Rh(0) nanoclusters in benzene hydrogenation catalysis: Kinetic and mechanistic evidence that a putative [(C8H17)3NCH3]+[RhCl4]- ion- pair catalyst is actually a distribution of C

Article abstract of DOI:10.1021/ja973045h

A reinvestigation is reported of a prototype literature arene hydrogenation system, one previously believed to involve a [(C₈H₁₇)₃NCH₃]⁺[RhCl₄]^- ion-pair catalyst. The methodolog

Full text of DOI:10.1021/ja973045h

J. Am. Chem. Soc. 1998, 120, 5653-5666
5653
Rh(0) Nanoclusters in Benzene Hydrogenation Catalysis: Kinetic and
Mechanistic Evidence that a Putative [(C8H17)3NCH3]+[RhCl4]-
Ion-Pair Catalyst Is Actually a Distribution of Cl- and
[(C8H17)3NCH3]+ Stabilized Rh(0) Nanoclusters
Katherine S. Weddle, John D. Aiken III, and Richard G. Finke*
Contribution from the Department of Chemistry, Colorado State UniVersity, Ft. Collins, Colorado 80523
ReceiVed August 29, 1997
Abstract: A reinvestigation is reported of a prototype literature arene hydrogenation system, one previously
believed to involve a [(C8H17)3NCH3]+[RhCl4]- ion-pair catalyst. The methodology employed to uncover the
true catalyst, and to deal with the classic and difficult mechanistic problem of “is it homogeneous or
heterogeneous catalysis?”, is the four-step mechanistic approach developed previously in our laboratories.
The data obtained (i) provide unequivocal TEM evidence that Rh(0) nanoclusters are formed under the reaction
conditions and (ii) provide kinetic evidence that the benzene catalytic hydrogenation reaction follows the
nucleation (A f B) and then autocatalytic surface-growth (A + B f 2B) mechanism elucidated recently for
metal(0) nanocluster growth. These latter results require that “A” (i.e., [RhCl4]-) is not the catalyst to within
the error limits (5-15%) of the fits of the data to the autocatalytic surface-growth mechanism; the kinetic
results also provide some of the strongest possible evidence that “B” is the true catalyst, “B” being the Rh(0)
nanoclusters. In addition, (iii) H/D exchange and (iv) Hg(0) poisoning data confirm that the Rh(0) nanoclusters
are the only active catalysts since added Hg(0) poisons the arene hydrogenation completely. The results reported
herein are of fundamental significance in five ways: (i) they are only the second use each of two new and
powerful methodologies that were required for the success of the studies reported, the (a) more general 4-step
methodology for testing “is it homogeneous or heterogeneous catalysis”, and (b) the pseudoelementary, catalytic
reporter methodology for following the nanocluster growth kinetics. In addition, (ii) they correct the claim
that [RhCl4]- is a benzene hydrogenation catalyst, and identify soluble Rh(0) nanoclusters as the true catalyst;
(iii) they call into question all previous claims of benzene hydrogenationsbut not anthracene or naphthalene
arene hydrogenationsby monometallic precatalysts; and (iv) they re-emphasize that, prior to any claim of a
homogeneous catalyst in a reaction (such as arene hydrogenation) where a facile heterogeneous M(0) catalyst
is well established, one must first rule out catalysis by even trace amounts of possibly highly active nanocluster
catalysts (e.g., by using the methods utilized herein and any other applicable method). Overall, the studies
presented herein (v) provide a definitive answer, at least for the specific Rh system studied, to the 34-year-old
question, one controversial for the past 17 years, of “is benzene hydrogenation homogeneous or heterogeneous?”.
Introduction
would, in turn, allow detailed catalyst composition and structure
determination, plus kinetic, spectroscopic, and other mechanistic
studies. Such key insights could then be employed to help in
the rational design of better, more selective arene hydrogenation
catalysts.
SelectiVe arene hydrogenation1 of, for example, benzene to
cyclohexene2 is a topic of considerable fundamental as well as
commercial interest.2o,p One major impediment in this area is
the lack of a well-defined, robust, and reproducible system that
is also soluble, especially one in which the catalyst is also
unequivocally identified. Such properties and information
Exciting both in the above context, and for hybrid homoge-
neous/heterogeneous catalysis in general, is the realization that
a few transition metal nanocluster systems have recently been
found to be stable enough that they can behave as isolable and
compositionally well-defined “soluble heterogeneous catalysts”sa
little precedented situation3 with exciting if not unparalleled
possibilities for anyone interested in heterogeneous catalysis and
in the rational development of mechanism-based, nanocluster
catalysts. Such soluble analogues of heterogeneous catalysts
are new enough, however, that isolable, high catalytic activity
examplessones proven to undergo tens of thousands of total
turnovers of even less demanding reactions, such as olefin
hydrogenation, in solution and without catalyst agglomeration
into bulk metalsare presently restricted to one example.3 That
system is [P2W15Nb3O62]9- polyoxoanion- and [Bu4N+]-
* E-mail: rfinke@lamar.colostate.edu.
(1) A literature search (Chemical Abstracts Service; CA file) of benzene-
(w)hydrogenation reveals 974 references, including 231 patents, while a
literature search of selective(w)benzene(w)hydrogenation reveals only 50
references (an incomplete search, since we already have in our files more
than 50 papers concerned with selectivity in benzene hydrogenation). Lead
references follow: (a) Struijk, J.; d’Angremond, M.; Lucas-de Regt, W. J.
M.; Scholten, J. J. F. Appl. Catal. A: General 1992, 83, 263. (b) Struijk,
J.; Moene, R.; van der Kamp, T.; Scholten, J. J. F., Appl. Catal. A: General
1992, 89, 77. (c) Struijk, J.; Scholten, J. J. F. Appl. Catal. A: General
1992, 82, 277. (d) Odenbrand, C. U. I.; Lundin, S. T. J. Chem. Technol.
Biotechnol. 1980, 30, 677. (e) Struijk, J.; Scholten, J. J. F. Appl. Catal.
1990, 62, 151. (f) Van der Steen, P. J.; Scholten, J. J. F. Appl. Catal. 1990,
58, 291. (g) Odenbrand, C. U. I.; Andersson, S. L. T. J. Chem. Technol.
Biotechnol. 1982, 32, 691 (h) Don, J. A.; Scholten, J. J. F. Faraday Discuss.,
Chem. Soc. 1982, 72, 145. (i) Odenbrand, C. U. I.; Lundin, S. T. J. Chem.
Technol. Biotechnol. 1981, 31, 660.
S0002-7863(97)03045-X CCC: $15.00 © 1998 American Chemical Society
Published on Web 05/28/1998

5654 J. Am. Chem. Soc., Vol. 120, No. 23, 1998
Weddle et al.
stabilized Ir(0)∼300 nanoclusters,3,4 a system proven to undergo
g18 000 turnovers of catalytic hydrogenation in solution.11
Fundamental studies of nanocluster mechanisms of formation,5
stabilization by polyoxoanions, by other anions,6a by various
R4N+ cations,6b and by different solvents,7 as well as the
synthesis and characterization of other examples of polyoxoan-
ion-stabilized nanoclusters (Rhxx, Pdyy, Ptzz), are in progress and
will be reported in due course. An overall goal of such studies
is the development, application, and full kinetic and mechanistic
understanding of nanocluster “soluble heterogeneous catalysts”.8
Benzene hydrogenation is well established using traditional,
oxide-supported metal particle heterogeneous catalysts.1,2 Given
this, and given that selective hydrogenation of benzene to
cyclohexene is an important commercial goal, we reasoned that
a perusal of the literature of benzene hydrogenations1,2 was likely
to reveal systems undergoing arene hydrogenation and claimed
to be a monometallic, “homogeneous catalyst”, but where the
true catalyst could instead be the transition-metal nanoclusters
formed from the monometallic catalyst precursors. Our key
initial goal was to identify a prototype benzene hydrogenation
system that we could, then, compare to our own efforts to
develop polyoxoanion-stabilized Rh(0), Ru(0), and other nano-
cluster benzene hydrogenation catalysts.
genation catalysts are for the more easily reduced polycyclic
aromatics (e.g., anthracene or naphthalene), but not from
monocyclics such as benzene; (b) that these kinetically docu-
mented anthracene or naphthalene homogeneous catalysis
systems do not, however, reduce benzene under conditions
where they remain undecomposed and homogeneous; (c) that
“a major question concerning these monocyclic arene hydro-
genation catalysts is whether the active catalysts are metal “sols”
in the 10-40 Å size range”;9a and (d) that, “Previously we saw
how difficult it is to determine the actual catalytically active
speciesseven with well-defined homogeneous olefin hydroge-
nation catalysts. In the instance of monocyclic arene hydro-
genation catalysts, this is more difficult”.9a Of historical interest
here is that the lore of catalysis once held that the observation
of benzene reduction could be used as a test for heterogeneous
catalysissthat is, that only heterogeneous catalysts could
hydrogenate benzene. In one sense, then, this paper returns to
attempt a more strenuous test of that issue. Note also that the
first report of a “homogeneous” monocyclic arene (o-xylene)
hydrogenation catalyst was in 1963 [NiII(2-ethylhexanoate) +
Et3Al; see ref 29 in Table 10.2 elsewhere9a], and that the first
paper presenting a test for the “is it homogeneous or hetero-
geneous catalysis” question was Maitlis’ 1990 paper (ref 55
cited elsewhere9a). The present paper returns, then, to this
between 17- and 34-year-old issue and attempts to answer it
definitively in the case of benzene hydrogenation and for the
Several systems arose from our literature search as
possiblesbut undemonstrated and thus previously unrecog-
nizedsnanocluster benzene (or other substrate) hydrogenation
catalysts.9 The insightful review and analysis of reported
benzene and other arene hydrogenation “homogeneous” catalysts
in a chapter written by Collman available elsewhere9a is
recommended as a first source of critically analyzed literature
in this area. Note especially Collman’s admonitions: (a) that
the only kinetically demonstrated homogeneous arene hydro-
-
RhCl4 precatalyst discussed below.
After a careful survey of the benzene and other arene
hydrogenation literature, both of the claimed benzene homo-
geneous hydrogenation catalysts and also of the established
heterogeneous catalysts, we chose to study a prototype literature
benzene hydrogenation system (hereafter, the “literature sys-
tem”) shown in Scheme 1.9c This system is partially based on
an earlier, important paper by a second group:10
Several features of the literature system9c caught our eye and
are highly suggestive of the presence of stabilized nanocluster
catalysts, at least in hindsight and with the advantage of recent
(2) (a) The selective hydrogenation of benzene to cyclohexene is of
synthetic and industrial interest2b-l since cyclohexene is a useful intermediate
material in the synthesis of commercially important products,2m such as
adipic acid.2n,o (b) Mitsui, O.; Fukuoka, Y. U.S. Patent 4,678,861, 1987, to
Asahi Kasei Kogyo Kabushiki Kaisha. (c) Niwa, S.; Mizukami, F.; Kuno,
M.; Takeshita, K.; Nakamura, H.; Tsuchiya, T.; Shimizu, K.; Imamura, J.
J. Mol. Catal. 1986, 34, 247. (d) Niwa, S.; Mizukami, F.; Isoyama, S.;
Tsuchiya, T.; Shimizu, K.; Imai, S.; Imamura, J. J. Chem. Technol.
Biotechnol. 1986, 36, 236. (e) Ichihashi, H.; Yoshioka, H. U.S. Patent
4,575,572, 1986, to Sumitomo Chemical Company. (f) Niwa, S.; Immamura,
J.; Mizukami, F.; Shimizu, K.; Orito, Y. U.S. Patent 4,495,373, 1985 to
Director-General of the agency of Industrial Science and Technology. (g)
Hideyuki, A.; Akio, K. U.S. Patent 4,197,415, 1980 to Toray Industries,
Inc. (h) See ref 1c. (i) Drinkard, W. C. Patent 1,381,48, 1975 to du Pont de
Nemours and Co. (j) Drinkard, W. C. U.S. Patent 3,767,720, 1973 to du
Pont de Nemours and Co. (k) Hartog, F. Patent 1,094,911, 1965, to
Stamicarbon N. V. (l) See ref 1g. (m) Hartog, F. U.S. Patent 3,391,206,
1968, to Stamicarbon N. V. (n) Nagahara, H.; Konishi, M. EP Patent
0220525, 1987, to Asahi Kasei Kogyo Kabushiki Kaisha. (o) G. W. Parshall
notes: “The hydrogenation of benzene to cyclohexene has been a major
target of industrial research because the oxidation of cyclohexene to adipic
acid may proceed more cleanly than the current cyclohexane oxidation.”
Parshall, G. W.; Ittel, S. D. Homogeneous Catalysis, John Wiley & Sons:
New York, 1992; pp 180-183. (p) We note that a plant for commercial,
selective hydrogenation of benzene to cyclohexene was brought on line by
Asahi in 1990: Chem. Eng. 1990, 97(10), 25.
(3) Aiken, J. D., III; Lin, Y.; Finke, R. G. J. Mol. Catal. 1996, 114, 29.
(4) Yin, L.; Finke, R. G. J. Am. Chem. Soc. 1994, 116, 8335.
(5) (a) Watzky, M. A.; Finke, R. G. J. Am. Chem. Soc. 1997, 119, 10382.
(b) Watzky, M. A.; Aiken, J. D., III, Widegren, J.; Finke, R. G. Chem.
Mater. Submitted (“A New Kinetic Method to Follow Transition-Metal
Nanocluster Formation Based on Catalytic Activity and the Pseudoelemen-
tary Step Concept”).
(6) (a) Aiken, J. D., III; Finke, R. G. Experiments in progress. (b) Aiken,
J. D., III; Finke, R. G. Experiments in progress. (c) Aiken, J. D., III; Finke,
R. G. J. Am. Chem. Soc., in press (Nanocluster Formation Synthetic, Kinetic
and Mechanistic Studies. The Detection of, and Then Methods to Avoid,
Hydrogen Mass-Transfer Limitations in the Synthesis of Polyoxoanion- and
(8) (a) However, it should be noted that other nanoclusters, such as the
Rh∼55 catalysts studied by Schmid8b were found of insufficient stability in
solution to be able to catalyze even >10 turnovers of hydroformylation.
Indeed, it is often stated that nanoclusters are insufficiently stable in solution
to be able to support catalysis without aggregation to bulk metal or without
supporting them on a solid-oxide support.8c (b) Schmid, G. In Aspects of
Homogeneous Catalysis; Ugo, R., Ed.; Kluwer: The Netherlands, 1990;
Vol 7, p 31. (c) Schmid, G.; Maihack, V.; Lantermann, F.; Peschel, S. J.
Chem. Soc., Dalton Trans. 1996, 589.
(9) (a) Collman, J. P.; Hegedus, L. S.; Norton, J. R.; Finke, R. G.
Principles and Applications of Organotransition Metal Chemistry; Univer-
sity Science Books: Mill Valley, CA, 1987. See pp 549-556, and Table
10.2, systems A-J and refs 29-38 therein, a section written by Professor
Collman (who conducted research for several years in the “is it homogeneous
or heterogeneous catalysis” area and specifically studying arene hydrogena-
tion catalysts). (b) Jones, R. A.; Seeberger, M. H. J. Chem. Soc., Chem.
Commun. 1985, 373. Note the induction periods and dark catalyst colors
described therein. (c) Blum, J.; Amer, I.; Vollhardt, K. P. C.; Schwarz, H.;
Hohne, G. J. Org. Chem. 1987, 52, 2804. (d) We also found a couple of
other systems9e,f that are ostensibly homogeneous Rh olefin or Ru aldehyde
hydrogenation catalysts, but where it might be an interesting and rigorous
test (perhaps of the method itself11), to see if these systems are actually
homogeneous. (e) Bergbreiter, D. E.; Chandran, R. J. Am. Chem. Soc. 1987,
109, 174. Note the effort by the authors to probe the “is it homogeneous or
heterogeneous” catalysis question by the methods available at that time
[e.g., their 31P NMR (e.g., gca. 90% catalyst recovery), g90% rate still
after 18 catalyst recycles, Table 1), and use of Collman’s three-phase test].
(f) Fache, E.; Senocq, F.; Santini, C.; Basset, J.-M. J. Chem. Soc., Chem.
Commun. 1990, 1776. Note that this interesting water-soluble system shows
little evidence for a nanocluster catalyst, save the induction period in Figure
1 therein, or the fact that the I- effect seen could possibly be explained by
I- stabilization of Ru nanoclusters. As such, it would be of some interest
to check the catalyst in this study too by especially the TEM method detailed
elsewhere11snot an inappropriate suggestion, since one does not prove a
mechanism, but only disproves alternative mechanisms.
Tetrabutylammonium- Stabilized 40 ( 6 Å Rh(0)
Nanoclusters).
to Rh(0)
∼1500
∼3700
(7) Aiken, J. D., III; Finke, R. G. Unpublished results.

Rh(0) Nanoclusters in Catalysis
J. Am. Chem. Soc., Vol. 120, No. 23, 1998 5655
Scheme 1. Summary of the Literature System9c under
Study, Including Key Experimental Observations
is well known, must consider and rule out any participation by
even a trace amount of a possibly highly active nanocluster
catalyst formed under the reaction conditions. This is first and
foremost simply the scientific method restatedsscience is an
exclusion process and thus is about disproof of all alternative
hypotheses. Also noteworthy here is that Collman made this
exact same point (about the possible contribution of trace
amounts of highly active “sols”) more than a decade ago and
specifically for benzene hydrogenation.9a The new point here,
then, is that we can now sharpen and amplify this key alternative
hypothesis for claimed benzene (and possibly other) homoge-
neous catalysts in light of the rapidly evolving literature of
nanocluster (“sol”) catalysts, a literature that was largely
unavailable even 5 years ago. The fundamental point to be
appreciated here is that very high rates are possible for very
small, high surface area, and thermodynamically higher energy
and thus kinetically highly reactive nanoclusters (vs bulk metal,
which is ca. 138 kcal/mol downhill vs the limit of a single
unsolvated Rh(0) atom13f). This means that, even cases where,
say, >99% of a homogeneous (pre)catalyst can be recovered
from a catalytic reaction, the true catalyst may still be a highly
active nanocluster formed from e1% of the precatalyst under
the reaction conditions, and in this hypothetical example.
This brings us to one more important, but historically
misinterpreted, property, that of the “highly reproducible” kinetic
behavior seen, at least for prefiltered solutions,9c a property
historically interpreted as compelling evidence that the true
catalyst must be a discrete, generally monometallic, homoge-
neous catalyst. This is a myth; one that has now been
disproved.3,4 In fact, such (e15% reproducibility is another
hallmark of nanocluster catalysts made under the hydrogen
autocatalytic growth, narrow-size-distribution-producing, nano-
cluster formation mechanism that has been recently elucidated.5
The observed effects of H2O,11 oxygen,13 or common PR3
literature (vide infra).11 These features are (i) the induction
periods seen; (ii) the shortening or complete removal of the
induction periods by pretreatment with H2 (or H2 plus arene
substrate); (iii) the observation of “small amounts of dark
particles (which dissolved in hot DMF or DMSO)”9c if both H2
and arene substrate are present [this solubility property is
characteristic of R4N+ X- (X- ) Cl-, Br-) stabilized nano-
clusters12]; and (iv) the “metallic rhodium” observed if the
[(C8H17)3NCH3]+ [RhCl4]- ion-pair precatalyst is prereduced
with H2 only (i.e., without arene present). Worth emphasizing
here are two key points. The first is the correct logic of the
observation of metallic Rh from a mononuclear precursor; this
demands that Rh nanoclusters must haVe been present in
solution, since there is no other way known, at least presentlysnor
is their any other easily imaginable waysto go from a single
metal reactant to the multi-metallic product except via smaller
metal(0) nanoclusters. The significance of this observation and
its proper interpretation was not pointed out until 1994,11 and
is still under appreciatedsbut is a powerful way in and of itself
to find literature systems claimed to be homogeneous, but which
are strong candidates for nanoclusters as the true catalysts.
Second, any claim for a homogeneous catalyst, in a reaction
where a very facile heterogeneous catalyst for that same reaction
14
impurities such as OdPR3 are also understandable in terms
of a nanocluster-based catalyst and the Rh nanocluster litera-
ture;15 even the long-chain tertiary amine stabilization seen with
the literature arene hydrogenation catalyst9c reexamined herein
has good precedent within the nanocluster literature.15a
In short, then, a critical reanalysis of the literature system,9c
in light of recent advances in the nanocluster literature3-5 and
(13) (a) Reetz, M. T.; Quasier, S. A.; Winter, M.; Becker, J. A.; Scha¨fer,
R.; Stimming, U.; Marmann, A.; Vogel, R.; Konno, T. Angew. Chem., Int.
Ed. Engl. 1996, 35, 2092. (b) Kolb, U.; Quaiser, S. A.; Winter, M.; Reetz,
M. T. Chem. Mater. 1996, 8, 1889. (c) Rothe, J.; Pollmann, J.; Franke, R.;
Hormes, J.; Bo¨nnemann, H.; Brijoux, W.; Siepen, K.; Richter, J. Fresenius
J. Anal. Chem. 1996, 355, 372. (d) Harada, M.; Asakura, K.; Ueki, Y.;
Toshima, N. J. Phys. Chem. 1992, 96, 9730. (e) Kiwi, J. Isr. J. Chem. 1979,
18, 369. (f) The ∆H(vaporization) for bulk Rh(0)n metal going to n Rh(0)
atoms is 138 kcal/mol. For further discussion of the relationship of this to
the intrinsic stability of different size nanoclusters, see elsewhere.4-6,11
(14) (a) A careful reading of the paper9c reexamined herein reveals that
OdPR3 is a likely culprit in the reported need9c to triply sublime P(OMe)3
to make Mutterties’ Co-based arene hydrogenation catalyst reproducible.
(b) Schmid, G.; Maihack, V.; Lantermann, F.; Peschel, S. J. Chem. Soc.,
Dalton Trans. 1996, 589. See the mention of OdPPh3 formation on p 590.
(c) Collman, J. P.; Hegedus, L. S.; Norton, J. R.; Finke, R. G. Principles
and Applications of Organotransition Metal Chemistry; University Science
Books: Mill Valley, CA, 1987; pp 255, 264, 340, 365, 649.
(10) (a) Januszklewicz, K. R.; Alper, H. Organometallics, 1983, 2, 1055.
(b) Note that these authors were aware that the actual catalyst had not been
identified, specifically stating on p 1056 of their paper that, “Finally, we
are not certain whether the phase-transfer process described herein involves
a soluble or insoluble rhodium catalyst”.
(11) Lin, Y.; Finke, R. G. Inorg. Chem. 1994, 33, 4891, and refs 16-27
therein to the earlier work of Maitlis, Whitesides, Laine, Crabtree, Collman,
Lewis and Lewis, and others on the research problem “is it homogeneous
or heterogeneous catalysis?”.
(12) (a) Reetz, M. T.; Helbig, W.; Quaiser, S. A.; Stimming, U.; Breuer,
N.; Vogel, R. Science 1995, 267, 367. (b) Bo¨nnemann, H.; Brijoux, W.;
Brinkmann, R.; Fretze, R.; Joussen, T.; Ko¨pper, R.; Korall, B.; Neiteler,
P.; Richter, J. J. Mol. Catal. 1994, 86, 129. (c) Bo¨nnemann, H.; Brinkmann,
R.; Neiteler, P. Appl. Organomet. Chem. 1994, 8, 361. (d) Reetz, M. T.;
Helbig, W. J. Am. Chem. Soc. 1994, 116, 7401. (e) Bo¨nnemann, H.; Brijoux,
W.; Brinkmann, R.; Dinjus, E.; Joussen, T.; Korall, B. Angew. Chem., Int.
Ed. Engl. 1991, 30, 1312.
(15) Other Rh nanoclusters: (a) Yonezawa, T.; Tominaga, T.; Richard,
D. J. Chem. Soc., Dalton Trans. 1996, 783. (b) Larpent, C. Brisse-Le Menn,
F.; Patin, H. New J. Chem. 1991, 15, 361. This Rh nanocluster also shows
facile H/D exchange with olefins and H2 (D2)/D2O (H2O). (c) Larpent, C.;
Patin, H. J. Mol Catal. 1990, 61, 65. Noteworthy is that the Rh nanocluster
shows H/D exchange with H2/D2O (or, alternatively, with D2/H2O). (d) See
ref 6c. (e) Aiken, J. D., III; Finke, R. G. Polyoxoanion- and Tetrabutylam-
monium-Stabilized Near Monodisperse, 40 ( 6 Å Rh(0)
Nanoclus-
∼2400
ters: Synthesis, Characterization and Hydrogenation Catalysis, submitted.
(f) Edlund, D. J. Ph.D. Thesis, University of Oregon, 1987.

5656 J. Am. Chem. Soc., Vol. 120, No. 23, 1998
Weddle et al.
a knowledge that the modern methods to rule out nanocluster
catalysis were not available previously, provides highly sug-
gestive, albeit not yet definitiVe, evidence that Rh(0) nanoclusters
may well be the true catalysts in the literature arene hydrogena-
tion catalyst beginning with the [(C8H17)3NCH3]+ [RhCl4]- ion-
pair precatalyst.
Herein we report the following results: (i) demonstration by
catalyst isolation and TEM that [(C8H17)3NCH3]+ [RhCl4]- is
a precatalyst, “A”, one converted under the reaction conditions
to a polydisperse distribution of Rh(0) nanoclusters; (ii)
demonstration that the isolated Rh(0) nanoclusters are kinetically
competent catalysts;16 (iii) demonstration that the benzene
hydrogenation reaction follows the nucleation (A f B) then
autocatalytic surface growth (A + B f 2B) mechanism for
nanocluster growth that we recently reported elsewhere, results
that require that A (i.e., RhCl4-) is not the catalyst to within
the 5-15% error limits of the fits to the kinetic data; and (iv)
demonstration that isolated Rh(0) nanoclusters hydrogenate
benzene with D2 to yield very similar H/D incorporation patterns
to those seen when one begins with the [(C8H17)3NCH3]+
[RhCl4]- precatalyst (and also H/D incorporation very similar
to that previously reported in the literature), “fingerprint” results
which argue strongly that all three reactions employ the same
catalyst. We also (v) find that the catalyst becomes completely
inactive when treated with Hg, the expected results for a Rh(0)
metal nanocluster catalyst. In short, we present very strong if
not compelling evidence that the true catalyst is the distribution
of [(C8H17)3NCH3]+ and Cl- stabilized Rh(0) nanoclusters
which are formed under H2 and the other, benzene hydrogena-
tion reaction conditions.
We wish to emphasize up-front that the authors of the
literature system reexamined herein were well aware of, and
tested for, the question of “is it homogeneous or heterogeneous
catalysis?”.9c They just did not have the advantages we now
have of a 1998 knowledge of nanoclusters or the advantage of
a new methodology for this “homogeneous or heterogeneous
catalysis problem”, one developed and thus intrinsically tested
on a nanocluster catalyst.11 Stated another way, the paper
describing the literature system reinvestigated herein is an
important paper, literally a “treasure trove” of experimental data
on arene hydrogenation.9c Hence, in no way do the studies
which follow diminish the value of the earlier, original work
which was focused toward the synthetic organic applications
of the ion-pair precatalyst, [(C8H17)3NCH3]+ [RhCl4]-.
Figure 1. A more general approach to distinguishing between a
“heterogeneous” colloid-nanocluster catalyst and a discrete, homoge-
neous catalyst as developed elsewhere.11
catalysis in that it was developed, and thereby intrinsically tested,
on what proved in the end to be a previously unidentified, high
activity and catalytically long-lived3 Ir(0)
nanocluster
∼300
catalyst. The four key steps to this method are shown in Figure
1: (i) catalyst isolation and then characterization by transmission
electron microscopy (TEM); (ii) kinetic studies testing the
kinetic competence or noncompetence of the isolated catalyst
(e.g., of any nanoclusters identified by TEM); (iii) quantitative
phenomenological tests (such as the classic Hg poisoning test
known to poison “heterogeneous” M(0) transition metal cata-
lysts); and (iv) then additional mechanistic studies as needed,
for example, in the present case of arene hydrogenation linking
the observed selectivities for H/D incorporation or exchange to
those independently observed for the isolated nanoclusters. That
is, a final but crucial step of the new approach to answering
the question “is it homogeneous or heterogeneous catalysis?”
is the strict adherence to the principle that the correct mechanism
and true catalysts will explain all of the data. A more complete
figure, showing the actual 12 specific experiments performed
in the four subcategories in Figure 1, is available as Figure 5
elsewhere.11 Of note is that the present study is just the second
use of the methodology in Figure 1 to uncover otherwise hidden
nanocluster catalysts, and the first time that the Figure 1
methodology has been combined with a new kinetic method
for following nanocluster growth kinetics.5a
Benzene Hydrogenation. Hydrogenation Apparatus and
“Standard Conditions”. The hydrogenation of benzene was
carried out using the two sets of conditions originally described
in the literature9c as Method A and Method B. Methods A and
B differ in that A is performed using a biphasic dichloroethane/
H2O solution whereas B is done in a monophasic THF solution.
The two methods were both carried out using a set of standard
conditions which include RhCl3‚3H2O, trioctylamine, Aliquat
336 (i.e., [(C8H17)3NCH3]+Cl-), benzene, and H2O at 31 ( 1
°C and under 687 ( 7 Torr hydrogen. The reaction was
followed via GLC to monitor the loss of benzene. The
hydrogenation apparatus used herein (Figure A, Supporting
Information) consists of a 25 mL side-armed Schlenk flask and
two hydrogen reservoirs (500 mL side-armed round-bottomed,
and a 100-mL round-bottomed flask) connected to a standard
Schlenk line equipped with a Hg manometer. Not surprisingly,
we found Method B to be easier to work with because of the
homogeneous nature of the reaction solution. But, as needed
to reinvestigate the reported system,9c we have reexamined both
of the reported methods, A and B.
Results and Discussion
Methodology Employed to Address the Question “Is It
Homogeneous or Heterogeneous Catalysis?”. First, we need
to remind the reader of the four main components of the
methodology used herein to address the central question “is it
homogeneous or heterogeneous catalysis?” The methodology
is that which we developed in 1994;11 it is an improved, more
general methodology especially for testing for nanocluster
(16) The literature system does not explain how the rates were
determined, nor does it define the term “rate”. (The failure to define “rate”
in terms of a specific differential equation that refers to a specific reaction
is one of the most common, yet most easily preventable, mistakes in
reporting kinetic data.) We were unable to calculate the cited maximum
rate (p 2805, column 1, line 62, k ) (3.51 ( 0.15) × 10-3 mmol L-1 min-1
)
given the data. Using data taken from the benzene conversion graph (p
2805, Figure 1) we calculated k ) 3.98 mmol L-1 min-1 from a first-order
plot (as the literature study describes their system as following first-order
kinetics). We found that the data had a better fit to an autocatalytic growth
curve, Figure 4, and expect, therefore, that the rate reported in the literature
study has no rigorous meaning because the data were fit to the wrong
mechanism.

Rh(0) Nanoclusters in Catalysis
J. Am. Chem. Soc., Vol. 120, No. 23, 1998 5657
show the presence of somewhat polydisperse, irregular-shaped
Rh(0) nanoclusters. Restated, TEMs confirm that Rh(0) nano-
-
clusters are indeed formed from RhCl4 in this system. The
clusters made in dichloroethane are 48 ( 18 Å, whereas the
clusters made in THF are 34 ( 13 Å, that is, of ca. (40% size
dispersion. These values correspond to the range of Rh
∼1000
(30 Å) to Rh
(66 Å) in dichloroethane, and the range
∼3900
∼11000
Rh
(21 Å) to Rh
(47 Å) in THF17san order of
∼350
magnitude spread in the number of Rh(0) atoms in the smallest
to the largest nanoclusters.
Kinetic Evidence That RhCl4- Is Not the Active Catalyst,
but That Rh(0) Nanoclusters Are the True Catalysts. Four
key experimental observations provide very strong evidence that
the observed Rh(0) nanoclusters are the true catalysts in the
literature system. First, a hydrogenation performed in THF
using RhCl3‚3H2O as the precatalyst began only after an ca.
20-30 min induction period, Figure 3a. At the end of this
period, the color of the reaction solution changed from clear
orange-red to an opaque, deep red/black. Such changes in color
are indicative of the formation of soluble metal particles.18
Second, hydrogenations performed using isolated Rh(0) nano-
clusters proceed immediately with no observable induction
periods, Figure 3b.
Third, and most significantly, essentially all the kinetic curves
observed that start with [RhCl4-], including a kinetic reanalysis
of the data reported in Figure 1 of the literature study9c (vide
infra, Figure 4), reveal induction periods plus a sigmoidal-shaped
curve that can be quantitatively curve fit, to within (5-15%
error bars, to the nucleation (A f B; rate constant, k1) then
autocatalytic surface-growth (A + B f 2B; rate constant, k2)
mechanism elucidated recently. Note that such a fit is the kinetic
signature for nanocluster formation from monometallic precur-
sors (A) and using H2 as the reductant.5 Specifically, Figure
3a shows the excellent curve fit according to this mechanism
for a reaction beginning with [RhCl4-]. Note that in Figure
3b, in which the isolated Rh(0) nanoclusters are employed as
the catalyst, no induction period is seen and the sigmoidal-
shaped curve has been replaced by the expected exponential
decay curve. Fourth, Figure 4 reveals that the kinetic data taken
directly out of the literature study9c (and which began with
[RhCl4-] in dichloroethane/H2O) reveals an induction period
(which was previously reported as a general observation)9c and
a sigmoidally shaped kinetic curve. This kinetic curve is also
closely fit by the A f B, then A + B f 2B nucleation plus
autocatalytic surface-growth mechanism.5a Moreover, Figure
F(a) of the Supporting Information shows a similarly shaped
kinetic curve, one also closely fit (to within (5-15% error bars)
by this mechanism. As noted in our earlier work, the ability of
the two-step nucleation plus autocatalysis mechanism to account
for a multistep nanocluster self-assembly reaction [one of g3900
Figure 2. (a) Top: TEM after reuse of isolated Rh(0) nanoclusters
(prepared from the monophasic reaction of RhCl3‚3H2O with H2 in
the presence of THF/Aliquat 336/H2O/trioctylamine/benzene). (b)
Bottom: Histogram of the Rh(0) nanocluster diameters. The mean
diameter is 34 Å, with a standard deviation of 13 Å, from a sample
population of 252.
Our two earlier publications on the use of pseudo-elementary
catalytic reporter reactions (benzene hydrogenation in the present
case) to follow nanocluster growth will allow the reader to
readily understand how we can (i) follow benzene hydrogenation
but (ii) still learn about the kinetics of nanocluster formation;
otherwise, this point will probably be obscure.5
steps (i.e., to g47 Å Rh(0)
nanoclusters, Figure 2)] is
∼3900
remarkable.5a It is also completely understandable, however,
Catalyst Isolation and Characterization. (1) Formation
(17) (a) The number (N) of Rh atoms in a Rh nanocluster within a given
diameter can be approximated using the following equation: N ) (N0FV)/
102.9, where N0 ) 6.022 × 1023, F ) 12.4 g/cm3, and V ) (4/3)π(D/2)3.
In dichloroethane, the nanoclusters with average diameter of 46 ( 17 Å
can be approximated as Rh3700 which is near the magic-number size cluster
of Rh3871. Similarly, 63 Å (i.e., 46 + 17) and 29 Å (i.e., 46 - 17)
nanoclusters can be approximated as Rh9500 and Rh930, respectively. In THF,
the nanoclusters with average diameter of 34 ( 13 Å can be approximated
as Rh1770 (the next magic-number size cluster is Rh2057). Similarly, 49 Å
(i.e., 36 + 13) and 23 Å (i.e., 36 - 13) nanoclusters can be approximated
as Rh4470 and Rh460, respectively. (b) See also elsewhere5b and references
therein.
-
and Isolation of Rh(0) Nanoclusters from RhCl4
. The
hydrogenation reaction was allowed to run ca. 4 h using the
standard conditions established in the literature9c for each
method (see the Experimental Section). The reaction flask was
removed from the hydrogenation apparatus and taken back into
a Vacuum Atmospheres drybox where it was evacuated to
dryness overnight to yield a black solid.
(2) Catalyst Characterization by TEM. To obtain TEM
images, a small amount of the isolated black solid was dissolved
in DMSO. The TEMs taken of isolated and then reused Rh(0)
nanoclusters from benzene hydrogenations in dichloroethane
(Figure D, Supporting Information) and THF (Figure 2) both
(18) (a) Bo¨nnemann, H.; Brijoux, W.; Brinkmann, R.; Dinjus, E.; Fretzen,
R.; Joussen, T.; Korall, B. J. Mol. Catal. 1992, 74, 323. (b) See also ref
16a. (c) Hamlin, J. E.; Hirai, K.; Millan, A.; Maitlis, P. M. J. Mol. Catal.
1980, 7, 543.

5658 J. Am. Chem. Soc., Vol. 120, No. 23, 1998
Weddle et al.
Figure 4. Literature data9c for the hydrogenation of benzene from the
reaction of RhCl3‚3H2O in the presence of dichloroethane/Aliquat 336/
H2O/trioctylamine/benzene. Also shown is a curve fit of the data
according to the A f B and A + B f 2B nucleation, then autocatalytic
surface-growth mechanism developed elsewhere.5 The resultant k1 )
7.9 ((0.6) × 10-4 min-1 and k2 ) 2.0 ((0.1) × 10-4 M-1 min-1
.
Figure 3. (a). The percent benzene conversion vs time from the
reaction of RhCl3‚3H2O with 693 Torr of D2 in the presence of THF/
Aliquat 336/H2O/trioctylamine/benzene at 31 ( 0.5 °C and its curve
fit (solid line) to the A f B nucleation plus A + B f 2B autocatalytic
surface growth nanocluster formation mechanism and analytic integrated
kinetic equations detailed elsewhere.5 The resultant rate constants from
the fit described in the text to the A f B (rate constant, k1), then A +
B f 2B (rate constant, k2) kinetic fits are: k1 ) 1.5 (( 0.2) × 10-3
min-1, k2 ) 6.7 (( 0.4) × 10-2 M-1 min-1. (b) The percent benzene
conversion vs time from the reaction of isolated Rh(0) nanoclusters
with 693 Torr of D2 in the presence of THF/H2O/benzene at 31 ( 1
°C and its approximate fit to an exponential decay. Note the absence
in (b) of the nucleation period seen in (a).
Figure 5. The percent benzene conversion vs time from the reaction
of isolated then reused Rh(0) nanoclusters in the monophasic solution
of THF/H2O/benzene. The solution was divided in two, with one-half
of the reaction solution being filtered (0) and the other half not ()).
Note that the nonfiltered solution ()) proceeds to 100% conversion
while the filtered solution (0) only shows activity to ca. 25%
conversion. [A data point for the filtered solution at longer reaction
times (1070 min) shows 24% (and thus no additional) conversion of
benzene, but was omitted from the graph for clarity.]
edly, that the filtered solution exhibits a lower reactivity in
comparison to the nonfiltered solution. This result requires that
the black solid filtered from the solution is also catalytically
active. As a control, a TEM image of the filtered solution
reveals that the nanoclusters in solution are still present and, as
expected, have not been detectably disturbed by this bulk metal
filtration process (Figure B, Supporting Materials). Note that
the black color of the solution makes it impossible to detect by
eye any black solid composed of agglomerated nanoclusters plus
bulk metal in the (black) solutions. Note also how this result
ties in to the point made elsewhere11 and cited in the Introduc-
tion: the observation of insoluble M(0) transition metal in a
reaction beginning with a monometallic precursor demands that
nanoclusters have been formed in solution.
since it is the A + B f 2B autocatalytic surface-growth step
which is the key, repetitive mechanistic step.5a In short, these
benzene-loss kinetic data, plus the direct TEM obserVation that
Rh(0) nanoclusters are the product (“B”) formed from [RhCl4]-
precatalyst, proVide compelling eVidence that Rh(0) nanoclusters
are the true catalysts for benzene hydrogenation. Recall also
that polymetallic Rh(0) particles, but not RhCl4-, are well-
precedented arene hydrogenation catalysts.1,2
One issue that comes up here is whether any agglomerated
Rh(0) nanoclusters or bulk, filterable Rh(0) metal, are contribut-
ing to the observed catalytic activity. To test this, two
hydrogenations were performed using a single sample of a
solution of Rh(0) nanoclusters, but with one-half of the black
solution being filtered through Whatman #1 paper while the
other half was not. The results (Figure 5) show, not unexpect-
The converse to the above observation is also generally true,
in that if nanoclusters are present in solution, then sufficiently

Rh(0) Nanoclusters in Catalysis
J. Am. Chem. Soc., Vol. 120, No. 23, 1998 5659
high temperatures and long enough times will cause them to
precipitate bulk metal. This latter observation provides a simple
experimental test for the presence of soluble nanocluster
catalysts: heat the solution; if metal precipitate is formed, then
obtain a TEM of the unheated reaction solution to see if
nanoclusters are present (and also do the control, as done herein,
to show that the TEM electron beam does not induce the
formation of nanoclusters from the monometallic precursor).
two experiments done herein and under nearly identical condi-
tions (k1 ) 4.28 ((1.54) × 10-3 min-1; k2 ) 1.63 (( 0.84) ×
10-2 M-1 min-1) shows that the k2 values from both this work
and from our kinetic analysis of the literature data (Figure 5
herein) are the same within the large error bars observed. [Our
k1 value is ca. 325% faster than that reported for the literature
system,9c but, again, the k1 values are the same within the large
error bars observed ((210% for k1, at 1σ). Note that the Method
A conditions detailed in the literature system’s9c experimental
section were used in these experiments, not those provided in
the figure caption for Figure 1 elsewhere.9c] In short, these
tests of the kinetic (ir)reproducibility of the present arene
hydrogenation system both (a) confirm, by the irreproducibility
seen, the predicted kinetic behavior for the range in size of
nanoclusters plus bulk metal present, and (b) show that the
literature system has the same kinetic k1 and k2 rate constants
within the large error bars seen.19c This provides another
significant piece of kinetic evidence in support of the conclusion
that the literature system’s catalyst, and the Rh(0) nanoclusters
characterized herein, are one and the same.
Testing the Kinetic Reproducibility of the Literature
System. The presence of agglomerated, insoluble Rh(0) nano-
clusters detected above made us retest the literature study’s
report9c of “highly reproducible” kinetics, at least when using
their prefiltered solutions (which their experimental section says
they did “if some precipitate was formed”9c). The results
presented in the previous section make it clear that the
experimental protocol of prefiltering the (black) reaction solu-
tions can now be seen to be a problematic, nonreproducible
experimental procedure; the black color of the solutions
precludes any reproducible visual determination of the (black)
Rh(0) precipitate. Based on results with Ir(0) nanoclusters and
when bulk metal is formed, a variability of at least 70% in rate
is expected due to the precedented variable surface area of bulk
metal precipitates.19,20 In addition, the TEM results reported
herein, and which show a (40% size distribution of nanoclus-
ters, predict that that the rates should vary by at least 2-fold
since the percentage of catalytically active Rh(0) surface atoms
decreases by 2-fold, from ca. 52% to ca. 26%, on going from
the smaller 21 Å to the larger 47 Å Rh(0) nanoclusters seen by
TEM. In short, the predicted kinetic (ir)reproducibility for
unfiltered or even filtered solutions is predicted to be g200%.
Two other conclusions are worth noting here. First, the
observations of such g( ca. 100-200% variable rates in
unfiltered solutions is another semiquantitative test that can now
be used to distinguish soluble nanocluster vs agglomerated
nanocluster plus bulk metal catalysissone can now add a kinetic
reproducibility of e15% or g100-2300% as semiquantitative
guidance to Maitlis’ classic filtration tests for distinguishing
“homogeneous” from “heterogeneous” catalysts18c (where, for
this filtration test, “heterogeneous” means agglomerated, bulk
metal catalysts). Note also here that this finding fortifies the
conceptual point made previously:11 it is crucial, in the third
step of the four-step method (Figure 1),11 to apply quantitatiVely
any phenomenological tests used to distinguish homogeneous
from heterogeneous catalysis. Second, the present Rh(0)
nanocluster system is not, because of its kinetic irreproducibility,
a useful one for studying the kinetics and mechanism of arene
hydrogenation in a soluble nanocluster system. However,
nanoclusters which show near-monodisperse3 distributions, and
hence which give rise to e15% reproducible rates,3-5 should
be of significant interest in this regard and are, therefore, under
separate investigation.
In five otherwise identical experiments (Method B, unfiltered
solutions), the kinetic (ir)reproducibility was ca. (210% for k1
and ca. (340% for k2. Moreover, a comparison of the k1 and
k2 values from the literature data in Figure 4 (k1 ) 7.9 ((0.6)
× 10-4 min-1; k2 ) 2.0 ((0.1) × 10-2 M-1 min-1) to those of
(19) (a) Clusters of 47 Å (34 +13 Å) correspond to Rh3945 which is
close to the magic number cluster size of Rh3871 (where the shell number19b
is n ) 10). To determine the number of Rh atoms on the surface of the
cluster, one needs to calculate how many atoms were added to the cluster
from the n - 1 cluster. Using the equation for calculating magic number
clusters, n ) 9 corresponds to Rh2869. Using this value one can calculate
that 1002 (3871 - 2869) Rh atoms were added to the surface on going
from the n ) 9 to n ) 10 cluster. This means approximately 26% [(1002/
3871) × 100)] of the total Rh atoms are on the surface of the cluster. The
same calculation can be done for the 21 Å (34 - 13 Å) clusters which
correspond to Rh352 which is close to the magic cluster size of Rh309 (n )
4), where it can be shown that 52% of the Rh atoms are on the surface of
a Rh561 cluster. Therefore, the surface number of Rh atoms decreases from
ca. 52% to ca. 26% (i.e., a 2-fold drop in the number of surface Rh atoms)
on going from a 47 Å to a 21 Å cluster. (b) Teo, B. K.; Sloane, N. J. A.
Inorg. Chem. 1985, 24, 4545. (c) Of course, and since kinetic studies were
not a focus of the earlier work, it is not even clear how the claim of “highly
reproducible” (p 2805, top left-hand column)9c was evaluated (no kinetic
data were actually reported to support the claim of reproducible kinetics,
nor were any experimental details provided for how the kinetics were
performed9c). Since the curves show an induction period and are sigmoidal
(i.e., as shown by the fit of the literature data in Figure 4 herein), but since
their autocatalytic nature was not appreciated previously, they could not
have been properly analyzed previously and, hence, any focus attempting
to further analyze the prior claim of highly reproducible kinetics is probably
a misguided effort.
(20) Elsewhere we have recently shown that H2 gas-to-solution mass-
transfer limitations (MTL) can occur in nanocluster formation reactions for
transition metals that are among the faster hydrogenation catalysts; in fact,
the specific case we demonstrated this for is Rh.6c We also showed that
variable rates and polydisperse nanoclusters are the result of such mass-
transfer limitations.6c Hence, we tested whether the literature conditions
are actually H2 mass-transfer limitation conditions or not. The results
described in the Experimental Section indicate that the system is not
influenced by the stirring rate in the region under study (Figure G,
Supporting Information). Fortunately, then, H2 mass-transfer limitations are
not a problem in the literature,9c nor the present, study.
Phenomenological Tests for Heterogeneous vs Homoge-
neous Catalysis. (1) D Incorporation and H/D Exchange
Studies. The literature system9c reports multiple deuterium
incorporation into cyclohexane (d3-d12), but found no incor-
poration of deuterium into unreacted benzene at 50% conversion.
This exchange pattern can be used as a “fingerprint”, one
characteristic of the true catalyst functioning in the literature
system.
Two experiments, one using the precatalyst and the second
using isolated Rh(0) nanoclusters, were done to determine the
deuterium incorporation both into the product and the H/D
exchange into the unreacted benzene starting material. The
hydrogenation reaction using the precatalyst under D2 was
sampled by GC-MS along the way to 100% conversion of the
benzene. GC-MS analysis at, for example, 44% conversion
showed multiple deuterium incorporation (d3-d9) into cyclo-
hexane, Figure 6, and when the electron impact mass spectra
were obtained at an ionizing voltage of 17 eV. No deuterium
incorporation into unreacted benzene is seen, consistent with
the literature system’s findings.9c In what proved to be a telling
experiment, we repeated the deuterium incorporation “finger-
print” experiment using isolated Rh(0) nanocluster catalyst, D2
and at 66% conversion and 17 eV ionizing voltage. The results

5660 J. Am. Chem. Soc., Vol. 120, No. 23, 1998
Weddle et al.
footnote b elsewhere9c). This both predicts and explains the
observed 1-2 amu lower peaks we see at the higher, 70 eV
ionizing voltage; they are due to D loss from the R+• formed
following (too energetic) electron impact. Rather clearly, then,
the higher amu pattern of D-incorporation peaks reflects the
true D-incorporation. On the other hand, the mass spectra at
70 eV are likely to be more sensitive and repeatable, especially
from lab to lab, and this is borne out in our results in entries 4
and 5, Table 1. Overall, these experiments serve as a useful
experimental reminder of a well-established22a-c (but perhaps
still under appreciated) point for anyone doing D-incorporation
analyses by mass spectroscopy: they should be done as a
function of ionizing voltage if electron impact methods are
employed.
(2) Mercury Poisoning Experiments. Mercury is a well
known and accepted heterogeneous catalyst poison, due to its
adsorption onto the catalyst surface or amalgam formation.23
The literature system9c reported that the addition of mercury
(an unreported and perhaps too small amount, vide infra) had
no effect on the observed reaction. However, we find that
sufficient Hg to contact all the catalyst reproducibly causes a
complete loss of catalytic activity. Literature precedent also
shows that large equivalents of Hg/equiv of metal catalyst have
generally been used in such Hg(0) poisoning studies,11,23b,c and
our results below strengthen the recommendation to examine
variable ratios of Hg(0) to catalyst in Hg(0) poisoning experi-
ments (from ca. 1.0 equiv to large, g300 equiv, excesses of
Hg(0)).
Three types of experiments were done to determine the effect
of Hg on the precatalyst as well as the isolated Rh(0) nanocluster
catalyst. First, a dark red/black reaction solution, made from
precatalyst, was stopped after 80% conversion and ca. 320 equiV
of mercury was added. After stirring for 1 h, the reaction
solution had changed color from dark red/black to yellow-
orange. The solution was reconnected to the hydrogenation
apparatus and, after 5 h, no further loss (e(10%) of benzene
was detected (Figure 7), indicating that Hg had completely
poisoned a previously functioning catalyst. (Solution color
changes following Hg reaction with either monometallic pre-
catalysts or nanoclusters are well established; in some cases,
reactions with the precatalyst ruins the applicability of the Hg
test. The interested reader is directed to the results and literature
summarized in Table A elsewhere.11)
Figure 6. Deuterium incorporation into cyclohexane taken at 44%
benzene conversion from the reaction of RhCl3‚3H2O with 657 Torr
of D2 in the presence of dichloroethane/Aliquat 336/H2O/trioctylamine/
benzene at 31 ( 0.5 °C. The MS shown are at 17 eV rather than 70
eV, since the former conditions avoid D fragmentation from the R+•
produced by electron impact. The m/z ) 89 corresponds, for example,
to C6H7D5.
again showed multiple deuterium incorporation into cyclohexane
(d1-d11) but no deuterium incorporation into unreacted benzene,
again quite consistent with the literature system’s results.9c
Noteworthy here is the fact that D-incorporation into benzene
is known in the benzene hydrogenation literature,21 indicating
that a different mechanism or active site may be responsible
for this process.
A comparison of our incorporation data to that reported in
the literature system,9c Table 1, entries 1-3, shows that our
data exhibit a quite similar but not identical pattern, at least at
a 17 eV ionizing voltage. This caused us to wonder what the
effects of the ionizing voltage might be, so we repeated our
experiments at 70 eV. Entries 4 and 5 in Table 1 provide the
second, and most telling, D-incorporation experiment in terms
of identifying the true catalyst. The results with either the
[RhCl4]- precatalyst (entry 4) or isolated Rh(0) nanoclusters
(entry 5) are identical within experimental error over the full
range of observed d1-d11 deuterium incorporation. These
results offer excellent, confirming, “fingerprint” mass spectral
evidence that (i) the Rh(0) nanoclusters are the active catalysts,
and (ii) that the literature catalyst formed from RhCl4-, and
the authentic Rh(0) nanoclusters characterized herein, are one
and the same.
Second, excess Hg completely poisoned a fully active
catalytic solution of isolated Rh(0) nanoclusters. A fresh
reaction solution beginning with isolated Rh(0) nanocluster
catalyst was allowed to go to 46% conversion, after which time
ca. 310 equiV of Hg was added and allowed to stir for 1 h. The
color of the solution changed from dark red/black to clear and
colorless (such color changes of transition M(0) nanocluster
A bit more comment is required here about the different
D-incorporation observed by mass spectroscopy at the 17 vs
the 70 eV ionizing voltages. It helps to realize first that the
literature mass spectroscopy work9c was done in the labs of
Professor H. Schwarz and thus under the guidance of a mass
spectroscopy expert.9c His choice of a low, 17 eV ionizing
voltage was obviously deliberate and clever, chosen to avoid
“hydrogen elimination from the molecular ion” (see Table 1,
(22) (a) Budzikiewicz, H.; Djerassi, C.; Williams, D. H. Mass Spec-
trometry of Organic Compounds; Holden-Day, Inc.: San Francisco, CA,
1967; pp 49-93. (b) McLafferty, F. W. Interpretation of Mass Spectra;
3rd ed.; University Science Books: Mill Valley, CA, 1980; pp 138-189.
(c) Biemann, K. Mass Spectrometry-Organic Chemical Applications;
McGraw-Hill: New York, 1962. (d) We also observed reverse fractionation
on the GLC column in our GC-MS studies, the well-established
phenomenon22e-g in which the more deuterated species elutes from the mass
spectrum before the least deuterated species (i.e., C6D11H1 comes out before
C6H12). (e) Foley, P.; DiCosimo, R.; Whitesides, G. M. J. Am. Chem. Soc.
1980, 102, 6713. (f) Possanzini, M.; Pela, A.; Liberti, A.; Cartoni, G. P. J.
Chromatog. 1968, 38. 492. (g) See, for example, pp 216 and 217
elsewhere.22c
(21) We surmise that the observation of D-incorporation into the
unreacted benzene seen in the following literature is indicating another
mechanism, and possibly a different active site, one perhaps able to undergo
a C-H oxidative-addition mechanism (i.e., to form Rh(D)2(H)(Ph) surface
intermediates) which can then lead to D-incorporation into the unreacted
benzene. (a) Moyes, R. B.; Wells, P. B. AdV. Catal. 1973, 23, 121. (b) Van
Hardeveld, R.; Hartog, F. AdV. Catal. 1972, 22, 75. (c) Siegel, S. AdV.
Catal. 1966, 16, 123. (d) Rooney, J. J.; Webb, G. J. Catal. 1964, 3, 488.
(e) Bond, G. C. Catalysis by Metals; Academic Press: New York, 1962;
pp 316-319.
(23) (a) Georgiades, G. C.; Sermon, P. A. J. Chem. Soc., Chem.
Commun. 1985, 975. (b) Whitesides, G. M.; Hackett, M.; Brainard, R. L.;
Lavalleye, J.-P. P. M.; Sowinski, A. F.; Izumi, A. N.; Moore, S. S.; Brown,
D. W.; Staudt, E. M. Organometallics 1985, 4, 1819. (c) Anton, D. R.;
Crabtree, R. H. Organometallics 1983, 2, 855.

Rh(0) Nanoclusters in Catalysis
J. Am. Chem. Soc., Vol. 120, No. 23, 1998 5661
Table 1. Comparison of Deuterium Incorporation into Cyclohexane from the Reaction of RhCl3‚3H2O with D2 in the Presence of
Dichloroethane/Aliquat 336/H2O/Trioctylamine/Benzenea
% deuterium incorporation into cyclohexane
system
m/z
d1
85
d2
86
d3
87
d4
88
d5
89
d6
90
d7
91
d8
92
d9
93
d10
94
d11
95
d12
96
lit. study9cb
1
1
8
5
9
4.5
15
17
15
17
17
38
25
24
22
30
30
20
23
17
17
9
10
10
11
11
6
8
7
9
9
1
7
6
6
6
3
1.5
present study (precatalyst)c 17 eV
present study [Rh(0)]d 17 eV
present study (precatalyst)c 70 eV
present study [Rh(0)]d 70 eV
1
1
2
3
5
4
1
3
2
1
1
a The values have been corrected for the natural 13C abundance in each M + 1 peak. b Data taken at 50% benzene conversion using RhCl3‚3H2O/
dichloroethane/H2O/trioctylamine/benzene/Aliquat 336 (Method A); note that a very low 17 eV ionizing voltage was deliberately used in this
literature study so that “hydrogen elimination from the molecular ions does not take place”.9c c Data taken at 44% benzene conversion using
Method A. d Data taken at 63% benzene conversion using isolated Rh(0) nanoclusters/dichloroethane/H2O/benzene.
(19.0 mg, 9.47 × 10-2 mmol) and 3.0 equiV of Hg (27.9 mg,
13.9 × 10-2 mmol), respectively, were added to the reaction
solutions. After stirring for 1 h, the solutions had not changed
color. After reconnection to the hydrogenation line, the
reactions proceeded until ca. 80% and 65% conversion, respec-
tively, then stopped (Figures H and I, Supporting Information).
Rather clearly, it seems that excess Hg(0) is needed to be sure
that the mercury physically makes full contact with all the active
Rh(0) catalyst.
These Hg poisoning experimentssthe complete stoppage to
within e(10% of the reaction by sufficient Hg to contact the
surface of the catalystsprovide confirming evidence that the
true catalyst in this system is the range of Rh(0) nanoclusters
present, a statement that is true quantitatively to within the
Figure 7. Hg poisoning of the catalyst in the reaction of RhCl3‚3H2O
with 686 Torr of H2 in the presence of THF/Aliquat 336/H2O/
trioctylamine/benzene at 30.5 ( 0.5 °C. After the reaction had gone to
88% conversion, 321 equiv of Hg was added and allowed to stir for 1
h. No further loss of benzene ((10%) was observed even after 5 h.
e(10% error limits of the complete poisoning observed.24
These experiments also warn against using small amounts of
Hg in attempts to do definitive Hg poisoning studies.
Can Rh(0) Nanoclusters as the True Catalysts Explain
All the Available Data? As noted in the Introduction and
Figure 1, one final key to the approach employed,11 and to
reliable mechanistic studies in general, is to ask the question
shown in part 4 of Figure 1, “Can the proposed mechanistic
hypothesis, in this case that nanoclusters are the true catalysts,
explain all the available data?” We reexamined the literature
paper9c with this question in mind, and specifically with our
list of the 35 observations or other pieces of data (citations of
previous literature, etc.) from the literature paper in mind (i.e.,
the list that we had collected and which is alluded to in footnote
a of Scheme 1). Actually, all but a couple of the key
observations have been dealt with already in the paper.
However, two interesting pieces of data from the literature
study merit discussion. First is the report9c that styrene
pretreatment leads to a 1.4 times faster catalyst. This observa-
tion, one unexplained in the literature report,9c has many
plausible explanations in the case of a Rh(0) nanocluster. A
couple of possible nanocluster-based explanations are: an effect
of styrene on the size or distribution of nanoclusters; an effect
of some polystyrene produced under the reaction conditions on
the size, stability, or activity of the Rh(0) nanoclusters (since
polymer stabilized nanoclusters are very well known; see the
references listed elsewhere3,4); or perhaps the effect of styrene
Figure 8. Hg poisoning of isolated Rh(0) nanoclusters with 687 Torr
of H2 in the presence of THF/Aliquat 336/H2O/trioctylamine/benzene
at 30 ( 0.5 °C. After the reaction had gone to 46% conversion, 312
equiv of Hg was added and allowed to stir for 1 h. No further loss of
benzene ((10%) was observed even after 2 h.
catalysts upon contact and reaction with Hg have precedent in
our earlier work4). No further catalytic activity (e(10%) was
seen after 2 h, Figure 8.
Third, a Hg poisoning (control) experiment was performed
to see the effect that a reduced amount of added Hg had on
catalytic activity of solutions starting with both RhCl3‚3H2O
and isolated Rh(0) nanoclusters. The reactions were run just
as in the experiments above except for in the reaction beginning
with isolated Rh(0) nanoclusters, where no Aliquat 336 or
trioctylamine was added to the isolated black solid. The reactions
were allowed to run to ca. 50% conversion, but now only 2.0
(24) (a) In principle, we would have liked to have repeated the literature
system’s use of a homogeneous catalyst poison (in this case, dibenzocy-
clooctatetraene, or dct) to support or refute their report that catalytic activity
is stopped when dct is added to the reaction solutions. However, upon further
reflection this was not attempted, since (i) the synthesis is known to be
unreliable and proceeds with unreasonably low yields,18b (ii) dct is no longer
commercially available, and, most importantly (iii) in light of the strong
evidence uncovered for Rh(0) nanocluster catalysis, the testing of a
“homogeneous” catalyst poison such as dct on Rh(0) nanoclusters is no
longer sensible or cost-effective. (b) Crabtree, R. H. Private communication.

5662 J. Am. Chem. Soc., Vol. 120, No. 23, 1998
Weddle et al.
coordinating to an adjacent, surface Rh(0) (something only
possible in a multimetallic Rh(0) catalyst). On the other hand,
it is much harder to see any precedented explanation for the
styrene or other olefin and acetylene effects reported9c in the
Summary and Conclusions
The major findings of the present study can be summarized
as follows.
-
(1) First, an up-front reanalysis of the data published for a
prototype arene hydrogenation literature system,9c and in light
of recent advances in nanocluster chemistry and in distinguishing
“heterogeneous from homogeneous catalysts”,3,4,5,11 allows the
recognition of several key experimental observations indicative
of nanocluster formation and catalysis [specifically, the literature
reports of (i) induction periods, (ii) disappearance of the
RhCl4 catalysis hypothesis.
A second observation is, in the case of naphthalene hydro-
genation done in a NMR pressure tube, the literature study
which reported9c three triplets and one unresolved multiplet that
are ascribed to Rh-H intermediates; that is, presumably a
Rh2-H intermediate in naphthalene hydrogenation [-22.35
ppm (JRh-H ) -7.6 Hz), -23.12 ppm (JRh-H ) 9.1 Hz), -24.29
ppm (JRh-H ) -7.6 Hz)] and a rather different signal at +22.78
ppm (unresolved multiplet, JRh-H ) -7.6 Hz, one that was
presented without further discussion in the literature study).9c
Unfortunately, no experimental details were provided for these
NMR experiments, so that they are not rigorously repeatable.9c
The well-resolved triplets with detectable coupling in the range
expected for RhI-H (i.e., smaller than 15-30 Hz)25 does indeed
seem to require the presence of discrete, presumably Rh2-H
hydrides and possibly a homogeneous hydrogenation system for
naphthalene (only), assuming that these hydride signals can be
connected to kinetically competent catalysis in a future study.
However, note that any eventual demonstration of such kinetic
data and thus homogeneous catalysis for this naphthalene system
does not apply logically to benzene hydrogenation. The
differences between benzene, naphthalene, and anthracene
hydrogenations were pointed out by Collman back in 1987.9a
Moreover, the unresolved multiplet at +22.78 ppm may be a
nanocluster-bound Rhn-H species, one with a Knight-shift26
contribution to its unusual chemical shift, an observation
therefore meriting further attention. Bradley’s excellent review
is recommended for anyone interested in the issues surrounding
the detection by NMR of nanocluster surface-bound intermedi-
ates,26 an important area for future nanocluster studies.
-
induction periods following pretreatment of the RhCl4 pre-
catalyst with H2, (iii) the observation of a dark, black color as
the reaction proceeds, and (iv) the formation of both soluble
and insoluble black (i.e., Rh(0)) metal)sall prima facie evidence
for the formation of Rh(0) nanoclusters in solution]. As such,
the present work may serve as a good example of what can
now be identified as “tell-tale” observables to look for in up-
front analyses of other systems in the literature, catalytic systems
presently believed to be mononuclear homogeneous catalysts
but which may, instead, be polymetallic, soluble nanoparticle
catalysts.28
(2) Second, the use of the most recent, more powerful four-
step methodology to distinguish “homogeneous” from “hetero-
geneous” catalysis worked effectively and efficiently in the
present case of a benzene hydrogenation catalyst. This is only
the second time this methodology has been applied.
(3) Third, very strong, if not compelling, evidence has been
gathered indicating that the true benzene hydrogenation catalyst
is a distribution of Rh(0) nanoclusterssa statement that is
quantitatively true to within the 5-15% error limits of the
kinetic fits or the complete deactivation within experimental
error ((10%) of the catalyst by added Hg(0). The key evidence
is (i) the direct TEM demonstration that the black reaction
solution consists of (40% polydisperse Rh(0) nanoclusters; (ii)
the sigmoidal-shape kinetic curves and their induction periods;
and especially (iii) the quantitative fit of the observed kinetic
curves to the nucleation plus autocatalytic surface-growth (A
f B, A + B f 2B) kinetic signature and mechanism first
elucidated elsewhere,5 kinetic evidence which requires that
RhCl4- (“A”) is not kinetically active, but that the product “B”
is the true catalyst (where B is the TEM-identified Rh(0)
nanoclusters plus insoluble agglomerated Rh(0) nanoclusters and
A closer look at Halpern’s kinetic and mechanistic studies
of, for example, anthracene hydrogenation merits mention here.27
Halpern has provided good evidence for homogeneous an-
thracene hydrogenation by, for example, a cationic27a Rh-H
or27b a Ru-Hx complex. Note, however, that the stated rate of
benzene hydrogenation is “negligible” (see p 841, right-hand
column27a) under mild conditions (e60 °C, 1 atm H2) where
the anthracene hydrogenation is, however, facile. Halpern points
out that the ease of (and thus greater rates of) reduction
anthracene > naphthalene . benzene reduction parallel the
Hu¨ckel approximation â localization energies of 1.16â, 1.21â,
and 1.53â, respectively (i.e., in the anthracene case for its
localization into the “diene + naphthalene”, higher energy
electronic configuration). In short, one needs to be cautious in
transition metal chemistry in extrapolating mechanisms from
one system to even seemingly closely analogous systems, a point
that is also made by Halpern in an earlier, classic study of metal
hydride reactivity.27 In short, we see no observable in the
literature study that is not bestsif not onlysexplained by Rh-
(0) nanoclusters as the true catalysts for benzene hydrogenation.
(28) (a) Also of some interest, as the literature study notes, is the “close
similarity between the rhodium- and (Muetterties’) cobalt-catalyzed arene
hydrogenation”.9c Although the task remains of applying the methods and
techniques11 used herein to the Co system (which we do not have plans to
do), the expectation based on the available evidence is that it, too, may
well be a nanocluster-based system, as there are certainly numerous
examples of Co nanoclusters. However, an important difference between
Co and Rh is that Co is not reducible, thermodynamically speaking, to Co-
(0) atoms, at least not by only 1 atm of H2.28b-i In this regard, it is interesting
that the literature9c reports that the Muetterties (allyl)Co[POMe3]3 precatalyst
system is “difficult to reproduce, ...extremely sensitive to the quality of the
reducing agent as well as the purity of the trimethyl phosphite” (the latter
an effect of OdPMe3 impurities?);14 it is also reported that one has to raise
the “temperatures to 60-80 °C” and the “hydrogen pressure to 3-6 atm”
to get the reported yields of benzene or toluene reduction. A cobalt
nanocluster catalyst needs to be carefully looked for in this system as well
and by the methods employed herein.11 (b) Osuna, J.; de Caro, D.; Amiens,
C.; Chaudret, B.; Snoeck, E.; Respaud, M.; Broto, J.-M.; Fert, A. J. Phys.
Chem. 1996, 35, 14571. (c) Hayashi, T.; Hirono, S.; Tomita, M.; Umemura,
S. Nature 1996, 381, 772. (d) Becker, J. A.; Scha¨fer, R.; Festag, J. R.;
Wendorff, J. H.; Hensel, F.; Pebler, J.; Quaiser, S. A.; Helbig, W.; Reetz,
M. T. Surf. ReV. Lett. 1996, 3, 1121. (e) Billas, I. M. L.; Chaˆtelain, A.; de
Heer, W. A. Science 1994, 265, 1682. (f) Gong, W.; Li, H.; Zhao, A.; Chen,
J. J. Appl. Phys. 1991, 69, 5119. (g) Chidress, J. R.; Chien, C. L. J. Appl.
Phys. 1991, 70, 5885. (h) Sapieszko, R. S.; Matijevic, E. Corrosion 1980,
36, 522. (i) Hess, P. H.; Parker, P. H., Jr. J. Appl. Polym. Sci. 1966, 10,
1913.
(25) (a) Mason, J. Multinuclear NMR; Plenium Press: New York, 1987;
p 552. The reported 1J (103Rh1H) of Rh(I) (RhHL4) is between 0 and 8 Hz.
(b) Harris, R. K.; Mann, B. E. NMR and the Periodic Table; Academic
1
Press: New York, 1978; pp 248, 249. These authors also note that the J
(
103Rh1H) for Rh(I) (RhHL4) are much less than those observed for Rh(III)
hydrides, the latter usually falling in the range of 15-30 Hz.
(26) Bradley, J. In Colloids and Clusters: From Theory to Applications;
Schmid, G., Ed.; VCH Publishers: New York, 1994; see pp 515-522.
(27) (a) Landis, C. R.; Halpern, J. Organometallics 1983, 2, 840. (b)
Wilczynski, R.; Fordyce, W. A.; Halpern, J. J. Am. Chem. Soc. 1983, 105,
2066. (c) Nappa, M. J.; Santi, R.; Halpern, J. Organometallics, 1985, 4,
34.

Rh(0) Nanoclusters in Catalysis
J. Am. Chem. Soc., Vol. 120, No. 23, 1998 5663
bulk metal). Confirming evidence includes (iv) the “fingerprint”
D-incorporation that is the same for either the isolated Rh(0)
nanoclusters or when beginning with the RhCl4 precatalyst;
work toward this goal is continuing and will be reported in due
course.
-
Experimental Section
(v) the lack of D-incorporation into the unreacted benzene,
consistent with what the literature system also reported; (vi)
the demonstration that the well-known heterogeneous catalyst
poison23 elemental mercury shuts down completely the catalytic
activity of previously active benzene hydrogenation catalyst
solutions; and (vii) the similar kinetic curves, curve-fits (i.e.,
and k1 and k2 rate constants), and similar fingerprint D-
incorporation patterns seen in the data taken from literature
system.9c One other, very important piece of data is (viii) that
Rh(0) metal is a well-established arene hydrogenation catalyst,1,2
while monometallic RhI/III will not hydrogenate benzene under
conditions where it reduces homogeneously the more easily
reduced anthracene.27c We have also considered (ix) whether
the Rh(0) nanocluster hypothesis can explain all of the available
data, and believe that it can, although a few issues remain which
will require their own, independent study (the effect of styrene,
the Rh2-H intermediates seen and whether they are catalysts,
and the possibility of a homogeneous Rh-H naphthalene
hydrogenation system). To our knowledge, no other hypothesis
for the nature of the catalyst is presently available which can
also explain all the available data.26
Materials. Rhodium(III) chloride hydrate (obtained from Strem
Chemicals), trioctylamine (98%), benzene (anhydrous, 99.8%), cy-
clooctane (99+%), dichloroethane (99.8%, HPLC grade), and tetrahy-
drofuran (anhydrous, 99.8%, inhibitor free) were obtained from Aldrich
and opened and then stored in a Vacuum Atmospheres drybox.
Alumina oxide (Aldrich, activated, acidic, Brockmann I), Aliquat 336
(i.e., [(C8H17)3NCH3]+Cl-, Aldrich), mercury (D. F. Goldsmith Chemi-
cal & Metal Corp., elemental grade, triply distilled), hydrogen (General
Air, 99.5%), and dichloromethane (Fisher Scientific, HPLC grade) were
used as received.
Analytical Procedures. Unless otherwise reported, all reaction
solutions were prepared under oxygen- and moisture-free conditions
using a Vacuum Atmospheres nitrogen atmosphere drybox. O2 levels
were maintained at less than 3 ppm as continuously monitored by a
Vacuum Atmospheres O2-level monitor.
Gas-liquid chromatography (GLC) was performed using a Hewlett-
Packard 5890 series II GLC with a FID detector equipped with a 30 m
(0.25 mm i.d.) Econo-cap Carbowax column (Alltech) and coupled to
a Hewlett-Packard 3395 integrator. Parameters were as follows: initial
temperature, 35 °C; initial time, 4.0 min; ramp, 15 °C/min; final
temperature, 200 °C; final time, 5 min; injector port temperature, 180
°C; detector temperature, 200 °C; injection volume, 2 µL.
Gas chromatography mass spectrometry (GC-MS) was preformed
using a Hewlett-Packard 5890 series II GC with a VG AutoSpec
equipped with a 30m DB-5 column (J&W Scientific). The ionizing
voltages of 17 eV and 70 eV were both employed for reasons discussed
in the main text. The GC parameters were as follows: initial
temperature, 0 °C (via dry ice); initial time, 6 min; solvent delay, 4
min; first ramp, 5 °C/min until 20 °C; second ramp, 25 °C/min until
100 °C; injector port temperature, 180 °C; detector temperature, 200
°C; injection volume, 1 mL. Mass marker calibration of the GC-MS
was performed using heptacosafluorotributylamine. The percent D-
incorporation was calculated according to the reaction C6H6 ) nD2 f
aC6H5D1 + bC6H5D2 + cC6H5D3 (and so on); that is, the sum of a +
b + c (etc.) relative to the total original concentration of C6H6 gives
the fraction (and thus percent) of D-incorporation.
Transmission electron microscopy (TEM) was performed on a JEOL
2000 EX-II operating at an accelerating voltage of 100 keV. Samples
were examined at magnifications between 100 and 400K, and in at
least three different places on the sample grid to ensure that the images
were representative of the sample as a whole. Samples were prepared
using type A (300 mesh) Formvar and carbon coated copper grids (Ted
Pella, supplier). Grids were suspended in chloroform for about 30 s
immediately prior to use to remove the Formvar coating and to expose
a fresh carbon surface. One drop of the solution (in DMSO) was placed
on the carbon-coated grid using a gas-tight syringe and allowed to air
dry. Particle size distributions were determined once the original
negative had been digitally scanned into Adobe Photoshop and
expanded to > 20 in. × 25 in. for more accurate resolution and
measurement.
A control experiment was done using a precatalyst solution in THF
(Method B) to show that deposition of the ion-pair precatalyst on a
TEM grid, followed by its subsequent TEM examination under the
conditions used to visualize the Rh(0) nanoclusters, did not yield any
(TEM-beam induced) Rh(0) nanoclusters. This control, which was
repeated, did not yield a clear image (as expected since no nanoclusters
are present) but is definitive in showing that the TEM beam does not
produce images characteristic of the nanoclusters seen in, for example,
Figure 2.
Hydrogenation Apparatus and “Standard Conditions”. The
hydrogenation apparatus used herein was modeled after a standard
catalytic hydrogenation apparatus for hydrogenations at atmospheric
pressure.30 This apparatus (Figure A, Supporting Information) consists
of a 25-mL side-armed Schlenk flask, a 500-mL side-armed round-
The present results have broader implications as well;
specifically they (4) verify the prediction that the use of the
pseudo-elementary reporter-reaction method5b is a more gener-
ally useful, powerful new method for following nanocluster
formation; (5) provide evidence fortifying the expectation5a that
the new mechanism uncovered recently of A f B nucleation
followed by A + B f 2B autocatalytic surface growth might
well be a more general, new mechanistic paradigm for transition
metal nanocluster formation under hydrogen; and (6) call into
question all previous claims of benzene hydrogenationsbut not
anthracene or naphthalene arene hydrogenationsby monome-
tallic precatalysts. The studies presented herein also (7) re-
emphasize that, prior to any claim of a homogeneous catalyst
in a reaction (such as arene hydrogenation) where a facile
heterogeneous M(0) catalyst is well established, one must first
rule out catalysis by even trace amounts of possibly highly active
nanocluster catalysts (e.g., by using the methods utilized herein
and any other applicable method), and (8) call for a reinterpreta-
tion of all of the other papers in the literature29 which used what
was previously believed to be a RhCl4- ion-paired catalyst under
hydrogen for catalytic reductions. This work also (9) offers a
reminder that electron impact mass spectroscopy studies of
D-incorporation need to be done at both high and low ionizing
voltages.
Overall, the studies presented herein provide a definitive
answer, at least for the specific Rh system studied, to the 34-
year-old question, one controversial for 17 years, of “is benzene
hydrogenation homogeneous or heterogeneous?”. The present
study is, however, just one of the early steps toward our broader
goal of developing nanocluster “soluble heterogeneous cata-
lysts”, including selective arene hydrogenation catalysts. Further
(29) (a) Blum, J.; Amer, I.; Zoran, A.; Sasson, Y. Tetrahedron Lett. 1983,
24, 4139 (hydrogenation of a variety of aromatics). (b) Amer, I.; Amer,
H.; Blum, J. J. Mol. Catal. 1986, 34, 221 (hydrogenation of naphthalene).
(c) Azram, J.; Buchman, O.; Amer, I.; Blum, J. J. Mol. Catal. 1986, 34,
229 (selective hydrogenation of R, â-unsaturated ketones and esters). (d)
Amer, J.; Bravdo, T.; Blum, J.; Vollhardt, K. P. C. Tetrahedron Lett. 1987,
28, 1321 (selective hydrogenation of unsaturated nitro compounds). (e)
Amer, I.; Amer, H.; Ascher, R.; Blum, J.; Vollhardt, K. P. C. J. Mol. Catal.
1987, 39, 185 (selective hydrogenation of polycyclic compounds). (f) Blum,
J. Russ. Chem. Bull. 1993, 42, 1619 (hydrogenation of aromatics).
(30) Augustine, R. L. Catalytic Hydrogenation; Marcel Dekker: New
York, 1965.

5664 J. Am. Chem. Soc., Vol. 120, No. 23, 1998
Weddle et al.
bottomed, and a 100-mL round-bottomed flask (the latter two were
used as H2 reservoirs) connected to a standard Schlenk line equipped
with a Hg-manometer.
dissolved in 0.5 mL (6.35 mmol) of dichloroethane, 50 µL (0.11 mmol)
of trioctylamine, 89 µL (1.0 mmol) of benzene, and 64 µL (0.47 mmol)
of cyclooctane. In a separate dram vial RhCl3‚3H2O (10.4 mg, 5.0 ×
10-2 mmol) was weighed and the colorless dichloroethane solution was
added using a disposable polyethylene pipet. While still in a drybox,
0.5 mL (28 mmol) of H2O was added using a 1-mL gastight syringe
and the mixture was stirred until the RhCl3‚3H2O had dissolved. The
percent benzene conversion vs time is summarized in Figure F(a),
Supporting Information. As the reaction proceeded, the solution
changed from a clear orange-red to an opaque, deep red/black. After
4 h the flask was brought back into the drybox. The solution was
evacuated overnight at room temperature to yield a black solid. A TEM
sample was prepared by dissolving ca. 2 mg of the black solid in ca.
3.5 mL of DMSO at room temperature to yield a clear brown solution.
A TEM image of this sample is shown in Figure C, Supporting
Information.
Unless otherwise reported, all hydrogenation experiments were
performed using the following “standard conditions”. Two general
reaction conditions taken from the previous literature9c were used: a
biphasic system (H2O/dichloroethane), Method A,9c and a monophasic
system (THF), Method B.9c Amounts of reagents identical with those
used in the literature9c were employed for both Method A and Method
B (for details see the sections which follow). The reaction solutions
were mixed in the drybox and placed into a 25-mL side-armed flask
1
containing a 5/16 in. × /2 in. Teflon-coated magnetic stir bar. The flask
was capped and sealed, taken out of the drybox, and connected to the
hydrogenation apparatus. The reaction solution was frozen at ca. -40
°C using a CO2/ethanol bath. Once frozen, the entire apparatus was
placed under ca. 10-3 Torr vacuum for ca. 2-3 min. The low-
temperature bath was removed and replaced by a silicone oil bath
thermostated at 31 ( 1 °C causing the reaction solution to thaw. The
entire apparatus was placed under 687 ( 7 Torr of hydrogen pressure.
Stirring of the reaction solution at 770 ( 30 rpm was started and the
time was set equal to zero at this point. [The literature reports stirring
at 400 rpm9c so control experiments were done at 350 ( 75 and 1100
( 100 rpm, reported in a section below, to test for possible H2 mass-
transfer limitations. The results show that there is no effect beyond
experimental error in the rate of arene hydrogenation due to the stirring
rate difference between 375, 770, and 1100 rpm (Figure G, Supporting
Information).]
Formation, Isolation, and Reuse of Rh(0) Nanoclusters Plus
Aliquat 336 in Dichloroethane. A “standard conditions”, Method A
hydrogenation was performed using Aliquat 336 (20.9 mg, 5.79 × 10-2
mmol), 0.5 mL (6.35 mmol) of dichloroethane, 50 µL (0.11 mmol) of
trioctylamine, 89 µL (1.0 mmol) of benzene, 64 µL (0.47 mmol) of
cyclooctane, RhCl3‚3H2O (10.3 mg, 4.92 × 10-2 mmol), and 0.5 mL
(28 mmol) of H2O. After 4 h, the reaction flask was disconnected
from the hydrogenation line and brought back into the drybox. The
reaction mixture was evacuated to dryness under vacuum overnight at
room temperature yielding a black solid. The black solid was
redissolved in 0.5 mL of dichloroethane, 50 µL of trioctylamine, 89
µL of benzene, 64 µL of cyclooctane, Aliquat 336 (21.7 mg, 5.37 ×
10-2 mmol), and 0.5 mL of H2O, and a second hydrogenation run was
begun with 682 Torr of hydrogen. Hydrogenation of benzene proceeded
without a detectable induction period and was allowed to continue for
4 h. The percent benzene conversion vs time is summarized in Figure
F(b), Supporting Information. A TEM of the reaction mixture is shown
in Figure D, Supporting Information.
Note also, and as discussed in the text, we deliberately did not use
the step used previously of filtering the catalyst through a fluoropore
filter “if some precipitate was formed”,9c since this step both depends
on the irreproducible identification of a black precipitate in a black
solution (i.e., it was neither used uniformly before, nor can it be
reproducibly applied by us or others) and since it would then remove
the catalytic contribution of agglomerated plus bulk Rh(0) that we have
shown is present in unfiltered solutions.
Hydrogenation of Benzene in THF Using RhCl3‚3H2O as a
Precatalyst and TEM Sample Preparation. Method B. This
reaction was carried out under the “standard conditions” described above
and is based on Method B described elsewhere.9c Aliquat 336 (21.8
mg, 5.39 × 10-2 mmol) was weighed into a 1-dram vial and dissolved
in 1.0 mL (12.3 mmol) of THF, 100 µL (0.23 mmol) of trioctylamine,
89 µL (1.0 mmol) of benzene and 64 µL (0.47 mmol) of cyclooctane.
RhCl3‚3H2O (10.3 mg, 4.92 × 10-2 mmol) was then dissolved in the
above THF solution and 50 µL (2.8 mmol) of H2O was added. The
resulting monophasic solution was placed into the reaction flask, sealed,
removed from the drybox, and connected to the hydrogenation apparatus
as detailed above. After 4 h, the flask was brought back into the drybox.
The solution was evacuated overnight at room temperature to yield a
black solid. A TEM sample was prepared by dissolving ca. 2 mg of
the black solid in ca. 3.5 mL of DMSO at room temperature to yield
a clear brown solution. A TEM image of the reaction mixture is shown
in Figure E, Supporting Information.
To monitor the loss of benzene, aliquots of the reaction solution
(1-2 drops, ca. 50 µL) were withdrawn via the septum-capped side
arm using a 9 in. long, 12-gauge metal needle attached to a 5-mL glass
syringe. For GLC analysis, 5 µL of this sample was added to 1 mL of
fresh solvent (either dichloroethane or THF) and (as described in the
literature, vide infra) filtered through ca. 200 mg (1.96 mmol) acidic
alumina (i.e., to remove water and filter out Rh(0) nanoclusters that
were present). This filtration step, which is based on the literature
procedure,9c proved necessary, as without it irreproducible GLC results
were obtained, probably due to the inhomogeneous (biphasic) nature
of the product mixture.
One ambiguity regarding the literature system9c is whether Aliquat
336 and trioctylamine were readded when the activity of the reported,
isolated black solid catalyst was studied. Because of this, when we
performed arene hydrogenations with the catalyst isolated under the
literature conditions, we did so both with and without added Aliquat
336 and trioctylamine. Fortunately, we find that the presence of this
additional aliquot of stabilizers has no effect on the activity of the
isolated catalyst; hence, no ambiguity is raised between the conditions
used in the literature and our need to repeat those literature conditions
as exactly as possible.
Formation, Isolation, and Reuse of Rh(0) Nanoclusters Plus
Aliquat 336 in THF. A “standard conditions”, Method B hydrogena-
tion was performed using Aliquat 336 (21.5 mg, 5.31 × 10-2 mmol),
1.0 mL (12.3 mmol) of THF, 100 µL (0.23 mmol) of trioctylamine, 89
µL (1.0 mmol) of benzene, 64 µL (0.47 mmol) of cyclooctane, RhCl3‚
3H2O (10.0 mg, 4.78 × 10-2 mmol), and 50 µL (2.8 mmol) of H2O.
After 4 h, the reaction flask was disconnected from the hydrogenation
line and brought back into the drybox. The reaction mixture was
evacuated to dryness under vacuum overnight at room temperature
yielding a black solid. The black solid was redissolved in 1.0 mL of
THF, 100 µL of trioctylamine, 89 µL of benzene, 64 µL of cyclooctane,
Aliquat 336 (22.0 mg, 5.44 × 10-2 mmol), and 50 µL of H2O, and
another hydrogenation run was performed under 685 Torr of hydrogen.
The reaction proceeded without a detectable induction period and was
allowed to run 4 h. A TEM image of the reaction mixture is shown in
Figure 2.
Hydrogenation of Benzene in Dichloroethane Using RhCl3‚3H2O
as a Precatalyst and TEM Sample Preparation. Method A. This
reaction was carried out under the “standard conditions” described above
and is based on Method A described elsewhere.9c In a drybox Aliquat
336 (23.4 mg, 5.8 × 10-2 mmol) was weighed into a 1-dram vial and
(31) (a) See Figure 13 and pp 114 and 115 of Che, M.; Bennett, C. O.
AdV. Catal. 1989, 36, 55. (b) Milone, C.; Neri, G.; Donato, A.; Musolino,
M. G.; Mercadante, L. J. Catal. 1996, 159, 253.
(32) The literature9c did not provide this number. The total turnover
number calculated is the maximum predicted. This number is calculated
assuming 100% conversion of benzene to cyclohexane and from a single
batch of material under standard conditions (see Experimental Section) in
THF. The total turnover number was corrected for the number of active
available Rh(0) atoms, assuming that, at a maximum, 31% of the Rh(0)
atoms were available and active for the average Rh(0) nanocluster size of
36 Å ( 13 [2057 Rh atoms].19
Filtration Experiments on Isolated and Then Reused Rh(0)
Nanoclusters in THF. These are detailed in the Supporting Informa-
tion.

Rh(0) Nanoclusters in Catalysis
J. Am. Chem. Soc., Vol. 120, No. 23, 1998 5665
Curve-Fitting of the Literature Data for Benzene Concentration
Ws Time. Curve-fitting of the benzene concentration vs time data given
in Figure 4 herein (of the data taken from Figure 1 of the literature
paper9c) and also of the data in Figures 3a and F(a) of the Supporting
Information was performed using the program and kinetic equations
described elsewhere, basically, the analytical integrated equations
corresponding to the kinetic steps of nucleation (A f B; rate constant,
k1), and then autocatalytic surface growth (A + B f 2B, rate constant,
k2).5 The resultant k2 values are not corrected for the “stoichiometry”
or “scaling” factors detailed elsewhere;5 that is, the values are
hydrogenation run was performed under 693 Torr of hydrogen. After
4 h the reaction flask was disconnected from the hydrogenation line
and taken back into the drybox. The solution was evaporated to dryness
overnight under vacuum at room temperature yielding a black solid.
The black solid was dissolved in 1.0 mL of THF, 89 µL of benzene,
64 µL of cyclooctane, and 50 µL of H2O. A second hydrogenation
run was begun under 645 Torr of D2. The reaction was allowed to run
until completion, 4 h. Every 30 min, samples were removed for GLC
analysis. A plot of loss of benzene vs time is shown in Figure 3b.
D Incorporation and H/D Exchange Studies under D2 Using
RhCl3‚3H2O as the Precatalyst in the Presence of Benzene-d0 and
Dichloroethane. A “standard conditions” hydrogenation was per-
formed using Aliquat 336 (20.5 mg, 5.07 × 10-2 mmol), 0.5 mL (6.35
mmol) of dichloroethane, 100 µL (0.23 mmol) of trioctylamine, 89
µL (1.0 mmol) of benzene, 64 µL (0.47 mmol) of cyclooctane,
RhCl3‚3H2O (9.6 mg, 4.59 × 10-2 mmol), and 1.0 mL (56 mmol) of
H2O. A hydrogenation run was begun under 686 Torr of D2. The
reaction was allowed to proceed until 44% conversion, approximately
90 min. Every 30 min, samples were removed for GLC analysis. At
90 min, a sample was removed for GC-MS analysis; specifically, 25
µL of the red/black solution was added to 0.5 mL (7.8 mmol) of
dichloromethane, and this sample was analyzed at ionizing potentials
of 17 and 70 eV. The data for deuterium incorporation into cyclo-
hexane are summarized in Table 1 and Figure 6. Reverse fractionation
is observed on the GLC column in the GLC-MS experiments.22d
k
2(curve-fit; uncorrected). Note also that, as in our earlier work,5 no possible
particle size effects were deconvoluted from the kinetic fits, an important
future research goal since the particle size effect (the “structure
sensitivity or insensitivity”) of benzene hydrogenation is an unsettled
issue.31
IR Spectroscopy on the Product Solutions Attempting to Show
-
That All the RhCl4 Precursor Has Been Converted to Rh(0)
Nanoclusters. This experiment is detailed in the Supporting Informa-
tion.
Probing for H2 Gas-to-Solution Mass-Transfer Limitations6c
under the 770 Rpm Stirring Rates Employed. These experiments
are detailed in the Supporting Information. These controls make it
clear that (i) there are no discernible differences between data obtained
at the 350, 770, and 1100 rpm stirring rates; and thus (ii) H2 mass-
transfer limitations are not an issue at the 770 stirring rate used in the
present system (although we have shown elsewhere that they can be
for very fast hydrogenation catalysts and at higher catalyst concen-
trations6c).
D Incorporation and H/D Exchange Studies under D2 Using
Isolated Rh(0) Nanoclusters and in the Presence of Benzene-d0 and
Dichloroethane. A “standard conditions” hydrogenation was per-
formed using Aliquat 336 (20.0 mg, 4.95 × 10-2 mmol), 0.5 mL (6.35
mmol) of dichloroethane, 100 µL (0.23 mmol) of trioctylamine, 89 µL
(1.0 mmol) of benzene, 64 µL (0.47 mmol) of cyclooctane, RhCl3‚3H2O
(10.0 mg, 4.78 × 10-2 mmol), and 1.0 mL (55.5 mmol) of H2O. A
hydrogenation was performed under 693 Torr of hydrogen. After 4 h,
the reaction flask was disconnected from the hydrogenation line and
taken back into the drybox. The solution was evaporated to dryness
overnight under vacuum at room temperature yielding a black solid.
The black solid was dissolved in 0.5 mL of dichloroethane, 89 µL of
benzene, 64 µL of cyclooctane, and 1 mL of H2O. A second
hydrogenation was begun under 690 Torr of D2. The reaction was
allowed to run to 63% conversion, approximately 2 h. Every 30 min
samples were removed for GLC analysis. At 120 min, a sample was
removed for GC-MS analysis; specifically 25 µL of the red/black
solution was added to 0.5 mL (7.8 mmol) of dichloromethane, and
this sample was analyzed at ionizing potentials of 17 and 70 eV. The
data for deuterium incorporation into cyclohexane are summarized in
Table 1. Reverse fractionation is observed on the GLC column in the
GLC-MS experiments.22d
Hydrogenation of Benzene under D2 in Dichloroethane Using
RhCl3‚3H2O as the Precatalyst. A “standard conditions” hydrogena-
tion was performed using Aliquat 336 (21.5 mg, 5.32 × 10-2 mmol),
0.5 mL (6.35 mmol) of dichloroethane, 100 µL (0.23 mmol) of
trioctylamine, 89 µL (1.0 mmol) of benzene, 64 µL (0.47 mmol) of
cyclooctane, RhCl3‚3H2O (9.8 mg, 4.68 × 10-2 mmol), and 1.0 mL
(56 mmol) of H2O. A hydrogenation run was begun under 657 Torr
of D2. The reaction was allowed to proceed until completion, 3 h.
Every 30 min, samples were removed for GLC analysis. A plot of
loss of benzene vs time is shown in Figure F(a), Supporting Information.
Hydrogenation of Benzene under D2 in Dichloroethane Using
Isolated Rh(0) Nanoclusters. A “standard conditions” hydrogenation
was performed using Aliquat 336 (21.6 mg, 5.34 × 10-2 mmol), 1.0
mL (12.7 mmol) of dichloroethane, 100 µL (0.23 mmol) of trioctyl-
amine, 89 µL (1.0 mmol) of benzene, 64 µL (0.47 mmol) of
cyclooctane, RhCl3‚3H2O (10.1 mg, 4.83 × 10-2 mmol), and 1.0 mL
(55.5 mmol) of H2O. A hydrogenation run was performed under 695
Torr of hydrogen. After 4 h, the reaction flask was disconnected from
the hydrogenation line and taken back into the drybox. The solution
was evaporated to dryness overnight under vacuum at room temperature
yielding a black solid. The black solid was dissolved in 0.5 mL (6.35
mmol) of dichloroethane, 89 µL of benzene, 64 µL of cyclooctane,
and 1.0 mL of H2O. A second hydrogenation run was begun under
662 Torr of D2. The reaction was allowed to run until completion, 4
h. Every 30 min, samples were removed for GLC analysis. A plot of
the percent benzene conversion vs time is shown as Figure F(b),
Supporting Information,
Hg Poisoning Tests Using RhCl3‚3H2O as the Precatalyst in the
Presence of Hg(0) and Benzene. A “standard conditions” hydrogena-
tion was performed using Aliquat 336 (21.7 mg, 5.37 × 10-2 mmol),
1.0 mL (12.3 mmol) of THF, 100 µL (0.23 mmol) of trioctylamine, 89
µL (1.0 mmol) of benzene, 64 µL (0.47 mmol) of cyclooctane,
RhCl3‚3H2O (10.0 mg, 4.78 × 10-2 mmol), and 50 µL (2.8 mmol) of
H2O. After 2 h (80% conversion) the reaction was stopped. The
reaction flask was disconnected from the hydrogenation line and taken
back into the drybox. The red/black solution was placed into a 2 dram
vial and stirred with Hg (3.0905 g, 15.4 mmol, 321 equiv) for 1 h. The
resulting clear yellow-orange solution was filtered through a Whatman
#1 paper to remove the excess elemental Hg and placed into a clean
reaction flask. The filter paper was then washed with ca. 0.5 mL (6.2
mmol) of THF and the washings were added to the reaction solution.
To compensate for any benzene lost during filtration, 40 µL (0.44 mmol)
of benzene was added to the reaction flask. A second hydrogenation
reaction was then started by placing the system under 689 Torr of
hydrogen. The previously fully active catalyst was completely
inactivated by Hg. Even 5 h 30 min after the Hg treatment, the catalyst
was still inactive. The percent benzene conversion vs time data is
shown in Figure 7.
Hydrogenation of Benzene under D2 in THF using RhCl3‚3H2O
as the Precatalyst. A “standard conditions” hydrogenation was
performed using Aliquat 336 (21.0 mg, 5.20 × 10-2 mmol), 1.0 mL
(12.3 mmol) of THF, 100 µL (0.23 mmol) of trioctylamine, 89 µL
(1.0 mmol) of benzene, 64 µL (0.47 mmol) of cyclooctane, RhCl3‚3H2O
(10.6 mg, 5.06 × 10-2 mmol), and 50 µL (2.8 mmol) of H2O.
A
hydrogenation run was begun under 693 Torr of D2. The reaction was
allowed to run to completion, 4 h. Every 30 min, samples were
removed for GLC analysis. A plot of loss of benzene vs time is shown
in Figure 3a.
Hydrogenation of Benzene under D2 in THF Using Isolated Rh(0)
Nanoclusters under D2. A “standard conditions” hydrogenation was
performed using Aliquat 336 (22.7 mg, 5.62 × 10-2 mmol), 1.0 mL
(12.3 mmol) of THF, 100 µL (0.23 mmol) of trioctylamine, 89 µL
(1.0 mmol) of benzene, 64 µL (0.47 mmol) of cyclooctane, RhCl3‚3H2O
A second Hg poisoning (control) experiment was performed to see
the effect that a reduced amount of added Hg had on catalytic activity.
(10.0 mg, 4.78 × 10-2 mmol), and 50 µL (2.8 mmol) of H2O.
A

5666 J. Am. Chem. Soc., Vol. 120, No. 23, 1998
Weddle et al.
The reaction was run just as above, the reaction was stopped at ca.
50% conversion, and only 2.0 equiv of Hg (19.0 mg, 9.47 × 10-2 mmol)
was added to the reaction solution. After stirring for 1 h, the black
solution had not changed color. After reconnection to the hydrogenation
line, the reaction proceeded until ca. 80% conversion, then stopped,
Figure H, Supporting Information. This control experiment warns
against using small amounts of Hg in poisoning studies.
Hg Poisoning Tests Using Isolated Rh(0) Nanoclusters Plus
Aliquat 336 in the Presence of Hg(0) and Benzene. These experiments
are detailed in the Supporting Information.
Supporting Information Available: Figure A, hydrogena-
tion apparatus; Figures B-E, transmission electron micrographs
of Rh(0) nanoclusters; Figure F, curve fits for isolated, and
isolated then reused, Rh(0) nanoclusters in dichloroethane;
Figure G, stirring rate controls; Figures H and I, Hg(0) poisoning
using reduced amounts of Hg(0); experimental details for
filtration experiments on isolated and then reused Rh(0)
nanoclusters in THF; experimental details for IR spectroscopy
-
on the product solutions (attempting to show that all the RhCl4
precursor had been converted to Rh(0) nanoclusters); experi-
mental probing for H2 gas-to-solution mass-transfer limitations
under the 770 rpm stirring rates employed; experimental details
for Hg poisoning tests using isolated Rh(0) nanoclusters plus
Aliquat 336 in the presence of Hg(0) and benzene (11 pages,
print/PDF). See any current masthead page for ordering
information and Web access instructions.
Acknowledgment. We thank Professors J. Blum and K. P.
C. Vollhardt for constructive criticism on earlier drafts of this
paper. Financial support was provided by the Department of
Energy, Chemical Sciences Division, Office of Basic Research,
via Grant DOE FG06-089ER13998. Don Dick in the Central
Instrumentation Facility at CSU is thanked for his expert help
with the GC-MS experiments.
JA973045H