an assumed binary compound is inferred from determination of a single component, the method is sensitive to the presence of impurities. Thus, if a sample of NaZS, contains an impurity such as water, the experimental value of x will be less than the true value, because, in effect, less than the specified amount of NaZS, is weighed out. Potentiometric assay of polysulfides is, therefore, not a substitute for complete analysis, but a technique which can be useful in determining whether a complete analysis is necessary. It should be useful for rapidly assaying the precise composition of polysulfides which are known to be free of impurities and for routine monitoring of polysulfides to detect decomposition or moisture pickup over periods of time.
I25 v)
NlOO
z
'75
> E
u- 5 0 Q
25
0
LITERATURE CITED I
2
3 X
in
4
5
No,S,
Figure 1. Potential of sulfide ion selective electrode as influenced by sodium polysulfide composition
The potentiometric assay should be applicable in general t o soluble metal polysulfides, the metal ion of which does not form strong complexes with sulfide or polysulfide ions. In common with other techniques in which composition of
(1) G. Schwarzenbach and A. Fischer, Heiv. Chim. Acta, 43, 1365 (1960). (2) N . H. Furman, Ed., "Scott's Standard Methods of Chemical Analysis", 2nd ed., Vol. 1, Van Nostrand, NY. 1962, p 1009. (3) P. Ahlgren, Sven. Paperstidn., 70, 730 (1967). (4) A. Teder, Sven. Paperstidn., 70, 197 (1967). (5) E. Rosen and R . Tegman, Acta Chem. Scand., 25,3329 (1971). (6) W. F. Hillebrand, G. E.F . Lundell, J. I. Hoffman and H . A. Bright, "Applied Inorganic Analysis", 2nd ed., Wiley, NY, 1953, pp 716-717. (7) B. Cleaver, A. J. Davies, and M. D. Hames, Nectrochim. Acta, 18, 719 (1973). (8) N. K. Gupta and R. P. Tischer, J. fiectrochem. Soc., 119, 1033 (1972)
RECEIVEDfor review January 16, 1975. Accepted February 11, 1975. Work performed under NSF Contract NSF C-805.
Optically Transparent Carbon Film Electrodes for Infrared Spectroelectrochemistry James S. Mattson and Carroll A. Smith Rosenstiel School of Marine and Atmospheric Science, University of Miami, Miami, FL 33 749
Internal reflection spectroscopy (IRS) has been used in conjunction with electroanalytical techniques to study reactions in the thin region adjacent to the prism-electrolyte interface since about 1966. Initial efforts a t combining these two techniques, employing UV-visible spectroelectrochemistry, were reported by Hansen et al. (1, 2 ) . They ( I , 2) restricted their initial investigations to the system glasstin oxide-solution, although they suggested that metalcoated, infrared-transparent internal reflection elements could be employed to extend spectroelectrochemistry into the infrared region. The first extension to the infrared was reported by Mark and Pons ( 3 ) ,employing a germanium prism as a combination electrode and internal reflection element (IRE). Laser and Ariel ( 4 ) examined the electrochemical oxidation of 1-naphthol a t Pt and Au optically transparent electrodes (OTE), using UV-visible spectroelectrochemistry, followed by examination of the resulting organic film by infrared IRS. Tallant and Evans (5) employed germanium as an OTE in examining the reduction of p - benzoquinone, using dimethyl sulfoxide (DMSO) as the solvent. Trifonov (6-8) used a single-reflection germanium prism to follow the electrochemically initiated polymerization of acrylonitrile a t the germanium surface. Reed (9) and Reed and Yeager (IO),in their study of electromodulation of the germanium space charge region, employed germanium as an OTE, and D20 as their solvent. Tallant and Evans ( 5 ) intended to go on to the logical step of making OTEs by depositing metal films on germa1122
ANALYTICAL CHEMISTRY, VOL. 47, NO. 7, JUNE 1975
nium or KRS-5, but were stopped by the total absorption of infrared energy by 50- to 100-8, metal films (11). Laser and Ariel (12), in an effort to get around the high absorption coefficients of metal films, demonstrated the applicability of a gold wire grid electrode (200, 1000, or 2000 lines per inch (lpi)) wrapped around a germanium IRE. They used the wire grid as the electrode and the germanium as the IRE, monitoring changes in the internal reflectance a t 1510 cm-I, for the oxidation and reduction of p - benzoquinone in DMSO. In an attempt to extend his studies to opaque metals, Rice (13) prepared ultra-thin metal films of gold, palladium, nickel, and aluminum on NaC1, KBr, and sapphire, and observed that films of sub-30-A thicknesses were generally infrared-transparent. I t is known ( 1 4 ) that cold vacuum deposition of gold, palladium, platinum, etc., in amounts equivalent to only a 20- to 30-A thickness, results in islands of metal, rather than a continuous film. These islands grow epitaxially, until they finally begin to coalesce. Such deposits do not conduct a t these thicknesses, and thus behave optically as though they were dielectrics, rather than metals. We have confirmed this observation in our laboratory with ultra-thin films of gold, platinum, aluminum, copper, and iron on germanium, KRS-5, and zinc selenide. The problem of making a thin film which will be representative of the metal, while retaining some infrared transmission, requires careful attention to both the vacuum deposition parameters and the later selection of internal reflection geometry and polarization.
COUNTER ELECTRODE CUAMBER (PLATINUM WIRE COIL I
CHAMBER O-RING 12-214) PORTS
a-
FRITTED DISC
c!a -MAGNETIC
COUNTER CHAMBER SALT ERIOGE PORT
CWNTER ELECTROOE
STIR B A R
O-RING 12-007)
GE PRISM ( 2 x 6 X l O m m l
-
PRESSURE PLATE
WMiKlNG ELECTROOE CHAMBE
LOCATING PINS
-!
PLATINUM FOIL 2 SOFT SPACER
Figure 1. Spectroelectrochemicalcell for 52.5 X 20-mm single-pass internal reflection elements.
Figure 2. Spectroelectrochemicalcell designed for microprisms. Fits in Harrick 4 X beam condensing accessory
Contact to the working electrode is made via a sponge-backed platinum foil (not shown) placed between the prism and the metal contact plate
Schmidt-Brucken and Schlapp (15) found t h a t excellent adhesion could be obtained between diamond and carbideforming metals via vapor deposition on a hot diamond substrate (-500 “ C ) , producing a metal-diamond bond which cannot be broken, I t was felt t h a t carbon, which has a wide useable potential range ( I @ , would have a low enough infrared absorption coefficient that films of 100- to 300-A thickness should be transparent enough to yield some useable energy even after several internal reflections (17). Following the reasoning of Schmidt-Brucken and Schlapp ( 1 5 ) , germanium, which has a diamond structure, should form a carbide bond with a carbon film if the germanium oxides are removed and the temperature of the germanium could be kept fairly high. This paper describes the properties of infrared-transparent, carbon-coated OTEs prepared in this fashion.
EXPERIMENTAL T h e carbon films were prepared by a commercial coating laboratory (Lebow Company, 1407 Norman Firestone Road, Goleta, CA 93017). Germanium internal reflection elements were initially polished and cleaned mechanically (Harrick Scientific), then cleaned with a detergent solution, followed by a trichloroethylene rinse. Surface oxides were then removed by cleaning with 40% HF, then glow discharge cleaning in the vacuum chamber (1-5 minutes, argon) while the prisms were “soaking” a t high temperature (450-500 ‘ C , quartz-iodine lamp). T h e carbon was evaporated by a n electron beam technique, using a n ultra-high purity glassy carbon source. T h e thickness of the deposited layer was monitored during deposition hy a quartz crystal thickness monitor, assuming a density of 2.2 for the carbon film. Identical films were deposited on glass slides for resistance measurements. In order to reduce the likelihood of pinholes penetrating the films, they were deposited in two consecutive runs, cleaning the surface between runs. Electrochemical measurements were carried out with an electroanalytical instrument of our own design, a Keithley model 414 micro-microammeter, a Keithley model 621 electrometer, and a . Moseley X-Y recorder. T h e cells employed are made of Teflon and Delrin, and are shown in Figures 1 and 2. Spectrophotometric measurements were made with a PerkinElmer model 180 infrared spectrophotometer, equipped with a wire grid polarizer and a Perkin-Elmer Standard Interface for digital d a t a transmission. Data acquisition and subsequent processing was done with a Data General NOVA 1220 minicomputer with 16K of core memory and three cassette tape transports. T h e spectrometer-computer system is completely described elsewhere (18).
RESULTS AND DISCUSSION Of ten OTEs prepared by this technique so far, all have been transparent over the entire spectral range of the sub-
strate. Large germanium prisms (52.5 X 20 X 2mm), that have thirteen reflections a t the germanium-carbon interface, yield transmissions ranging between 4 and 8%. Germanium microprisms (10 X 6 X 2mm), with only three reflections a t the carbon film, transmit about 15%. Bare germanium prisms transmit between 25 and 30%. The first three films prepared as single coatings were pinholed extensively, so that subsequent films were made in two layers, adding a second cleaning step between the two depositions. The pinhole problem has not been observed with the films prepared in this fashion. One germanium deposition failed to produce adherent films, and it appeared that the polishing process carried out prior to the chemical cleaning and ion bombardment was inadequate. A thorough pitch-polishing step thus appears t o be a prerequisite to the preparation of adherent carbon films. There are several trade-offs involved in the selection of either the large surface area prisms or the microprisms. Figure 1 illustrates the spectroelectrochemical cell constructed for use with large prisms, in which the working electrode area is 6.2 cm2. Even with the relatively low transmittances observed with the larger prisms, computeraveraging and smoothing of spectra can produce excellent signal-to-noise ratios (S/N). Figure 2 illustrates the cell designed for use with microprisms, in a 4X beam condensing accessory (Harrick Scientific). Contact to the working electrode is made via the platinum foil pressed against the “back side” of the prism just as it is in the cell shown in Figure 1, and the working electrode area is about 0.12 cm2, Since both the spectral contrast and the background absorption due to the carbon film are directly proportional to the number of reflections, it is possible to improve the signal level by using microprisms, and the spectral contrast can be restored by employing the appropriate ordinate expansion factor with the final spectrum. The S/N levels initially observed with some of the large prisms appeared to be good enough that some degradation could be allowed without seriously affecting the final spectra, so the carbon films on the microprisms were made somewhat (-20%) thicker than they are on the large prisms. The thicker carbon films exhibit lower resistances and better mechanical properties than the thinner films. The resistances of several carbon films simultaneously deposited on glass slides were from 2000 to 5000 fl/o for films between 270 and 320 thick. Such resistances correspond t o a resistivity of about 1.5 X ohm-cm, or about 10 times that of bulk graphite. ANALYTICAL CHEMISTRY, VOL. 47, NO. 7, JUNE 1975
1123
I
0.3c
0.15 E 0
2
0.10
c. 0.05
z
/Carbon
0.2t 0.3
4
jGe
OTE
I
;
0.00
3 0
I
l
0.9
0.6
l
I
I
I
0.3 0.0 -0.3 -0.6 -0.9
I
I
I
-1.2
4.5
-1.8
aos
VOLTS. l o SCE
Figure 3. Background voltammetric scans for platinum foil, germanium prism, and carbon OTE, l O m M sodium borate solution, in microprism cell
-02 -0.3 -0.4 -0.1 -0.6 -0.7 -0.8 -0.9 -1.0 -1.2 VOLTS, to SCE
Figure 5. Attempts to use germanium OTE for cyclic voltammetry of AQS in sodium borate
0:
02 I
j N
.
0
+ .
,
..
.
E 01
2E
2000
0.2
I /
I
-04 -06 -0.8 VOLTS, t o SCE
Figure 4. Cyclic voltammograms for 1O-3M anthraquinone-2-sulfonate in lO-'Msodium borate on Pt foil and carbon OTE, 46 mV/sec
1800
1600 14W WAVENUMBER 'CM
Figure 6. Deposition of protein film face
at the
1200
IMX)
carbon OTE-solution inter-
(1) Average of 25 base-line spectra at rest potential of -340 mV (to SCE), (2) 25 spectra after 120 min at -200 mV, (3) 25 spectra after 37 min at 0 mV, (4) 10 spectra after 90 min at 0 mV. (5)10 spectra after 128 minutes at 0 mV. Solution was 0.36% fibrinogen (90% ciottable) in 0.1M phosphate buffer, pH 7.4. Microprism had 320 A carbon OTE, one reflection at prismsolution interface, all spectra 1OX expanded, reference beam attenuated by chopper to 10% T, parallel polarization
Table I. Background Potential Rangesa Electrode
Anodic '
(VE
SCE)
.10/.20 ma-cm-2
Cathodic (vs SCE)
.10/.20 ma-cm-2
P t Foil +1.20/+1.24V -.go/-.95V -0.28/-0. 19 -1.33/-1.44 Germantum Carbon OTE +0.54/+0.73 -1.45/-1.68 Pyrocarbonb +0.8 -1.7 10 mM sodium borate. bFrom B. D . Epstein, General Atomic Company, San Diego.
The electrode capabilities of these films have been examined to some extent using 320-A films deposited on germanium microprisms. Figure 3 illustrates the background potential ranges obtained for a carbon OTE, a germanium microprism, and a platinum foil electrode, in 0.01 M sodium borate solution. Table I lists the anodic and cathodic cutoff potentials for current densities of 100 and 200 palcm2, plus values for isotropic pyrolytic carbon electrodes. The anodic potential range of a carbon OTE is a t least 0.8 V greater than pure germanium, while the cathodic range is from 0.1 to 0.2 V better than germanium. In comparing the values obtained for the carbon-film electrodes to those reported for isotropic pyrolytic carbon, it appears that the carbon OTE surface is similar to that of isotropic pyrolytic carbon. Figure 4 illustrates cyclic voltammograms for 10-3M anthraquinone-2-sulfonate (AQS) in 10-2M sodium borate solution, on a platinum foil electrode and a carbon OTE, in 1124
ANALYTICAL CHEMISTRY, VOL. 47, NO. 7, JUNE 1975
the cell shown in Figure 2. For comparison purposes, Figure 5 illustrates the results of attempts to use an uncoated germanium prism in the same system, with the erratic results illustrating the superiority of the carbon film OTE over uncoated germanium. Figure 6 illustrates constant potential deposition of a negatively charged protein (fibrinogen) a t the surface of a carbon OTE on a microprism, using the cell shown in Figure 2. The actual spectra are shown (marked 1 through 5), and are unsmoothed, averaged spectra taken a t the speed of 10 cm-'/sec, with either 10 or 25 spectra averaged together, as indicated in the caption. The reference beam was attenuated to 10Y0 T with a precision chopper attenuator (Harrick Scientific), and the spectra are shown a t 1OX scale expansion. The difference spectra a t the bottom of Figure 6 show that no adsorption occurred a t -200 mV ( t o SCE) after 120 minutes of potentiostating. The amide peaks a t 1640 and 1540 cm-' become pronounced in the difference spectrum (3 - 1) taken while potentiostating a t 0 mV. One spectrum was taken every 3 minutes, from 2000 to 1000 cm-', in 2-cm-l increments, a t a constant slit width of 2.2 mm. Additional potentiostating a t 0 mV to SCE produced no change in the amide peak intensities after 90 minutes. This could be accounted for by assuming that deposition within the region sampled by the evanescent field had been completed, and that additional deposition could not be detected by internal reflection spectrometry.
SUMMARY These carbon film OTEs are the first such films t o be produced with conductivity high enough t o be employed as electrodes, while still retaining sufficient infrared transparency for internal reflection spectrometry. Other infraredtransparent materials with the diamond crystal structure should also make good carbon OTE substrates. Cadmium telluride, with its longer infrared window, is a particularly attractive candidate, although one attempt to coat CdTe has failed. Getting carbon to adhere to CdTe will require proper selection of coating temperature, and perhaps a modified cleaning procedure. T h e procedure for putting metal films on diamond was described in the introduction, and diamond is a very attractive candidate as an OTE substrate.
LITERATURE CITED (1)W. N. Hansen, R. A. Osteryoung, and T. Kuwana. J. Am. Chem. Soc., 88, 1062 (1966). (2) W. N. Hansen. T. Kuwana. and R . A. Osteryoung, Anal. Chem., 38, 1910 (1966). (3)H. 8. Mark, Jr. and 5. S.Pons, Anal. Chem., 38, 119 (1966). (4)D. Laser and M. Ariel, Nectroanal. Chem. Interfacial Hectrochem., 35, 405 (1972). (5)D. R. Tallant and D. H. Evans, Anal. Chem., 41, 835 (1969).
(6)A. Trifonov, B. Jordanov, and M. Poneva, Comm. Dept. Chem., Bulg. Acad. Sci., 4, 131 (1971). (7)A. Z.Trifonov and I. D. Schopov, Nectroanal. Chem. Interfacial Necfrochem., 35, 415 (1972). (8) A. 2 . Trifonov. T. Popov, and 6. Jordanov. J. Mol. Struct., 15, 257 (1973). (9)A. H. Reed, Diss. Abstr. Int. B., 29,4561 (1969). (IO)A. H. Reed and E. Yeager, flectrochim. Acta, 15, 1345 (1970). (11) D. Evans, Univ. of Wisconsin, private communication, January 1973. (12)D. Laser and M. Ariel, J. Nectroanal. Chem., 41, 381 (1973). (13)R . W. Rice, Ph.D. Thesis, Yale University, 1972. (14)J. V. Sanders, in "Chemisorption and Reactions on Metallic Films", Vol. 1, J. R. Anderson, Ed., Academic Press, London, 1971,pp 1-39. (15)H. Schmidt-Brucken and W. Schlapp, Z. Angew. fhysik., 32, 307 (1971). (16)B. D. Epstein. D. Daiie-Molle. and J. S.Mattson. Carbon, 9,609 (1971). (17)J. S.Manson, Anal. Chem., 45, 1473 (1973). (18)J. S. Mattson and C. A. Smith, Chapter 2 in "Computers in Chemistry and Instrumentation", Vol. 7,J. S. Mattson, H. B. Mark, Jr., and H. C.
MacDonald. Jr.. Ed. Marcel Dekker, Inc., New York (in press). Preprints will be available from the authors until volume is published.
RECEIVEDfor review December 23, 1974. Accepted February 18, 1975. This research was supported by the National Institutes of Health, National Heart and Lung Institute Grant No. HL-15919-01A1, and a Cottrell research grant from Research Corporation, Atlanta, GA. Contribution from the Rosenstiel School of Marine and Atmospheric Science, University of Miami, FL.
New Ambient Temperature Molten Salt Solvent for Electrochemistry: Triethyl- n-hexylammonium Triethyl- nhexylboride Warren 7 . Ford Department of Chemistry, Roger Adams Laboratory, University of Illinois, Urbana, lL 6 180 7
Preparation of a series of tetraalkylammonium tetraalkylborides which are liquid a t room temperature makes available a new class of solvents. ( I ) . The least viscous of these molten salts, triethyl-n-hexylammonium triethyl-nhexylboride (N222&2226), has been used as a solvent for kinetics (2, 3) and NMR spectroscopy ( 4 ) . N2226B2226 is not appreciably soluble in alkanes or water, but it is completely miscible with many organic solvents ranging from benzene to methanol. Experiments reported here demonstrate its promise as a solvent for electrochemistry. Only one other liquid salt, tetra-n- hexylammonium benzoate hemihydrate, has been reported as suitable for room temperature electrochemical experiments (5).
EXPERIMENTAL All experiments were performed in a nitrogen atmosphere. Conductance of liquid N 2 2 2 6 B 2 2 2 6 was measured in a 4-ml cell with platinum black plates and a constant of 0.686 cm-'. Temperatures were accurate to f O . O l O , and conductances were identical a t 1- and 3-kHz oscillator frequencies. Cyclic voltammograms were obtained in a 2-ml cell. T h e working electrode was a cross-section of 18 gauge platinum wire sealed in glass. A platinum wire served as the anode, and a n aqueous Ag/ AgC1, saturated KC1 electrode placed about 5 m m from the working electrode served as the reference. Because of an appreciable iR drop, the cathodic and anodic peak potentials usually varied with scan rate over a 0.05-V range, and their separation was larger than t h e theoretical 58 mV for a reversible one-electron process a t 22'. Peak potentials a t a single scan rate in repeated scans all fell with-
in a 0.02-V range. Slight leakage from the reference electrode left 0.1 wt % water in the N 2 2 2 6 B 2 2 2 6 after a typical series of experiments.
RESULTS AND DISCUSSION The specific conductance of N2226B2226 increases by a factor of 7.8 when it is warmed from 25' to 75' (Table I). Its specific conductance a t 25' is the same as that of 0.0035M aqueous KCl. The marked temperature dependence is a consequence of its high viscosity and the nearly equal sizes of its cation and anion. The activation energies for conductance and shear viscosity of N2226B2226 both lie in the 8-10 kcal/mol range a t 25-75'. Coetzee and Cunningham (6) have recommended tetraisoamylammonium tetraisoamylboride as a reference electrolyte for the evaluation of single ion conductivities because its anion and cation would have nearly equal mobilities. Carbon-13 spin-lattice relaxation times of neat N2226B2226 indicate that internal rotational motion in the anion is slightly faster than in the cation, perhaps because the B-C bonds are longer than the N-C bonds ( 7 ) . If the B2226 anion is slightly larger than the N2226 cation, the anion should have a slightly lower mobility in the neat liquid. Typical cyclic voltammograms of anthracene and benzophenone in N2226B2226 a t 2 2 O are shown in Figure l. The reduction of anthracene (assumed to be a one-electron process by analogy to many previous investigations of anthraANALYTICAL CHEMISTRY, VOL. 47, NO. 7, JUNE 1975
1125
