the free acid dispersed in a KBr
pellet. The upper portion shows a glancing angle reflection
spectrum of the limiting (Le., highest coverage seen by
ellipsometry) monolayer of this same acid.33 It is imme-
diately apparent, even on a qualitative level, that these
(31) These materials are discussed in detail: Allara, D. L.; Nuzzo, R.
G. Langmuir, following paper in this issue.
(32) Whitesides, G . M., unpublished results.
(33) The accompanying paper describes the vibrational spectroscopy
of n-alkanoic acid monolayers, especially as it relates to the determination
of structure and orientation, in depth.
Spontaneously Organized Molecular Assemblies. 1
spectra show significant differences in relative band in-
tensity. The symmetric and asymmetric methylene
stretching modes (2851 and 2919 cm-l, respectively) are,
as expected, the predominant bands in the isotropic
spectrum. In the monolayer spectrum, these features are
lower in intensity than those assigned to the methyl C-H
stretching modes, seen here at 2879 and 2938-2966 cm-1.33
As we shall show in the accompanying paper, the alkyl tails
in these acids are best modeled as being at, full chain ex-
tension, showing little tilt from the surface normal. In this
context then, we discuss the data presented above.
One of the most striking features of this study was the
observation of a minimum chain length for the formation
of well-oriented, closest packed structures. As evidenced
by the ellipsometry and wetting data, closest packed, low
surface free energy phases were obtained only for chain
lengths greater than ell. The reasons underlying this
observation are not immediately evident. Careful control
experiments did not reveal any obvious or atypical im-
purities in these lower carbon number preparations. It is
significant to note that, even above C12,the contact angle
data shows a pronounced trend toward higher values, in-
dicating the formation of “denser’! (presumably lower
defect density) structures, as one adsorbs chains of in-
creasing length. Thus, it would appear that the absolute
magnitude of the enthalpy associated with chain-chain
packing interactions is an important effect, a conclusion
that also can be strongly inferred from the Langmuir-
Blodgett literature.ls The observation of nonstatistical
selection of stearic-do acid over stearic-d,, acid further
illustrates this point. If the structure of the adsorbed layer
were determined solely by the interaction of the head
group with the substrate (as a result of electrostatic in-
teractions or a preferred surface bonding site), it is hard
to imagine how either chain length or isotopic substitu-
t i ~ could
n ~ ~provide a significant perturbation to gross
structure.35 Our feeling, at this time, is that the chain-
length discontinuity discussed above reflects in part the
complex balance between the head-group bonding and
tail-packing interactions, forces which we believe tend to
expand and contract the layer, respectively.
The formation kinetics we have observed illustrates
another interesting and technically important feature of
self-assembly. First and most significant is that these
kinetics can be extremely slow, viz., on the order of days.
The tremendous inconsistency seen for similar samples in
the same solutions (Figure 2) argues strongly for an im-
portant role for surface and/or monolayer defects, the
nature of which we are not able to determine. Our
qualitative observation is that the slowest and most un-
predictable part of the assembly occurs during the incor-
poration of the last 20-25% of the monolayer. These
findings must also be viewed in the context of ligand-ex-
change chemistry. While most of our work involved the
preadsorption of acetate (see above), the final results we
obtain, in terms of packing and structure, in no way de-
pend upon this added step. Rather, it was noticed that
the kinetics of assembly became much more erratic in the
absence of a specific preadsorbed monolayer. The highly
active sorption characteristics of the clean native oxide
(34) Isotope effects on crystal packing are well documented in the
polymer literature. For example, deuterated polyethylene does not co-
crystallize with ita protio analogue (Schelten, J.; Wignall, G. D.; Ballard,
D. G. H.; Longman, G. W. Polymer 1977,18,1111-1120). We have not
seen any evidence that militates either for or against phase separation
in these mixed monolayers.
(35) One experimental observation we have made offers a perplexing
and incompletely understood contradiction to this suggestion. Hexa-
noic-dll acid consistently yielded oleophobic monolayers, a property
rarely observed for any acid below CI2.
Langmuir, Vol. 1, No. 1, 1985 51
surface, we feel, make it an extremely difficult material
to handle in the laboratory ambient. It was our experience
that some form of reversible chemical passivation-in this
case adsorption of acetic acid-was absolutely necessary
to ensure that closest packed monolayers would be ob-
tained in reasonable lengths of time and that the data from
such simple probes as ellipsometry could be routinely
compared.
Implicit in the above discussion is the most significant
aspect of this study as regards the nature of adsorbed
carboxylic acid monolayers on native aluminum oxide
overlayers. Simply stated, these are dynamic structures;
ligands, once adsorbed, can be displaced by other materials
in solution and presumably in the laboratory ambient as
well. The observed exchange of stearic acid by its per-
deuterated analogue deserves specific comment in this
regard. First, the exchange reaction does not appear to
require that the exchanging ligand have a much higher
formation constant for coordination to the surface. In this
system, where the ligands are structurally similar, the
process is driven simply by mass action. Qualitative ob-
servations suggest that this is a general property of the
system for most if not all pairings of the linear carboxylic
acids studied; longer acids will displace shorter ones and
visa versa as long as the concentrations of the displacing
ligand in solution are high enough. Second, the exchange
process is not kinetically simple. As we have described
above, a simple analysis predicts that the kinetics of the
approach to equilibrium should exhibit first-order be-
havior. This clearly is not consistent with the observed
data.36 If we assume that the observed exchange kinetics
truly reflect two competing first-order processes, two
reasonable explanations come to mind. First, the exchange
mechanism may reflect differing rates related to a heter-
ogeneous distribution of bonding sites on the support (that
is, some sites are “stickier” than others). A second pos-
sibility is that the exchange mechanism is dominated by
surface or monolayer defects. Exchange of material at or
near these defects is expected to be fast. The diffusion
of material further removed from these selective desorption
sites might be slow, and thus kinetically distinguishable,
at these high surface coverages. We cannot, at present,
distinguish these and other possible mechanisms. Our
inclination, given the complicated microscopic topography
of the substrate, is to conclude that structural defects must
play an important mechanistic role in the exchange reac-
tions d e s ~ r i b e d . ~ ’ ~ ~ ~
(36) Attempts to fit the data to second- and zero-order rate depen-
dences yielded equally poor correlations. Analyses based on a postulated
rate-determining dissociation of the carboxylic acid dimer in hexadecane
were found to be incompatible with the kinetic data.
(37) Assuming first-order rate behavior and an Arrhenius preexpo-
nential factor of 1013s-1,the activation energy for exchange in the limiting
slow rate regime is calculated to be -15 kcal/mol, a value consistent with
selective ’displacement” a t defects.
(38) The work of Torney and McConnell (Torney, D. C.; McConnell,
H. M. Proc. R. SOC.London, Ser. A 1983, 387, 147-170) suggests an
intriguing mechanistic possibility which, unfortunately, the precision of
our data prevents us from explicitly testing. If, in fact, exchange a t
defects were preferred, then the observed rate of exchange may be dom-
inated both by the density of these defects as well as the rate of diffusion
of the adsorbate along the surface to these same sites. Processes of this
sort,involving mass-transport-limited rates in two dimensions, need not
be characterized by a time-independent rate constant. Thus, the data
do not necessarily have to reflect complex, multiprocess kinetics. The
fact that diffusion in two-dimensional organic assemblies can be ex-
tremely varied-behaviors ranging from liquid- to solidlike-has been
broadly shown, for example, in lipid multibilayers and monolayers. See:
Seul, M.; Weis, R. M.; McConnell, H. M. Biophys. J. l983,41,212a. Wu,
E.; Jacobson, K.; Papahadjopoulos, D. Biochemistry 1977,16,3936-3941.
Smith, B. A.; McConnell, H. M. Proc. Nutl. Acud. Sci. U.S.A. 1978, 75,
2759-2763. Suel, M.; Eisenberger, P.; McConnell, H. M. Proc. Nutl. Acud.
Sci. U.S.A. 1983, 80, 5795-5797 and references cited therein.
52
Taken together, the data show that closest packed, or-
iented monolayers of n-alkanoic acids can be formed by
adsorption from dilute solution. The formation of these
quasi-two-dimensional organic surface phases is charac-
terized by complicated kinetics in which surface and/or
monolayer defects, as well as impurities, seemingly play
important roles. The structures so obtained are dynamic
in nature in that they undergo rapid exchange with com-
parable ligands in solution. These results, and those
presented in the following paper, further suggest an im-
Langmuir 1985,1, 52-66
portant limitation to closest-packed, self-assembly on this
substrate which is related directly to the length of the alkyl
chain. The data further show that oleophobicity is an
insufficiently sensitive probe to determine the formation
of equilibrium closest packed structures. In a companion
paper we describe these structures and their relationship
to many of the above observations in detail.
Registry No. n-Cl9H&O2H, 506-30-9; aluminum oxide,
1344-28-1; stearic acid, 57-11-4; hydrogen, 1333-74-0.

Spontaneously Organized Molecular Assemblies. 2.

Quantitative Infrared Spectroscopic Determination of
Equilibrium Structures of Solution-Adsorbed n -Alkanoic
Acids on an Oxidized Aluminum Surface
David L. Allara*
Bell Communications Research, Murray Hill, New Jersey 07974
Ralph G. Nuzzo*
AT&T Bell Laboratories, Murray Hill, New Jersey 07974
Received August 16, 1984

Infrared reflection spectroscopy has been applied to the determination of the structures of adsorbed
monolayer films of n-alkanoic acids on oxidized aluminum substrates. Acids of 16-22 carbons, terminated
by methyl, vinyl, or propargyl groups, were absorbed from hexadecane solution at 25 O C , using immersion
times up to several days. These results,together with those for perdeuterated acids, indicate that close-packed
-
assemblies are formed with extended alkyl tails oriented with their chain axes tilted away from normal
to the surface with a value of 10” for the longer chains. Frequency shifts to higher values are observed
for CH stretching modes of the vinyl and propargyl terminal groups measured in monolayer films compared
to the bulk acids. These shifts are of similar magnitude to gas-liquid phase shifts for these groups and
imply the environment of the ambient-monolayer interface exhibits very diminished intermolecular in-
teractions for these terminal groups compared to the bulk. The spectra clearly show that chemisorption
occurs by proton dissociationto form carboxylatespecies. Examination of both line width and peak positions
of the carboxylate stretching modes suggests a variety of binding geometries of the carboxylate groups
to the surface exists. These results lend support to the general concept of forming stable, oriented and
ordered two-dimensional organic films by solution adsorption and show the utility of applying surface
vibrational spectroscopy to structural determination in these films.

1. Introduction cated molecular organic films. A very elementary problem

The structures, formation, and properties of organic
molecular monolayer films are the subjects of much current
interest. One system of recent particular interest to us has
been spontaneously adsorbed n-alkanoic acid monolayer
films on ambient metal surfaces. In a previous paper’ we
presented extensive results on the formation, properties,
and dynamics of solution-adsorbed n-alkanoic acid films
on oxidized aluminum substrates. Central to a compre-
hensive understanding of those results is the development
of the details of molecular structure. Such detail includes
the nature of the individual molecular species, e.g., ion
formation, surface orientations, and surface-adsorbate
bonding, and the nature of the total monolayer assembly,
especially as regards ordered packing arrangements. While
a number of surface analysis tools are now available due
to extensive developments in the surface physics commu-
nity, in actual fact, few are applicable to the analysis of
the subtle features of the chemical structures of compli-
(1) A h a , D. L.; Nuzzo, R. G. Langmuir, preceding paper in this issue.
is that most popular surface techniques require impinging
ions or electrons as probes and this usually causes damage
to organic molecules. The techniques that have been
generally found useful for organic structures are those
using photon probes, in particular, X-ray photoelectron,
infrared, and Wan spectroscopies. Additionally, inelastic
electron tunneling and electron energy loss spectroscopies
have had success for organic monolayers. Past studies of
the adsorption of medium- to long-chain n-alkanoic acids
on ambient metal surfaces have utilized electron diffrac-
t i ~ n ,inelastic
~ - ~ electron tunneling (IETS),
(2) A review of earlier work can be found in: Bowden, F. P.; Tabor,
D. “The Friction and Lubrication of Solids”; Oxford University Press:
London, 1968; Part 11, Chapter 19.
(3) Menter, J. W.; Tabor, D. Proc. R . SOC.London, Ser. A 1957,204,
514-524.
(4) Chapman, J. A,; Tabor, D. Proc. R. SOC.London, Ser. A 1957,242,
96-107.
(5) Brockway, L. 0.; Karle, J. J. Colloid Sci. 1947,2, 277-287.
(6) Cass, D. A.; Straws, H. L.; Hansma, P. K. Science (Washington,
D.C.)1976, 192, 1128-1130.

0743-7463/85/2401-0052$01.50/0 0 1985 American Chemical Society