3162
J. Phys. Chem. 1980, 84, 3162-3172
at 150 “C, suggesting its stable activity at that temperature. Reactivity of nitric oxide on CoTPP/Ti02 in the absence of hydrogen at 50 and 150 “C is summarized also in Table I. A small but distinct decomposition of nitric oxide to nitrous oxide over CoTPP/Ti02 should be noted at 50 “C without hydrogen, although adsorption of nitric oxide was dominant for its consumption. Such a reaction has never been observed on unsupported CoTPP.16 This reaction was further enhanced a t 150 “C. The major product of this reaction was nitrous oxide. Nitrogen was a minor product especially a t 50 OC, indicating its successive formation through nitrous oxide. Neither nitrogen dioxide nor oxygen was detected in the gas phase, suggesting the adsorption of oxygen in any form on the catalyst. Further analyses of the reaction are now in progress. Such a significant enhancement of catalytic activity of CoTPP supported on TiOz may be explained in terms of the modified electronic structure of CoTPP described in the early part of this paper as well as the enlarged active surface of CoTPP, The partial reduction of cobalt ion may be favorable for its back-donation ability to produce an active species of anionic nitric oxide which may accelerate the adsorption and N-O bond cleavage of nitric oxide. The adsorption of hydrogen on CoTPP/Ti02 which could never be observed on CoTPP alone may be related to the activation of hydrogen. This adsorption appears to be connected to the porphyrin ring because H2TPP/TiO2could also adsorb hydrogen. Further study of the interaction of nitric oxide and hydrogen with supported CoTPP is now in progress. In conclusion, the strong interaction of CoTPP and TiOz
revealed in the present study may promise novel aspects of heterogeneous catalysis and photocatalysis.
Acknowledgment. We thank Professors A. Ohyoshi and S. Kusumoto and Dr. S. Sakaki, Kumamoto University, for their helpful suggestions regarding the Experimental Section and the Discussion of this paper.
References and Notes K. M. Smith, Ed., “Porphyrins and Metalloporphyrins”, Elsevier, Amsterdam, 1975. P. A. Leermakers, H. T. Thomas, L. D. Weis, and F. C. James, J. Am. Chem. sot., 88, 5075 (1966); M. L. Unlant and J. J. Freeman, J. Phys. Chem., 82, 1036 (1978). F. F. Fan and A. J. Bard, J . Am. Chem. SOC.,101, 6139 (1979). I. Mochida, K. Takeyoshi, H. Fujltsu, and K. Takeshita, Chem. Lett., 589 (1976). P. B Weisz, J. Chem. Phys., 21, 1531 (1953); J. Cunningham, J. J. Kelly, and A. L. Penny, bid., 74, 1992 (1970); H. Hattori, M. Itoh, and K. Tanabe, J. Cafal., 38, 172 (1975). S. J. Tauser, S.C. Fung, and R. L. Garten, J . Am. Chem. SOC., 100, 170 (1978); M. A. Vannice, R. L. Graten, J. Catal., 56, 236 (1979). A. D. Adler, F. L. Longo, F. Kampas, and J. Kim, J . Inorg. Nucl. Chem., 32, 2443 (1970). F. A. Walker, J. Am. Chem. SOC.,95, 1154 (1973). J. M. Assour, J. Chem. Phys., 43, 2477 (1965). (a) R. H. Ball, G. D, Dorough, and M. Calvin, J . Am. Chem. SOC., 68, 2278 (1946); (b) R. H. Feiton and H. Linschitz, ibid., 88, 1113 (1966). A. Woiberg and J. Manassen, J. Am. Chem. SOC.,92, 2982 (1970). H. Kageyama, M. Hidai, and Y. Uchida, Bull. Chem. SOC.Jpn., 45, 2898 (1972). G. Peychal-Heillngand G. S.Wilson, Anal. Chem., 43, 550 (1971). C. G. Pierpont, D. G. Van Deveer, W. Durland, and R. Eisenberg, J. Am. Chem. SOC.,92, 4760 (1970). K. TsuJl, H. Fujitsu, K. Takeshita, and I. Mochida, J . Mol. Catal., in press.
An Inelastic Electron Tunneling Spectroscopy Study of Some Iron Cyanide Complexes K. W. Hipps” and Ursula Mazur Department of Chemistry and Chemical Physics Program, Washington State University, Pullman, Washington 99 164 (Received: Februaty 25, 1980; In Flnal Form: July 14, 1980)
Inelastic electron tunneling spectroscopyis used to study the nature of the surface complex formed when several cyanide-containing iron complexes are adsorbed on freshly prepared alumina and hydrous alumina surfaces. The complexes studied are of the form M(3,4)Fe(CN)6X where M = H, Li, Na, K, or Cs and X = CN-, HzO, or NO. Complexes are incorporated in tunnel diodes via adsorption from low concentration [10-g-104 MI aqueous solutions. It is determined that complexes which contain iron(II1)in solution are reduced to the plus two form in every case. Cation effects on the vibrational spectra observed indicate that charge compensation occurs by adsorption of the complex plus the requisit number of cations to form a relatively discrete surface unit. Computer fitting of the current-voltage characteristics of these doped diodes clearly shows that submonolayer adsorption is being studied. Further, by comparison of the barrier parameters with the tunneling spectra observed, a coherent picture of the nature of the adsorbed species is obtained. The tunnel diode structures used in this study are of the form A1-A1203-M’, where M’ = Pb, Sn, or Ag. Effects due to diffusion of Sn and Ag into the oxide layer on the vibrational spectra of the adsorbed layer ions are observed.
Introduction Improved understanding of metal complexes as adsorbates on oxide and hydrous oxide surfaces is critical to many problems in surface chemistry. Such diverse topics as oxide-supported catalysts, surface-induced chargetransfer processes, and transport of metal-containing pollutants in natural waters will benefit from a thorough understanding of the nature of metal complexes on surfaces. Although a large body of literature relating to the stoichiometry and thermodynamics of metal complex ad-
sorption on oxide surfaces exist~,l-~ little spectroscopic information is available. ESR spectroscopy has been shown to be useful for ellucidating the site symmetry of adsorbed paramagnetic however, it provides little other information about the nature of the adsorbed species and is inappropriate for diamagnetic complexes. The venerable infrared method is useless in the metal-ligand region of the spectrum due to the opacity of the subs t r a t e ~ . ~Raman - ~ ~ spectroscopy provides a wider free spectral range for observing vibrational states of asorbed
0022-3654/80/2084-3162$01.00/00 1980 American Chemical Society
IETS of Some Iron Cyanide Complexes
The Journal of Rbysical Chemistry, Vol. 84,No. 24, 1980 3163
n
complexes and is probably superior to IR spectroscopy in many cases. It has the disadvantage of requiring an optically transparent substrate, little or no substrate fluorescence, large surface area samples, and adsorbates which will not be decomposed in the intense laser radiation.12-16 Electron energy loss spectroscopy is another vibrational ,v= 0 spectroscopy which is potentially useful for studying electrically neutral complexes on single-crystal~ u r f a c e s . l ~ - ~ ~ The relatively low resolution of contemporary instruments, about 80 cm-l, and the necessity of submonolayer coverages of a single-crystal surface under ultrahigh vacuum conditions make its present usefulness rather limited. Optical spectroscopy in the visible and near-UV regions of the spectrum has also been used with some success.11,21t22 Unfortunately, the electronic transitions of many metal complexes arc insensitive to the detailed structure of the complex. The technique is primarily useful for studying changes in the oxidation state of the metal and in observing coarse changes in the site symmetry at the metal. During the past decade, a new surface spectroscopy called inelastiic electron tunneling spectroscopy (IETS) has been introduced. Tunneling spectroscopy can provide information about vibrational and electronic transitions of adsorbed materials over the 0.5-13 000-cm-l range and e v > Iiw a useful resolution of about 8 cm-1.23-28Although the utility Flgure 1. Schemati’c representation of a simplified potential energy of IETS for studying adsorbed organics has been extensurface for MIM diodes under various applied voltages. sively demonstrated, only a few reports of metal complex studies are a ~ a i l a b l eand , ~ ~these ~ ~ have ~ ~ ~dealt ~ with Fermi energy and the minimum energy required to remove neutral species whose spectra were essentially those of the an electron from the metal is the work function, W. free ligand. Recently, we reported the first tunneling I-V Characteristic. The wavefunctions of the electrons spectrum of a metal complex ion, the ferrocyanide in the metals are undamped oscillating functions of posThat letter was restricted to the demonstration of the ition, In the insulating region, however, they are expoappropriateness of the technique for studying metal comnentially damped. If the barrier is not too high or too wide, plex ions. In this paper we will provide a detailed study this exponential tail of the electronic wavefunction proof the nature of the adsorbed species as discerned from vides a finite probability that the electron will be transthe tunneling spectra of a number of different cyanidemitted through the barrier. Transmission occurs without containing anionic complexes of iron. The role of cations a loss of energy to the barrier and is classically forbidden. will be thoroughly explored and isotopic substitution will This process is elastic tunneling. At zero bias, i.e., when be utilized to help confirm spectral assignments. Further, the metals are at the same potential, their Fermi surfaces we will make use of computer modeling of the observed must coincide because there can be no net change in current-voltage characteristics of the diodes used in these electrochemical potential for an electron transported from studies to help develop a model for the detailed nature of one metal to the next via the external circuit. No net the oxide-complex unit. tunneling processes occur, therefore, and no current flows. Theory As the right-hand metal becomes more positive, electrons The theory of opera tion of metal-insulator-metal (MIM) from the left metal may elastically tunnel into vacant states tunneling junctions and tunneling spectroscopy has been of the right metal and current flows. This current is thoroughly r e v i e ~ e d . ~While ~ - ~ it~ is fair to say that many roughly proportional to the number of vacant states with difficulties with the theory are still unresolved, the major the appropriate energy which, in turn, is roughly proporfeatures of the tunneling process are well understood. In tional to the applied voltage. At low bias voltages, this section we will briefly review the concepts required therefore, the device has a current-voltage (I-V) characfor an understanding of the tunneling process. The initeristic which is ]predominantly resistive. At high bias tiated may proceed directly to the next section. voltages, the I-V curve becomes distinctly nonlinear. The basic object of study of IETS is the MIM tunnel Although this crude model gives a satisfactory qualitadiode. Typically this three layer structure is composed of tive description of the elastic tunneling processes at low a bare metal layer of about 1000 A in thickness, an insubias voltages, it is unsatisfactory a t higher bias and lating region between 8 and 50 A in thickness, and a top quantitatively unsatisfactory at all bias voltages. Many metal of about lo00 8, in thickness. The metals are usually improvements to this simple model have been suggested. deposited in a crossed configuration to yield junctions of The growth of the insulating layer on the bare metal often roughly 1 mm2 area. The simplest theoretical model for generates an internal field gradient of the order of 1V.355,36 these structures is depicted in Figure 1. One replaces the In the case of dissimilar metals, the difference in work metal electrodes by a semiinfinite “jellium” model and the functions for the two metals generates an additional change insulator by a vacuum space. Filling in electrons consistent in the top side barrier height of the order of AV = ( W , with the Pauli principle and electrical neutrality, one ob- WB)/e.37The existence of defects in the insulator which tains fully occupied orbitals up to the energy of the fermi serve as electron traps can contribute to a nonlinear varsurface (assuming the temperature is absolute zero). The iation in the shape of the barrier height.38 Image fields hatched areas in Figure 1depict the energy regions of fully due to the presence of the conductors also affect the model occupied orbitals. The top of this electron “sea” is the potential.39 To the best of our knowledge, all of the bar-
n
n
”‘*1
3164
The Journal of Physical Chemistry, Vol. 84, No. 24, 1980
Hipps and
Mazur
if distances are measured in angstroms, u annd 4 are measured in electron volts, and $(O,O) = &, at the fermi surface as shown in Figure 2. If one assumes that the voltage falls linearly across the barrier, for the situation where the right metal is positive x d)+(x,v) = 41 + (42 - 4,), -
XV
x v o+s
4+(x,v) = 4 3 - TV for D
for 0 I x I D
Ix ID
+S
(4a)
In the case where the left metal is positive
Flgure 2. TRPSQR barrier model used in current-voltage characteristic
calculations.
riers proposed to date have some unsatisfactory features. Recently, Coleman and his associate^^^*^^ have revived and modified a barrier model originally proposed by Brinkman et al.39This model is not completely satisfactory but it is capable of providing a very good fit to experimental data taken from some remarkably nonlinear MIM junctions. Coleman has been attempting to relate tunneling spectra to the obtained barrier parameters for tunnel diodes containing organic molecules. We feel it is appropriate to contrast the parameters obtained from our junctions doped with inorganic ions with his. As a consequence, we will use his TRPSQR model in the discussion that follows. The salient features of the TRPSQR barrier are presented in Figure 2. The adjustable parameters are the barrier heights &, &, and 43and the widths D and 5. The I-V characteristic of the junction is calculated by assuming the free electron model for the electron incident from the metal, and the WKB approximation is used to determine the transmission probability. Further, one assumes that the metal-insulator boundaries are perfectly sharp. The appropriate equation is
l o
dx d E , l m [ f ( E )- f(E + eV) dE, (1) where j is the current per cm2 area of the junction, and all quantities are in cgs units. E = E, + E, is the total energy of the electron, E, is the energy due to motion in the x direction, a' is the height of the barrier as measured from the bottom of the conduction band of the left (base) metal, f(E) is the Fermi function, and x = 0 at the left metal-insulator interface. Because all reported data were taken a t temperatures below 4 K, the Fermi function integral was evaluated in the T 0 limit to give
-
I = 1.6185 x 101l[VJiD(u)du
+ J v0 D ( u ) u du)
(2)
where I is the current in milliamps per mm2, V is the bias voltage in volts, u = q - E,, q = fermi energy measured from the bottom of the conduction sand of the left-hand metal, and
D(u) = exp(-0.683124i
D+S
( @ ( x , V ) + ~ ) ' / ~ ( 3 / 2dx) ) (3)
for S 5 x I D
+S
(sa)
where eq 5 assumes that x is measured from right to left with x = 0 at the right metal surface. The potential specified in eq 4 and 5 allows eq 3 to be integrated exactly to provide a closed form expression for D(u). Equation 2 is then integrated numerically to obtain the I-V curve as a function of the barrier parameters of Figure 2. It was empirically found that, for voltages less than 2, the integration from V to q could be approximately terminated at 6 eV and very accurately terminated (
