Solution-liquid-solid growth of indium phosphide fibers from organometallic precursors: Elucidation of molecular and nonmolecular components of the pathway

Article abstract of DOI:10.1021/ja9640859

Methanolysis of {t-Bu₂In[μ-P(SiMe₃)₂]}₂ (1) in aromatic solvents gives polycrystalline InP fibers (dimensions 10-100 nm x 50-1000 nm) at 111-203°C. The chemical pathway consists of a molecular component, in which precursor substituents are eliminated, and a nonmolecular component, in which the InP crystal lattices are assembled. The two components working in concert comprise the solution-liquid-solid (SLS) mechanism. The molecular component proceeds through a sequence of isolated and fully characterized intermediates: 1 → [t-Bu₂In(μ-OMe)]₂ (2) → [t-Bu₂In(μ-PHSiMe₃)]₂ (3) → 2 → [t-Bu₂In(μ-PH₂)]₃ (4). Complex 4, which is alternatively prepared from t-Bu₃In and PH₃, undergoes alkane elimination, the last steps of which are catalyzed by the protic reagent MeOH, PhSH, Et₂NH, or PhCO₂H. In the subsequent nonmolecular component of the pathway, the resulting (InP)(n) fragments dissolve into a dispersion of molten In droplets, and recrystallize as the InP fibers. Important criteria are identified for crystal growth of covalent nonmolecular solids from (organic) solution. The outcomes of other solution-phase semiconductor syntheses are rationalized according to the functioning of molecular and nonmolecular pathway components of the kind identified here.

Full text of DOI:10.1021/ja9640859

2172
J. Am. Chem. Soc. 1997, 119, 2172-2181
Solution-Liquid-Solid Growth of Indium Phosphide Fibers
from Organometallic Precursors: Elucidation of Molecular and
Nonmolecular Components of the Pathway
†
†
†
‡
Timothy J. Trentler, Subhash C. Goel, Kathleen M. Hickman, Ann M. Viano,
†
†
‡
Michael Y. Chiang, Alicia M. Beatty, Patrick C. Gibbons, and
William E. Buhro*,†
Contribution from the Departments of Chemistry and Physics, Washington UniVersity,
St. Louis, Missouri 63130
X
ReceiVed NoVember 25, 1996. ReVised Manuscript ReceiVed January 2, 1997
Abstract: Methanolysis of {t-Bu2In[µ-P(SiMe3)2]}2 (1) in aromatic solvents gives polycrystalline InP fibers (dimensions
1
0-100 nm × 50-1000 nm) at 111-203 °C. The chemical pathway consists of a molecular component, in which
precursor substituents are eliminated, and a nonmolecular component, in which the InP crystal lattices are assembled.
The two components working in concert comprise the solution-liquid-solid (SLS) mechanism. The molecular
component proceeds through a sequence of isolated and fully characterized intermediates: 1 f [t-Bu2In(µ-OMe)]2
(2) f [t-Bu2In(µ-PHSiMe3)]2 (3) f 2 f [t-Bu2In(µ-PH2)]3 (4). Complex 4, which is alternatively prepared from
t-Bu3In and PH3, undergoes alkane elimination, the last steps of which are catalyzed by the protic reagent MeOH,
PhSH, Et2NH, or PhCO2H. In the subsequent nonmolecular component of the pathway, the resulting (InP)n fragments
dissolve into a dispersion of molten In droplets, and recrystallize as the InP fibers. Important criteria are identified
for crystal growth of covalent nonmolecular solids from (organic) solution. The outcomes of other solution-phase
semiconductor syntheses are rationalized according to the functioning of molecular and nonmolecular pathway
components of the kind identified here.
Introduction
organometallic precursors than were employed in eq 1.4 Here
we provide a full account of the conversion of the precursors
Solution-phase organometallic syntheses of semiconductors
are used for the growth of high-quality quantum dots,
{t-Bu2In[µ-P(SiMe3)2]}2 (1, eq 1a) and t-Bu3In to polycrystalline
1
-3
and
InP fibers. This study identifies important mechanistic criteria
for crystal growth of covalent nonmolecular solids from
organometallic (or metalloorganic) precursors in solution.
are potentially useful for the deposition of semiconductor
materials on thermally sensitive substrates such as polymers or
glasses. However, organometallic reactions that support the
crystal growth of semiconductors or other covalent nonmolecular
solids directly from solution are presently rare. We report the
synthesis of crystalline InP from the solution-phase organome-
tallic reaction 1a conducted at low temperatures. Our elucida-
Molecular-precursor syntheses of semiconductors have been
5
investigated since the 1960s. Reactions conducted in solution
6
yield three principal outcomes. (1) Generally, elimination
reactions precipitate incompletely converted molecular or oli-
gomeric species containing the semiconductor elements and
5
,7-11
residual precursor substituents.
The reactions of trialky-
lgallanes or -indanes and phosphine or arsine are prototypical
5
,7
examples (eq 2).
(2) Some precursor-based syntheses pre-
cipitate the target semiconductors from solution, but in amor-
1
2-14
phous rather than crystalline form.
Examples are the
alcoholyses of disilylphosphido and disilylarsenido precursors
(
5) (a) Didchenko, R.; Alix, J. E.; Toeniskoetter, R. H. J. Inorg. Nucl.
tion of the chemical pathway revealed unexpected factors
responsible for InP formation and crystallization, prompting
discovery of the solution-liquid-solid (SLS) crystal-growth
Chem. 1960, 14, 35. (b) Harrison, B. C.; Tompkins, E. H. Inorg. Chem.
1962, 1, 951.
(
6) Buhro, W. E. Polyhedron 1994, 8, 1131.
7) Cowley, A. H.; Harris, P. R.; Jones, R. A.; Nunn, C. M. Organo-
(
4
mechanism. The resulting insights enabled solution-phase
metallics 1991, 10, 652.
synthesis of the crystalline III-V (13-15) compounds InP,
InAs, and GaAs under comparably mild conditions from simpler
(8) (a) Wells, R. L.; Pitt, C. G.; McPhail, A. T.; Purdy, A. P.; Shafieezad,
S.; Hallock, R. B. Chem. Mater. 1989, 1, 4. (b) Wells, R. L.; Self, M. F.;
McPhail, A. T.; Aubuchon, S. R.; Woudenberg, R. C.; Jasinski, J. P.
Organometallics 1993, 12, 2832. (c) Aubuchon, S. R.; McPhail, A. T.;
Wells, R. L.; Giambra, J. A.; Bowser, J. R. Chem. Mater. 1994, 6, 82. (d)
Wells, R. L.; Aubuchon, S. R.; Kher, S. S.; Lube, M. S.; White, P. S. Chem.
Mater. 1995, 7, 793.
†
Department of Chemistry.
‡
Department of Physics.
X
Abstract published in AdVance ACS Abstracts, February 1, 1997.
(1) Murray, C. B.; Norris, D. J.; Bawendi, M. G. J. Am. Chem. Soc.
1
993, 115, 8706.
2) Bowen Katari, J. E.; Colvin, V. L.; Alivisatos, A. P. J. Phys. Chem.
994, 98, 4109.
3) Micic, O. I.; Sprague, J. R.; Curtis, C. J.; Jones, K. M.; Machol, J.
(9) Healy, M. D.; Laibinis, P. E.; Stupik, P. D.; Barron, A. R. J. Chem.
Soc., Chem. Commun. 1989, 359.
(10) Stuczynski, S. M.; Opila, R. L.; Marsh, P.; Brennan, J. G.;
Steigerwald, M. L. Chem. Mater. 1991, 3, 379.
(
1
(
L.; Nozik, A. J.; Giessen, H.; Fluegel, B.; Mohs, G.; Peyghambarian, N. J.
Phys. Chem. 1995, 99, 7754 and references therein.
(11) (a) Matchett, M. A.; Viano, A. M.; Adolphi, N. L.; Stoddard, R.
D.; Buhro, W. E.; Conradi, M. S.; Gibbons, P. C. Chem. Mater. 1992, 4,
508. (b) Goel, S. C.; Buhro, W. E.; Adolphi, N. L.; Conradi, M. S. J.
Organomet. Chem. 1993, 449, 9.
(4) Trentler, T. J.; Hickman, K. M.; Goel, S. G.; Viano, A. M.; Gibbons,
P. C.; Buhro, W. E. Science 1995, 270, 1791.
S0002-7863(96)04085-1 CCC: $14.00 © 1997 American Chemical Society

SLS Growth of Indium Phosphide Fibers
eqs 1b,c, 3, and 4).12,14 In some of these cases, the semicon-
J. Am. Chem. Soc., Vol. 119, No. 9, 1997 2173
(
ductors are obtained as nanoparticles that may be too small to
express regular crystal lattices. (3) As noted above, a few
solution-phase organometallic (and metalloorganic) syntheses
yield crystalline semiconductors directly. These include elimi-
nation reactions of the types shown in eqs 5 and 6, which are
Figure 1. XRD Patterns: (a) eq 1a; (b) thermolysis of a solution of 4
at 111 °C in a heating mantle; (c) thermolysis of a solution of [t-Bu
In(µ-PH )] (4) in the presence of MeOH catalyst (10 mol %) at 111
C in a heating mantle. Peak intensities provide a qualitative estimation
2
-
2
3
°
of the crystalline content in each sample.
Results
Growth of InP Fibers. Addition of methanol to toluene
slurries of {t-Bu2In[µ-P(SiMe3)2]}2 (1) gave colorless homo-
geneous solutions that yellowed slowly at room temperature,
and became orange rapidly upon reflux. When refluxing was
conducted in a heating mantle (see below), a black precipitate
formed. Spectroscopic monitoring evidenced the clean genera-
tion of the molecular byproducts t-BuH (isobutane) and MeO-
SiMe3 (eq 1a). X-ray powder diffraction (XRD) established
that the air-stable precipitate contained the crystalline phases
InP (major) and metallic In (minor; Figure 1a). The crystalline
coherence lengths calculated by the Scherrer equation were
conducted in high-boiling coordinating solvents at >200 °C and
produce semiconductor crystallites with dimensions of up to
ca. 10 nm (100 Å).1
-3,15,16
Other cases, such as eq 1a, afford
4
,17-19
larger crystallites at lower solution temperatures.
We now
present a rationale for the differing outcomes of solution-phase
semiconductor syntheses, that is, for the extent of elimination
achieved in solution and for the crystallinity (or noncrystallinity)
of the resulting products.
9
-16 and 28-71 nm for InP and In, respectively. Combustion
analyses indicated retention of low levels of residual carbon
1.02%) and hydrogen (0.08%). This residue was likely traces
(
The chemical pathway followed by eq 1a consists of two
components: a molecular component comprising transforma-
tions of molecular intermediates, and a nonmolecular component
comprising the steps in which the nonmolecular InP lattices are
assembled. We show that elimination catalysis by protic
reagants to generate (InP)n fragments is the critical aspect of
the molecular component. To our knowledge, Theopold and
co-workers were first to suggest such a role for protic reagents
of hydrocarbon solvents adsorbed onto the high-surface-area
InP/In powder.
Transmission-electron-microscope (TEM) images showed the
principal microstructural feature in the precipitate to be crooked
fibers 10-100 nm in width and 50-1000 nm in length (Figure
2
a,b). Energy dispersive X-ray spectroscopy conducted on
individual fibers in the TEM gave 1:1 In:P ratios, within
experimental limits. A wide-area (multiple-fiber) electron
diffraction pattern (Figure 2c) gave diffraction rings that indexed
12a
in organometallic reactions like eq 1.
We further show that
4
operation of the SLS mechanism is the critical aspect of the
nonmolecular component, which induces growth of polycrys-
talline InP fibers from a dispersion of nanometer-sized In metal
droplets. Some of the results contained herein have been
reported in preliminary form.4
to InP, with a slight apparent lattice contraction (observed, a )
20
0
.572 nm; expected, a ) 0.587 nm), which was likely due to
imperfect calibration of the TEM camera length rather than to
a real contraction. The XRD data (Figure 1a) on the bulk
sample showed no such contraction. The electron diffraction
patterns of three individual fibers gave a values ranging from
,6
(
12) (a) Byrne, E. K.; Parkanyi, L.; Theopold, K. H. Science 1988, 241,
32. (b) Douglas, T.; Theopold, K. H. Inorg. Chem. 1991, 30, 594. (c)
Douglas, T. Ph.D. Thesis, Cornell University, 1991.
13) Stuczynski, S. M.; Brennan, J. G.; Steigerwald, M. L. Inorg. Chem.
989, 28, 4431.
14) Goel, S. C.; Chiang, M. Y.; Buhro, W. E. J. Am. Chem. Soc. 1990,
12, 5636.
15) Olshavsky, M. A.; Goldstein, A. N.; Alivisatos, A. P. J. Am. Chem.
Soc. 1990, 112, 9438.
16) Uchida, H.; Matsunaga, T.; Yoneyama, H.; Sakata, T.; Mori, H.;
Sasaki, T. Chem. Mater. 1993, 5, 716.
17) Ramli, E.; Rauchfuss, T. B.; Stern, C. L. J. Am. Chem. Soc. 1990,
12, 4043.
18) Hepp, A. F.; Andras, M. T.; Bailey, S. G.; Duraj, S. A. AdV. Mater.
Opt. Electron. 1992, 1, 99.
19) Brennan, J. G.; Siegrist, T.; Carroll, P. J.; Stuczynski, S. M.;
0.570 to 0.579 nm. Thus, the fibers were identified as crystalline
3
InP, the primary constituent of the bulk product.
(
Higher-resolution TEM lattice images revealed the polycrys-
tallinity of the fibers. Numerous regions exhibiting corduroy
patterns of 〈111〉 lattice fringes were evident. These regions
formed a patchwork pattern of individual crystalline domains
1
(
1
(
(grains), between which the lattice fringes were oriented in
(
different directions. Unfortunately, the images were not of
sufficient contrast for reproduction here. The domains appeared
to have lateral dimensions of 10-20 nm, consistent with the
average coherence length calculated from the line broadening
in the XRD pattern (see above), and larger than the crystalline
(
1
(
(
Reynders, P.; Brus, L. E.; Steigerwald, M. L. Chem. Mater. 1990, 2, 403.
In this study nanosized crystallites were produced, but at lower solution
temperatures.
(20) West, A. R. Solid State Chemistry and Its Applications; Wiley: New
York, 1984; p 237.

2174 J. Am. Chem. Soc., Vol. 119, No. 9, 1997
Trentler et al.
Figure 2. TEM data for InP fibers: (a, and b) dark-field images at two magnifications; (c) electron diffraction pattern. The indices refer to Bragg
reflections in InP.
Figure 4. Thermal-ellipsoid plot of [t-Bu
Hydrogen atoms were omitted for clarity. Selected bond distances
2
In(µ-PHSiMe
3 2
)] (3).
Figure 3. Thermal-ellipsoid plot of [t-Bu
2
2
In(µ-OMe)] (2). Hydrogen
atoms were omitted for clarity. Selected bond distances (Å): In(1)-
O(1), 2.153(2); In(1)-O(1a), 2.153(2); In(1)-C(1), 2.188(6); In(1)-
C(4), 2.191(5); O(1)-C(7), 1.398(4). Selected bond angles (deg):
C(1)-In(1)-O(1), 109.0(1); C(1)-In(1)-O(1a), 109.0(1); O(1)-In-
(
(
(
Å): In(1)-P(1), 2.683(1); In(1)-C(1), 2.190(5); In(1)-C(5), 2.198-
6); In(1)-P(1a), 2.637(1); P(1)-In(1a), 2.637(1); P(1)-Si(1), 2.239-
2). Selected bond angles (deg): P(1)-In(1)-C(1), 108.2(1); P(1)-
In(1)-C(5), 112.9(1); C(1)-In(1)-C(5), 123.5(2); P(1)-In(1)-P(1a),
3.9(1); C(1)-In(1)-P(1a), 113.7(1); C(5)-In(1)-P(1a), 107.8(1); In-
1)-P(1)-In(1a) 96.1(1); In(1)-P(1)-Si(1), 127.7(1).
(
1)-O(1a), 75.2(1); C(4)-In(1)-O(1), 111.5(1); C(4)-In(1)-O(1a),
8
(
1
11.5(1); C(1)-In(1)-C(4), 128.2(1); In(1)-O(1)-In(1a), 104.8(1);
In(1)-O(1)-C(7), 127.6(1); In(1a)-O(1)-C(7), 127.6(1).
produced the net outcome of (3).12b,c As described below, the
last-detected, molecular intermediate was not unique to precursor
domain sizes obtained in other solution-based metalloorganic
3,8-10,12,15
syntheses of III-V semiconductor crystallites.
Poly-
1
, and its identity did not account for the successful growth of
crystalline fibers, as opposed to single-crystal whiskers, are able
to incorporate bends during growth, allowing the irregular,
crooked fiber morphologies.
crystalline InP. This result led to identification of a catalytic
role for MeOH in the elimination steps that ultimately produced
InP.
To our knowledge, the procedure summarized above and in
eq 1a gave crystalline InP under the mildest known conditions.
The unusual InP growth morphology, low growth temperature,
and unprecedented solution-based growth conditions prompted
us to determine the growth mechanism, as described below.
Elucidation of the Molecular Component of the Pathway.
We observed and isolated three molecular intermediates on the
eq 1a pathway between precursor 1 and InP, revealing two
cycles of In-P bond cleavage and re-formation. Addition of
MeOH to precursor 1 was expected to produce one or more of
three initial outcomes: (1) methanolysis of In-C bonds to
generate indium alkoxide species, (2) methanolysis of In-P
bonds to generate indium-alkoxide species, or (3) methanolysis
of P-Si bonds to generate new phosphidoindium species having
P-H bonds. We establish below that the initial outcome of
each methanolysis reaction was (2), and that subsequent steps
Two products were obtained from reactions of 1 and 2 molar
equivalents of MeOH. After a short reaction time (2 min), the
methoxide-bridged dimer [t-Bu2In(µ-OMe)]2 (2) was isolated
in 59% yield. The unexceptional structure of 2 is shown in
21
Figure 3; it is isostructural to [t-Bu2Ga(µ-OMe)]2 and similar
22
to the previously determined structure of [t-Bu2In(µ-OEt)]2.
After a longer reaction time (3 h), the phosphido-bridged dimer
[t-Bu2In(µ-PHSiMe3)]2 (3) was isolated in 80% yield. Com-
pound 3 was produced as an equal mixture of cis and trans
isomers; the structure of the trans isomer is shown in Figure 4.
Dialkylindane dimers like 3 possessing phosphido bridges are
common, and Wells and co-workers have previously character-
(
21) Cleaver, W. M.; Barron, A. R.; McGufey, A. R.; Bott, S. G.
Polyhedron 1994, 13, 2831.
22) Bradley, D. C.; Frigo, D. M.; Hursthouse, M. B.; Hussain, B.
Organometallics 1988, 7, 1112.
(

SLS Growth of Indium Phosphide Fibers
J. Am. Chem. Soc., Vol. 119, No. 9, 1997 2175
2
were observed to form by spectroscopic monitoring. After
longer reaction times (3 h) the trimer [t-Bu2In(µ-PH2)]3 (4) was
isolated in 53% yield. The structure of 4 is shown in Figure 5;
4
is isomorphous and isostructural to [t-Bu2Ga(µ-PH2)]3, which
was prepared and characterized by Cowley, Jones, and co-
7
7
workers. Like [t-Bu2Ga(µ-PH2)]3, 4 has a characteristically
3
1
complex proton-coupled P NMR spectrum (see Figure 6).
Spectroscopic monitoring established that 2 and 4 were formed
by the sequential reactions in eqs 9 and 10. Thus, precursor 1
was converted to trimer 4 by the eqs 7-10 sequence, comprising
two cycles of In-P bond cleavage and re-formation.
Figure 5. Thermal-ellipsoid plot of [t-Bu
2
2 3
In(µ-PH )] (4). Hydrogen
atoms were omitted for clarity. Selected bond distances (Å): In(1)-
P(1), 2.627(3); In(1)-P(1a), 2.627(6); P(1)-In(1a), 2.627(3); In(1)-
C(1), 2.200(15); In(1)-C(1a), 2.200(15). Selected angles (deg): P(1)-
In(1)-C(1), 106.6(2); P(1)-In(1)-P(1a), 101.2(3); C(1)-In(1)-P(1a),
1
06.6(2); P(1)-In(1)-C(1a), 106.6(2); C(1)-In(1)-C(1a), 126.7(8).
We found that 4 was generated directly from t-Bu3In and
PH3 (eq 11), by analogy to the synthesis of [t-Bu2Ga(µ-PH2)]3
7
reported by Cowley, Jones, and co-workers. Precursor 1 was
Figure 6. 1H-coupled 31P NMR spectrum of [t-Bu
2
2 3
In(µ-PH )] (4).
ized such a derivative containing a µ-PHSiMe3 ligand.23
Spectroscopic monitoring established that 2 and 3 were formed
by the sequential reactions in eqs 7 and 8. Thus, the In-P bonds
in 1 were initially cleaved by methanolysis, and were subse-
quently re-formed.
also prepared from t-Bu In (see the Experimental Section). The
direct production of 4 by eq 11 constitutes a savings of at least
3
two procedures over its preparation via 1 and eqs 7-10.
Methanolysis product 4 was observed to form during spec-
troscopic monitoring of eq 1a, and was the last isolable or
detectable, discrete, molecular intermediate on the pathway.
However, isolated or independently generated (eq 11) 4 did not
yield crystalline InP by simple thermal decomposition at the
temperatures used in eq 1a. Gradual decomposition of 4 ensued
immediately in room-temperature solutions, as evidenced by the
generation of isobutane (t-BuH) and the appearance of a very
3
1
1
broad P{ H} NMR resonance (-230 to -260 ppm) that we
assigned to complex In-P oligomers. When these solutions
were refluxed at 111 °C, the resulting precipitates retained
significant quantities of residual alkyl substituents, and analyzed
for the approximate empirical formula [(t-Bu)0.30InPH0.30]. The
XRD patterns contained broad, amorphous features punctuated
by sharper, low- to moderate-intensity reflections for InP,
indicating varying amounts of a crystalline fraction (coherence
length e12 nm) within a primarily noncrystalline product (see
Figure 1b).
Solution thermolysis of 4 at 203 °C produced InP of
comparable purity and crystallinity to that obtained from eq 1a.
Residual carbon and hydrogen were reduced to near the levels
observed in eq 1a, and the average crystalline coherence length
in the InP product (9 nm) was within the range achieved in eq
1a. Thus, the production of crystalline InP by thermal decom-
As in the case for 1 just described, two products were obtained
from reactions of 3 and 2 molar equivalents of MeOH. At short
reaction times (ca. 2 min) significant amounts of the methoxide
(23) Wells, R. L.; McPhail, A. T.; Self, M. F. Organometallics 1993,
1
2, 3363 and references therein.

2176 J. Am. Chem. Soc., Vol. 119, No. 9, 1997
Trentler et al.
Table 1. Survey of Catalysts for Conversion of 4 to InP (Eq 12)
the temperature at which complete elimination was achieved,
enabling crystal growth as long as other necessary conditions
were met. Thus, eqs 7-10 and 12 constituted the molecular
component of the eq 1a pathway. Presumably, InP was
generated initially from eq 12 as noncrystalline (InP)n clusters
or fragments, because the insolubility of InP in conventional
solvents precludes direct, solvent-phase assembly of the regular
solution
temp
(°C)
InP
InP
heating
coherence
InP Anal.
yield
catalyst
method length (nm) (% C, % H, % N)b (%)c
a
none
PhCO2H
PhSH
111
111
111
111
111
111
111
111
111
111
111
203
203
203
203
M
M
M
M
M
M
M
M
O
O
O
O
M
M
O
e2
18
11
e2
10
e2
16
e2
e2
e2
e2
9
9.90, 1.39, 0.00
1.43, 0.05, 0.00
1.95, 0.18, 0.00
0d
42
78
0
70
0
69
0
0
0
0
88
77
71
86
H2O
4
MeOH
t-BuOH
Et2NH
Et3N
PhCO2H
PhSH
MeOH
none
PhSH
1.04, 0.10, 0.00
InP crystal lattice (see the Discussion). The molecular
component provided the nutrient species for the crystal-growth
process, which occurred in the nonmolecular component of the
eq 1a pathway as established below.
0.79, 0.02, 0.00
12.53, 2.36, 0.09
12.42, 1.88, 0.00
10.56, 1.48, 0.00
12.24, 1.15, 0.00
1.24, 0.14, 0.00
Elucidation of the Nonmolecular Component of the
Pathway. As previously noted, some metallic In always formed
as a side product with crystalline InP in eq 1a. A series of
observations established a critical role for In metal in InP
crystallization. Organometallic analogs to 1 or to t-Bu3In that
failed to produce In as a byproduct also failed to give crystalline
InP. TEM studies revealed frozen, nanometer-sized In droplets
attached to the ends of polycrystalline InP fibers; our results
indicated that these In droplets were the sites from which the
InP fibers grew. Mechanistic studies verified that the poly-
crystalline fibers grew only under conditions in which the In
droplets were molten (liquid), and confirmed that elemental
phosphorus had sufficient solubility in molten In at temperatures
below 200 °C to generate crystalline InP. The combined results
25
9
9
MeOH
MeOH
a
b
M ) heating mantle; O ) oil bath. Average values of multiple
c
analyses. Yields were calculated by dividing the mass of the InP/In
product by the theoretical mass of InP (assuming complete conversion
d
of indane precursor to InP). Yields of crystalline InP were assumed
to be zero in products having small mean coherence lengths (e2 nm).
position of 4 required solution temperatures considerably higher
than those used in eq 1a (111 °C).
The failure of isolated or independently generated (eq 11) 4
to produce wholly crystalline InP from solution at 111 °C was
consistent with ample precedent. As stated in the Introduction,
solution-phase reactions of trialkylindanes or -gallanes and
phosphine or arsine have not produced crystalline semiconduc-
4
indicated a three-phase SLS crystal-growth pathway in which
solution-phase reactions supplied In and P to the liquid In
droplets, which became supersaturated, precipitating solid
(polycrystalline) InP. In this manner, a covalent nonmolecular
5
,7
tors (see eq 2).
Trimer 4 was an intermediate in such a
solid was crystallized in conjunction with organometallic
reactions conducted in solution. The details supporting the
proposed nonmolecular component of the eq 1a pathway are
given below.
reaction (see eq 11). Obviously a condition (or conditions)
present during decomposition of 4 in eq 1a was not reproduced
in the direct thermal decomposition of isolated or independently
generated 4.
The ability of several organoindanes to produce crystalline
InP under conditions comparable to eq 1a was evaluated. The
methanolysis of {Me2In[µ-P(SiMe3)2]}2, the methyl homolog
of precursor 1, gave amorphous material containing large
amounts of residual C and H. Douglas and Theopold previously
reported that methanolysis of the (trimethylsilyl)methyl and
neopentyl homologs of 1 gave amorphous and/or nanoparticulate
The missing condition was the presence of MeOH. In eq
a, 4 was generated and decomposed in a mixture containing
1
at least traces of MeOH. When independently synthesized 4
was heated at 111 °C with a catalytic amount (10 mol %) of
MeOH, crystalline InP was obtained as in eq 1a. Residual
carbon and hydrogen levels in the product were low, and the
XRD pattern was sharp (see Figure 1c; coherence length 10
nm). We surmised that MeOH catalyzed the elimination
reaction in eq 12. The variable crystalline fractions produced
1
2b,c
InP by eq 1.
None of these, except the reaction of 1,
produced In metal in amounts detectable by XRD. Among this
homologous series of compounds, 1 was unique in producing
In metal, and in producing crystalline InP.
We also examined reactions between PH3 and aryl and alkyl
analogs to t-Bu3In under conditions comparable to the eqs 11
and 12 sequence. These experiments were undertaken to
establish the dependence of reaction pathway on the In sub-
stituent, and to possibly identify an alternative precursor to t-Bu3-
In, which is somewhat difficult to prepare and decomposes
gradually upon long-term storage. The compounds (2,4,6-
Me3C6H2)3In and PH3 were allowed to react at 111 °C with a
catalytic amount of MeOH. The compounds Et3In and PH3 were
allowed to react at 203 °C without a protic catalyst. In both
cases amorphous material was produced having an XRD pattern
similar to that in Figure 1b, rather than crystalline InP. In
neither case was In metal detected by XRD.
at 111 °C in the absence of MeOH were presumably due to
adventitious catalysis by trace protic impurities.
We then surveyed a series of other potential catalysts (Table
2
2
1
). Intermediate 4 was generated by eq 11, and 0.1 molar
equivalent (based on t-Bu3In) of the potential catalysts was
added. As shown in Table 1, H2O, t-BuOH, and NEt3 were
ineffective for promoting eq 12. The compounds PhCO2H,
PhSH, MeOH, and NEt2H were effective, and yielded coherence
lengths of 10-18 nm at 111 °C when mantle heating was
employed. Note that the effective catalysts were protic reagents
with pKa values over a wide range (4-36); we concluded that
the alkane-elimination steps contained within eq 12 were weak-
acid-catalyzed. Catalyst performance as indicated by InP yield,
purity, or coherence length was not evidently correlated with
catalyst pKa. A mechanistic rationale is given in the Discussion.
Complete alkane elimination is a necessary but insufficient
condition for InP crystal growth. Effective catalysts decreased
In contrast to the above, (Me2EtC)3In and PH3 gave crystalline
InP (coherence length 10 nm) at 203 °C without a protic catalyst;
In metal was also formed in this reaction. Thus, each time
crystalline InP was obtained in the present work, so was In
metal. Bradley and co-workers reported that t-Bu3In and other
t-Bu-substituted indanes decomposed photochemically (in day-
light) to indium metal, isobutane, and isobutene among other
2
2
products. We found that t-Bu3In and (Me2EtC)3In were also
thermally unstable in solution to give In metal, unlike the

SLS Growth of Indium Phosphide Fibers
J. Am. Chem. Soc., Vol. 119, No. 9, 1997 2177
Figure 8. Solution-liquid-solid (SLS) crystal-growth mechanism.
which whisker crystals are grown from flux droplets that are
fed from the vapor phase rather than a solution phase. Retro-
VLS, or SLV, processes are also known, in which crystals are
consumed by flux droplets and their constituent elements
2
5
liberated to the vapor phase.
An SLV mechanism was
operating during TEM examination of our fibers at higher
e-beam current densities, where they redissolved into liquid In
with the release of volatile P4 (or P2) vapor to the high-vacuum
TEM environment (see above). That is, the reverse of the
proposed crystal-growth (nonmolecular) process was directly
observed in real time. Thus, liquid In served as a flux for InP
crystal growth or crystal decomposition when in contact with
an appropriate solution phase or vacuum.
To verify the proposed role of In, we sought to demonstrate
that it was indeed molten under our reaction conditions. The
normal melting point of In is 157 °C, and In is not known to
form a (lower-melting) eutectic by dissolution of P.26 Conse-
quently, reaction temperatures above 157 °C should have been
required to produce crystalline InP by the proposed SLS
mechanism. As stated above, crystalline InP was obtained from
reactions conducted in refluxing toluene (111 °C), provided that
a heating mantle was employed. In those experiments we found
significant amounts of the black In/InP product adhering to the
walls of the reaction vessels. When the heating mantle was
replaced by an oil bath held at approximately 145 °C, the
reactions failed to produce crystalline InP. When the reaction
was conducted in refluxing mesitylene (1,3,5-Me3C6H3, bp 164
Figure 7. TEM image of InP fibers showing In spherules (arrows) at
fiber tips.
indanes lacking tertiary alkyl substituents. Therefore, the source
of metallic In in eq 1a and related procedures was likely thermal
decomposition of 4 or other tertiary-alkyl-containing intermedi-
ates.
The intermediates derived from reaction of Et3In and PH3
lacked tertiary alkyls, were consequently stable with respect to
In formation, and produced amorphous products (see above).
We repeated the experiment by adding PhSH as a protic catalyst
and t-Bu3In as an In source after the initial reaction between
Et3In and PH3. The mixture produced both crystalline InP
(coherence length 7 nm) and metallic In at 203 °C. Thus,
metallic In was strongly implicated in the InP crystal-growth
process.
The distribution of In in the TEM images of the InP samples
was expected to provide clues about the role of In. Note that
metallic In was not evident in the Figure 2 images although
metallic In was certainly present in the sample. Under normal
TEM operating conditions, In was readily melted under the
electron beam and splattered out of the field of view.
Undisturbed In metal was successfully imaged at a lower
e-beam current density. Indium spherules were often found
affixed to the ends of isolated InP fibers (see Figure 7). Metallic
In was also found in other configurations. Fibers often formed
porous, tangled bundles having their fiber tips soldered together
at numerous In vertices within the bundles. When the e-beam
current density was increased, the spherules melted and some
InP fibers dissolved into growing pools of liquid In (see below).
This principal microstructural featuresspherulitic In particles
having one or more InP fiber tips embedded into themswas
the key indicator of the crystal-growth mechanism.
°
C) in a 170 °C oil bath, crystalline InP was again produced.
These results suggested strongly that molten In was necessary
for InP crystal growth.
We inferred that mantle heating produced a larger thermal
gradient at the walls of the glass reaction vessels than did oil-
bath heating, and that temperatures g157 °C were required in
either case. As stated above, oil-bath temperatures slightly
above the In melting point gave crystalline InP whereas oil-
bath temperatures below the In melting point did not. With
mantle heating, In nanoparticles collecting on the vessel walls
apparently experienced hot-spot temperatures above 157 °C,
even in refluxing toluene (bp 111 °C).
Finally, we sought to confirm that P was sufficiently soluble
in liquid In at our reaction temperatures to support InP crystal
growth. Phosphorus is only minimally soluble in liquid In;
extrapolation of thermodynamic data collected at higher tem-
The In spherules (droplets) were strongly implicated as the
crystallization medium for growth of InP by the SLS mecha-
nism. The distinguishing features of SLS growth are (1) a
solution dispersion of small liquid flux droplets, (2) a solution
phase that feeds the elements of the crystal phase into the flux
droplets, and (3) pseudo-one-dimensional growth of the crystal
phase from the flux droplets after supersaturation is achieved
2
6
peratures indicates that the equilibrium atomic fraction of P
-8
in a saturated In(l) solution is ∼10 at temperatures below 200
°
C. To establish if this low solubility was consistent with InP
crystallization in quantities and at rates comparable to those
achieved in eq 1a, we studied the direct reaction of liquid In
and P4 under comparable conditions. A dispersion of submi-
crometer-sized In droplets was prepared by high-intensity
4
see Figure 8).4 The SLS mechanism is closely analogous to
the well-known vapor-liquid-solid (VLS) mechanism, in
4
(
2
4
sonication of In at 164 °C in mesitylene. A P solution was
(
24) (a) Wagner, R. S. In Whisker Technology; Levitt, A. P., Ed.;
(25) Retro-VLS or SLV processes are also referred to as “negative
whisker growth”. (a) Wagner, R. S. J. Cryst. Growth 1968, 3-4, 159. (b)
Schoonmaker, R. C.; Buhl, A.; Lemley, J. J. Phys. Chem. 1965, 69, 3455.
(26) Kuphal, E. J. Cryst. Growth 1984, 67, 441.
Wiley: New York, 1970; Chapter 3. (b) Givargizov, E. I. In Current Topics
in Materials Science; Kaldis, E., Ed.; North-Holland: Amsterdam, 1978;
Vol. 1, Chapter 3.

2178 J. Am. Chem. Soc., Vol. 119, No. 9, 1997
Trentler et al.
Scheme 1. Direct Intermolecular Elimination-Condensationa
of [t-Bu2In(µ-PH2)]3 (4) to a Zinc-Blende Cluster Fragment
a
Peripheral t-Bu and H cluster substituents are omitted for clarity.
The dotted lines are bonds formed by intermolecular elimination; the
solid lines are bonds preexisting in 4.
the low temperatures typically employed in solution-phase
chemistry. We have argued that the observed growth of nearly
defect-free semiconductor nanocrystals at higher temperatures
1
-3
in coordinating solvents
is allowed by a special occurrence
of solid-state diffusional mobility that disappears with increasing
27
crystallite size. Crystalline domains as large as were observed
in this study should not be directly accessible by low-
temperature solution chemistry. Indeed, crystal growth under
our conditions occurs only indirectly from the organic-solvent
phase through the intervention of a liquid-metal crystallization
phase, a special provision in the SLS pathway.
The SLS mechanism provides a novel means for activating
reversible adspecies adsorption-desorption and likely also
surface diffusional mobility at low solution temperatures. As
indicated above, these processes are inactive at liquid-solid
interfaces between semiconductor crystals and reaction solutions
in conventional organic solvents. However, such processes do
operate, presumably with monoatomic In and P solutes and
adspecies, at the liquid-solid interface between an In flux
droplet and an InP crystallite (Figure 8). This allows SLS InP
growth to proceed in organic-solvent dispersions at temperatures
considerably below those required by organometallic chemical
Figure 9. Representative In spherule coated with InP crystallites
generated by reaction of an In dispersion with P . The dark central
region is In; the lighter, irregular periphery is crystalline InP.
4
added to the hot In dispersion, which darkened immediately
with the conversion of a significant fraction of the In to
crystalline InP as confirmed by XRD (InP coherence length 50
nm). A TEM image revealed In spherules coated with InP
crystallites (Figure 9), strongly suggesting that P had dissolved
into the droplets and InP had subsequently crystallized out at
the droplet surfaces. Indium dispersions and P4 did not react
at temperatures below the melting point of In. The solubility
of P in liquid In under our reaction conditions was thus sufficient
for InP crystal growth at readily observable rates, supporting
the proposed SLS mechanism.
We concluded that the InP crystal lattices were assembled in
a nonmolecular component of the eq 1a pathway that comprised
crystal-growth steps operating in conjunction with the previously
described molecular (ligand-elimination) component of the
pathway. Liquid In as small flux droplets was the true crystal-
growth solvent. The SLS growth mechanism coordinated the
dynamics of three phases in intimate contactsthe organometallic
reaction solution, the In crystallization liquid, and the growing
InP solid fibers.
28
vapor deposition (g500 °C, surface diffusion) or melt crystal-
lization (g1000 °C, surface diffusion and/or adsorption-
2
9
desorption).
Role of Protic Catalyst. As stated above, solution-phase
reactions of R3In or R3Ga and PH3 or AsH3 typically do not
afford complete elimination to the binary semiconductor (eq
2). We found that t-Bu3In and PH3 gave, at 111 °C without a
catalyst, a product corresponding to elimination of ∼2.70 of
the expected 3.00 equiv of t-BuH. Apparently, the protic
catalyst is required for promoting elimination of the last ∼0.30
equiv. Our proposed elimination-condensation model below
accounts for the difficulty in liberating the last small amounts
of alkane, and the role of the catalyst in assisting it.
In the eq 12 elimination-condensation process, t-BuH
elimination is coupled with In-P bond formation, bringing about
the oligomerization (condensation) of molecular intermediates.
The resulting cluster fragments should likely adopt core
structures resembling the stable, zinc-blende structure of crystal-
line InP. Because the six-membered metallacycle 4 is the last-
detected molecular intermediate, one may propose that t-BuH
elimination proceeds by direct intermolecular condensation of
Discussion
Requirements for Crystal Growth. As detailed else-
4
,27
where, one of two conditions must exist for uniform crystal-
lattice construction to occur. Adspecies from a fluid nutrient
phase (vapor, melt, or solution) must either be capable of
reversible adsorption-desorption from a crystal-growth surface
or have rapid diffusional mobility upon a growth surface. These
conditions allow mispositioned adspecies to find proper sites
in growing crystals.
30
4
(Scheme 1). As shown in Scheme 1, condensation of chairlike
In3P3 rings can produce zinc-blende fragments. A similar
condensation process has been proposed by Gladfelter and co-
Neither crystal-growth condition typically exists in solution-
phase semiconductor syntheses. The elimination reactions used
to deposit semiconductor elements are generally chemically
irreversible (eqs 1-6), and no soluble species are generally
desorbable from semiconductor growth surfaces. The cascades
of covalent-bond-making and -breaking events necessary to
support atomic surface diffusion are energetically precluded at
(
28) Ludowise, M. J. J. Appl. Phys. 1985, 58, R31.
(
29) Dugue, M.; Goullin, J. F.; Merenda, P.; Moulin, M. In PreparatiVe
Methods in Solid State Chemistry; Hagenmuller, P., Ed.; Academic: New
York, 1972; pp 309-360.
(30) (a) Dance, I. G.; Choy, A.; Scudder, M. L. J. Am. Chem. Soc. 1984,
1
06, 6285. (b) Herron, N.; Calabrese, J. C.; Farneth, W. E.; Wang, Y.
Science 1993, 259, 1426. (c) Vossmeyer, T.; Reck, G.; Katsikas, L.; Haupt,
E. T. K.; Schulz, B.; Weller, H. Science 1995, 267, 1476. (d) Vossmeyer,
T.; Reck, G.; Schulz, B.; Katsikas, L.; Weller, H. J. Am. Chem. Soc. 1995,
117, 12881.
(
85.
27) Buhro, W. E.; Hickman, K. M.; Trentler, T. J. AdV. Mater. 1996, 8,
6

SLS Growth of Indium Phosphide Fibers
J. Am. Chem. Soc., Vol. 119, No. 9, 1997 2179
Scheme 2. Proposed Catalytic Cycle for Stereochemically
Disfavored Elimination-Condensation Steps
(CH2-t-Bu)4 (within the conformationally restricted surface
environment), despite the stronger thermodynamic driving force
for the latter. This was attributed to the kinetic facility of proton
transfer to an O (X) lone pair relative to a metal-alkyl σ bond
in the surface-confined reaction environment. Likewise, we
assert that the proton transfer to X in 6 as depicted in Scheme
a
2
should be considerably faster than a corresponding proton
transfer to R in 5. Thus, conformationally disfavored alkane
eliminations may be catalyzed by appropriate protic reagents.
We believe that the crucial last steps of catalyzed alkane
elimination occur at the interface between the reaction solution
and the In flux droplet (see Figure 8). Indeed, when intermedi-
ate 4 was decomposed (eq 12) at 111 °C in an oil bath in the
presence of the protic catalysts, but in the absence of In droplets,
alkane elimination was incomplete (see Table 1). This suggests
that catalysis must occur upon the surface of the In droplets, at
least for the final elimination steps.
The molecular, elimination-condensation pathway we have
proposed may possibly generate large, zinc-blende cluster
fragments, but would be incapable of producing the extended
structural periodicity (crystallinity) and the large InP fiber
dimensions for the reasons given in Requirements for Crystal
Growth (see above). The cluster fragments should have external
faces terminated by residual In-t-Bu and P-H groups. Elimi-
nation of these last t-Bu and H substituents should be completed
on the flux-droplet surfaces so that the resulting naked (InP)n
clusters may immediately dissolve (presumably with dissocia-
tion) into the metallic flux. Naked (InP)n clusters could have
only a short lifetime in solution (in a noncoordinating solvent)
prior to agglomeration and precipitation as amorphous solid.
That crystalline InP was obtained suggests that naked (InP)n
clusters are generated in close proximity to (or on) the In flux
droplets. Once In and P are deposited into the flux the
nonmolecular component of the pathway, described above,
assembles the polycrystalline fibers.
a
The cycle (step) is depicted for a hexagonal-net fragment derived
from [t-Bu
2 2 3
In(µ-PH )] (4); solid lines represent bonds preexisting in
molecules of 4. Only the participating In and P centers are labeled;
other atom labels and peripheral substituents are omitted for clarity.
workers to account for conversion of the six-membered cyclo-
trigallazane [H2GaNH2]3 to the wurtzitic GaN.
31
Initially, intermolecular elimination-condensation of 4 should
occur readily, because conformationally mobile metallacycles
should easily achieve the mutual orientations necessary for four-
center elimination of t-BuH. Indeed, elimination-condensation
begins in solution even at room temperature, and in the absence
of added protic catalysts (or In particles; see above).
However, as elimination-condensation proceeds and the
conformational rigidity of the resulting fragments increases,
situations arise for which four-center elimination is highly
disfavored. For example, structure 5 (Scheme 2) depicts a
hexagonal-net fragment resulting from condensation of five
molecules of 4. Note that in 5 the depicted In and P centers
have an incorrect regiochemistry for a four-center elimination
to give an In-P bond; the In and P atoms are anti rather than
syn in the four-center arrangement. These and related situations
should arise whether elimination-condensation occurs by the
straightforward intermolecular condensation of 4 as suggested
in Scheme 1, or by a more complex pathway. We propose that
protic catalysis is particularly vital for promoting elimination
steps that, as in 5, cannot proceed by a direct four-center process.
Such elimination steps likely correspond to a significant
component of the last ∼0.30 equiv of t-BuH in eq 12.
Rationale for Outcomes of Solution-Phase Semiconductor
Syntheses. The three outcomes of metalloorganic semiconduc-
tor syntheses noted in the Introduction are (1) incomplete
elimination, (2) production of an amorphous semiconductor, and
(3) production of a crystalline semiconductor. These outcomes
are given in order of decreasing frequency and increasing
importance. No preexisting rationale accounts for the outcome
achieved by any specific synthesis.
We propose that the outcome depends on the functioning of
molecular and nonmolecular pathway components of the kind
identified here. To produce crystalline material, both compo-
nents must function effectively. If the molecular component
operates but the nonmolecular component does not, then
amorphous material is produced; precursor substituents are
eliminated completely, but no means exist for assembling the
nonmolecular crystal lattices. If the molecular component fails,
then incomplete elimination obtains.
Protolytic cleavage of a t-Bu-In bond by a catalyst XH is
the first step in our proposed catalytic cycle (see Scheme 2).
The catalysts we identified are sufficiently reactive to cleave
3
2
indium-alkyl bonds with attendant In-X bond formation,
generating intermediates like 6. Proton transfer from P to a
lone pair on X, perhaps with intermolecular participation of a
second molecule of XH, produces a zwitterion with a good
leaving group (7). Loss of XH followed by collapse of the
zwitterion regenerates the protic catalyst and forms the In-P
bond (8).
In a recent review we analyzed organometallic syntheses
represented by eqs 1 and 2, finding many mechanistic similari-
6
ties between them. Despite these similarities, the reactions give
consistently different outcomes, which were inexplicable when
the review was written. We now understand that reactions
corresponding to eq 2 give incomplete elimination for lack of
an elimination catalyst and the consequent inefficiency of the
molecular component of the pathway. Reactions corresponding
to eq 1 generally give amorphous or nanoparticulate materials
because a protic catalyst is present and the molecular component
is successful. When the eq 1 precursor contains t-Bu substit-
uents some metallic In forms during the reaction, activating the
nonmolecular component of the pathway and producing crystal-
line InP. We propose that all organometallic (or metalloorganic)
Schwartz and co-workers studied metalloorganic reactions on
3
3
hydroxylated surfaces that are relevant to Scheme 2. They
found that proton transfer from a surface hydroxyl to a surface-
bound Zr(O-t-Bu)4 is more rapid than to a surface-bound Zr-
(
31) Hwang, J.-W.; Campbell, J. P.; Kozubowski, J.; Hanson, S. A.;
Evans, J. F.; Gladfelter, W. L. Chem. Mater. 1995, 7, 517.
32) Tuck, D. G. In ComprehensiVe Organometallic Chemistry; Wilkin-
son, G.; Stone, F. G. A., Abel, E. W., Eds.; Pergamon: Oxford, 1982; Vol.
(
1
, Chapter 7, p 711.
33) Miller, J. B.; Schwartz, J.; Bernasek, S. L. J. Am. Chem. Soc. 1993,
15, 8239.
(
1

2180 J. Am. Chem. Soc., Vol. 119, No. 9, 1997
Trentler et al.
semiconductor syntheses conducted in solution at low temper-
atures may now be similarly rationalized. The outcomes depend
on the success or failure of (1) molecular pathways for
elimination-condensation (possibly catalyzed) and (2) nonmo-
lecular pathways for assembling the covalent crystal lattices.
sample probe in a N
protective sleeve with an O-ring fitting. The probe was then inserted
into the microscope port through a N -filled glovebag. The analyses
were performed on a JEOL 2000 FX transmission electron microscope,
fitted with a Noran Voyager X-ray spectrometer for energy-dispersive
spectroscopic (EDS) elemental analyses.
2
-atmosphere glovebox and sealed by an air-tight
2
Formation of InP by Methanolysis of {t-Bu
1). Compound 1 (1.10 g, 1.35 mmol) was suspended in toluene (30
2
In[µ-P(SiMe
3 2 2
) ]}
Conclusion
(
Low-temperature, solution-based growth of crystalline InP
proceeds by the SLS pathway, which comprises an effective
substituent-elimination mechanism working in concert with an
effective crystallization mechanism. These are the necessary
and sufficient conditions for organometallic synthesis of crystal-
line covalent nonmolecular solids under any conditions (includ-
ing those for CVD). The key to reducing growth temperatures
is identifying novel crystal-growth pathways that function at
lower temperatures. Further developments in this area may
bring the synthesis of nonmolecular inorganic materials into the
regime of conventional, solution-phase organometallic chem-
istry.
mL), and MeOH (0.22 mL, 0.17 g, 5.4 mmol) was added at room
temperature, resulting in a clear colorless solution within 3-4 min.
Stirring was continued for 10 h whereupon the color changed to yellow
orange. The reaction mixture was then refluxed for 24 h in a heating
mantle. The color of the solution changed to dark orange within the
first 10 min at reflux, and then polycrystalline InP precipitated slowly
as a black solid. The mixture was allowed to cool, and the InP was
collected by filtration, washed with hexane (2 × 10 mL), and dried in
vacuo (yield 0.31 g, theoretical InP weight yield 0.40 g). Anal.
Found: C, 1.02; H, 0.08. EDS (atom %): calcd for InP, In, 50; P, 50;
found, In, 46; P, 49.
The XRD pattern of the black solid indicated an average InP
crystallite coherence length of ca. 11 nm (Figure 1a). A small amount
of indium metal was also detected. The coherence lengths of InP varied
from 9 to 16 nm and the In coherence lengths varied from 29 to 71 nm
with decreasing stirring time prior to refluxing.
Experimental Section
General Methods. All ambient-pressure procedures were carried
2 3
out under dry N using standard inert-atmosphere techniques. t-Bu -
Preparation of [t-Bu
2 2
In(µ-OMe)] (2) by Methanolysis of 1.
2
2
In,34 and (2,4,6-Me
35
In, (Me
2
EtC)
3
3
C
6
H
2
)
3
In were prepared by the
Compound 1 (0.66 g, 0.81 mmol) was suspended in hexane (5 mL).
MeOH (66 µL, 0.052 g, 1.62 mmol) was added at room temperature
with stirring, whereupon a homogeneous solution formed in 2 min.
The solution was stirred at room temperature for 2 min and then stored
at -75 °C for 10 h. Colorless crystals of 2 were deposited. The
supernatant was removed by cannula, and the crystals of 2 were dried
in vacuo (yield 0.25 g, 0.48 mmol, 59%). Mp: 121 -122 °C. Anal.
procedures of Bradley and Barron. Et
3
In was used as received from
Strem. {t-Bu
from t-Bu In and HP(SiMe
zene-d
was prepared similarly to the method of Steigerwald and co-workers.
PH was used as received from Matheson. Caution! Phosphine (PH
2
In[µ-P(SiMe
3
)
3
2
]}
2
6
(1) was prepared as a white precipitate
3
31
3
)
2
in hexane. P{H} NMR (ppm, ben-
3
7
6
): -189.3 (s). Mp: 152-3 °C dec. {Me
2
In[µ-P(SiMe
3
)
2
]}
2
10
3
3
)
is highly toxic and must be handled in accordance with proper safety
Calcd for C18
H
42
O
2
In
2
: C, 41.56; H, 8.14. Found: C, 41.03; H, 8.04.
): 3.46 (s, 6 H, OMe), 1.39 (s, 36 H, t-Bu). IR
3
8
measures. We used PH
3
from a lecture bottle by bubbling small
1
H NMR (δ, benzene-d
(
6
amounts through reaction solutions in a well-Ventilated hood, and by
-1
cm , KBr): 2945 s, 2925 s, 2829 vs, 2761 w, 2703 w, 1465 vs, 1360
passing the effluent gas from the reaction solutions through aqueous
m, 1248 w, 1191 w, 1158 m, 1051 s, 1012 m, 836 w, 808 s, 625 w,
39 s.
3 2
Preparation of [t-Bu In(µ-PHSiMe )] (3) by Methanolysis of 1.
38
sodium hypochlorite to destroy unreacted PH
Fisher), PhSH (Aldrich), Et NH (Fisher), and Et
as received. Indium metal shot and P (white phosphorus) were used
as received from Strem. Methanol and t-BuOH were dried over Mg
activated by I , and distilled. Hexane, toluene, mesitylene, and THF
were distilled from sodium-benzophenone ketyl. 1,3-Diisopropyl-
benzene was sequentially washed with concentrated H SO , deionized
O, dilute NaOH, and deionized H O, then sequentially dried over
CaH and KOH (s), and finally sparged with N . NMR solvents were
sparged with N and stored over type 4A sieves.
Melting points were measured under N . C, H, and N analyses were
3
.
The reagents PhCO
2
H
4
(
2
3
N (Aldrich) were used
2
4
Compound 1 (1.47 g, 1.81 mmol) was suspended in hexane (4 mL) at
room temperature. MeOH (147 µL, 0.116 g, 3.62 mmol) was added
at room temperature with stirring, whereupon a homogeneous solution
formed in 2 min. The solution was stirred at room temperature for 3
h and then stored at -23 °C for 5 h. Colorless crystals of 3 were
deposited. The supernatant liquid was removed by cannula, and the
crystals were dried under reduced pressure. A second crop of 3 was
obtained similarly after reducing the volume of the supernatant to 2
mL (total yield 0.97 g, 1.45 mmol, 80%). Dec pt: g110 °C (slow
2
2
4
H
2
2
2
2
2
2
performed by Oneida Research Services, Whitesboro, NY. NMR
decomposition). Anal. Calcd for C22
Found: C, 39.74; H, 8.53.
56 2 2 2
H P Si In : C, 39.53; H, 8.44.
1
spectra were recorded at a field corresponding to 300 MHz for H.
XRD patterns were recorded on a Rigaku DmaxA diffractometer using
Cu KR radiation (λ ) 1.541 845 Å) and Materials Data Inc. (MDI)
software. Experimental powder patterns were compared to the JCPDS
reference patterns for InP (32-0452) and In metal (05-0642). Coherence
lengths (crystalline-domain dimension) were estimated by the JADE
X-ray powder data processing program, which uses the Scherrer
formula. KR2 features and background counts were stripped from the
data, and fwhm values were determined by peak integration; a term
correcting for instrumental broadening was included. The reported
coherence lengths are averages of the values obtained from the three
major InP reflections, or the 100% reflection of In.
Compound 3 was soluble in THF and hydrocarbon solvents. 1H
1
NMR (δ, toluene-d
8
): 1.53 (d, JH-P ) 231 Hz, 0.50 × 2 H, PH of cis
1
or trans isomer), 1.45 (d, JH-P ) 234 Hz, 0.50 × 2 H, PH of trans or
cis isomer), 1.45 (s, 0.50 × 18 H, t-Bu of cis isomer), 1.41 (s, 0.50 ×
3
0
6 H, t-Bu of trans isomer), 1.37 (s, 0.50 × 18 H, t-Bu of cis isomer),
31
.30-0.27 (complex m, 36 H, SiMe
3
of trans and cis isomers).
P
1
3
NMR (ppm, toluene-d ): -237.0 (dd, JP-H ) 163 Hz, JP-H ) 71
8
1
Hz, PHSiMe
3
of cis or trans isomer), -239.5 (dd, JP-H ) 164 Hz,
3
-1
J
3
P-H ) 71 Hz, PHSiMe of trans or cis isomer). IR (cm , KBr):
2942 m, 2920 m, 2826 m, 2757 w, 2700 w, νP-H 2301 w, 1461 m,
1360 m, νSi-Me 1248 s, 1154 m, 1012 m, νSi-Me 839 vs, 807 s, 771 m,
700 w, 625 vs, 436 vs.
Specimens for TEM analyses were prepared from pyridine suspen-
sions of the product powders generated in an ultrasonic cleaning bath.
A few drops of the suspensions were applied to the TEM substrates
Preparation of [t-Bu
61 µL, 0.048 g, 1.5 mmol) was added to a toluene (2 mL) solution of
(0.50 g, 0.75 mmol) at room temperature. The reaction mixture was
2
In(µ-PH
2 3
)] (4) by Methanolysis of 3. MeOH
(
3
(placed on absorbent tissues). The substrates were holey, amorphous
carbon films on copper grids. Some specimens were loaded into the
stirred for 3 h at room temperature, and then stored at -23 °C
whereupon pale yellow crystals of 4 deposited in 15 h. The supernatant
was removed by cannula, and the crystals were dried (yield 0.21 g,
0.27 mmol, 53%). Dec pt: 115-120 °C. The thermal instability of 4
precluded elemental analysis.
(34) Stoll, S. L.; Bott, S. G.; Barron, A. R. J. Chem. Soc., Dalton Trans.,
in press.
(
(
(
(
35) Leman, J. T.; Barron, A. R. Organometallics 1989, 8, 2214.
36) B u¨ rger, H.; Goetze, U. J. Organomet. Chem. 1968, 12, 451.
37) Goel, S. C.; Cha, D.; Chiang, M. Y.; Buhro, W. E. To be published.
38) (a) Fluck, E. Fortschr. Chem. Forsch. 1973, 35, 1. (b) Braker, W.;
Compound 4 was soluble in hydrocarbons, but decomposition began
immediately in solutions prepared at room temperature as evidenced
Mossman, A. L. Effects of Exposure to Toxic Gases - First Aid and Medical
Treatment; Matheson Gas Products: East Rutherford, NJ, 1970; pp 37-
1
1
by H NMR. H NMR (δ, benzene-d
6
): 1.60 -1.30 (br m, 0.77 × 54
3
8, 86-96.
H, t-Bu), 1.34 (s, 0.23 × 54 H, t-Bu). Resonances for the PH
2
protons

SLS Growth of Indium Phosphide Fibers
J. Am. Chem. Soc., Vol. 119, No. 9, 1997 2181
Table 2. Crystallographic Data for 2-4
Formation of InP by Reaction of Dispersed Indium and P
4
. A
stock solution of P (0.862 g, 6.96 mmol) in mesitylene (50 mL) was
4
2
3
4
prepared at room temperature and stirred overnight. Indium shot (1.015
g, 8.840 mmol) was melted in refluxing mesitylene (50 mL, 164 °C)
and then dispersed by high-intensity ultrasonic irradiation with a
titanium immersion horn, forming a suspension of micrometer-sized
particles. Ultrasonic irradiation was then discontinued, and a portion
chemical formula
color
a (Å)
b (Å)
c (Å)
C
18
H
42In
2
O
2
C
22
H
56Si
2
P
2
In
2
C
24
H
54
P
3
In
3
colorless
colorless
8.643(1)
11.181(2)
18.237(2)
94.19(1)
1757.7(4)
2
pale yellow
11.327(3)
14.598(6)
10.814(5)
9.748(3)
129.11(2)
1194.0(8)
2
16.962(7)
of the P
4
solution (16.0 mL) was injected rapidly into the hot indium
â (deg)
3
)
dispersion, resulting in a near instantaneous darkening of the light-
gray color. The mixture was allowed to cool and settle, and the solid
component was collected by filtration, washed with hexane, and dried
in vacuo to give a dark-gray powder.
The XRD pattern for the solid contained only reflections for In and
InP; the data indicated an average coherence length of 59 nm for InP.
The experiment was repeated in the manner described above, except
in refluxing toluene (111 °C) instead of refluxing mesitylene as the
reaction medium. The In dispersion used was first sonicated in hot
mesitylene, cooled, washed, dried, and added to toluene. Analysis by
XRD confirmed that no crystalline InP was formed.
Crystallographic Procedures. Suitable crystals of 2-4 for dif-
fraction analysis were obtained directly from their respective reaction
mixtures at -23 °C. Crystals were mounted in thin-wall glass
capillaries in an inert-atmosphere glovebox. Intensity data were
collected on a Siemens R3m/V diffractometer.
V (Å
1884.6(15)
2
780.0
P 6h 2m
-100
0.710 73
1.375
19.28
0.0582
0.0792
0.98
Z
formula weight
space group
T (°C)
520.2
C2/m
22
0.710 73
1.447
668.4
P2 /n
1
22
0.710 73
1.263
λ (Å)
3
F
calcd (g/cm )
-
1
µ (cm
)
19.36
14.56
a
R(F
o
)
0.0132
0.0148
1.01
0.0293
0.0452
0.80
w o
R (F
)b
GOF
a
b
2
R(F
o
) ) Σ||F
o
| - |F
c
||/Σ|F |.
2 -1
o
R
w
(F
o
) ) (Σ
w
||F
o
| - |F || /
c
2
1/2
2
Σw|F
o
| ) ; w ) [σ (F
o
o
) + g(F ) ] .
could not be assigned with certainty. Resonances for isobutane (t-
BuH), a primary decomposition product, were observed in every
Crystallographic calculations were done with the Siemens SHELXTL
PLUS program package. Crystal data are summarized in Table 2.
1
spectrum of isolated 4: 0.85 (d, JH-H ) 6.7 Hz, t-Bu of isobutane).
3
1
P{H} NMR (ppm, benzene-d
P NMR spectrum is shown in Figure 6.
6
): -271.4 (s, PH
2
). The proton-coupled
39
Neutral-atom scattering factors were used. All structures were solved
31
by direct methods. The space groups assigned on the basis of systematic
absences were C2m for 2 and P2 /n for 3. Five space groups were
1
Solution-Phase Decomposition of [t-Bu
2
In(µ-PH
2
)]
3
(4) at 111 °C.
In
0.372 g, 0.001 30 mmol) in toluene (15 mL) was prepared at room
Compound 4 was generated in situ as follows. A solution of t-Bu
(
3
possible for 4 (i.e., P622, no. 199; P6mm, no. 183; P 6h 2m, no. 189;
Pm2, no. 187; and P6/mmm, no. 191); successful refinement confirmed
the choice of space group P 6h 2m (no. 189). Refinement results for each
are described in the Supporting Information. Except for 4 (see below),
all non-hydrogen atoms were refined anisotropically. All hydrogen
atoms were placed in calculated positions (dC-H ) 0.96 Å) and included
in the final structure factor calculations by using a riding model, except
for H(1) in compound 3 for which positions and U were refined.
In 4, the asymmetric unit contained two trimeric units, one of which
exhibited disordered Me groups on the t-Bu substituents. The
disordered carbon atoms were fixed at their found positions, and their
isotropic thermal parameters were refined. The In and P atoms for the
disordered unit were refined anisotropically. Hydrogen atoms on the
disordered moiety were omitted. Final positional parameters are given
in the Supporting Information.
temperature. Excess PH
3
(g) was then bubbled into the solution. (In
In to 4 was
was removed by
, and then the solution was refluxed for
4 h in a heating mantle to yield a dark brown-black insoluble solid,
separate experiments the complete conversion of t-Bu
confirmed by NMR monitoring.) The excess PH
sparging the solution with N
3
3
2
2
which was collected by filtration, washed with hexane, and dried in
vacuo (yield 0.192 g). Anal. Calcd for [(t-Bu)0.30InPH0.30]: C, 8.90;
H, 1.87. Found: C, 8.90; H, 1.44.
The XRD pattern for the solid contained both broadened (major)
and sharper (minor) components corresponding to InP. Figure 1b shows
the pattern for a sample prepared by this procedure.
MeOH-Catalyzed Decomposition of [t-Bu
C. Compound 4 was generated in situ in toluene from t-Bu
g, 1.29 mmol) and PH as described above. MeOH (5.5 µL, 0.0044 g,
.14 mmol) was added, and the reaction mixture was stirred at room
2
In(µ-PH
2
)]
3
(4) at 111
In (0.370
°
3
3
Acknowledgment. We are grateful to Dr. Douglas M. Ho
Princeton University) for crucial assistance with the crystal
0
(
temperature for 12 h. The mixture was then refluxed for 24 h in a
heating mantle, whereupon InP was precipitated as a black solid. The
product was collected by filtration, washed with hexane, and dried in
vacuo (yield 0.132 g, theoretical InP weight yield 0.19 g). Anal.
Found: C, 1.04; H, 0.10.
The XRD pattern of the black solid indicated an average InP
crystallite coherence length of ca. 10 nm (Figure 1c). A small amount
of indium metal was also detected.
structure of 4 and to Professor Andrew R. Barron (Rice
University) for sharing results prior to publication. Funding
was provided by an NSF Presidential Young Investigator Award
(Grant CHE-9158369), the Eastman Kodak Co., the Emerson
Electric Co., the Exxon Education Foundation, and the Monsanto
Co. K.M.H. was supported by a Department of Education
GANN grant. Washington University’s X-ray Crystallography
Facility was funded by the NSF Chemical Instrumentation
Program (Grant CHE-8811456). The Washington University
High-Resolution NMR Service Facility was funded in part by
NIH Biomedical Research-Support Shared-Instrument Grants
RR-02004, RR-05018, and RR-07155, and a gift from the
Monsanto Co.
General Procedure for Formation of InP from Catalyzed
Reactions of t-Bu
described above, in either toluene or 1,3-diisopropylbenzene. The
normal reaction used 0.370 g (1.29 mmol) of t-Bu In and 15.0 mL of
3 3
In and PH . Compound 4 was generated in situ as
3
solvent; moderate variations in scale and concentration did not appear
to alter the results. A protic catalyst (or potential catalyst; see Table
1
3
) was then added in 10-11 mol % relative to t-Bu In. The resulting
solution was stirred for 12 h at room temperature prior to heating;
moderate variations in this interval did not appear to alter the results.
The reaction mixture was then refluxed for approximately 24 h. Heat
was applied to some reactions by a heating mantle (with a consistent
variac setting), and to others by an oil bath. Yields, crystalline
coherence lengths, and elemental analyses are summarized in Table 1.
Supporting Information Available: Tables listing details
of the crystallographic data collection, atomic coordinates, bond
distances, calculated hydrogen atom parameters, and anisotropic
thermal parameters (and text giving additional experimental
details (27 pages). See any current masthead page for ordering
and Internet access instructions.
(
39) International Tables for X-ray Crystallography; Hahn, T., Ed.;
Kynoch Press: Birmingham, England, 1974; Vol. IV, pp 99, 149.
JA9640859