J . Phys. Chem. 1988, 92, 1013-1016 aAZ -at
regards electron transport in an electric field the same second-order rate law as previously established for simple electron hopping will apply.9 The overall rate law in terms of fluxes is thus the same as is eq 1 where A is replaced by AZ and DE by Dc as derived from eq 2 or 3. In electrochemical applications where a chemical potential gradient is created by the reduction of AZ or the oxidation of B at an electrode surface, the electroinactive counterion Z- must be transported through the redox polymer film (from the electrode toward the solution in the first case and vice versa in the second). It follows that the charge-transport mechanism then follows the overall foregoing reaction scheme where electron hopping between A+ and B sites and physical displacement of Z - coexist with electron-ion hopping between AZ and B sites. The rate laws are thus given by the following relationships (by analogy with previous treatments6)
1013
-
aA _ at
introducing three diffusion coefficients: Dc = kCAX2CoE, DE = kEhXZcE,and QI = kIAx2(the definitions of the various symbols are shown in Scheme I).
(9) Since the potential energy curves are symmetrical and in the case the electrical potential variation between two adjacent redox sites are small, the electron-ion transfer coefficient (i.e., the symmetry factor) can be considered as close to 0.5 as required by eq 1.4‘ For large spatial variations of the electrical potential the linearization implied in eq 1 may not remain valid and the full exponential variation of the electron-ion transfer kinetics with potential may have to be considered.
Acknowledgment. Discussions with C. P. Andrieux (UniversitE Paris 7) on the matter of the present paper were, as always, very helpful.
Preparation of Ultrafine Amorphous Fe,, C, Alloy Particles on a Carbon Support Jacques van Wonterghem and Steen Marup* Laboratory of Applied Physics II, Technical University of Denmark, DK- 2800 Lyngby. Denmark (Received: October 29, 1987)
Iron pentacarbonyl has been thermally decomposed on a carbon support at 353 K. By use of Mossbauer spectroscopy it is shown that the reaction leads to formation of ultrafine amorphous Fq8C22 alloy particles with a mean diameter of about 3.9 nm. The particles exhibit a superparamagnetic behavior at 80 K.
Introduction Thermal decomposition of iron carbonyls on various supports has been studied e~tensively.l-~The aim of most of the work has been to form ultrafine metallic particles which may have interesting catalytic properties. Recently, we have studied the thermal decomposition of iron pentacarbonyl in organic liquids with a view to prepare ferrofluids containing metallic particles with a large It was found that the decom(1) Phillips, J.; Clausen, B. S.;Dumesic, J. A. J. Phys. Chem. 1980, 84, 1814. (2) Phillips, J.; Dumesic, J. A. Appl. Surf. Sci. 1981, 7 , 215. (3) Lazar, K.;Matusek, K.;Mink, J.; D o h , S.;Guczi, L.; Vizi-Orosz, A.; Marko, L.; Reiff, W. M. J. Catal. 1984, 87, 163. (4) Schay, Z.; Lazar, K.;Mink, J.; Guczi, L. J . Catal. 1984, 87, 179. ( 5 ) Rojas, D.;Bussiere, P.;Dalmon, J. A.; Choplin, A.; Basset, J. M.; Olivier, D. Surf. Sei. 1985, 156, 516. (6) Trautwein, A. X.;Bill, E.;Bl&, R.; Doppler, G.; Seel, F.; Klein, R.; Gonser, U.Surf. Sci. 1985, 156, 140. (7) Doppler, G.; Bill, E.;Gonser, U.; Seel, F.;Trautwein, A. X . Hyperfine Interact. 1986, 29, 1307. (8) van Wonterghem, J.; Merup, S.;Charles, S. W.; Wells, S.;Villadsen, J. Phys. Rev.Lett. 1985, 55, 410. (9) van Wonterghem, J.; Msrup, S.; Charles, S. W.; Wells, S.; Villadsen, J. Hyperfine Interact. 1986, 27, 333. (10) Msrup, S.;Christensen, B. R.;van Wonterghem, J.; Madsen, M. B.; Charles, S . W.; Wells, S . J. Magn. Magn. Mater. 1987, 67, 249. (1 1) van Wonterghem, J.; Msrup, S.; Charles, S.W.; Wells, S.J. Colloid Interface Sci., in press.
0022-3654/88/2092-lO13$01 S O / O
position of the iron pentacarbonyl led to formation of amorphous Fel,CX alloy particles. We have also shown that amorphous particles of Fe-B, Fe-Co-B, and Fe-Ni-B alloys can be formed by chemical reactions at a low temperature.12 Amorphous alloys are normally prepared as thin ribbons or films by the liquid quench technique or by vapor deposition. In order to avoid crystallization during the preparation, the material must be cooled rapidly to a temperature below the glass transition temperature. However, when an alloy is formed by a chemical method at a low temperature, the material may be amorphous if the chemical reaction takes place below the glass transition temperature of the alloy.8~12 In this Letter we show that ultrafine amorphous Fel& particles can be prepared on a carbon support by thermal decomposition of Fe(CO),. Such supported particles are of great interest because they can be used for studies of surface and catalytic properties of amorphous alloys. Experimental Section The sample was prepared in a sample holder consisting of a copper tube closed in both ends with heat resistant plastic film (Kapton). A hole was drilled in the copper tube through which the sample material could be injected. Before impregnation, the ~
~
~~~
~~~
~~
~
(12) van Wonterghem, J.; Msrup, S.; Koch, C. J. W.; Charles, S.W.; Wells, S.Nature (London) 1986, 322, 622.
0 1988 American Chemical Society
1014
Letters
The Journal of Physical Chemistry, Vol. 92, No. 5, 1988
.,
'. .*.'
. .... '
.. '*
,
'
-8
-12
-4 0 4 Velocity (mm/s)
8
Figure 2. Mossbauer spectra of Fe(CO)5on a carbon support obtained after various times of heating at 353 K. The spectra were obtained at 80 K in an applied magnetic field of 0.8 T.
27k
r---l-l
26
,' ! - 8 - 4 0 4 Velocity (mm/s)
8
- c
v
D
Figure 1. Mossbauer spectra of Fe(CO)Son a carbon support obtained after various times of heating at 353 K. The spectra were obtained at 80 K in an applied magnetic field of 0.01 T.
carbon support (Black Pearls 2000, 1475 m2 g-I, Cabot) was treated in a flow of pure hydrogen at 675 K. After cooling it was filled into the sample hdlder, and Fe(CO)5 (Merck) was injected by use of a syririge. Immediately after, the sample was frozen in liquid nitrogen. Because Fe(CO)5 is a poisonous liquid at room temperature, the impregnation was performed in a fume cupboard. The frozen sample was mounted in the Mitssbauer in situ cell described previ0us1y.l~ During the experiments the sample was kept in a flow of pure hydrogen in order to avoid oxidation. The sample was heated several times at 353 K for periods of 2 h. After each heating a Mossbauer spectrum was obtained at 80 K. This procedure was repeated until all Fe(CO)5 was decomposed, Le., after a total heating time of 12 h. Mixssbauer spectra were measured with a constant-acceleration spectrometer. Magnetic fields between 0.01 (the remanence field of the electromagnet) and 0.8 T were applied perpendicularly to the y-ray direction. Velocities and isomer shifts are given relative to the isomer shift of a-Fe at room temperature.
Results and Discussion Mossbauer spectra of the sample, obtained at 80 K in the remanent field of the magnet (0.01 T) after various heating times, are shown in Figure 1. Before the first heating the spectrum consists of a quadrupole doublet with a splitting AEQ = 2.45 f 0.02 mm s-I and an isomer shift 6 = -0.09 0.02 mm s-l, This component is due to Fe(CO)5. Upon heating, two new components appear, Le., a Fez+ quadrupole doublet with AEQ = 2.40 f 0.02 mm sd and 6 = 1.10 f 0.02 mm s-l and a broad singlet component
*
(13) Clausen, B. S.; Msrup, S.; Nielsen, P.; Thrane, N.; Topsm, H. J . Phys. E 1979, 12, 439.
25-
24
-
23'
1:o
2'0
310 410 8-'(T-')
510
bo
'
Figure 3. Induced magnetic hyperfine field at 80 K as a function of the reciprocal applied magnetic field for the sample prepared by thermal decomposition of Fe(CO)Sfor 12 h.
with an isomer shift of 0.25 f 0.03 mm s-'. For heating times exceeding 6 h the singlet line develops shoulders a t about f4.5 mm s-I. No intermediate iron carbonyl complexes were observed. Figure 2 shows spectra obtained a t 80 K with an applied magnetic field of 0.8 T after various heating times. The application of the magnetic field leads to a substantial magnetic splitting of the singlet component. This result shows that the singlet component is due to small magnetic particles which are superparamagnetic at 80 K.'"Ig The line widths of the six-line component are considerably larger than the natural line width. For large applied magnetic fields the induced magnetic hyperfine field in a superparamagnetic particle is given b ~ ' ~ , ' ~ - ' * (14) Msrup, S.; Dumesic, J. A.; Topsee, H. In Applications of Mbsbauer Spectroscopy; Cohen, R. L., Ed.;Academic: New York, 1980; Vol. 11, p 1. (15) Merup, S.; Topm, H.;Clausen, B. S. Phys. Scr. 1982, 25, 713. (16) Merup, S . J. Magn. Magn. Mater. 1983, 37, 39. (17) Merup, S. In Mossbauer Spectroscopy Applied to Inorganic Chemistry; Long, G . J., Ed.; Plenum: New York, 1987; Vol. 11, p 89. (18) Christensen, P. H.; Merup, S.; Niemantsverdriet, J. W. J . Phys. Chem. 1985, 89, 4898.
Letters
The Journal of Physical Chemistry, Vol. 92, No. 5, 1988 1015
The value of Bind is found by adding the applied magnetic field to the total magnetic field at the nucleus, estimated from the Mossbauer s p e ~ t r a . l ~ Bo - ' ~is the saturation hyperfine field, k is Boltzmann's constant, T is the temperature, p is the magnetic moment of the particle, and B is the applied magnetic field. Equation 1 is derived on the assumption that the magnetic anisotropy is negligible. However, for a sample with random orientation of the easy directions of magnetization, eq 1 can be used even in the presence of a significant magnetic anisotropy.Ig The sample which had been heated at 353 K for 12 h was studied by Mossbauer spectroscopy at various applied magnetic fields, and the value of Bid was plotted as a function of B' (Figure 3). The experimental results were fitted with a straight line in accordance with eq 1. From the intercept we found that Bo = 27 f 0.5 T at 80 K, which is smaller than the hyperfine field of 4 The quadrupole shift of the magnetically split a-Fe ( ~ 3 T). component is negligible. In an earlier study it was found that even very small (2.5 nm) particles of a-Fe had essentially the same saturation hyperfine field as bulk a-Fe.I8 Therefore, it can be concluded that the present particles do not consist of a-Fe. The broad lines in the Miissbauer spectra and the relatively large isomer shift support this conclusion. The Mossbauer parameters are also different from those of crystalline iron carbides.2021 The magnetic hyperfine field, the isomer shift, the line widths, and the (negligible) quadrupole shift are, however, very similar to the values found in Mossbauer studies of particles prepared by thermal decomposition of Fe(CO)5 in organic Studies of these particles by X-ray diffraction and the fact that they transform into a mixture of crystalline iron carbides and a-Fe during heating a t 523 K showed that they consisted of an amorphous Fel,C, alloy.s The carbon-supported particles are very pyrophoric and can therefore not easily be studied by, for example, X-ray diffraction. However, the fact that the Mossbauer spectra are clearly different from those of crystalline particles of iron carbides and a-Fe and the similarity to the spectra of the amorphous Fel,C, particles prepared by thermal decomposition of Fe(CO), in organic liquids show that the carbon-supported particles also consist of an amorphous Fel& alloy. By comparing the saturation hyperfine field and the isomer shift with the values obtained by Bauer-Grosse et al.22923for amorphous Fel-,C, films prepared by sputtering, we find that x 0.22. Our results therefore show that the thermal decomposition of Fe(CO)5 on the carbon support lead to formation of small superparamagnetic amorphous particles of Fel-,C, with x = 0.22. The carbon atoms probably enter into the particles during the preparation as a result of chemisorption and disintegration of CO on the surface of the particles. Iron metal catalysts are carburized by such a mechanism during FischerTropsch s y n t h e s i ~ . ~ ~ , ~ ~ From the slope of the straight line in Figure 3 we find an average magnetic moment of the particles, p = (4.2 f 0.5) X J TI.The mean magnetic moment per iron atom in amorphous Fe&22 is about 1.9 pLeeZ3 Thus, the average number of iron atoms per particle is about 2400. Assuming that the density is identical with that of crystalline Fe,C, we find an average diameter of 3.9 f 0.2 nm assuming spherical particle shape. The present carbon-supported particles exhibit a longer superparamagnetic relaxation time (T = 5 X s) at 80 K than
-
(19) Msrup, S.;Christensen, P. H.; Clausen, B. S. J . Mugn. Mum. Muter. 1987, 68, 160. (20) Le Catr, G.; Dubois, J. M.; Pijolat, M.; Perrichon, V.; Bussiere, P. J. Phys. Chem. 1982,86, 4799. (21) Ma, C.-B.; Ando, T.; Williamson, D. L.; Krauss, G. Metall. Trans. A. 1983, 14, 1033. (22) Bauer-Grosse, E.; Le Caer, G.; Fournes, L. Hyperfine Interact. 1986, 27, 297. (23) Bauer-Grosse, E.; Le Catr, G. Philos. Mug. B 1987, 56, 485. (24) Amelse, J. A.; Butt, J. B.; Schwartz, L. H. J . Phys. Chem. 1978,82, 558. (25) Raupp, G. B.; Delgass, W. N. J . Coral. 1979, 58, 348.
n
.'c/\ 0: 30
.
K HEATING TIME
I (h)
Figure 4. Relative areas of the three components of the spectra, shown in Figure 1, as a function of the heating time.
the particles in the ferrofluid studied earlier.I0 This can be seen from the presence of broad shoulders at about f4.5 mm s-l in the Mossbauer spectrum (Figure 1). Such shoulders were not observed in the spectrum of the particles in the ferrofluid at 80 KIo although the magnetic moment of these particles, and therefore also the volume, was about 50% larger. The superparamagnetic relaxation time increases with increasing values of the product of the volume, V, and the magnetic anisotropy energy constant, K.Iel7 The difference in relaxation times for the two samples therefore shows that the carbon-supported particles have a considerably larger magnetic anisotropy energy constant than the particles in the ferrofluid. From the value of 7 at 80 K one can estimate that K = (0.9 f 0.3) X lo5 J m-3,10,18 The difference in magnetic anisotropy energy constants is probably due to the fact that particles prepared in a liquid have nearly spherical shape,' and therefore the shape anisotropy is small. Particles which are prepared on a support may be nonspherical and may therefore have significant shape anisotropy. Moreover, the interaction with the support may also result in contributions to the stress anisotropy and surface anisotropy.14 The kinetics of the decomposition of Fe(C0)5 is illustrated in Figure 4 which shows the relative areas of the three components in the Mossbauer spectra of Figure 1 as a function of the heating time. If we assume identical f factors for the three iron species, Figure 4 illustrates the relative amounts of iron in the three species. After approximately 8 h, essentially all Fe(CO)5 has decomposed and about 90% of the iron atoms are found in the amorphous Fe78C22particles. The remaining 10% of the iron atoms are in Fez+ complexes. It is interesting to compare our results with the results of previous Mossbauer studies of decomposition of iron carbonyls on various supports. Phillips et al.'J found that the decomposition of Fe(CO), on grafoil led to a phase which had a six-line Mossbauer spectrum with broad lines and a magnetic hyperfine field below the value of bulk a-Fe. The isomer shift was slightly larger than that of a-Fe. The authors proposed that the component was due to a-Fe particles, and the reduction in the magnetic hyperfine field was explained by the influence of collective magnetic excitations. Rojas et ale5decomposed Fe3(CO),, on a silica support in hydrogen at 393 K. This resulted in a magnetic component with very broad lines and a room-temperature hyperfine field of about 23 T. The component was attributed to iron carbide particles. Trautwein et aL6s7have studied the decomposition of Fe(CO)5 in zeolites of the faujasite type. These authors found that after intercalation and thermal decomposition of Fe(C0)S the 4.2 K spectra of a Na-X zeolite contained small amounts of a-Fe and a broad component which was tentatively attributed to superparamagnetic a-Fe. In view of our results for decomposition of Fe(CO)Son a carbon
J. Phys. Chem. 1988, 92, 1016-1018
1016
support, it is likely that the components with broad lines found in these earlier studies of decomposition of iron carbonyls may at least partly be attributed to the presence of amorphous Fe,,C, particles. Our studies of chemically prepared amorphous alloy particles suggest that whenever alloy particles are formed at a low temperature the particles may be amorphous if the reaction takes place below the glass transition temperature.
Acknowledgment. The work was supported by the Danish Technical Research Council and the Danish Natural Science Research Council. We are grateful to G. Le Caer for sending a copy of ref 23 prior to publication. Registry No. Fe(CO)5, 1528 1-98-8; C, 7440-44-0; Fe78C22, 9384477-0.
Lumlnescence Decay of Ruthenium( I I ) Complexes Adsorbed on Metal Oxide Powders In Vacuo: Energy Gap Dependence of the Electron-Transfer Rate K. Hashimoto,t M. Hiramoto,+A. B. P. Lever,$and T. Sakata*+ Institute for Molecular Science, Myodaiji, Okazaki 444, Japan, and Department of Chemistry, York University, Downsview, Ontario, M3J I P3 Canada (Received: November 9, 1987)
Luminescence decays of R ~ ( b p y ) , ~and + R ~ ( b p z ) , ~adsorbed + on powdered S O 2 , SrTi03, Ti0, (anatase), Ti02 (rutile), ZnO, and SnO, were measured in vacuo. The decay curves depended on the substrates and were fitted with the sum of four exponentials. Electron-transfer rates from the excited state of the Ru(I1) complexes to semiconductors were evaluated from the decay rates.
Introduction Photochemical reactions on solid surface currently attract attention from a practical viewpoint for uses such as solar energy conversion and preparation of thin films. One of the important and interesting problems concerning such systems is electron transfer (ET) from photoexcited states of adsorbed dye molecules to semiconductors. E T between molecules and semiconductors has been less well-studied compared to that among molecules in homogeneous solution.' We previously reported luminescence decays and time-resolved spectra for Ru(bpy)?+ and its derivatives adsorbed on powdered TiO, and SiO, in vacuo,z in the presence of water vapor: and in several solvents! Here we measured the luminescence decays of R ~ ( b p y ) , ~and + R u ( b p z ) p adsorbed on various powdered semiconductors and insulators in vacuo and observed an interesting semiconductor dependence of E T rate. Experimental Section For sample preparation, the particles were allowed to stand in contact with a water solution of the Ru(I1) complex in which the ratio of particle to complex weights was controlled and then dried by evacuation Torr) at 50 OC for a few days. The surface coverage was calculated by assuming that the Ru complexes were spheres with a radius of 5 A and were adsorbed uniformly on the surface. The samples were set in an optical cryostat (Oxford Instruments Limited, C F 1104) a t room temperature and were evacuated to less than Torr for more than 20 h before each measurement. The decay curves were observed at 580 nm for Ru(bpy)?+ and 570 nm for R ~ ( b p z ) , ~ +Decay . curves were measured with some differing time scales (subnanosecond to microsecond region) by using nano- and picosecond photon counting apparatus in the instrument center of IMS, and decay curves were analyzed by the method reported p r e v i ~ u s l y . ~ . ~ Results and Discussion Luminescence Decay. The luminescence from Ru(I1) complexes adsorbed on metal oxide powders decayed nonexponentially. When the surface coverage was varied from 1/ 10 to 1/ 100 of a monolayer for the semiconductors, from 1/ 10 to 1/ 1000 of a monolayer Institute for Molecular Science. *York University.
0022-3654/88/2092- 1016$01.50/0
TABLE I: Lifetimes of Four Decay Components and Relative Weights of Preexponential Factors of the Luminescence from Adsorbed (a) Ru(bpy)32+and (b) Ru(bpz)d+, and Calculated ET Rates
ns (1,) k,, s-l
il,
substrate
ns (1,) k2, s-'
ns
i3, (1,)
72,
k,,
(a) 0.8 (0.70) 1.3 x 109 ZnO 1.4 (0.73) 7.1 x io8 Ti02 (rutile) 1.8 (0.74) 5.6 X IO8 Ti02 (anatase) 3.6 (0.64) 2.8 X lo8 SrTi03 2.0 (0.35) 5.0 X lo8
Ru(bpyh2+
Sn02
17 (0.22) 5.8 x 107 36 (0.20) 2.7 x 107 36 (0.19) 2.7 X lo7 37 (0.23) 2.6 X lo7 51 (0.18) 1.9 X lo7
(b) 1.8 (0.52) 5.6 X lo8 Ti02 (rutile) 1.9 (0.61) 5.3 X lo8 Ti02 (anatase) 3.4 (0.37) 2.9 X lo8
RU(bPz)i2+
Sn02
SrTiOpa
24 (0.36) 4.1 X IO7 36 (0.19) 2.7 X lo7 73 (0.29) 1.3 X lo7 85 (0.28) 1.1 x 107
s-l
80 (0.07) 1.2 x 107 90 (0.05) 1.0 x 107 130 (0.05) 6.9 X lo6 140 (0.10) 6.4 X lo6 210 (0.32) 4.0 X lo6 140 (0.09) 6.6 X lo6 130 (0.05) 7.1 X lo6 280 (0.22) 3.0 X lo6 350 (0.49) 2.3 x io6
7,.
ns (1,)
k,, s-' 510 (0.01) 1.2 x io6 500 (0.02) 1.2 x 106 490 (0.02) 1.3 X lo6 490 (0.03) 1.3 X IO6 730 (0.15) 6.0 X lo5 590 (0.03) 1.1 x 106 670 (0.02) 9.0 x 105 860 (0.07) 5.7 x 105 1100 (0.23) 3.2 x 105
"See ref 5
for SO,, and from 1/ 100 to 1/ 10000 of a monolayer for PVG, the luminescence decays were found to be independent of coverage. Therefore, in this concentration region, concentration quenching can be neglected as it is clear that the emission arises from mo(1) For example: (a) Ke-stner, S.R.; Logan, J.; Jortner, J. J . Phys. Chem. 1974, 78, 2148. (b) Kakitani, T.; Mataga, N. Chem. Phys. 1985, 93, 381. (c) Closs, G. L.; Calcaterra, L. T.; Green, N. J.; Penfield, K. W.; Miller, J. R.J . Phys. Chem. 1986, 90, 3673 and references therein. ( 2 ) Kajiwara, T.; Hashimoto, K.; Kawai, T.; Sakata, T. J. Phys. Chem. 1982, 86, 4516. (3) Hashimoto, K.; Hiramoto, M.; Kajiwara, T.; Sakata, T. J. Phys.
Chem., submitted for publication. (4) (a) Takemura, H.; Saji, T.; Fujihira, M.; Aoyagui, S.;Hashimoto, K.; Sakata, T. Chem. Phys. Left. 1985, 122,496. (b) Hashimoto, K.; Hiramoto, M.; Sakata, T.; Muraki, H.; Takemura, H.; Fujihira, M. J. Phys. Chem. 1987, 91, 6198.
0 1988 American Chemical Society