J. Phys. Chem. 1982, 86, 2818-2827
2818
There is no reason not to expect significant progress regarding nonequilibrium fluctuations in the near future. Both theory and measurement are making good progress, and novel systems are being studied. Light-scattering measurements for systems subjected to thermal gradients also have bearing’” on long-range order in nonequilibrium systems, long-time tails in autocorrelation functions, and the nonanalytic behavior of the virial expansion for transport coefficients. The phenomenon which is studied in this way is closely related to the emergence of long-range
spatial order in reaction-diffusion systems3’ which are driven to steady states. If the theory can be confirmed in the hydrodynamic setting, then more confidence can be felt for its application to the reaction context. Acknowledgment. This work was supported by NSF Grant PHY 7921541. (31) D. A. McQuarrie and J. E. Keizer, Theor.Chem.: Adu. Perspect., 6A,165 (1981).
ARTICLES Electron Paramagnetic Resonance and Infrared Studies of the Genesis and Reactivity toward Carbon Monoxide of Ni+ Ions in a NiCa-X Zeolite M. Kermarec,* D. Ollvler, M. Richard, M. Che, Laboratoire de Chlmie des SolMes, ER 133-CNRS, Universit6 Pierre et Marie Curie, 75230 Paris Cedex 05, France
and
F. Bozon-Verduraz
Laboratoire de Catalyse et Spectro&trIe, Universtti?Park V l l , 7525 1 Paris Cedex 05, France (Receive& M y 11. 198 1; In Final Form: December 28, 198 1)
Ni+ ions have been generated by reduction of an anhydrous NiCa-X zeolite with molecular hydrogen activated either thermally or photochemically or with atomic hydrogen produced by a microwave discharge. These ions have been characterized by their EPR signals and are found to occupy Sir, or SII,sites in the sodalite units. On adsorption of carbon monoxide at room temperature, these ions migrate from sodalite units to the supercages where they form carbonyl species. Using 13C-enrichedCO, it is possible from the hyperfine structures of the EPR spectra to identify Ni(C0)2+species (gl = 2.056, g2= 2.071, g3 = 2.206) at low pressures (a few torr) and Ni(C0)3+(gl = 2.012, g, = 2.114, g3 = 2.176) at higher pressures with a reversible change, one into the other, as a function of pressure. It is also possible from the IR results to identify Ni(CO)+(2115 cm-’1, Ni(C0)2+ (2145 and 2100 cm-’), and Ni+(C0)2Ni+species (2015 and 1955 cm-’), as well as CO adsorbed on atoms (2085 cm-’) or aggregates (2060 cm-’) of nickel. These aasignmenta have been confiimed by using 12CO-13C0mixtures. Upon outgassing at moderate temperatures (below about 423 K), the carbonyl species gradually loose their ligands to finally recover their initial location in the sodalite units. The formation of the Ni+ unusual oxidation state is explained from the reverse disproportionation reaction Ni2+ + Nio 2Ni+ and from the stabilizing effect of the ligands (H2,CO). The above equilibrium is found to be reversible as long as the nickel atoms do not agglomerate to form large particles. Provided this condition is fulfilled, the genesis and the destruction of the carbonyl complexes, characterized by EPR and IR, can be reversible to a large extent. Introduction Since the pioneering work of Rabo et al.,’ several authors have reported on the formation of Ni+ ion^.^-^ Recently attention has been focused on the formation of Ni+ ions in nickel-loaded zeolites, and various studies, involving EPR, UV-visible, and infrared spectroscopies, have appeared.- Reducing a Ni(72%)-Y zeolite by sodium va(1) J. A. Rabo, C. L. Angell, P. H. K a d , and V. Shomaker, Discuss.
Furuday SOC.,41, 328 (1966).
(2) L. Porri, G. Vittuli, and M. C. Gallazzi, Angew. Chem., Znt. Ed. Engl., 6,452 (1967). (3) E. Dinjue and R. Kirmse, 2.Chem., 16,286 (1976). (4) (a) C. Ammo and S. Fujiwara, Bull. Chem. SOC.Jpn., 46, 1379 (1973); (b) C. Amano, T. Watanabe, and S. Fujiwara, ibid., 46, 2586 (1973); (c) C. Amano and S. Fujiwara, ibid.,49, 1817 (1976). ( 5 ) E. Garbowski,M. V. Mathieu, and M. Primet, Chem. Phys. Lett., 49,247 (1977). (6) E. Garbowski and J. C. V W e , Chem. Phys.Lett., 48,550 (1977). (7) P. H. K d , R. J. Bishop, and D. McLeod, Jr., J.Phys.Chem., 82, 279 (1978). 0022-365418212086-2818$01.25/0
por, Rabo et al.’ observed a gmall g, EPR component at 2.065 assigned to Ni+ at Sn sites, while a Ni(5%)-Y sample exhibited a sharp signal at g, = 2.094 ascribed to Ni+ at SI sites. For a NiCa-Y sample reduced by molecular hydrogen at 473 K, a g, component at 2.096 was assigned to Ni+ ions in “accessible position”6while in NiCa-X a species giving the same g, value was identified, with the help of ENDOR experiments, as Ni(H2)+located at SI,sites? It follows that information concerning the influence of the reducing conditions on the nature and location of Ni+ species is still much needed. In addition, the observations concerning the reactivity of these species toward CO are Contradictory; infrared bands observed at 1930 and 2140 cm-’ on a reduced NiCa-Y zeolite, first ascribed to Ni2(CO)and Ni(8)E. Garbowski, M. Primet, and M. V. Mathieu, ACS SymD. - . Ser., No. 40,281 (1977). (9) D. Olivier, M. Richard, M. Che, F. Bozon-Verduraz, and R. B. Clarkson, J. Phys. Chem., 84,423 (1980).
0 1982 American Chemical Society
Ni+
Ions in a NCa-X Zeolite
(CO)+ species, respectively,8 were attributed later to a Ni(CO),+ complex,S while detailed experiments performed on dehydrated Ni-mordenite’ showed that such a complex is responsible for bands at 2094 and 2140 cm-’. In the case of the Ni-Y sample reduced at 350 “C, Ione et al.l0 assigned a band located at 2145 cm-l to Ni(CO)+ and bands observed in the 2020-2120-cm-’ range to NiO(CO). Moreover, very recent results obtained in our laboratory gave evidence for the role of ion migration and the intricacy of the mechanisms involved in the CO-NiCa-X zeolite interaction.’l Undoubtedly the influence of factors such as CO pressure and outgassing conditions upon the migration of nickel ions and the nature of carbonyl species did not receive enough attention; these considerations prompted us to reinvestigate the reduction of NiCa-X zeolites and their interaction with carbon monoxide, by using conjointly EPR and IR spectroscopies with a particular attention to the experimental conditions. The reduction was carried out by molecular hydrogen either thermally or photochemically activated, and hydrogen atom beams were also employed.12 The role of CO pressure and outgassing conditions was examined with a special emphasis.
Experimental Section The starting material, a Linde Na-X synthetic zeolite, was successively exchanged with Ca2+and Ni2+ ions by using 0.1 M CaC12and Ni(NOJ2 solutions and then heated at 873 K in air. Chemical analysis of the sample led to the unit cell formula NiloCa20Na4H22(Si02)los(A102)861 which will be referred to as NiCa-X. The activation procedure consisted of pretreatments at 773 K f i t in oxygen (160 torr, 3 h) and then in vacuo (15 h) and of a reduction treatment by hydrogen followed by outgassing for 15 min at the reduction temperature. In EPR experiments, this reduction was performed in 0 of H2, different ways: (i) thermally at 473 K ( 1 ~ 5 0 torr overnight); (ii) photochemically by UV irradiation in hydrogen (200 torr) at 77 K using a helical high-pressure mercury lamp Claude 360VA; (iii) by an atomic hydrogen beam produced by a 2450-MHz microwave discharge,12the cold plasma of hydrogen atoms being formed outside the sample. The spectra were recorded at 77 K on a CS-E 109 Varian spectrometer equipped with a dual cavity in the X-band mode and a subharmonic generator (Telmore Instruments) for recording third derivative spectra, using multiple field modulation. Carbon monoxide (up to 500 torr) was introduced onto the sample either at room temperature or at 77 K. In IR studies, the sample was pressed (3 ton cm-2) into pellets weighing about 6 mg cm-2. The reduction was performed only thermally (100-500 torr of H2 for 4 h) in the 473-573 K range. In some experiments, longer outgassing times ( 1 2 h) than the standard one (15 min) were employed. The spectra were recorded on a Perkin-Elmer 521 spectrophotometer at room temperature; the effective sample temperature in the IR beam, which could be called the “beam temperature”, was about 315 K. CO adsorption experiments were carried out at this temperature, a reference cell being used for gas-phase compensation. 13Cenriched CO (90%) supplied by the Commissariat 11’(10)K.G.Ione, V. N. Romannikov, A. A. Davydov, and L. B. Orlova, J. Ccrtal., 57, 126 (1979). (11)D.Olivier, M.Richard, and M. Che, Chem. Phys. Lett., 60,77 (1978).For Ni(C0)2+the hyperfine tensor should read (p 80)us = 25 G , a2 al = 31 G. (12)M. Che, M. Richard, and D. Olivier, J. Chem. SOC.,Faraduy Trans. 1 , 76, 1526 (1980).
-
The Journal of Physical Chemistry, Vol. 86, No. 15, 1982 2810
Figure 1. EPR spectra obtained upon exposing species A (see text) to 20 torr of ‘*CO (1) at 77 K; (2) at room temperature.
Energie Atomique (France) was used without further purification. All other gases were high-purity grade and obtained from Air Liquide (France).
Results and Discussion (1)EPR Spectroscopy. (1.1)Influence of the Reducing Conditions on the Nature of Ni+ Complexes in NiCa-X Zeolite. Evidence was given that reduction of NiCa-X, by any of the three procedures described above, produced Ni+ species. Reduction by molecular hydrogen at 473 K gave rise to a Ni(H2)+speciesg characterized by an EPR signal (A) with g , = 2.096 and a line width peak to peak AHpp= 50 G at 77 K. The photochemical method at 77 K led to an axially symmetric EPR signal (B) with g , = 2.069 and a doublet assigned to adsorbed Ho atoms. When the sample was warmed to room temperature, the EPR signal of Ho disappeared and B changed into a new signal (A’) characterized by g , = 2.090 and AHpp= 40 G at 77 K. The value of the gllcomponents could not be reasonably estimated because of the presence of iron impurities. Reduction by a beam of hydrogen atoms at 273 K produced directly the species A’ provided that the concentration of Ho atoms was kept low enough to avoid the formation of NiO aggregates (by controlling the microwave generator power and the distance between the sample and the microwave cavity). (1.2)Adsorption of Carbon Monoxide on Ni+ Species. Influence of the Pressure on the Stability of the Complexes Formed (Table I). Adsorption of CO on species A was carried out under 20 torr at 77 K and followed as a function of time. The signal A decreased rapidly and was partly replaced by a signal (B’) at g , = 2.069 and g , = 2.16 (Figure 1, spectrum 1). When the sample was warmed to room temperature, a new signal (D)increased at the expense of B’, which finally disappeared (Figure 1, spectrum 1). Signal D has been recently assigned to Ni(CO),+(ref 6 and 7). The use of the third derivative accessory and 13C-enrichedCO led to the following g and hyperfine tensors (Figure 2): gl = 2.056, g2 = 2.071, and g, = 2.206; A I = A2 = 33 G and A , = 27.5 G . When CO was adsorbed at a higher pressure (100 torr) at 77 K, A decreased whereas B’ increased conjointly with another complex signal (C) which further increased with time at the expense of B’ (Figure 3). A detailed study of the intensity and shape of C vs. temperature and microwave power revealed that it was due to the complex superimposition of three main signals, C1, Cz, and Cs. Cz, which is a 2g signal (2.226 and 2.006), is due to a transitory species since it disappeared irreversibly when
2820
The JOUrnal of phvslCal Chemisby, Vol. 86, No. 15, 1982
Kermarec et al.
173K
,A3,
2 .‘056 Figure 2. EPR spectra (thkd derivative) recorded after contacting species A (seetext) at room temperature with 20 torr of ‘%O or 13W.
c2
/
,
c3
i
2.069
Figwe 3. EPR spectra obtained on contacting species A at 77 K with 100 torr of ‘%O (full line, upper spectrum) or 100 torr of I3CO (dotted line, lower spectrum).
the sample was heated to room temperature after CO adsorption at 77 K. Because of its transitory nature, Cz will not be further discussed. C1 and C3are stable species, since they remained in the spectrum after heating to room temperature. Their relaxation times however are quite different, since at 173 K the 3g signal C1 with gl = 2.012, gz = 2.114, and g3= 2.175 was broadened beyond detection leaving C3apparently as the only remaining signal (Figure 4). From the temperature dependence (Figure 4) and from the relative intensities of the various peaks and shoulders present in the composite signal obtained with variow samples in slightly different conditions, signal C3 itself was found to be the superimposition of two 2g signals, C3aand Cgb,whose g components were measured to be 2.072 and 2.205 for C3a and 2.015 and 2.16 for the latter. Attempts to better clarify the composite signal C by using the Q band were not satisfactory. The same experiments performed by using 90% 13Cenriched CO showed that the composite signal C was split into several lines whereas B’ was not modified (Figure 31, showing that B’ was not a carbonyl species. To identify the carbonyl species, we studied the composite signal C after the sample contacted with 13C-en-
Figure 4. EPR spectra obtained upon exposing species A at room temperature to 100 torr of ‘%O variable temperature recording. 83
d
Figure 5. EPR spectra recorded upon exposing species A to 100 torr of ’%o at room temperature.
riched CO at 77 K was warmed to room temperature in order to eliminate signal Cz. From the spectrum shown in Figure 5, it is seen that each component of the g tensor is split into a number of lines. The spectrum can be interpreted as being the superimposition of a well-resolved hyperfine structure belonging to C1 and a poorly resolved hyperfine structure due to CB. The difference in resolution between the two hyperfine structures is essentially linked to the narrow line width of C1 as compared to the larger one observed for C3 (Figure 5 ) . The nature of the carbonyl species can be obtained from the relative intensities of the hyperfine lines. These intensities can be predicted from the probability13 P J m ) = (n/m)(l - q)n-mqm n/m = n!/[m!(n - m)!] of having one, two, or three 13C0or l2C0 ligands coordinated to each Ni+, where n refers to the total number of possible ligands, m to the number of labeled ligands, and q to the percentage of 13C in CO. From this formula and with a 90% 13Cenrichment, one would expect a hyperfine structure with about the following relative intensities: 4.5:1:4.5 for Ni(CO)+,21:41:2 for Ni(CO)z+with equivalent (13) H. Margenau and G . M. Murphy, ‘Mathematics of Physics and Chemistry”, Van Nostrand, New York, 1956, p 438.
The Journal of phvslcsi Chemistry, Vol. 86, No. 15, 1982 2821
NI+ Ions In a NICa-X Zeolite
TABLE I C0,473K, 100 T o r r
CO, 77K, 1 00
TOrr
c,scz+c3
/\
CO, 293K, 20 T o r r
A'
Cor 293K
vacuo 423K
CO,
293K, 10-
EPR signal or species
'
b Torr
A'
A
B = B'
D,c,, c,
E
c,
a
a
gl
2.090
2.096
species assignment localization a See text.
Nit
SI#or Sui
Ni(H,)+ Ni+ or Ni(H,)+ Ni(CO)+ Ni(CO),+ or Ni(CO),+" Ni(CO),+ SI^ Sn' SI1 sn Sn
carbons, 1.5:1:5:2:5:1:1.5for Ni(C0)3+ with equivalent carbons where Ni+ is also bonded to three framework oxygen ions. Thus, if the carbonyl species contains an odd number of CO ligands, the hyperhe structure will be characterized by a central peak smaller in intensity than its nearest satellites in contrast to the case where there is an even number of CO ligands. From the relative intensity distribution and the number of hyperfine lines, one can conclude that C1 is probably due to a tricarbonyl species Ni(CO),+. It must be noticed that the observed relative intensity distribution does not fit completely the theoretical one; this model, indeed, does not take into account that the electron-nucleus broadening varies with MIand implies that the line due to 12C(I= 0)-labeled carbonyl species is not broadened compared to the other hyperfine lines. Analysis of Figure 5 gives the following hyperfine tensor: Al = 45 G and A2 N A3 = 49 G, in good agreement with earlier data." As for species C3, comparison between Figures 3 and 5 shows that the low-field peak of C3 is broadened when 13C-enrichedCO is used, demonstrating that C3 is also a carbonyl species. The use of higher pressures did not change significantly the overall shape of the signal composed of C1 and C3 except that the relative importance of C1 grew as a function of the CO pressure. In the case of C3, the central peak of the hyperfine structure centered around gl = 2.205 for C3nis larger in intensity than the other peaks; this indicates that C3 is either a dicarbonyl species with equivalent carbons (see above) or a tricarbonyl species with only two equivalent carbons. In the latter case, indeed, a tricarbonyl entity generated by 90% I3Cenriched CO will exhibit a hyperfine structure with relative intensities 21:4.4:1.2:5:1.2:4.41:2assuming that the smaller hyperfine constant is half that of the two equivalent carbons. The central peak has then the largest intensity. As, in addition, C3 is observed under a high CO pressure, in a domain where dicarbonyl species are no longer formed, it should be assigned to a tricarbonyl entity. Nevertheless, attempts to confirm this conclusion by using the Q band were not successful. By varying the CO pressure, one could change reversibly the low-pressure (-20 torr) carbonyl species D into the
2.069
2.40
/ 2.090
/
v 0 .1 5
,
2.020 2.b56
Figwe 6. EPFl spectra recorded upon outgassing species D (see text): (1) at room temperature; (2) at 423 K.
high-pressure carbonyl species C, the Ni(C0)3+appearing above 50 torr (Table I). The carbonyl species described above (D, C1, and C,) have been directly obtained through reduction of the NiCa-X sample by CO at 473 K or by adsorption of CO at 293 K on species A' (Table I). But in the latter case, we have to point out that at 77 K CO adsorption did not transform A' into B' even after 4 h of contact. Roomtemperature outgassing partly converted signal D-Ni(C0)2f-into a broad anisotropic signal (E) (Figure 6, spectrum 1). E is a composite signal with a broad per2.40 and two resolved gl, pendicular component at g, components at 2.015 and 2.020. Use of 13C-enrichedcarbon monoxide did not lead to any resolved hyperfine structure for the E signal. Outgassing at 423 K transformed E into a 3g signal E' (gl = 2,020, g, = 2.35, g3 = 2.42) (Figure 6, spectrum 2) and A' (gl = 2.090). A subsequent evacuation at 473 K for 3 h removed all of the EPR signals. This process was completely reversible, and introducion of CO at room temperature regenerated all of the signals previously described with the same intensities. If outgassing was prolonged overnight at 473 K, or performed at higher temperature, a broad ferromagnetic
-
2822
The Journal of physicel Chemktry, Voi. 86, No. 15, 1982
Kermarec et al.
TABLE IP
symmetry type ground state g II gl re1 order deformed tetrahedral with 8 d, , d,l-yl b b b 90" < p < 109" (degenerate) octahedral; ligands in a plane 8 d,z-y2 ge - 8h/(Ed,~-~z - Edxy) ge - 2h/(Edx2-y~- Edxzdyz) gll > g l > ge perpendicular to the symmetry axis dominate trigonal bipyramidaP or e dzz ge deformed tetrahedral with p > 109" octahedral; ligand along the 6 d,z ge symmetry axis dominates a See ref 17a and 18. h is the spin-orbit coupling constant of nickel; @meanscompressed and 6: elongated. Temperature-dependent EPR spectrum, because of the degeneracy of the ground state likely to be removed by Jahn-Teller effect. TABLE I11
4'
Pi
L4
M =cation ligands
L1 , L * , Lg , Lq
Figure 7. Tetrahedral symmetry with trigonal distortion.
signal appeared which was characteristic of a NiO metallic phase. (1.3) Discussion. The experimental results are summarized in Table I. Rabo et al.' have reported earlier the presence of two Ni+ sites in Ni-Y zeolite reduced by sodium vapor: one a t SI(gL = 2.094), the other at SII (9, = 2.065). In dehydrated Ca-Faujasite, most of the Ca2+ cations occupy sites I and 11,14and X-ray diffraction data have shown that, in the NiCa-X zeolite studied here, the initial Ni2+ions are mostly distributed between sodalite cavities and super~ages.'~Hence, the predominant sites available for Ni+ in anhydrous NiCa-X appear to be SI!, SII,,or SII. They have a C3"symmetry and are shown in Figure 7, where L1,L2, and L3 represent the oxygen neighbors located on the six-ring window and L4is a host ligand.lB Taking the threefold axis as the z axis, the trigonal distortion of the tetrahedral crystal field is characterized by the parameter /3 &ML4 angle) by moving M along the symmetry axis. The g values may be calculated for a 3d9 ion by using a simple crystal field modeZ7J74le(Table 11). Ni+ Species. The g tensors obtained for A, A', B, and B' are all axially symmetric and their respective perpendicular components are very close to those found for Ni+ ions in an Q-type crystal field (Table 111). All of these signals were not split in the presence of 13C-enrichedCO; hence, they do not contain carbonyl ligands. Since CO does not enter the sodalite cages,lgand as SI (14)J. M.Bennett and J. V. Smith, Mater. Res. Bull., 3,633(1968). (15)J. Jeanjean, unpublished results. (16)R. Kellerman and K. Klier, "Surface and Defects Properties of Solids", Vol. 4,The Chemical Society, London, 1975. (17) (a) A. Abragam and B. Bleaney, 'F&onance Paramagngtique fillectronique de8 Ions de Transition",Preases Universitaires de France, Paris,1971,p 460; (b) ibid., p 457. (18)H.G.Hecht and T. S. Johnston, J. Chem. Phys., 46,23 (1967). (19)T.A. Egerton and F. S. Stone, Trans. Faraday SOC.,66, 2364 (1970).
g components Ni' ions in a type-& crystal field g II gl ref Ni+ in LiF 2.53-2.62 2.10-2.11 17b Ni+at Sn in Ni-Y zeolite 2.065 1 2.094 1 Ni+at SIin Ni-Y zeolite Ni+ complexes in glassy 2.47-2.31 2.064-2.078 4a solvents
sites are filled with Ca2+cations, the species corresponding to A, A', and B must be located in the sodalite cages. Consequently, the conversion of A to B' upon adsorption of CO should reflect the variation of the Ni+ environment within the sodalite units from SI,to Sn,under the influence of CO trapped in the supercages. From the similarity of the signals B and B', we conclude that they correspond to similar species: Ni+ or Ni(HJ+ species located at Sn.sites. In the case of species B (prepared by photochemical reduction at 77 K), the stabilization of the Ni+ ion at Sn.may be explained by the presence of OH or H20 trapped in the supercage at this very low reduction temperature (Table I). It is possible to prepare the carbonyl species C1 and C3 by reduction of NiCa-X by CO (200 torr) at 473 K. As A' may be obtained through evacuation of these carbonyl species, A' cannot contain hydrogen. Since we have never observed the conversion of A' into B, A' may be located either in SUI or at SI,. Ni+ Carbonyl Species. Species D. From the hyperfine tensor components A, = 33 G and A,, = 27.5 G, the isotropic part a and the anisotropic traceless part b are found to be a ( W ) = 31.16 G and b ( W ) = 1.84 G from the relations (A,,AL,All) = a + (-b,-b,2b) a = (1/3)(2AL
+ Ail)
using a method recently reviewed.21 The molecular orbital coefficients C22 and Cm2,which characterize the spin density of the unpaired electron in the 2s and 2p carbon orbitals, are then calculated from the relations
a = C2;AO
b = Czp2Bo
using the theoretical values A. = 1110.8 G and Bo = 32.4 G20 for pure 2s and 2p carbon orbitals. This leads to (222 = 0.028 Czp2 = 0.057 Hence, when two CO molecules are bonded to one Ni+ ion, 17% of the unpaired spin density is found on the carbon atoms and thus 83% on Ni+, neglecting spin density on oxygen atoms. (20)D.H. Whiffen, J. Chem. Phys., 61, 1589 (1964). (21)J. H. Lunsford, Adu. Catal., 22, 276 (1972).
The J m i of Physical Chemistry, Vol. 86, No. 15, 1982 2823
Ni+ Ions in a NICa-X Zeolite
TABLE IV d9 ions in a type-d crystal field
g components gl
g II
ref
2.05 22 23, 24 2.01-2.04 2.00-1.99 25 2.002 7a 2.015 this work 2.020 Ni+ 2.35 2.06-2.08 17b Ni+ 2.364 2.012 4b Estimated values from EPR spectra given in ref 7. Pd+ cu+
2.30
{ 2.24-2.25 2.32-2.30
Species C1. The same calculation in the case of Ni(CO),+ leads to a = 47.7 G giving C%2= 0.04 and b = 1.3 G giving CW2= 0.04. Thus, when three CO molecules are linked to one Ni+ ion, 24% of the unpaired spin density is found on the carbon atoms and thus 76% on Ni+, neglecting spin density on oxygen atoms. Species E and E ! The decomposition of carbonyl species (Table I) may be described by the following sequence:
~
b2
I
, 2100
2000
Cm-'
1903
Figure 8. Influence of the pretreatment condins on the I R spectrum of CO adsorbed on a reduced NICa-X. Sample I: (- -) sample reduced by H, at 573 K for 4 h and then outgassed for 15 min at 573 K; (-) immediately after CO introduction (18 torr); (- - -) after 15 h. Sample 11: (- -) sample reduced by H, at 573 K for 4 h and then outgassed for 2 h at 573 K (-) Immediately after CO introduction (160 torr); (--) after 4 h.
TABLE V
E and E' were obtained through outgassing of Ni(C0)2+. Under the same experimental conditions, IR spectroscopy shows the presence of Ni(CO)+ entities (cf. section 2.3) nevertheless, we were unable to detect any hyperfine structure in the presence of 13C0. This contradiction can be resolved by assuming that the nickel orbital containing the unpaired electron has not the correct symmetry to overlap with the molecular orbital of carbon monoxide so as to lead to the detection of a superhyperfine interaction. From Table IV,it can be seen that the g values observed for Cu2+ions in a Y zeolite held in the trigonal window with one NH3" or two H20%molecules in the axial position (S, sites) are very similar to the g values of the E species. It is thus reasonable to locate species E at SIIsites with one CO molecule as seen by IR. Table IV also gives other examples of de ions in a tetrahedral field trigonally distorted (labeled c$ crystal field in Table 11). (2)IR Spectroscopy. (2.1)Preliminary Considerations. Preliminary experiments were performed on anhydrous unreduced samples (only pretreated at 773 K). X-ray data indicated that there are about two Ni2+ions and eight Ca2+ ions in the supercages per unit cell.15 Upon adsorption of CO, two IR bands, a, and a2, appeared at 2215 and 2195 cm-l, which may be assigned unambiguously to the stretching vibrations of CO bonded to Ni2+and Ca2+,respectively, by comparison with the spectra of CO adsorbed on NiIo-X and Ca-X. The intensities of a1 and a2 were shown to depend upon CO pressure (sharply in the case of az) but not on the contact time. The nature and the concentration of carbonyl species formed over reduced samples were influenced by (i) the reduction temperature, (ii) the duration of the outgassing (22)M.Che, J. Dutel,P. Gdezot, and M. Primet, J. Phys. Chem., 80, 2371 (1976). (23)C . Naccache and Y. Ben T w i t , Chem. Phys. Lett., 11,ll (1971). (24)E. F. Vansant and J. H. Lunsford, J. Phys. Chem., 76, 2860 (1972). (25)R. G.Herman, Inorg. Chem., 18,995 (1979). (26)G.A. Senyukova, I. D. Mikheikin, and K. I. Zamaraev, Zh.Stmkt. Khim., 11, 23 (1970). (27)R. P.Eischens and W. A. Pliskin, Ado. Catal., 10,1 (1958). (28)M. Primet, J. A. Dalmon, and G. A. Martin, J. Catal., 46, 25 (1977). (29)J. C. Bertolini and B. Imelik, Surf. Sci., 80, 586 (1979).
bands a1 a,
C
d e
fI
g
wavenumbers, cm-' 2215 2198 2145 21001 2115 2085 2060 2015 19551
assignmenta Ni2+-CO Ca2+-C0 F0 Nl+ 'co
Ni+-CO Nio-CO NiO-CO (aggregates) co
"Q'
All species in the supercages.
procedure following the reducing treatment, (iii) the CO pressure, and (iv) the contact time. The reduction temperature was generally kept at 1573 K to avoid complicationsarising from the presence of large amounts of metallic nickel, but at 1523 K to allow the detection of Ni+ complexes. In addition, the presence of water, which was shown to influence the migration conditions of carbonyl species as well as the CO band positions, was carefully avoided. (2.2)Effect of the Outgassing Time Following the Reducing Treatment. After reduction by hydrogen at a given temperature, the sample was outgassed at the same temperature either for a short time (15 min/sample I) or for a long time (2 h/sample 11). Besides a, and a2,the spectra presented in Figure 8 exhibit, in both cases, a couple of bands bl and b2, together with bands f (only as a weak shoulder on sample 11) and g. The frequencies and the assignments are noted in Table V. It was observed that, immediately after CO introduction, the intensity of bl and b2 was much lower for sample I1 than for sample I. Furthermore, the relative intensity of g vs. b, or bz was greater for sample 11. For sample I, the intensity of bl, b2, f, and g increased vs. adsorption time, the b,/b2 and g/f intensity ratios being invariant. For sample 11, growing of these bands with time was still observed, but estimation of these ratios was delicate, a strong broadening of b2,indeed, could be observed with a shoulder e located at about 2060 cm-' and overlapping f. Brief outgassing at 373 K eliminated all bands. If CO was introduced again, the intensity of bl, b2, f, and g was
2024
The Journal of Fhysiicel Chemistry, Voi. 86, No. 15, 1982 sample
Kermarec et ai.
I
. ! L
2300
2200
z i o u 2000 I Y O O
lnnu
a
Figure 0. Effect of increasing CO pressure upon the nature of the 18, (- - -) 50, and (- -) 200 torr.
carbony! complexes (sample I reduced at 573 K): (-)
equal to that observed after a long contact time during the
adsorption on freshly reduced samples (presented in Figure 8 dashed line). Hence, the growing of these bands with time during the first addition cannot be ascribed to an intergranular diffusion process of CO. Since this intensity increase did not occur a t the expense of al, it must be concluded that CO induces the migration of the reduced species from hidden sites (CO cannot enter the sodalite unit or hexagonal prisms a t room temperat~rel~). (2.3)Influence of Pressure on the Nature and Stability of Carbonyl Complexes. For sample I, Figure 9 shows a noticeable increase of azwith the CO pressure (from 18 to 200 torr), a slight enhancement of bl and bz, f, and g, and a weak broadening of b2 in the low-frequency range. On the other hand (Figure lo), when the pressure was lowered from 150 to 2 torr, a narrowing of b2 was reversibly observed. The intensity of al and a2 diminished greatly, the decrease being more important for a2, while g and f increased (spectra 1 and 2). The main resulta are the invariance of the intensity ratio of bz to bl (about 3) and g to f (about 5) on either increasing or decreasing the pressure (2-200 torr). Let us now examine the lower pressure range. Below 5X torr, the intensity of all bands decreased whereas two new bands, c and d, appeared at 2115 and 2085 cm-l, respectively (Figure 10, spectrum 3). At 5 X lod torr, only c and d were detected (spectrum 4). These bands were removed through brief evacuation a t 100 "C. When CO was introduced again under 4 X lo-' torr, bl, bz, d, f, and g were regenerated but d disappeared when the pressure reached 1 torr (Figure 10, spectra 6 and 7). These results show that c and d behave differently and thus do not belong to the same species (Table V). For sample I1 (long outgassing time before CO adsorption), a similar influence of the CO pressure was noted, but lowering it progressively below 1 torr led to the resolution of a supplementary band, e, near 2060 cm-I which was observed together with c and d (Figure 11). Experiments were also performed on sample I using '2CO-'3C0 mixtures of different compositions (Figure 12 and Table VI). The presence of Ca2+cations rendered the attribution somewhat difficult because of the overlapping of band a2/ (CaZ+-l3CO)with b,; however, in the
II
B I
2300
2200
2100
2 0 0 0 1900
cm
_ I
1800
Flguro 10. Influence of CO pressure. (A) Effect of decreasing CO pressure through a first adsorption (sample I reduced at 553 K): (1) 150, (2)2, (3) 5 X lo4, and (4) 5 X lod torr; (5)after evacuation for 1 h at room temperature. (B) After desorption at 473 K for 5 min and readsorption of CO: (8) 4 X lo-' and (7)8 X lo-' torr. sample
I1
I
3
25
"1,
E
2
.-0
u?
1
.-lil E u? C
2
I
B cm-' 2300
2200
2100
2000
1900
1800
EffeclotdecreaslngCOgressue~hroughaRrstadsorption (sample 11): (1) 0.15,(2) 4 X 10- , and (3)4 X lO-'torr.
-11.
1-100-torr range, a i was more pressure sensitive than b,, and lowering progressively the total CO pressure allowed
The Jownal of phvslcel Chemktry, Vol. 86, No. 15, 1982 2025
Ni+ Ions In a NCa-X Zedlte
sample
I
B'
Figwe 12. IR spectra obtained after exposing sample I to various 1-13C0 mixtures (total pressure: 0.5 torr): (1) 50-50, (2) 25-75, and (3)60-40 1%O-13C0.
TABLE VI bands wavenumbers, cm-' 2165 a1 2148 a,
:
ul C
m
+ L
assignment Nj'+-lSCO Ca2+-WO
a2
cm 2300
C'
2067
Ni+-13C0
2200
2100
-,
2000 1900
Figure 13. Influence of the desorption temperature upon the concentram of speck produced through a second CO adsorption: (-) sample I submitted to a first CO adsorption (200 torr); (--) after outgassing at 573 K for 15 h and subsequent CO adsorptkn (200 tom); after outgassing at 673 K for 15 h and subsequent CO adsorptkn (200 torr). All spectra are recorded immediately after CO introduction. (-e-)
1910 lg70t
E: t a better observation of the effect of isotopic mixing. Figure 12 represents the results obtained under 0.5 torr of I2C& 13C0mixtures. The striking points are the following: (i) Six bands are observed in the 2150-2060-cm-' range-bl, b2, b{, b i , bl", b{-instead of two when pure l2C0 was used. (ii) Six other bands are detected in the 20301900-cm-' range-f, g, f', g', f" (broadening off near 2000 cm-9, g"-instead of two when pure l2C0 was present. In addition, after a brief (15 min) outgassing below lo4 torr at room temperature, only two bands, c and c', remained a t 2115 and 2067 cm-l, respectively (not represented in the figure). (2.4)Influence of the Desorption Temperature upon the Concentration of the Species Produced through a Second Adsorption. When the carbonyl complexes were desorbed in the 523-573 K range for 2 h, a subsequent CO adsorption led to bands bl, b2, f, and g presenting a weaker intensity than after the first adsorption. However, the growing of these bands with the contact time was still observed. After desorption for a long time (15 h) in the 573-673 K range followed by CO readsorption at room temperature, the bands bl, b2, f, and g were even weaker whereas the a, intensity was increased. The higher the temperature and the longer the desorption time, the greater the al enhancement (Figure 13). Furthermore, the increase of bl, b2, f, and g with contact time was no longer observed. (2.5)Discussion (Tables V and VI). For sample I, the invariance of the intensity ratio of bl to b2with adsorption time and pressure shows that these bands are related to the same species, that is, a dicarbonyl species, the band positions (2145 and 2100 cm-l, respectively) indicating that the sites involved should be Ni+ ions. Experiments per-
formed with 13c&12c0mixtures, which show the coupling of oscillators, confirm the dicarbonyl assignment: the wavenumbers recorded for the six b bands observed are well accounted for by the existence of Ni(13C0)2+and Ni(13C0)(12CO)+ along with Ni(12C0)2+(Tables V and VI). Bands c and c' detected only below torr (and not presented in Figure 12) may be ascribed to monocarbonyl Ni(12CO)+and Ni(13CO)+species generated by decomposition of the dicarbonyl complexes. Very similar results were previously obtained on a Ni-mordenite by Kasai et al.,' who found the presence of Ni(C0)2+and Ni(CO)+. Also similar were the spectra recorded by Garbowski et al.6$ on a NiCa-Y zeolite,but these authors ascribed bands at 2140 and 1930 cm-' to Ni(C0I2+and a band at 2095 cm-' to NiO-CO while peaks observed at low pressure near 2105 and 2085 cm-l were not assigned. This interpretation, which was not supported by I3CO IR experiments, may be considered as questionable. Intensity variations vs. time and pressure have shown that f and g belong to the same entity, and isotopic mixing experiments gave evidence for a CO-CO interaction. As the concentration of this entity is growing when the CO pressure is decreased from 150 to 2 torr (Figure lo), this structure involves likely a bridged carbonyl coupled with another CO. Species like CO\N,+/CO\N,+
or Co \N
I+/CO
\N
,+/CO
should be ruled out because (i) the position of the coupled terminal CO (f, 2015 cm-l) seems too far from that of the "free" one (c, 2115 cm-l) and (ii) the stability of the Ni-CO bonds should be different, which is in contradiction with the conjoint elimination off and g at low pressure. Hence,
2826
The Journal of phvslcel Chemistry, Vol. 86, No. 15, 1982
f and g could be better ascribed to a dinuclear doublebridged entity
which may be written as [Ni(CO)+],. While the f and g positions are not unexpected for such a complex, they are higher than in the zero-valent double-bridged species.30 Their intensity ratio (f/g) may be found to be rather low ("0.2); this can be explained by a large angle between oscillators. Going further in the discussion now requires us to account for the differences between samples I and I1 and to ascribe bands d and e. In a previous paper: the reduction of NiCa-X by molecular Hz was described as follows: (i) formation of nickel atoms through reduction of some Ni2+ located in the supercages
+
Ni2+ H2
-
or
Ni2+
+ 202- + H2
Nio + 2H+
(1)
-
(1')
NiO
+ 20H-
(ii) migration of NiO and diffusion of Hz into the sodalite
units where they react with Ni2+ions located in these units Ni2+ + NiO + 2H2 + 2Ni(H2)+
(2)
the right-hand-left-hand reaction, occurring in vacuo, being a disproportionation. If the outgassing treatment following the reduction is short (sample I), the disproportionation does not occur to a significant extent and Ni+ entities are present in the sodalites. Upon CO adsorption, these species may migrate (rapidly at 293 K) in the supercages. As each supercage is connected to four sodalites, dinuclear complexes (eq 4) as well as mononuclear entities (eq 3) should be formed: Ni(H2)++ 2CO 2Ni(H2)++ 2CO
-
-
Ni(C0)2++ H2
(3)
[Ni(CO)+], + 2H2
(4)
If the outgassing treatment following the reduction is prolonged (sample 11), the disproportionation reaction (reverse of eq 2) is favored and the number of Ni+ ions available decreases, which explains the lower intensity (compared to sample I) of bl, b2, f, and g immediately after CO introduction (Figure 8). On the other hand, the growing of these bands with contact time, more important than for sample I, suggests that CO has induced, to some extent, the migration of NiO and Ni2+-initially located in the sodalites-toward the supercages where they recombine to give Ni+ species: NiO + Ni2+ + 2CO + [Ni(CO)+],
(5)
NiO + Ni2+ + 4CO == 2[Ni(CO)2+]
(6)
Such a mechanism implies that Ni+ ions are generated by pairs in the supercages and reaction 6, which involves the conversion of such pairs into mononuclear species, should be less probable than reaction 5 ; it follows that the (f,g)/(bl,b2) intensity ratio should be greater in sample I1 than in sample I where individual Ni+ entities may attain the supercages (eq 3). This is actually observed in Figure 8. Another peculiarity of sample I1 is that the bl/b2 ratio is not constant; b2 is greater than expected and a shoulder e appears on the low-frequency tail. This result suggests (30) P. S. Braterman, 'Metal Carbonyl Spectra", Academic Press, New York, 1975, pp 193-8.
Kermarec et ai.
that some metal atoms not recombined with Ni2+ ions migrate and agglomerate and form linear NiO-CO species responsible for the shoulder e and the abnormally high intensity of b2 compared to b1.27928 In opposition with e, the d band (2085 cm-l) is detected for both samples (I and 11), only in the 10-1-10-3-torr range (Figures 10 and ll), Le., just as the bridged complex [Ni(CO)+],is evolved (reverse of eq 4). It could be ascribed to an intermediate NiO-CO species formed in the disproportionation reaction: [Ni(CO)+], + NiO(C0) + Ni2++ CO + Nio + Ni2++ 2CO Let us now examine the results obtained through a second CO adsorption (Figure 13). Desorption at 573 K following the first CO adsorption at room temperature should induce the transformation of Ni+ carbonyl complexes to NiO and Ni2+ (reverse of eq 5 and 6), which explains the significant growing of the al band (ascribed to Ni2+(CO))upon the second adsorption, compared to that recorded upon the first one. However, no band related to any NiO(C0) species is detected, in contradiction with EPR experiments which showed a broad ferromagnetic signal of metallic nickel. This lack of reactivity of nickel particles could arise from their poisoning, i.e., the formation of a surface carbide through the reaction 2Nio(CO) Ni2C + C02
-
which waa already detected on Ni single crystals at low CO pressure and 413 K.28 The scheme presented a k v e must finally be confronted with the proposals of Kasai et al.' These authors, who did not use hydrogen as a reducing agent, were able to detect Ni+ complexes when the sample (Ni-mordenite), previously heated in vacuo at 673 K, was exposed to CO at room temperature. Their views are summarized by the following equations 2Ni2++ H 2 0 2Ni+ + 2H+ + '/*O, (7)
-
Ni2++ Nio (8) +2co, Ni2++ NiO + 2CO + 2Ni(CO)+ 2Ni(C0)2+ (9) 2Ni+
-
and call for the following comments: (i) Direct generation of Ni+ ions from.Ni2+need not necessarily be postulated. In fact, as Kasai et al. did not pretreat their samples in oxygen, reduction of Ni2+to NiO (upon heating in vacuo) by hydrocarboned pollutants incidentally adsorbed on the solid cannot be discarded. However, as these authors did not observe any band below 2080 cm-', it may be inferred that all Nio atoms produced were transformed into Ni+ through reaction 9 or placed in inaccessible positions. (ii) No bridged entities were detected. (iii) Equation 9 is in agreement with our scheme, but our proposals underline that Ni+ ions are stable only when linked to appropriate ligands (H2, CO). (3) General Discussion and Comparison between EPR and IR Results (Tables I and V). The differences observed between EPR and IR results should first be explained by the higher sensitivity of EPR spectroscopy. For instance, Ni+ ions are detected by EPR on samples reduced at 473 K by Hz while the minimum reduction temperature required for infrared observation was 523 K. EPR has shown that the Ni+ species generated by thermal reduction are located in the sodalite. According to eq 2, these species are produced by pairs. As the AM, = 2 transition corresponding to a triplet state was never observed, the absence of significant dipolar broadening indicates that the Ni+Ni+ distance is larger than 350 pm, suggesting that each
J. phys. Chem. 1982, 86, 2827-2835
Ni+ occupies a SI,site (distance E 535 pm). Concerning the mononuclear Ni+ polycarbonyl complexes, EPR evidences unambiguously the presence of a tricarbonyl species C1 (and possibly C,) whereas IR spectroscopy detects only a dicarbonyl entity, even with the help of isotopic mixing experiments. This discrepancy may be explained by differences in recording conditions. In IR experimenta, the CO pressure is exactly known; in EPR investigations, the sample is cooled down from room temperature to 77 K in CO atmosphere before recording. Hence, physical adsorption occurs, which led to an increase of CO pressure in the supercages, favoring the formation of c1. On the other hand, both techniques have shown that CO induced the migration of Ni+ species from the sodalite cavities to the supercages. At room temperature, this migration is rapid enough to allow the immediate observation of carbonyl complexes in the supercages: (i) dicarbonyl entities have been observed by both spectroscopies, the species detected by IR corresponding to the D stable complex appearing in the EPR spectra (while Cz is a transient species detected only upon adsorption at 77K). (ii) The monocarbonyl Ni(CO)+ entity evidenced by IR (band c) may be related to the paramagnetic signal E generated through decomposition of D at low CO pressure. (iii) The double-bridge dinuclear complex [Ni(CO)+],is not detected by E P R this is not surprising when considering the structure of this entity; as the Ni+-Ni+ distance could not be much larger than 250 pm,90the EPR lines should suffer a strong dipolar broadening (AHN 400 G) preventing their observation. Concerning the nature of the sites, it must be noticed that both Ni+ ions of this complex cannot be placed in crystallographic positions (Sn and/or Sm)because the shortest distance between them is larger than 400 pm.15 Finally, the variation of Ni+ concentration upon outgassing was shown to be governed by the disproportionation reaction: the removal of the Ni+ paramagnetic signal, indeed, was observed conjointly with the increase of the Ni2+concentration in the supercages detected by IR spectroscopy.
2827
Conclusions It has been shown that the NiloCam-X zeolite is a suitable material for the stabilization of Ni+ ions. In this sample, indeed, there are only two Ni2+ions (mean number) distributed over the eight supercages of a unit cell.15 Hence, reduction of these ions by Hz gives rise to isolated NiO atoms (eq 4) which may migrate in the sodalite where they combine with the remaining Ni2+ions to generate Ni(Hz)+ species (eq 2). On the other hand, when the concentration of Ni2+ is higher-as in Ni3,Na-X, for example-several NiO atoms are formed in a supercage, which favors their aggregation.,l In the case of NiloCam-X, the location of Ni+ species produced through reduction by Hz depends upon the procedure used: the thermal reduction leads to the preferential occupancy of SI,sites while the Ni+ ions generated photochemically are placed in Snt. The adsorption of CO induces the migration of Ni+ ions from these sites toward the supercages and the nature of the complexes formed is sharply pressure dependent: mono-, di-, and tricarbonyl mononuclear species as well as dinuclear Ni(1) entities have been detected. In addition, the role of disproportionation reaction
+
2Ni+ + Nio Ni2+ has received consistent evidence; this reaction is reversible as long as the temperature is kept low enough to avoid metal agglomeration. Finally, the conjoint use of EPR and IR spectroscopies has allowed us to throw some light on the intricate questions of genesis, migration, and reactivity of monovalent nickel species. Acknowledgment. We thank L. Bonneviot for recording 13Cthird derivative spectra (Figure 2) and M. Bonnet for technical assistance. We dedicate this work to the memory of Juri Kukk, Estonian Professor of Chemistry, who died in a Soviet labor camp on March 27, 1981, at the age of 40. (31) M.F.Guilleux, D. Delafosae, G. A. Martin, and J. A. Dalmon, J. Chem. SOC., Faraday Trans. I, 75,165 (1979).
Effect of Structure on the Electrical Conductivity of Selenium K. E. Murphy, 6. 6. Wunderllch, and Bernhard Wunderllch' Depament of Chemisby, Rensseleer Polytechnic Institute, Troy, New York 12181 (Received: September 8, 1981; In Final Form: March 15, 1982)
The structure and changes in structure of liquid, supercooled liquid, glassy, trigonal, and monoclinic selenium were studied from 300 to 600 K through thermal analysis coupled with measurement of dc electrical conductivity at field strengths below 100 V cm-'. Isothermal and scanning experiments (between 1and 15 K m i d heating and cooling) were performed. Long-time changes in the conductivity of polycrystalline and liquid selenium were observed which are probably due to chemical structure changes. The time scale of the ring-chain equilibrium in amorphous selenium was found to be faster than generally believed. When trigonal or monoclinic selenium is melted, practically instantaneous establishment of the ring-chain equilibrium occurs. The conductivity measurements support a floor temperature below the glass transition temperature, rather than the prior proposed 356 K.
pr0perties.l
In t h e two monoclinic &&al forms2J the
(1) R. A. Zingaro and W. C. Cooper, Eds.,"Selenium",Van Nostrand, New York, 1974.
0022-3654/82/208&2827$01.25/0
(2) R.E. Marsh, L. Pauling, and J. D. McCullough, Acta Crystallogr., 6, 71 (1953). (3) P. Unger and P. Cherin in "The Physics of Selenium and Tellurium", W C. Cooper, Ed., Pergamon Press, Oxford, 1969, p 223.
0 1982 American Chemical Society
