3 Organic Monolayer Studies Using Fourier Transform Infrared Reflection Spectroscopy D. L. ALLARA
Downloaded by SUNY STONY BROOK on October 15, 2014 | http://pubs.acs.org Publication Date: November 26, 1980 | doi: 10.1021/bk-1980-0137.ch003
Bell Laboratories, Murray Hill, NJ 07974
Vibrational spectroscopy of adsorbed surface species is a rapidly developing field. Although the traditional method has been infrared spectroscopy a number of competing techniques now exist including surface enhanced Raman, infrared surface wave, electron energy loss, inelastic electron tunneling and neutron scattering. These methods are treated in detail elsewhere and will not be discussed further here. Transmission infrared spectroscopy is one of the simplest methods to use experimentally but is insufficiently sensitive for detecting monolayers on smooth surfaces. The reflection infrared techniques on the other hand are sufficiently sensitive for flat monolayers. Internal reflection spectroscopy (IRS) is quite sensitive because of the large number of multiple reflections which can be used. However, the method requires a suitable transparent infrared material for the reflection element which also serves as the basic support for the sample. In some cases this provides obvious difficulties in studying phenomena such as adsorption on bulk metals. The subject of internal reflection has been reviewed in detail by Harrick. External reflection spectroscopy (ERS) is useful for detecting spectra on reflective substrates,^ such as bulk metals, thus is an obvious complement to internal reflection. The most popular use of ERS to date has been the study of CO on metals.^ In contrast there are very few reports of studies of monolayers of larger organic molecules on metals and metal oxides. Francis and Ellison were among the pioneers in the field with their report of spectra from oriented monolayers of barium stéarate on metal mirrors.^ Since that time there have been a number of reported spectra of thin films of organic molecules on metal mirrors but often the films have been many monolayers thick with no measurement of the actual thickness consequently these reports have not really involved examination of isolated surface species. Some studies which appear to involve very thin films, if not monolayers, have dealt with formic acid on copper - ~ ^ aluminum*-* and nickel^* and nitric oxide and isoamyl nitrite on copper, nickel and iron/-* The first study above^ indicates that formic acid only chemisorbs in the presence of oxygen but other results with 10~ torr vacuum suggest otherwise.*^ Earlier work on the chemisorption of acetic acid on copper^* (with a cuprous oxide overlayer) indicates that adsorption of carboxylic acids can build up vacuum stable, multilayer films so any oxide formed on the copper in the above studies^*^ could result in multilayer coverages. Boerio and Chen^ have obtained spectra from ~15A thick films of an epoxy polymer deposited onto iron and copper surfaces (oxide covered). These authors*—* have also reported spectra ôf monolayer thicknessfilmsof long chain fatty acids and alcohols on iron and copper surfaces (oxide covered). A study of ethylene adsorption on silver and platinum surfaces in (1)
(1)
(2)
(
),(
),(
9
0-8412-0585-X/80/47-137-037$05.00/0 © 1980 American Chemical Society
In Vibrational Spectroscopies for Adsorbed Species; Bell, A., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 1980.
38
VIBRATIONAL
which spectra were examined as a function of coverage has been reported.*^ In contrast to the minimal activity in infrared reflection studies the technique of inelastic electron tunneling spectroscopy (IETS) recently has contributed a large amount of information on monolayer adsorption of organic molecules on smooth metal oxide surfaces, ^ — mostly thin (20-30A) aluminum oxide layers on evaporated aluminum. These results indicate that a variety of organic molecules with acidic hydro gens, such as carboxylic acids and phenols chemisorb on aluminum Oxide overlayers by proton dissociation^'*^and that monolayer coverage can be attained quite reproducibly by solution doping techniques.^ The IETS technique is sensitive to both infrared and Raman modes.*—* However, almost no examples exist in which Raman ^ and or infrared spectra have been taken for an adsorbate/substrate system for which IETS spectra have been observed. The present study was initiated to provide a direct comparison of IETS and IR spectra for an identical molecule adsorbed on an aluminum oxide covered, evaporated aluminum substrate. Further, it was of interest to see if a weakly acidic C-H bond, such as that present in 1,3-dialkanediones, would show dissociative chemisorption simi lar to the well-known chemisorptions of Bronsted acids containing acidic O-H bonds (see above). The molecules chosen for this study were acetic acid and 2,4pentanedione. Both oxide covered copper and aluminum were used as substrates in order to see the effects of substrate oxide on the chemisorption spectra. (
Downloaded by SUNY STONY BROOK on October 15, 2014 | http://pubs.acs.org Publication Date: November 26, 1980 | doi: 10.1021/bk-1980-0137.ch003
SPECTROSCOPIES
)(
)
Theory
A brief description of the physical principles of the reflection experiment is in order because of the significant differences between optimum experimental conditions and theoretical calculations for reflection and the more usual technique of transmission. A representation of the reflection experiment is given in Figure 1. The theory was origi nally developed in detail by Hansen*— and Greenler.*—* The principles developed for the IR indicate*—* that maximum signal strength of monolayers on a metallic substrate should be obtained for a p-polarized beam of light incident at near glancing angles. These conditions give the highest value of the surface Ε field. The latter almost com pletely consists of the Ζ component and consequently only oscillators which have a Ζ component to their transition dipole moment will absorb energy at their excitation fre quencies. This selection rule can be of use in determining the orientation of adsorbed molecules although there have been no quantitative demonstrations of this reported. For optimum conditions a reflection spectrum of a thin film on a metal surface is gen erally 10-50 times stronger than the corresponding transmission spectrum obtained (at normal incidence) for the samefilmsupported on a transparent substrate. Quantitative calculations of reflection phenomena can be made from the boundary value solutions of Maxwell's equations for the interaction of a propogating infinite plane wave with a sys tem of parallel layers.*—*'*—* A discussion of the theoretical and experimental aspects of bond intensities and shifts for thin films on reflective surfaces has been reported*-^ recently. In general, strong bands are likely to show some distortion in reflection spec tra. For monolayers this usually amounts to a small upward shift in frequency of the band maximum.*—* For example, a band with a line width (at half height) of 25 cm should show a shift of — h 9 cm . This type of shift must be taken into account when carefully comparing IR reflection spectra of monolayers with high resolution vibrational spectra obtained by other means. )
-1
-1
In Vibrational Spectroscopies for Adsorbed Species; Bell, A., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 1980.
3.
ALLARA
FTIR Reflection Spectroscopy
39
Experimental
Sample Preparation Metal films were prepared by thermal deposition of pure metals (>99.99%)at pressures in the 10~ torr range using polished Si substrates (surface roughness ~50A). Film thicknesses were measured with a quartz thickness monitor and were generally about 200 nm. After deposition the vacuum system was backfilled with pure oxygen gas. The samples were quickly removed, flooded with ethanolic solutions of the selected organic adsorbate (usually 0.01 M) and spun horizontally at several thousand RPM using a standard photoresist spinner. These are representative of the standard conditions used for preparation of IETS samples by solution doping.^ Once formed thefilmswere quite stable to further exposure to the ambient environment. Downloaded by SUNY STONY BROOK on October 15, 2014 | http://pubs.acs.org Publication Date: November 26, 1980 | doi: 10.1021/bk-1980-0137.ch003
8
Spectra Spectra were obtained using a Digilab 15-B Fourier transform infrared spectrome ter operating at 2 cm resolution. A diagram of the optical system is shown in Fig. 2. The source output is processed through a mirror and aperture system (A) to give a parallel beam of light entering the interferometer. The exiting beam is stopped down through a variable aperture (1-2 cm) (Α') and processed through another mirror system to give an ~f60 beam focussing to a 2 mm beam spot (determined by A) at a position where the center of the sample is placed, the s-component is removed by a polarizer placed at Ρ and/or P'. The reflected beam isfinallyfocused onto a liquid nitrogen cooled mercury cadmium telluride detector. The general details of Fourier transform infrared spectroscopy are discussed elsewhere.^ All spectra were taken in a thoroughly dry, N purged atmosphere and samples were stored between spectra in tightly closed, individualfluoro-polymercontainers. The angle of incidence (φ, see Fig. 1) was 86 degrees for all the reported spectra. -1
2
Results and Discussion
Acetic Acid Acetic acid chemisorption has been previously studied using IETS by Lewis, Mosesman and Weinberg^ for oxide covered aluminum surfaces. Using reflection IR Tompkins and Altera^ have reported spectra for adsorption on oxidized copper and Hebard, Arthur and Allara^ for adsorption on oxidized indium. All these studies demonstrate that chemisorption from the gas phase involves proton dissociation since the observed spectra are those of acetate ion species. The present IR results with solution doping are shown in Figs. 3 and 4 for adsorp tion onto oxide covered aluminum and copper, respectively. For aluminum oxide the peaks at 1590, 1405 and 1335 cm match up in frequency with the major IETS peaks reported^ at 1589, 1403 and 1331 cm . However, the intensity pattern is significantly different since the weakest of the above peaks in IETS is at 1589 c m whereas this is the strongest IR peak. The IETS peaks, respectively, have been attri buted to—> C=0 and C-0 stretching vibrations, C-H asymmetric deformation and C-H symmetric deformation. Although partly a matter of terminology, we attribute the IETS peak at 1589 cm to the asymmetric carboxylate stretch. The IETS peak at 1452 cm should then correspond to the symmetric stretch. The appearance of the 1452 cm mode only in IETS is possible on the basis that its symmetry makes it only weakly -1
-1
-1
(
-1
-1
-1
In Vibrational Spectroscopies for Adsorbed Species; Bell, A., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 1980.
40
VIBRATIONAL SPECTROSCOPIES
UNEXCITED
p(ll)
Downloaded by SUNY STONY BROOK on October 15, 2014 | http://pubs.acs.org Publication Date: November 26, 1980 | doi: 10.1021/bk-1980-0137.ch003
/ s(-L)'
11c ι / \
EXCITED PHASE c=o /
AMBIENT
1
ORGANIC
7777777X77777777777 METAL
Ik Figure 1.
Description of the single reflection experiment. The C = 0 oscillator is shown to demonstrate the surface selection rule.
FIXED MIRROR
A (2mm)
-2mm
x
'SOURCE
(-f 60) BEAM
SAMPLE
-RETRO-MIRROR Figure 2. Simplified diagram of the optical layout of the FTIR spectrometer; A and A' are apertures, Ρ and P' are polarizers, and the area inside the dashed box is the actual interferometer assembly
In Vibrational Spectroscopies for Adsorbed Species; Bell, A., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 1980.
ALLARA
FTIR Reflection Spectroscopy
1
1
1
C H C 0 H / A I (OXIDE COVERED)
Downloaded by SUNY STONY BROOK on October 15, 2014 | http://pubs.acs.org Publication Date: November 26, 1980 | doi: 10.1021/bk-1980-0137.ch003
3
2
IETS: 15J39
1403 I 14J52 J 1331
1590 I
#Λ 1500 J 1 1465 I Ι 11550 1 \J Λ 1405
CM
Ο
—
1
ill ^
2000
ι 1800
ι 1600
11335
»
1400 1300
WAVENUMBERS Figure 3. Reflection spectrum of an oxidized aluminum film exposed to 0.010M acetic acid in ethanol; also shown are the IETS peak positions (13)
In Vibrational Spectroscopies for Adsorbed Species; Bell, A., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 1980.
42
VIBRATIONAL
SPECTROSCOPIES
IR active but Raman active and IETS selection rules allow Raman modes.However, since these stretching modes probably do not have ideal symmetry near the surface one might expect additional IR intensity for the symmetric mode. Another factor to be considered is the orientation. If the acetate ions were oriented with their rotational symmetry (C ) axis normal to the surface, the surface selection rule would favor the symmetric mode with its dipole derivative perpendicular to the surface in contrast to the asymmetric mode with its dipole derivative parallel to the surface and thus the symmetric/asymmetric ratio would be enhanced. Most IETS evidence*—* suggests that carboxylate ions are oriented away from the aluminum oxide surface with variable angles of tilt, as the structure in Fig. 5 suggests and accordingly some symmetric/asymmetric enhancement would be expected. It seems unlikely that the 1465 cm~ peak in in the present experiments is the 1452 cm~ symmetric carboxylate IETS mode shifted up in frequency although some small optical up shifts could occur in the reflection experiment^) but certainly less than 13 cm . It is possible that the 1465 cm~ peak in the IR spectrum could correspond to a CH bending mode of a coadsorbed ethoxide species formed by adsorption of ethanol which is used as the solvent for the acetic acid solution. Evans and Weinbeg^D report an IETS peak at 1472 for adsorbed ethanol on aluminum oxide at ambient temperatures and attribute it to the CH bending mode. However, the 1465 cm~ peak does not occur for adsorption on copper or for adsorption of 2,4-pentanedione on aluminum oxide, although a weak 1460 cm~ peak is present in the latter, which suggests that co-adsorption of ethanol may not be the correct explanation for the above acetic acid results. The current evidence suggests that a different acetate species (or mixture of species) is observed in the present experiments than was observed by IETS. This conclusion is strengthed by the appearance of the 1500 cm~ peak which is unobserved in IETS and which is presently unassigned to any specific mode. Further the 1550 cm~ shoulder observed in Fig. 3 may be the asymmetric stretching mode of a second surface species of acetate ion (see below). 2v
Downloaded by SUNY STONY BROOK on October 15, 2014 | http://pubs.acs.org Publication Date: November 26, 1980 | doi: 10.1021/bk-1980-0137.ch003
l
x
-1
]
2
l
2
x
x
l
The IR spectrum for adsorption on oxidized copper (Fig. 4) exhibits a different relative intensity pattern than the spectrum for the oxidized aluminum case although the peak positions are roughly the same. These peaks match up in frequency with the major peaks reported for the gas-phase adsorption on thick Cu 0 films^ except for the 1630 cm~ peak ^ which is not observed in the present study. The gas-phase results showed that vacuum stable multilayer films were formed and that the species present varied from run to run as evidenced by the variability in the intensity of the 1630, 1590 and 1550 cm~ peaks. The spectrum shown in Fig. 4 is probably that of a monolayer or near-monolayer coverage judged by the weak intensity, A=—log(R/R ) The high frequency modes (1730, 1710 cm ) present in the AcAcH transmission spectrum are absent in reflection and generally there is little direct correspondence between the spectra of the pure and adsorbed material. On the other hand there is a better match between the peak positions of the adsorbed species and the Al(AcAc) complex. This strongly suggests that AcAcH forms the AcAc~ enolate ion on chemisorption and that some sort of aluminum acetylacetonate complex is formed on the surface as depicted in Fig. 5, the exact orientation and bonding of such a complex is presently undetermined and Fig. 5 is only meant to convey a possible type of structure. Chemisorption on oxidized copper gives a roughly similar spectrum to that on oxidized aluminum but with a down shifted high frequency peak, a new peak at 1435 cm and no observable peak at 1460 cm , as shown in Fig. 7. The transmission spectrum of the copper acetylacetonate complex exhibits major peaks at 1577, 1552, 1529, 1461, 1413, 1353 and 1274 cm in the spectral range of Fig. 7. Again the absence of the "free" C=0 high frequency stretch modes of the pure diketone indicate that an AcAc~ enolate type of structure is adsorbed but the failure of the surface spectrum to match the Cu(AcAc) spectrum indicates that the surface species is not simply a layer of the complex salt. Exposure of a fresh substrate to the adsorption solution for 4 hours gives no changes in the reflection spectrum except for some loss of resolution and a suggestion of a shoulder forming at —1580 cm . Thus film growth does not appear to occur under these conditions. The intensity of the peaks in Fig. 7 (-log (R/R )