8260
J. Phys. Chem. 1995,99, 8260-8269
99Mo(P-)99TcSpectroscopy of MoS2, Lithiated MoS2, Coo.sMoS2, and MoS2 Surface Species Peter Mottner:** Tilman Butz,t$@ Anton Led,# Gerard0 Ledezma29l' and Helmut Knozinger*JI Physik Department, Technische Universitat Munchen, 85748 Garching, Germany, Walther-Meissner-Institutfur Tieftemperaturforschung der Bayerischen Akademie der Wissenschaften, 85748 Garching, Germany, and Institut fur Physikalische Chemie, Universitat Munchen, Sophienstrasse I I , 80333 Munchen, Germany Received: October 24, 1994; In Final Form: February IO, 1995@
Nuclear quadrupole interactions (NQI) were measured for several molybdenum-sulfur compounds by time differential perturbed angular correlation (TDPAC) spectroscopy using the 99Mo@-)99Tcprobe. Compounds having the composition MoS2 of different origins and having different degrees of crystallinity, Li-intercalated MoS2, C00.5MoS2, and exfoliated single MoS2 slabs, were investigated. Crystalline and well-ordered 2HMoS2 exhibits a characteristic NQI-frequency of 111 Mrads and axial symmetry. Dispersed MoSz prepared by thermal decomposition of various precursor compounds shows lower frequencies (70-95 Mrads), and this effect is attributed to the presence of part of the Mo in octahedral coordination. Co0,sMoSz and Liintercalated MoS2 with Mo in octahedral coordination exhibit typical frequencies around 50 Mrad/s. Single MoS2 slabs prepared by exfoliation of Li,MoSz also give rise to a low frequency of 56 Mrads, this being indicative for octahedral Mo coordination. Highly disperse MoS2 samples typically contained additional minority species that have NQI-frequencies in the range 400 < w < 650 Mrads. These fractions were identified as Mo centers that have sulfur vacancies which are located along the edges of MoS2 platelets. These data are relevant for the interpretation of TDPAC spectra of sulfided supported Mo/Al2O3 catalysts. We show that formation steps of the sulfided phase and aging processes can be followed by TDPAC. Moreover, the observation of high-frequency fractions at 400 < w < 650 Mrads represents the first direct experimental detection of coordinatively unsaturated MoX+sites at edges and comers of MoS2 slabs in these catalysts.
I. Introduction
are probably oriented normal to the alumina support surface. At higher sulfidation temperatures, the average size of the MoS2 Sulfided molybdenum-based catalysts, which are typically slabs grows. The slabs are then detached from the alumina and supported on alumina, play a major role in industrial applicaprobably lie flat on the surface. tions, particularly in refining processes that involve hydrotreating Coordinatively unsaturated (cus) low-valent M d + ions, which of crude oil. The technological importance of this class of are exposed along the edges of the MoS2 slabs, may act as catalysts has thus motivated a great number of research activities coordination and perhaps active sites. Kasztelan et al.27 and to characterize the catalyst structure and the nature of the Wambeke et al.28have developed a geometrical model for highly coordination sites.'-8 It is now generally accepted that MoS2dispersed sulfided MoIAl203 catalysts according to which the like structures in the form of highly dispersed platelets are found morphology of the supported MoS2 can, for example, be in the finished sulfided catalysts. Experimental evidence for described as hexagonal platelets. These platelets expose the this structural description has come from laser Raman spectroscopy (LRS) 9-11 and X-ray absorption spectro~copy.'~-'~ (1070) and (7010) edge planes in addition to the (0001) basal planes. Different types of sulfur and molybdenum ions having Single and stacked MoS2 slabs have been seen directly in different local environments along the edges and corners of the electron micrographs,'*-** and their dimensions and degree of slabs can be distinguished.6 Surface relaxation may occur so stacking depend on the sulfidation conditions. The presence as to reduce the degree of coordinative unsaturation of exposed of highly dispersed and defective MoS2 in typical catalysts has Md;f ions.29 Upon reduction, the low-coordinated S2- and SHalso been detected by time differential perturbed angular ions at edge and comer positions are removed and cus MoX+ correlation (TDPAC) s p e c t r o ~ c o p y . ~ ~ ions in low oxidation states are then e x p o ~ e d . ~Wambeke ~,~~ The genesis of the sulfided catalysts during H2/H2S treatment et al.28 have suggested that the most likely formal oxidation of the oxide precursor has been studied by LRS9.'O and state of the cus MoX+sites is x = 2. Experimental evidence temperature programmed ~ u l f i d a t i o n . ~ ~The - ~ ~picture that for Mo2+ sites in various coordination environments (depending emerges from these studies can be briefly summarized as on the pretreatment and sulfidation conditions of the catalyst) follows. Highly dispersed MoS2-like slabs, typical lengths 13 on sulfided Mo/Al2O3 catalysts has recently been reported, and nm, are the dominant molybdenum species in sulfided catalysts it is based on the carbonyl infrared spectra of adsorbed C0.30 on alumina supports. Molybdenum is thought to have trigonal Despite these rather detailed model structures that have been prismatic coordination by sulfur atoms. These slabs are developed for sulfided Mo/Al2O3 catalysts, additional experistabilized by Mo-0-A1 linkages, which are formed after mild mental information using complementary methodologies is sulfidation at temperatures 1670 K, and in this state, the slabs required to further describe these materials at an atomic level. Thus, it is not clear whether the coordination of molybdenum ' Technische Universitat Miinchen. * Bayerische Akademie der Wissenschaften. in the supported MoS2 slabs remains trigonal prismatic or Present address: Fakultat fir Physik und Geowissenschaften, Universitat whether the MoX+ions can adopt a distorted octahedral sulfur Leipzig, LinnCstrasse 5, 04103 Leipzig, Germany. environment, perhaps at least under certain circumstances. The I' Universitat Miinchen. Abstract published in Advance ACS Abstracts, April 15, 1995. possible occurrence of distorted octahedral coordination in @
0022-3654/95/2099-8260$09.OQ/O 0 1995 American Chemical Society
J. Phys. Chem., Vol. 99, No. 20, 1995 8261
99Mo@-)99TcSpectroscopy of Surface Species monolayers of MoS2 has been demonstrated by Qin et by means of E M S and STM. Also, cus Mo"+ sites have only been detected indirectly by adsorbed probe molecules such as C0.30 To the best of our knowledge, direct detection of these cus M d + sites has hitherto not been reported although their concentration may amount to 20-30% of the total molybdenum Some preliminary TDPAC results23 appear to indicate that these edge atom sites can be detected. Nuclear quadrupole interaction (NQI) parameters are sensitive to the local environment of the considered atomic probe. It has been demonstrated that NQI parameters of molybdenumcontaining materials, including supported catalysts, can be favorably measured using 99Mo@-)99Tc perturbed angular correlation s p e c t r o ~ c o p y . ~This ~ - ~ ~spectroscopy allows us to measure the NQI parameters that characterize the interactions of the wMo(P-)wTc nuclei with an electric field gradient (EFG), which is produced by a nonspherical distribution of all extranuclear charges. The local chemical environment can thus be deduced from the NQI parameters measured on an unknown sample, provided sufficient fingerprint data for structurally wellcharacterized reference materials are available. The measurements can be carried out under in situ conditions, at elevated temperatures, and in controlled atmospheres. Therefore, the technique is ideally suited for studies of catalytic materials under controlled environment condition^.^^.^^ The goal of this investigation is to determine fingerprint data for a variety of molybdenum- sulfur compounds having molybdenum in either a trigonal prismatic (MoS2) or a distorted octahedral (COo.SMOS2) coordination environment. Several preparation routes for MoS2 have been used. Furthermore, NQI parameters for Li-intercalation compounds and for exfoliated MoS2 are reported. The effect of thermal treatments in various atmospheres on the nature of polycrystalline MoS2 has been studied. Based on the TDPAC data obtained for these reference materials, the possible local environments of molybdenum in the MoS2 slabs of alumina-supportedcatalysts and of M d + edge sites are discussed.
11. Experimental Methods A. Time Differential Perturbed Angular Correlation, TDPAC, Spectroscopy. I. Physical Background. TDPAC spectroscopy requires excited nuclei that decay via the successive emission of y-rays. For the experiments described, 99Mo (T112 = 67 h) was used as a probe which was obtained by irradiation of molybdenum (natural abundance) with a thermal neutron flux of 1.1 x l o f 3n/(cm2 s) for 2 h.34,3599Modecays via /3--emission to an excited state of 99Tcand feeds both the 740-181 keV (positive anisotropy) and the 740-(41)-141 keV cascades (negative anisotropy) in 99Tc. Both cascades were detected simultaneously. The constraint of angular momentum conservation produces an angular correlation between the two y-quanta of each cascade, which can be perturbed by interactions of the nuclear moments of the 99Tcintermediate state (T1/2 = 3.6 ns) with effects of extranuclear fields, such as electric field gradients in the present case. The intermediate level with I = 5 / 2 splits into three states (m = H2, f 3 h ,& l / ~ ) , and three spin precession frequencies, corresponding to the three energy differences between these states, are observable. These spin precession signals modulate the exponential decay of the 99Tc intermediate level. In the TDPAC measurements, the time distributions of the coincidence count rate of the two successively emitted y-rays are determined. The coincidence count rate for powder samples, assuming a static NQI (I = 5 / 2 ) , can be written as37
in which A2 is the anisotropy of the cascade, PZis the Legendre polynomial, 8 is the angle between y l and y2, and t~ is the lifetime of the q c intermediate level. The perturbation function G2(t) can be written as
G2(t)= a,
+ al cos(wt) + a2 cos(w't) + u3 cos(w"t)
(2)
+
in which a'' = w 0'. This means that the Fourier transform of G2(t) yields three lines which are due to one unique NQI at one unique probe site. In the following, the fundamental w (together with the asymmetry parameter 17, see below) is quoted as being characteristic for the NQI at a specific site. For axial symmetry, the a coefficients assume the values, ao = 0.2, al = 0.371, a2 = 0.268, and a3 = 0.143, o' = 2 0 with w = (3n/lO)v~,and VQ = e2qQ/h (eQ is the nuclear quadrupole moment and eq the largest component of the EFG (=V2J). This leads to a periodic precession signal. . For nonaxial symmetry the intensities ai and frequencies w and w' depend on the asymmetry parameter q = (Vu - VYJ V,.38 In this case, the relation between w and VQ is given for nuclear spin I = jI2 by eq 3.39
v'QQ == 11-
1ow Am sin(arccos p
(3)
)
with
a=
J-
and p =
-r2)
80( 1
a2
The frequency VQ thus depends on the asymmetry parameter 17 and assumes a value of VQ = 1.0610w for q = 1. The finite time resolution of the spectrometer was taken into account as described elsewhere.4O With the exception of Li,MoS2, for which a broad frequency distribution around zero was required to fit the data, we never needed a line width (or damping) parameter to describe the discrete frequencies in the spectra, Le., there was no inhomogeneous broadening. This, of course, is partly due to the poor frequency resolution of the 99Mo@-)99Tcprobe. For a frequency distribution around zero frequency with q = 0 one obtains
cui 3
G2(t)=
exp(-irt)
(4)
i=o
where r denotes the distribution width at half maximum. When more than one nonequivalent nuclear probe site is postulated, a superposition of perturbation functions is required to represent the NQI's. We shall denote the fundamental frequencies by 01, w2, etc. The corresponding fractions are denoted by aiin Tables 1 to 4. A more detailed description of the physical background and the theory of 99Mo@-)99Tc TDPAC measurements is given by Butz et al.34 2. Spectrometer and Data Reduction. The TDPAC spectrometer was a conventional fast-slow-coincidence setup with NaI(T1) detectors placed in a plane at intervals of 90". It is described in detail elsewhere.35The experimental time resolution was ~ 2 . ns. 5 Since each detector served as a start and a stop detector, time distributions for eight out of twelve possible coincidences (four at 90" and four at 180") for each cascade were stored simultaneously. An R(t) function was determined as A2G2(t),
8262 J. Phys. Chem., Vol. 99, No. 20, 1995
4 z/w13w31w24w42 L 4
,
J W 1 3 w3 1 w24 w42
Mottner et al.
~~,
- sw14w41w23w32
\JJ
+ 2$w14
(neutron flux 1.1 x lOI3 n/(cm2 s) for 2 h) according to Brauer.M. A 100 mg sample of APM was dissolved in 2 mL of a 25% aqueous NH3 solution. H2S gas was bubbled into this solution through a capillary. The resulting ATM crystals were filtered off and washed with ethanol and water and thereafter decomposed in a glass reactor in vacuum at 673 K for 18 h to yield MoS2-D. The sample was measured in vacuum without exposure to air. A commercial sample, denoted MoS2-E, was obtained from Aldrich. The average particle size of this material was ca. 4 4 pm. MoS2-E was irradiated by a neutron flux of 1.1 x lOI3 n/(cm2 s) for 2 h and measured by TDPAC at room temperature in air after 1-2 days, after the short-lived foreign activities had decayed. 2. Coo.,-MoS2. This ternary compound was prepared from the elements according to a modified literature procedure.45 Quantities of 36.86 mg of cobalt, 80.22 mg of sulfur, and 120 mg of irradiated molybdenum (neutron flux 1.1 x lOI3 n/(cm2 s) for 1 h) were mixed and heated at 1273 K for 16 h in an evacuated quartz ampule. The resulting material was crushed and powdered and then further heated at 1373 K for 3 days in an evacuated quartz ampule. The product was identified as Co0.5MoS2by comparison of its X-ray diffraction pattern to that reported in the l i t e r a t ~ r e .The ~ ~ sample was measured at room temperature in air. 3. Alumina-Supported Catalyst. An alumina-supported catalyst, denoted Mo8A1, was prepared in its oxide precursor state by impregnation (incipient wetness method) of the y-Al2O3 support with an aqueous solution of neutron-irradiated APM, which contained the appropriate amount of APM to yield a catalyst containing 8.2 wt % Moos. The y-Al2O3 (BET surface area, 160 m2/g) was obtained from CONDEA PUFL4L SB by calcination in air at 1023 K for 16 h. The impregnated material was dried and calcined at 773 K for 2 h in air. The resulting oxide precursor was sulfided in a flow of H2/H2S (9:l) at conditions which are specified below.
w4, w23 w32
(with Wi,, denoting the coincidence between detector i ( y l ) and detectorj (y2), corrected for the chance coincidence background) for each cascade separately, and finally both values of R(t) were combined in a weighted sum.37 R(t) is henceforth called the TDPAC spectrum and plotted as “anisotropy” vs time. By adjustment of parameters (frequencies w,, and their fractions a,,frequency distributions r around zero frequency, and frequency ratios w,‘/w, (which are a function of 7,)) theoretical perturbation functions are fitted to the TDPAC spectra. In addition, fast Fourier transforms of the TDPAC spectra were performed as described by Butz et B. Sample Preparation. For the preparation of polycrystalline MoS2 and Co0.5MoS2, the starting molybdenum compounds were activated in a flux of thermal neutrons at Forschungsreaktor Munchen, Garching. The active compounds were then used for the synthesis of the target compounds. A MoS2 single crystal and commercial polycrystalline MoS2 (Aldrich) were irradiated and used for TDPAC measurements without any thermal treatment to anneal possible radiation damage. It has previously been shown23that radiation damage in MoS2 is not detectable by TDPAC in the temperature range 300-673 K. 1. Polycrystalline MoS2. Polycrystalline MoS2-A was prepared from the elements by an iodine transport reaction in an evacuated quartz ampule at 1273 K. The resulting material was irradiated for 2 h in a neutron flux of 1.1 x l O I 3 n/(cm2 s) and then measured at room temperature. A sample of ( N H ~ ) ~ [ M O ~ S ( S ~ ) was ~ ] . ~prepared H ~ O from activated ammonium paramolybdate APM, (NH4)6[Mo&]~H20 according to method 1 as described by Mottner et al.36 The resulting Mo3-cluster compound was then decomposed in a glass reactor at 673 K according to Diemann et al.42 to yield polycrystalline MoS2-B. The TDPAC measurements on this sample were carried out in vacuum. Another variant, denoted MoS2-C, was obtained by successively heating radioactive ( N H ~ ) ~ [ ( M O ~ S ( S ~ )in~ ]vacuum .~H~O at a heating rate of 10 Wmin to 373,463, and 600 K, at which temperatures the sample was held for several days. The sample was then measured under argon without contact with air. MoS2-D was prepared via thermal decomposition of ammonium thiomolybdate ATM, (NH&MoSq, as described by Prasad et al.43 ATM was synthesized from radioactive APM
MoS,
111. Results A. Molybdenum Disulfide. For comparison purposes, a MoS2 single,crystal was measured in various orientations of the largest EFG tensor component V2,relative to the position of the four detectors and the plane defined by them. The TDPAC spectra are shown in Figure 1. Depending on the orientation, either the fundamental frequency w or the first or second overtone is observed with maximum possible intensity. When V, is oriented perpendicular to the detector plane (Figure la), the fundamental frequency w is dominant. The first
SINGLE CRYSTAL
10
-$ 5 L
*
c1
0
E5
0
2I 0 a
-5
0
10
20
30
0
10 20 TIME lnsecl
30
0
10
20
30
Figure 1. TDPAC spectra for the NQI of 99Mo@-)9?c in a MoSz single crystal for various crystal orientations. The insets show the orientation (vector C) of the largest component of the EFG (V,) to the detector plane. The solid lines are the result of least squares fit analyses of the experimental data points.
99Mo(j3-)99Tc Spectroscopy of Surface Species
10
J. Phys. Chem., Vol. 99, No. 20, 1995 8263
I
w = 72 M r a d l s
TABLE 1: NQI Parameters of Polycrystalline and Highly Disperse MoS2 and of Coo.sMoSz w2r
WI.
T/K Mrads al,"% V I Mrads a2,% 72 G, %
sample MoS2-A MoS2-B MoS2-C MoS2-D
300 111(1) 100 673 97(3) 90(3) 673 72(5) 66(2) 300 93(1) 100 MoS2-E 300 93(2) 97(1) COosMoSz 300 fit 1 38(1) 100 fit 2 58(1) 100
10
I
1
I
I
I
I
i
w = 92 Mrad/s -4
ai abundance of fraction i. Mrads. n
x
0 0 210(17) lO(3) 0 0 148(9) 29(5) 0 0 0
5(3) 3(1)
0 1
cl, abundance of fraction with w = 0
0
-1
I
Y
%
e
0 QI
I-
-,(I--
O v> w = l l l Mradls
H
z I
0
10
TIME
20
30
[NSECI
Figure 3. TDPAC spectrum for the NQI of 99Mo@-)99Tcin
c00.5-
MoS2. The solid line is the result of a least squares fit analysis of the experimental data points with 17 = 0.
The TDPAC spectra of MoS2-B and MoS2-C are shown in parts b and c of Figure 2. In contrast to MoS2-A, the spectra of both of these polydisperse MoS2 samples contain two fractions that deviate from the reference value of 111 Mrads. 0 5 10 15 20 25 30 35 The main fraction for MoS2-B (continuous heating) has a frequency w1 = 92(3) Mrads, while the main fraction for TIME [nsec MoS2-C (stepwise heating) has a frequency wl = 72(5) Mrad Figure 2. TDPAC spectra for the NQI of 99Mo@-)99Tc in MoSz s. The minority fractions have frequencies of w2 = 210(17) powder samples. (a) Obtained from the elements by iodine vapor Mrads (10(3)%) and 02 = 148(9) Mrads (29(5)%), respectransport at 773 K (MoS2-A). The 740-(41)-141 keV cascade was tively, and are due to 99Mo@-)99Tcprobes in different environmeasured only. Therefore, the ordinate is reversed as compared to spectra b and c. (b) Thermal decomposition of ( N I & ) z [ M o ~ S ( S Z ) ~ ~ . ~ H ~ Oments. When polydisperse MoS2 was prepared by thermal decomat 673 K, heated continuously (MoS2-B). (c) Same as b but heated in steps (MoS2-C). The solid lines are the result of least squares fit analyses position of ATM (sample MoS2-D), a single fraction spectrum of the experimental data points. (not shown) was observed which gave NQI parameters of w = 93( 1) Mrads (100%) and 7 = 0. These values are identical to overtone has maximum contribution when V, lies in the detector those found for the main fraction in sample MoS2-B. plane and is oriented centrally between two neighboring detectors (Figure lb). Finally, when V,, is inclined toward the Comparable NQI parameters were also measured for the detector plane by 45" and its projection onto the plane is oriented commercial MoS2-E (see Figure 5a), namely w = 93(2) Mrads (97(1)%) and 7 = 0. A second fraction at low abundance centrally between two neighboring detectors (Figure IC), the second overtone has maximum intensity. The best-fit analysis (3(1)%) has a frequency distribution around 0 Mrads. of the experimental spectra is obtained when a single NQI is All of the NQI parameters for the various MoSz samples are used, Le., a fit that involves a single, chemically distinct crystal summarized in Table 1. It is obvious that the main fractions site which has the characteristic NQI parameters w = 111 of all preparations that yield highly dispersed MoS2, namely Mrads and 7 = 0. MoS2-B, MoS.2-C, MoS2-D, and the commercial MoS2-E, have frequencies that are significantly lower than the reference value The same NQI parameters, Le., w = 111(1) Mrads (100% of 111 Mrads. abundance) and 7 = 0, were obtained for polycrystalline MoS2A. In this case, only the 740-(41)-141 keV cascade of the B. COO.JMOS~. The TDPAC spectrum of COO.JMOSZ is 99Mo(P-)99Tcdecay was measured. The TDPAC spectrum is shown in Figure 3, indicating that the anisotropy decays very shown in Figure 2a. The NQI parameters obtained for these slowly in this case and only a small part of one precession period two samples (namely w = 111 Mrads and 7 = 0) appear to be is measurable. characteristic for well-crystallized MoS2 and are used subseAccording to van den Berg,45 the coordination of the Mo quently as reference values. atoms in this compound is distorted octahedral; hence, the
I
8264 J. Phys. Chem., Vol. 99, No. 20, I995
Mottner et al.
a
Fit 1
Fit 2 b w=454 Mrsd/s (8%)
'0
10
TIME
20
30
[NSECI
Figure 4. TDPAC spectra for the NQI of 99Mo@-)99Tc in MoS2, prepared from APMIATM at 673 K with subsequent H2 (at 573 K) and vacuum treatment (at 673 K). (a) Least squares fit analysis with one frequency component. (b) Least squares fit analysis with two frequency components (appearance of a high-frequency component). (c) After a subsequent H2S treatment at 300 K. The solid lines are the result of least squares fit analyses of the experimental data points. asymmetry parameter may deviate from zero. However, an accurate determination of 7 is not possible here because of the slow decay of the anisotropy as mentioned above. The singlefraction fit analysis for the two extreme cases, namely 7 = 0 and 7 = 1, yields frequency values of 38( 1) and 58(1) Mrads, respectively. These NQI parameters are also included in Table 1. Low frequencies and finite 7-values (0 < 7 1) appear to be typical for Mo probes in distorted octahedral sulfur environments. C. Thermal Treatments of MoS2. Thermal treatments in hydrogen and subsequent evacuation were carried out to test
the possibility of detecting defect sites, Le., cus Mo centers. MoS2-D was placed in a glass reactor and flushed with argon. The sample was then treated at 573 K for 1 h in flowing H2, again flushed with argon and then evacuated for 1 h at 673 K. The TDPAC measurement was then carried out in an argon atmosphere at room temperature with the sample in the reactor. The result of this thermal treatment is a highly dispersed almost certainly sulfur-deficient MoS2-, phase, which develops a BET surface area > 50 m2/g. The experimental spectrum and two different fit analyses are shown in Figure 4. In fit 1, a single fraction was assumed, while two independent fractions were used in fit 2. The frequency of the main fraction is 94(2) Mrads and, thus, is again significantly lower than the reference frequency and compares well with the corresponding value for the as-prepared MoS2-D sample (see Table 1). Fit 2 gives a second high-frequency fraction at 454(22) Mrads with 8(3)% abundance. Note that the frequency resolution of 99Mo@-)99Tc measurements is poor. Nevertheless, the detection of even small contributions of a high-frequency fraction is significant. The NQI parameters obtained in this experiment are summarized in Table 2, where they are compared with those of the as-prepared MoS2-D sample and those determined after a subsequent experiment, namely exposure of the H2-treated sample to H2S. In this experiment, the sample that had been thermally treated in H2 was exposed to an H2S atmosphere (1 bar), and the TDPAC spectrum was measured while the sample was under H2S at room temperature. The spectrum is shown in Figure 4c. It was expected that H2S would coordinate to possible cus MoX+sites; thus, it would complete their coordination spheres. The TDPAC spectrum measured after exposure of the sample to H2S is distinct from that for the H2-treated sample, particularly in the range 10-15 ns. A good fit to this spectrum using two independent frequency fractions, which would represent two different probe environments, is no longer possible. The best fit NQI parameters are w = 108(3) Mrads (100%) and 11 = 0.33(7). A high-frequency fraction is not detectable. D. Lithium Intercalation and MoS2 Exfoliation. Unsolvated lithium can be intercalated into MoS2 either electrochemicallf6 or by reaction with n-butyllithium in n-he~ane.4~ For lithium intercalation in the present case, the commercial MoS2-E was reacted with n-butyllithium (BuLi). Initially, 10 mg of irradiated MoS2 was added to a 1.6 M BuLi n-hexane solution, and then the progress of the intercalation reaction was followed by TDPAC. The spectrum shown in Figure 5b was taken during the first 12 h of reaction. It characterizes the final state of the intercalated MoS2, since no further changes could be detected at longer reaction times. Although the anisotropy in Figure 5b decays slowly, a fit of the experimental spectrum using one single fraction is not possible. For the best fit, we used a frequency distribution around w = 0 Mrads (71(4)%) with a width at half maximum r = 71( 11) Mrads and a second low-frequency fraction with a discrete frequency of 52(4) Mrads (29(4)%). The X-ray diffraction pattern of the present Liintercalated sample is identical to that reported for an electrochemically intercalated LilMoS2$6 in which Mo is octahedrally coordinated by sulfur ligands. The Li-intercalated MoSz is the precursor for the MoS2 e x f ~ l i a t i o n .For ~ ~ exfoliation experiments, 2.5 mg of MoS2-E (Aldrich) was irradiated at a neutron flux of 1.9 x lOI3 n/(cm2s)
TABLE 2: NQI Parameters of MoS2-D Measured in Situ after Various Thermal Treatments treatment TIK fit w l , Mrads a], % 7) as prepared 93(1) 100 0 300 H2 at 573 K and evacuation at 673 K
300
subsequent H2S adsorption at 300 K
300
1 2
94( 1) 9G.9 108(3)
100 92(5) 100
0 0 0.33(7)
02,Mrads
a2,%
r72
454(22)
8(3)
0
J. Phys. Chem., Vol. 99, No. 20, 1995 8265
99MoCg-)99TcSpectroscopy of Surface Species 0 c
a m
0
0 c
In
0
-
0
x -
cl
Y
> - m
n. 0 CY k
o
0
v, U
=
a o-
d m
0
-
0
e
m
0
? '0
20
10
TIME
30
[NSECI
Figure 5. TDPAC spectra for the NQI of 99Mo@-)99Tc in educts, intermediate, and final products of the MoS2 exfoliation process and in restacked and Al203-adsorbed MoS2 layers: (a) MoSz (Aldrich, 325 mesh); (b) MoS2 (Aldrich, 325 mesh) BuLi; (c) MoS2 (Aldrich, 325 mesh) BuLi exfoliated; (d) MoS2 (Aldrich, 325 mesh) BuLi exfoliated and restacked; (e) MoS2 (Aldrich, 325 mesh) + BuLi
+
+
+
exfoliated and adsorbed onto AlzO3. The solid lines are the result of least squares fit analyses of the experimental data points.
for 6 h and then reacted with BuLi as described above. After 3 days the resulting Li-intercalated MoS2 was separated from the solution, washed with n-hexane under argon, and dried. Subsequently, as described by Murphy and 10 mL of H20 containing ca. 15 drops of glycerine as a surfactant was added, and then the suspension was treated ultrasonically for 14 min. The resulting brown solution was stable for several days without any detectable sedimentation of MoS2. TDPAC spectra of the solution containing (mono)disperse MoSz were measured at room temperature. A typical spectrum is shown in part c of Figure 5, in contrast to the fits obtained for the Li-intercalated precursor compound (vide supra). The best-fit analysis for the exfoliated single MoS2 layers gives a major fraction (88(7)%) with a frequency w = 54(4) Mrads. In addition, a minority fraction of ca. 12% with zero frequency is also present. It should be noted that the frequency of the dominant fraction, which is 56 Mrads, is comparable to the frequency of the minority fraction (52(4) Mrads) for the Liintercalated precursor material. The NQI parameters of Liintercalated and exfoliated MoS2 samples are summarized in Table 3. Note that the tumbling motion of the single MoS2 slabs in solution is sufficiently slow to yield static spectra, Le., the lateral dimension of the slabs are likely in the range 0.1- 1 Pm. The exfoliated single slabs of MoS2 can be restacked in the aqueous solution (with no surfactant present) when the pH is reduced by addition of hydrochloric acid. The restacking is indicated by flocculation and sedimentation of MoS2 particles. After removal of the water, the restacked material was measured at room temperature in air. A typical spectrum is shown in part d of Figure 5, and the corresponding NQI parameters are given in Table 3. The experimental spectrum can be represented well by only one fraction that has a set of three discrete frequencies. Performing fits with the limiting asymmetry parameters q = 0 and q = 1, one obtains frequencies of 84(1) and 138(2) Mrads, respectively. However, the best fit is obtained with r being treated as a free parameter. The resulting NQI parameters are then w = 107(5) Mrads (100%) and r = 0.53(11). The restacked MoS2 was not checked by XRD and is likely to be turbostratic. However, the stacking disorder is not expected to be responsible for this large q value. Miremadi and Morrison50have shown that exfoliated MoS2 can be adsorbed on alumina from the aqueous solution. In the present series of experiments, 100 mg of y-AhO3 (BET surface area 160 m2/g) was added to 5 mL of an aqueous solution containing surfactant and 2 mg of exfoliated MoS2. The suspension was stirred while hydrochloric acid was added for restacking and adsorption of the MoS2 on the alumina surface. After removal of the water, the MoSz/A1203 sample was measured at room temperature in air. A typical TDPAC spectrum of this material is shown in part e of Figure 5. It is impossible in this case to determine the asymmetry parameters with high precision. The fit analyses clearly show that a main fraction that has a set of three discrete frequencies and a less abundant high-frequency fraction are required to reproduce the experimental spectrum. Three different fit analyses were carried out in which the asymmetry parameter for the main fraction was constrained to either 171 = 0 or 71= 1 or was allowed to vary freely. The asymmetry parameter q 2 for the high-frequency fraction was fixed at 7 2 = 0. The resulting NQI parameters for the three fits are summarized in Table 3. The free q parameter fit procedure appeared to give a slightly better fit than the other two, as was the case with the exfoliated MoS2 prior to adsorption. The corresponding NQI parameters for the main fraction (80(6)%) are 01 = 109(11) Mrads and 71 = 0.32(23). They compare fairly well to those determined for
Mottner et al.
8266 J. Phys. Chem., Vol. 99, No. 20, I995
TABLE 3: NQI Parameters of MoS2-E after Li-Intercalation, Exfoliation, Restacking, and Adsorption on Alumina TIK
sample MoS2-E MoSzBuLi MoSZBuLi, exfoliated restacked, previous sample
300 300 300 300
exfoliated, adsorbed on A1203
300
fit
wl,Mrad/s
al, %
tll
o2,MradJs
a2, %
72
a,
1 2 3
9W) 52(4) 56(4) 84( 1) 138(2) 107(5) 97m 155(4) 109(11)
97(1) 29(4) 88(7) lOO(2) 96(1)
0 0 0 0 1 0.53(11) 0 1 0.32(23)
440(10) 437(13) 442(10)
19(2) 17(4) 18(3)
0
0 6(2) 2(3)
1
2 3
100 81(2) 77(6) 80(6)
the exfoliated MoSz prior to adsorption. The minority fraction has a frequency in the range 400-500 Mrads. The histogram of Figure 6 summarizes the majority frequencies and the abundances of the fractions for the starting material, for the intermediate and final products of the MoSz exfoliation process, for the restacked and Al203-supported MoSz-layers, and for COO.SMOS~ for comparison. For the Li-intercalated MoS2, the frequency of the minority fraction is given in Figure 6. Since the precession frequency vp depends less strongly on the asymmetry parameter than the w values, the former frequency is plotted on the abscissa of Figure 6. For the relation between VQ and w see eq 3. The widths of the bars in the histogram represent the error limits and (where appropriate) the frequency range detemined for the limiting cases q = 0 and q = 1. Frequencies above 120 MHz are not shown in Figure 6. They are listed in Tables 1 to 4 and are discussed in the text.
E. Supported Catalysts. The TDPAC results on y-AhO3 impregnated with an APM solution during impregnation, drying, and calcination have been reported previously34 and can be summarized as follows. (i) For catalyst precursors prepared with about 9 wt % Moo3 up to 15 wt % and by the incipient wetness method, predominantly polymolybdates are adsorbed, which very closely resemble the condensed octahedra of APM in the crystalline state. (ii) Drying at 373 K for 2 h does not lead to major changes. It is noteworthy that part of the molybdate in solution that is not adsorbed appears to remain mobile, Le., the drying conditions seem to be too mild in order to remove all water. (iii) Calcination at 773 K for 2 h leads to a significant line broadening but leaves the overall NQI parameters unchanged suggesting that the polymolybdate structure remained essentially intact. Sulfidation at 623 K for short times (typically 2 h) leads to precession frequensies greatly exceeding that for
-
Charge transfer to the MoSz lattice
No charge transfer to the MoSz lattice
n
4
3 100
I I
-
80
E Q * .r(
8
0 0
r,Mrad/s
/
5
-
1
60 -
40-
2
G 20 -
0-
1
30
50
1 : CO~.(MOS~ 2 : MoSz (Aldrich, 325mesh) 3 : MOSS(Aldrich, 325mesh) 4 : MOSS(Aldrich, 325mesh) 5 : MoS2 (Aldrich, 325mesh) 6 : MoSz (Aldrich, 325mesh)
70
T
7 90
110
+ BuLi + BuLi exfoliated + BuLi exfoliated and restacked + BuLi exfoliated and adsorbed onto A I 2 0 3
Figure 6. Histogram of the respective main NQI frequencies ( V Q , quoted in MHz) and their fractions for educts, intermediate, and final products of the MoS2 exfoliation process, restacked and A1203-adsorbedMoS2 layers together with Coo.sMoS2.
J. Phys. Chem., Vol. 99, No. 20, 1995 8267
99Mo(B-)99TcSpectroscopy of Surface Species
TABLE 4: NQI Parameters for Sulfided Oxidic Precursors Prepared by APM-Impregnation of y-Al203 (MoSAl) under Various Conditions sample treatment measuring temperatureK, atmosphere o1,Mrads a]," % 71 02, Mrads a2,O % 72 Q ,% ~ #1 sulfidation at 623 W25 h plus v a c u u d l day at 673 K plus v a c u u d l h at 573 K #2 sulfidation at 1023 W2 h plus under H2 at 573 W1 h plus vacuum at 673 W1 h
a, abundance of fraction i.
623, H2/H2S 673, vac. 300, vac. 300, vac. 300, Hz 300, vac. Q
llO(fix) 81(3) 109(3) 122(5) 109(10) 139(10)
77(3) 79(2) 73(2) 81(2) 83(3) 70(4)
0.30(10) O(fix) O(fix) 0.39(8) 0.54(24) O(fix)
433(20) 523(21) 419(19) 333(15) 471(41)
12(2) 19(2) 19(2) 17(3) 17(3)
l(fix) 0.49(5) 0.46(6) l(fix) 0.49(13)
23 9 8 13
abundance of fraction with o = 0 Mrads.
MoS2 (about a factor of 2), indicating that under these conditions the sulfidation is far from being complete.23 On the contrary, sulfidation at 623 K for 25 h (or alternatively, at 1023 K for 2 h) leads to precession frequencies W I in the vicinity of that for bulk MoS2 and below (see Table 4). In fact, any long term treatment (> 1 day) under H2/H2S (9: 1) or during thiophene conversion5' leads to spectra closely resembling that of MoS2 which show a progressive trend toward lower frequencies when given prolonged treatment and which show the complete absence of high-frequency components. This result suggests that all Mo atoms are coordinatively saturated by sulfur. On the other hand, evacuation at elevated temperatures as well as exposure to H2 at 573 K for 1 h immediately produces minority fractions between 10 to 20% with precession frequencies ranging from 333 Mrads (under H2) up to 523 Mrads (see Table 4). It should be mentioned that the asymmetry parameters for these minority sites are difficult to determine, but they are large and lie in the range 0.5-1.0. Similarly, the frequency w z is also subjected to large uncertainties. Nevertheless, this means, together with rather large frequencies w2, that the Mo environment is rather distorted from the regular trigonal prismatic coordination (or octahedral, if existent). A very plausible interpretation would be in terms of a coordinatively unsaturated Md[+ species lacking one (or more) sulfur neighbor and a possible lattice relaxation. The striking similarity between highly dispersed unsupported MoS2 and supported MoS2, as far as all hyperfine parameters are concerned ( W I , V I , al;02, 72, a2), very strongly supports the idea that very similar particles, as far as size, Mo coordination, and number of cus Mo are concerned, are present in both cases. IV. Discussion The NQI parameters for MoS2 of varying provenance (precursor compound, synthesis route, thermal treatment, and degree of dispersion) show a fairly broad range for the fundamental frequency corresponding to the main site fraction detected in the TDPAC spectra. The reference values measured for a MoSz single crystal and for a polycrystalline product are w = 111 Mrads (100%) and q = 0 (Figure 1, part a of Figure 2, and Table 1). These NQI parameters coincide perfectly with those reported earlier for well-crystallized polycrystalline MoSz sample^,^^,^^ which exist in the 2H-MoS2 modification with Mo in trigonal prismatic coordination. Hence, the mentioned reference NQI parameters must be considerd as characteristic for this type of coordination. All other highly dispersed MoS2 samples gave main fractions that have fundamental frequencies that are smaller than the reference value of 111 Mrads; these frequencies fall into the range 70-100 Mrads (see Table 1). It is interesting to note that similar low-frequency fractions were also observed for the Li-intercalated MoS2 (see Table 3). We propose two possible reasons for the occurrence of frequencies < 111 Mrads, namely a change of the Mo coordination andor a charge transfer into the MoS2 host lattice in intercalation compounds. The compound Coo.sMoS2, in which Mo is located in a distorted octahedral sulfur environment, gave frequencies in the
range 38-58 Mrads for 0 5 q I 1 (Figure 3 and Table 1). Hence, NQI parameters within these limits are to be considered as characteristic for distorted octahedral Mo coordination. The observation of average frequencies in the range 50 Iw I100 Mrads may thus indicate the presence of Mo in both trigonal prismatic and distorted octahedral coordinations. This situation may exist in both highly dispersed MoS2 and Li-intercalated MoS2. As a matter of fact, a 2H-MoS2 (trigonal prismatic coordination) to 1T-MoS2 (Octahedral coordination) phase transition has been reported to occur when the intercalation compound LilMoS2 had been synthesized electr~chemically.~~ A thermal 2HllT phase transition has been observed for TaS2 accompanied by a decrease of the NQI frequency.52 The presently available data do not permit us to determine the amount of intercalate necessary for the phase transformation within the MoS2 layers to occur. It is, however, known that intercalation compounds A,MoS2 do exist, in which the Mo coordination remains unchanged relative to that of the starting MoS2, provided x I0.3.53 One may thus argue that a LixMoS2 phase with x I 0.3 forms during the initial stages of the intercalation reaction. At increasing intercalate concentration (0.3 < x < l), the 2H modification is then progressively transformed into the 1T modification, which gives rise to the reduced NQI frequencies between 50 and 100 Mrads depending on the relative amounts of the 1T modification formed. It was also observed that an increasing contribution of a fraction that has a frequency distribution around 0 Mrads was present and ultimately became the most abundant fraction as the intercalation reaction proceeded (Table 3). Although supporting evidence from complementary methods is lacking, we believe that the most likely origin of these low frequencies is the formation of additional LixMoS2 at high degrees of intercalation. Analogously, phases have been observed that gave rise to low NQI frequencies relative to that of pure TaS2, when Li-intercalation compounds with high intercalate concentrations were prepared by reaction of 2H TaS2 with B u L ~ . ~ ~ At least with the intercalation compounds, charge tranfer to the matrix of the host lattice may also contribute to the frequency decrease. It has in fact been argued 46 that the charge transfer would be the driving force for the 2HllT phase transformation since the ligand field stabilization of Mo3+(d3)in an octahedral sulfur environment would be more favorable than that in trigonal prismatic coordination. It is interesting to compare the observed change in NQI for lithiation of MoS2 with that observed for the formation of the bronze HxM003 from Moos. In the former case, Mo4+is reduced Mo3+upon intercalation with concomitant reduction of the NQI, in close analogy to observations for intercalation into 2H-TaS2 (and 1T-TaS2).54s55In contrast, the insertion of protons into Moo3 with a reduction of Mo6+ to Mo5+ is accompanied by an increase of the NQI.56 High NQI's are presumably characteristic for pentavalent m o l y b d e n ~ m . ~ ~ , ~ ~ Exfoliation of the Li,MoSz intercalation compound into individual MoS2 slabs occurs when it is suspended in water.48 For this process to occur, Li,MoSz must be decomposed in water under H2 evolution. Schollhorn and Weisss9 have shown that swollen polyelectrolyte phases LiX(H20)&loS2 have very limited
Mottner et al.
8268 J. Phys. Chem., Vol. 99, No. 20, 1995
stability and do exist only at x 5 0.1. The TDPAC spectra of the dispersed single slab MoS2 are dominated (88%) by an NQI frequency of w = 56 Mrads (see Table 3), which is almost identical within the error limits to the discrete frequency observed for the intercalated starting material and which was attributed to Mo in distorted octahedral coordination. This suggests that the sulfur coordination sphere around Mo is predominantly octahedral in the single slabs after exfoliation. The almost quantitative disappearance of the frequency distribution around 0 Mrads (which was attributed to the presence of material with high and possibly nonuniform intercalate concentration) must then be related to the quantitative decomposition of the corresponding LixMoS2 phase into single slabs, in which distorted octahedral coordination seems to prevail. However, it cannot entirely be excluded at present that adsorbed Lif ions andor solvent molecules may have an effect on the NQI parameters of Mo despite their relatively remote location relative to the Mo position. When restacking of the exfoliated MoS2 was induced by addition of HC1, the NQI parameters were found to be w = 107 Mrads (100%) and q = 0.53. This frequency value is very close to the reference value, and therefore, this result suggests that the 2H modification is present in the restacked MoS2 despite the deviation from axial symmetry. This observation is not unexpected, since the treatment with hydrochloric acid must lead to a quantitative removal of Li+ ions from the MoS2. Supporting evidence for the 2H structure of the restacked MoS2 was reported by Joensen et a1.60 These authors found Mo-Mo distances that are characteristic for MoS2 in the 2H modification. Hence, the available data suggest that the 1T modification with Mo in octahedral coordination is stabilized by the presence of intercalate. Additionally, high-frequency minority fractions were additionally observed for most of the highly dispersed MoS2 samples (see Table 2), and these fractions were particularly abundant when the samples had been thermally treated in H2 and evacuated. Frequencies in the range 150 to 200 Mrads are probably caused by small contributions from MoS3 (w = 178(6) Mrads), which occurs as an intermediate in the synthesis of MoS2. H2 treatments and evacuation are thought to create sulfur vacancies, which should be concentrated along the edges of the dispersed MoS2 particles where their reactivity is located.61.62The removal of sulfur atoms from the edges would create cus M d + sites, and they are expected to give rise to NQI frequencies higher than the reference value for bulk Mo atoms of 111 Mrads. In fact, the high-frequency fractions that were extracted from the respective TDPAC spectra are typically in the range between 400 and 650 Mrads (assuming q = 0) and their abundance corresponds to 10-20% of the total number of Mo atoms in the specimen. The high frequency is consistent with a low-coordinated Mo center, and the abundance of 1020% is consistent with the density of edge atoms that can be estimated for the MoS2 particles taking into account their dimension^.^^ It should be noted that a frequency around 500 Mrads was observed for Mo sites that have nearby oxygen vacancies in Moos, which had been created by neutron irradiati01-1.~~ Ni et al.63reported NQI parameters of w = 550 Mrads and q = 0.75 for Mo centers that have neighboring oxygen vacancies in y-Mo4011. The available information thus strongly suggests that the high-frequency fraction can in fact be attributed to cus Mo centers. That these are predominantly surface centers is demonstrated by their disappearance when H2S is adsorbed on MoS2. We believe that these data provide the first direct detection of cus Mo surface (edge) centers in highly dispersed MoS2. The exact frequency corresponding to the defect sites depends sensitively on temperature, on the
duration of the thermal treatment, and on the applied atmosphere. This dependence leads to a rather broad frequency spread for differently treated samples. Thus, the nature of defects, such as edge or comer location, single or multiple vacancy, crystallographic orientation of edge plane etc., should depend on the treatment conditions. An unequivocal assignment of experimentally observed frequencies to individual defect types is not possible at present. It should be mentioned that El-Kateb et aLM have recently reported TDPAC measurements on MoS2 and Li-intercalated MoS2 that also include high-frequency fractions. These authors reported, however, data quite different from those reported here. Their experimental spectra were taken over a very narrow time range and were characterized by rather poor statistics. A singlecrystal study was not included. Therefore, these deviating results of El-Katab et al.@ could have been caused by sample preparation problems, and they may have overinterpreted the results of their fits. The NQI frequencies that have been observed for the highly dispersed MoS2 are also found for supported sulfided Mo/Al203 catalysts. Here, the abundance of the corresponding fractions is strongly dependent on the sulfidation and subsequent pretreatment conditions. Frequencies w < 110 Mrad/s occur when the oxide precursor is sulfided under mild conditions for short periods of time and also when thiophene hydrodesulfurization is performed under mild conditions. Since the tranformation of the oxide precursor into the sulfided form of the supported catalyst is preceded by an O/S exchange, the Mo species formed under mild conditions do contain oxygen and sulfur in the coordination sphere?*24Such structures may be responsible for the low-frequency fraction observed in the TDPAC spectra. Alternatively, the low frequencies may also be caused by the presence of octahedral Mo centers in the initial stages of sulfidation, as discussed above for highly dispersed unsupported MoS2. When the catalysts were treated in sulfiding atmospheres for extended periods of time and at elevated temperatures, the lowfrequency fractions progressively disappeared and a discrete frequency of 110 Mrads with q = 0 was observed. This result is consistent with the k n o ~ n ' 9particle ~ growth and restacking of the supported MoS2 slabs. It is interesting to note that measurements on the material that was prepared by adsorption of exfoliated MoS2 gave the same frequencies. High-frequency fractions in the range 400 < w < 600 Mrads were also observed for supported sulfided Mo/A1203 catalysts and for the adsorbed exfoliated MoS2. The abundance of the corresponding species in the sulfided Mo/Al203 was in the range of 10-30% of the total molybdenum content in the catalyst. This value is perfectly consistent with the number of edge and comer atoms present on supported MoS2 catalysts having typical dispersion^,',^,^^ and the values agree well with the number of sites that was estimated by CO a d ~ o r p t i o n . ~Therefore, ~ . ~ ~ we suggest that the frequencies between 400 and 600 Mrads in supported sulfided catalysts are indicative of the presence of cus Mo sites, which are located at comers and edges of the MoS2 slabs that are anchored to the A1203 support surface. These sites should be of the same type as those described above for highly dispersed unsupported MoS2. The present experiments on supported catalysts demonstrate that in situ TDPAC permits us to follow structural changes and aging processes of supported sulfided Mo species. Moreover, cus Mo sites on the supported MoS2 slabs have been detected directly.
V. Conclusions Crystalline 2H-MoS2 with Mo in trigonal prismatic coordination has a characteristic NQI frequency of 111 Mrads and q =
J. Phys. Chem., Vol. 99, No. 20, 1995 8269
99Mo@-)99TcSpectroscopy of Surface Species
0. In contrast, highly dispersed MoS2 preparations give rise to lower frequencies in the range of 70-100 Mrads. This result is attributed to the presence of Mo in a distorted octahedral sulfur environment (lT-MoS2) in addition to the 2H-MoS2 modification. This interpretation is supported by the observation of characteristic frequencies near 50 Mrads for compounds that have Mo in octahedral coordination such as Coo.5MoS2 and Liintercalated MoS2. High-frequency fractions which were observed for highly dispersed MoS2 in the range of 400-650 Mrads are identified as cus Md'+ sites (sulfur vacancies) located at the edges and comers of the MoS2 particles. These results are relevant for the interpretation of TDPAC spectra of supported sulfided MO/A1203 catalysts. We have shown that TDPAC can experimentally follow the genesis of the active catalysts during sulfidation procedures and can also detect aging processes. In their initial stages, the Mo surface species contain oxygen and sulfur in the coordination sphere. Also, octahedrally coordinated Mo sites may be present. Severely sulfided or aged catalysts have a NQI frequency at 110 Mrads that suggests that relatively well-ordered MoS2 structures have been formed. High-frequency fractions in the range 400-650 Mrads are also detected, and their relative abundance is comparable to the percentage of surface (edge and comer) Mo atoms in the catalysts for given dimensions of the MoS2 platelets. We suggest that these sites are similar to those detected on dispersed unsupported MoS2 and that they correspond to the cus M d + sites that can be probed by CO chemisorption. This measurement may be the first direct detection of the cus M d + coordination sites which may be involved in the catalytic hydrodesulfurization reaction. Acknowledgment. The neutron irradiations were carried out at Forschungsreaktor Munchen (FRM).We gratefully acknowledge the interest and support by Professor Dr. G. M. Kalvius and Professor Dr. K. Andres. The work was financially supported by the Deutsche Forschungsgemeinschaft, by the Bundesminister fiir Forschung und Technologie, and by the Fonds der Chemischen Industrie. References and Notes (1) Prins, R; de Beer, V. H. J.; Somorjai, G. A. Catal. Rev. Sci. Eng. 1989, 31, 1. (2) Topsoe, H.; Clausen, B. S.; Topsoe, N.-Y.; Pedersen, E. Ind. Eng. Chem. Fundam 1986, 25, 25. (3) Knozinger, H. Proc. In?. Congr. Catal., 9th, Calgary, 1988; Phillips, M. J., Teman, M. Eds.; The Chemical Institute of Canada: Ottawa, Ontario, 1988; p 20. (4) Grange, P. Catal. Rev. Sci. Eng. 1980, 21, 135. (5) Chianelli, R. R. Catal. Rev. Sci. Eng. 1984, 26, 361. (6) Ratnasamy, P.; Sivasanker, S. Catal. Rev. Sci. Eng. 1980,22,401. (7) Delmon, B. Chemistry and Uses of Molybdenum, Proc. Climax 3rd Int. Conf.; Barry, H. F., Mitchell, P. C. H., Eds.; Climax Molybdenum Comp.: Ann Arbor, Michigan, 1979; p 73. (8) Mitchell, P. C. H. Catal. R. SOC.Chem. London 1980, 4, p 175. (9) Schrader, G. L.; Cheng, C. P. J. Catal. 1983, 121, 369. (10) Payen, E.; Kasztelan, S . ; Houssenbay, S . ; Szymanski, R.; Grimblot, J. J. Phys. Chem. 1989, 93, 6501. (11) Polz, J.; Zeilinger, H.; Muller, B.: Knozinger, H. J. Catal. 1989, 120, 22. (12) Parham, T. G.; Merrill, R. P. J. Catal. 1984, 85, 295. (13) Chiu, N.-S.; Bauer, S. H.; Johnson, M. F. L. J. Catal. 1986, 98, 32. (14) Bauer, S. H.; Chiu, N.-S.; Johnson, M. F. L. J. Phys. Chem. 1986, 90, 4888. (15) Clausen, B. S.; Topsoe, H.; Candia, R.; Lengeler, B. Proc. Int. Con$ EXAFS and Near Edge Structure III; Hodgson, K. O., Hedman, B., PennerHahn, J. E., Eds.; Springer: Berlin, Heidelberg, New York, Tokyo, 1984; p 181. (16) Clausen, B. S . ; Topsoe, H.; Candia, R.; Villadsen, J.; Lengeler, B.; Als-Nielsen, J.; Christensen, F. J. Phys. Chem. 1981, 85, 3868.
(17) Boudart, M.; Dalla Betta, R. A,; Foger, K.; Loffler, D. G.; Samant, M. G. Science 1985, 228, 717. (18) Delannay, F. Appl. Catal. 1985, 16, 135. (19) Pratt, K. C.; Sanders, J. V.; Christov, R. J. Catal. 1990, 124, 416. (20) Sanders, J. V. J. Electron Microsc. Tech. 1986, 3, 67. (21) Villa-Garcia, M. A,; Lindner, J.; Sachdev, A,; Schwank, J. J. Catal. 1989, 119, 388. (22) Korgnyi, T.; Manninger, I.; Paal, Z.; Marks, 0.; Gunter, J. R. J. Catal. 1989, 116, 422. (23) Vogdt, C.; Butz, T.; Lerf, A,; Knozinger, H. Polyhedron 1986, 5, 95. (24) Amoldv. P.: van den Heikant. J. A. M.; de Bok. G. D.: Mouliin, J. A. J. Catal. 1985, 92, 35. (25) Vissers, J. P. R.; Scheffer, B.; de Beer, V. H. J.; Mouliin, J. A,; Prins, R. J. Catal. 1987, 105, 277. (26) Stuckly, V.; Zahradnikova, H.; Beranek, L. Appl. Catal. 1987, 35, 23. (27) Kasztelan, S.; Toulhoat, H.; Grimblot, J.; Bonnelle, J. P. J. Catal. 1984, 13. 127. (28) Wambeke, W.; Jalowiecki, L.; Kasztelan, S . ; Grimblot, J.; Bonnelle, J. P. J. Catal. 1988, 109, 320. (29) Farragher, A. L. Adv. Colloid Interface Sci. 1979, 11, 3. (30) Muller, B.; van Langeveld, A. D.; Moulijn, J. A,; Knozinger, H. J. Phys. Chem. 1993, 97, 9028. (31) Qin, X. R.; Yang, D.; Frindt, R. F.; Irvin, J. C. Phys. Rev. B 1991, 44, 3490. (32) MaugC, F.; Lavalley, J. C. J. Catal. 1992, 137, 69. (33) Muller, B. Doctoral Thesis, University of Munich, 1993. (34) Butz T.; Vogdt, C.; Lerf, A.; Knozinger, H. J. Catal. 1989, 116, 31. (35) Lerf, A,; Butz, T. Angew. Chem. 1987, 99, 113; Angew. Chem. Int. Ed. 1987, 26, 110. (36) Mottner, P.; Led, A.; Butz, T.; Knozinger, H.; Muller, A,; Wittneben, V.; Krickemeyer, E. Chem. Phys. 1992, 160, 327. (37) Steffen, R. M.; Frauenfelder, H. In Alpha-, Beta- and Gamma Ray Spectroscopy; Siegbahn, K., Ed.; North Holland: Amsterdam, 1965; Vol. 2, p 997. (38) Butz, T. Hyperfine Interact. 1989, 52, 189. (39) Gerdau, E.; Wolf, F.; Winkler, H.; Braunsfurth, J. Proc. R. SOC. London 1969, 311, 197. (40) Rogers, J. D.; Vasquet, A. Nucl. Instrum. Methods 1976, 130, 539. (41) Butz, T.; Vasquez, A.; Lerf, A. J. Phys. C 1979, 12, 4059. (42) Diemann, E.; Muller, A.; Aymonino, P. Z. Anorg. Allg. Chem. 1981, 479, 191. (43) Prasad, T.; Diemann, E.; Muller, A. J. Inorg. Nucl. Chem. 1973, 35, 1895. (44) Brauer, G. Handbuch der praparativen Anorganischen Chemie Ilk Enke Verlag: Stuttgart, 1981; p 1551. (45) van den Berg, J. M. Inorg. Chim. Acta 1968, 2, 216. (46) Py, M. A.; Haering, R. R. Can. J. Phys. 1983, 61, 76. (47) Dines, M. B. Mater. Res. Bull. 1975, 10, 287. (48) Joensen, P.; Frindt, R. F.; Morrison, S . R. Mater Res. Bull. 1986, 21, 457. (49) Murphy, D. W.; Hull, G. W. J., Jr. Chem. Phys. 1975, 62, 973. (50) Miremadi, B. K.; Momson, S . R. J. Catal. 1991, 131, 127. (51) Vogdt, C. Doctoral Thesis, University of Munich, 1986. (52) Butz, T.; Lerf, A.; Hubler, A.; Gierisch, H.; Saibene, S.; Besenhard, J. 0. Hyperfine Interact. 1983, 15/16, 925. (53) Somoano, R. A.; Woolam, J. A. In Intercalated Layered Materials; Levy, F., Ed.; Reidel: Dordrecht, 1979. (54) Butz, T.; Lerf, A.; Besenhard, J. 0. Rev. Chim. Miner. 1984, 21, 556. (55) Butz, T.; Lerf, A. Rev. Chim. Miner. 1982, 19, 496. (56) Butz, T.; Led, A,; Vogdt, C.; Eid, A. M. M. Hyperfine Interact. 1983, 16, 915. (57) Ni, X.; Sun, G.; Butz, T.; Lerf, A. Chem. Phys. 1988, 123, 455. (58) Sun, G.; Ni, X.; Butz, T.; Lerf, A. Chem. Phys. Lett. 1988, 151, 54. (59) Schollhom R.; Weiss, A. J. Less-Common Met. 1974, 36, 229. (60) Joensen, P.; Crozier, E. D.; Albering, N.; Frindt, R. F. J. Phys. C 1987, 20, 4043. (61) Tanaka, K; Okuhara, T. J. Catal. 1982, 78, 155. (62) Suzuki, K.; Soma, M.; Onishi, T.; Tamaru, K. J. Electron Spectrosc. Relat. Phenom. 1981, 24, 283. (63) Ni, X.; Butz, T.; Lerf, A. Hyperfine Interact. 1989, 52, 131. (64) El-Kateb, S.; Martin, P. W.; Mulhem, P. J. Hyperfine Interact. 1990, 60, 781. (65) Vrinat, M.; Breysse, M.; Geantet, C.; Ramirez, J.; Massoth, F. Caral. Lett. 1994, 26, 25. JP94287 1Z