Gas-phase thermolysis of a guanidinate precursor of copper studied by matrix isolation, time-of-flight mass spectrometry, and computational chemistry

Article abstract of DOI:10.1021/ic902247w

The fragmentation of the copper(i) guanidinate [Me₂NC(N/Pr) ₂Cu]₂ (1) has been investigated with time-of-flight mass spectrometry (TOF MS), matrix-isolation FTIR spectroscopy (Ml FTIR spectroscopy), and density functional theory (DFT) calculations. Gas-phase thermolyses of 1 were preformed in the temperature range of 100-800 °C. TOF MS and Ml FTIR gave consistent results, showing that precursor 1 starts to fragment at oven temperatures above 150 °C, with a close to complete fragmentation at 260 °C. Precursor 1 thermally fragments to Cu( _S), H₂(g), and the oxidized guanidine Me ₂NC(=N/Pr)(N=CMe₂) (3). In TOF MS experiment, 3 was clearly indentifled by its molecular ion at 169.2 u. Whereas H₂₊ was detected, atomic Cu was not found in gas-phase thermolysis. In addition, the guanidine Me₂NC(N/Pr)(NH/Pr) (2) was detected as a minor component among the thermolysis products. Ml thermolysis experiments with precursor 1 were performed, and species evolving from the thermolysis oven were trapped in solid argon at 20 K. These species were characterized by FTIR spectroscopy. The most indicative feature of the resulting spectra from thermolysis above 150 °C was a set of intense and structured peaks between 1600 and 1700 cm-¹, an area where precursor 1 does not have any absorbances. The guanldine 2 was matrix-isolated, and a comparison of its FTIR spectrum with the spectra of the thermolysis of 1 indicated that species 2 was among the thermolysis products. However, the main IR bands In the range of 1600 and 1700 cm-¹ appeared at 1687.9,1668.9,1635.1, and 1626.6 cm-¹ and were not caused by species 2. The oxidized guanldine 3 was synthesized for the first time and characterized by¹H NMR and FTIR spectroscopy. A comparison of an FTIR spectrum of matrix isolated 3 with spectra of the thermolysis of 1 revealed that the main IR bands in the range of 1600 and 1700 cm -1 are due to the presence of 3. The isomers exhibit the NMe₂ group cis or trans to the /Pr group, with c/s-3 being significantly less stable than trans-3. At higher temperature secondary thermal fragments had been observed. For example at 700 °C, TOF MS and Ml FTIR data showed that species 2 and 3 both eliminate HNMe₂ to give the carbodilmides /PrNCNPr (CDI) and /PrNCN[C(=CH ₂)Me] (4), respectively. A DFT study of the decomposition of compound 1 was undertaken at the B3LYP/6-31+G(d,p) level of theory employing dispersion-correcting potentials (DCPs). The DFT study rationalized both carbodilmide delnsertion and β-hydrogen elimination as exergonic decomposition pathways (AG = -44.4 kcal/mol in both cases), but experiment showed β-hydrogen elimination to be the favorable route.

Full text of DOI:10.1021/ic902247w

2844 Inorg. Chem. 2010, 49, 2844–2850
DOI: 10.1021/ic902247w
Gas-Phase Thermolysis of a Guanidinate Precursor of Copper Studied by Matrix
Isolation, Time-of-Flight Mass Spectrometry, and Computational Chemistry
§
,‡
,†
Jason P. Coyle,†,‡ Paul A. Johnson,‡,§ Gino A. DiLabio, Sean T. Barry,* and Jens Muller*
ꢀ
€
†Department of Chemistry, University of Saskatchewan, 110 Science Place, Saskatoon, Saskatchewan,
S7N 5C9 Canada, ‡Department of Chemistry, Carleton University, 1125 Colonel By Drive, Ottawa, Ontario,
§
K1S 5B6 Canada, and National Institute for Nanotechnology, National Research Council of Canada,
11421 Saskatchewan Drive, Edmonton, Alberta, Canada T6G 2M9
Received November 13, 2009
The fragmentation of the copper(I) guanidinate [Me2NC(NiPr)2Cu]2 (1) has been investigated with time-of-flight mass
spectrometry (TOF MS), matrix-isolation FTIR spectroscopy (MI FTIR spectroscopy), and density functional theory
(DFT) calculations. Gas-phase thermolyses of 1 were preformed in the temperature range of 100-800 °C. TOF MS
and MI FTIR gave consistent results, showing that precursor 1 starts to fragment at oven temperatures above 150 °C,
with a close to complete fragmentation at 260 °C. Precursor 1 thermally fragments to Cu(s), H2(g), and the oxidized
guanidine Me2NC(=NiPr)(N=CMe2) (3). In TOF MS experiment, 3 was clearly indentified by its molecular ion at
169.2 u. Whereas H2þ was detected, atomic Cu was not found in gas-phase thermolysis. In addition, the guanidine
Me2NC(NiPr)(NHiPr) (2) was detected as a minor component among the thermolysis products. MI thermolysis
experiments with precursor 1 were performed, and species evolving from the thermolysis oven were trapped in solid
argon at 20 K. These species were characterized by FTIR spectroscopy. The most indicative feature of the resulting
spectra from thermolysis above 150 °C was a set of intense and structured peaks between 1600 and 1700 cm-1, an
area where precursor 1 does not have any absorbances. The guanidine 2 was matrix-isolated, and a comparison of its
FTIR spectrum with the spectra of the thermolysis of 1 indicated that species 2 was among the thermolysis pro-
ducts. However, the main IR bands in the range of 1600 and 1700 cm-1 appeared at 1687.9, 1668.9, 1635.1, and
1626.6 cm-1 and were not caused by species 2. The oxidized guanidine 3 was synthesized for the first time and
characterized by 1H NMR and FTIR spectroscopy. A comparison of an FTIR spectrum of matrix isolated 3 with spectra
of the thermolysis of 1 revealed that the main IR bands in the range of 1600 and 1700 cm-1 are due to the presence
of 3. The isomers exhibit the NMe2 group cis or trans to the iPr group, with cis-3 being significantly less stable
than trans-3. At higher temperature secondary thermal fragments had been observed. For example at 700 °C,
TOF MS and MI FTIR data showed that species 2 and 3 both eliminate HNMe2 to give the carbodiimides
iPrNCNiPr (CDI) and i PrNCN[C(=CH2)Me] (4), respectively. A DFT study of the decomposition of compound 1
was undertaken at the B3LYP/6-31þG(d,p) level of theory employing dispersion-correcting potentials (DCPs). The
DFT study rationalized both carbodiimide deinsertion and β-hydrogen elimination as exergonic decomposition
pathways (ΔG = -44.4 kcal/mol in both cases), but experiment showed β-hydrogen elimination to be the favorable
route.
Introduction
deposition (ALD), including β-diketonates,1-4 β-diketimi-
nates,5 amidinates,6-9 and guanidinates.10
Thin films of copper metal are of utmost interest in
microelectronics, specifically for use as an interconnect
material. Several copper(I) precursors have been reported
for chemical vapor deposition (CVD) and atomic layer
(4) Wenzel, T. J.; Williams, E. J.; Haltiwanger, R. C.; Sievers, R. E.
Polyhedron 1985, 4, 369–378.
(5) Park, K. H.; Marshall, W. J. J. Am. Chem. Soc. 2005, 127, 9330–9331.
(6) Li, Z. W.; Rahtu, A.; Gordon, R. G. J. Electrochem. Soc. 2006, 153,
C787–C794.
(7) Li, Z. W.; Barry, S. T.; Gordon, R. G. Inorg. Chem. 2005, 44, 1728–
1735.
(8) Lim, B. S.; Rahtu, A.; Gordon, R. G. Nat. Mater. 2003, 2, 749–754.
(9) Lim, B. S.; Rahtu, A.; Park, J. S.; Gordon, R. G. Inorg. Chem. 2003,
42, 7951–7958.
*To whom correspondence should be addressed. E-mail: sbarry@
connect.carleton.ca (S.T.B.), jens.mueller@usask.ca (J.M.).
(1) Chen, T. Y.; Omnes, L.; Vaisserman, J.; Doppelt, P. Inorg. Chim. Acta
2004, 357, 1299–1302.
(2) Bollmann, D.; Merkel, R.; Klumpp, A. Microelectron. Eng. 1997,
37-8, 105–110.
(3) Lagalante, A. F.; Hansen, B. N.; Bruno, T. J.; Sievers, R. E. Inorg.
Chem. 1995, 34, 5781–5785.
(10) Coyle, J. P.; Monillas, W. H.; Yap, G. P. A.; Barry, S. T. Inorg. Chem.
2008, 47, 683–689.
r
pubs.acs.org/IC
Published on Web 02/12/2010
2010 American Chemical Society

Article
Inorganic Chemistry, Vol. 49, No. 6, 2010 2845
Scheme 1
The guanidinate dimer of copper(I), [Me2NC(NiPr)2-
Cu]2 (1), deposits copper in a CVD process above 225 °C
as a nanocrystalline film;10 similar reactivity is known for the
acetamidinatocopper(I) dimer.9 An interesting aspect of the
deposition is that these precursors act as single source
precursors for copper, meaning that they do not require a
second, reducing precursor to produce copper metal. The
solution-phase thermolysis of the guanidinate compound 1
showed the production of diisopropylcarbodiimide (CDI),
and thus we suggested a mechanism whereby CDI deinserted
from the ligand to produce a copper amide intermediate,
Cu2(NMe2)2, which then eliminated an imine and an amine
to produce metallic copper (Scheme 1).10
Figure 1. Overview MS of the copper guanidinate 1 with thermolysis
oven at 130 °C (no thermal fragmentation; bottom spectrum; Mþ of 1
enlarged) and at 260 °C (complete thermal fragmentation of 1; top
spectrum).
We wanted to further investigate this mechanism, particu-
larly in light of a recent report of β-hydrogen elimination as a
potential thermolysis route.11 In addition, we recently un-
raveled the fragmentation of an alumina ALD precursor, the
aluminum guanidinate [Me2NC(NiPr)2]Al(NMe2)2, by time-
of-flight mass spectrometry (TOF MS) and matrix-isolation
FTIR spectroscopy (MI FTIR spectroscopy).12 Using these
techniques, we have shown that this precursor eliminates
CDI to give the monomeric aluminum amide Al(NMe2)3
above 300 °C.12 Thus, we employed these techniques to
further investigate the gas-phase thermolyses of the copper
precursor 1 in the temperature range of 100-800 °C. We
report on results of TOF MS, MI FTIR spectroscopy, and
DFT calculations.
Chart 1. Protonated and Oxidized Ligand with Calculated Masses
TOF MS. The mass spectrum at the bottom of Figure 1
shows the typical ion pattern of precursor 1 in the absence
of any thermal decomposition; all fragments of the mole-
cular ion were caused by ion impact ionization at 70 eV of
the precursor 1. An oven temperature of 130 °C was used
to avoid any build-up of 1 in the apparatus through
condensation. The highest detectable masses in this spec-
trum are due to Mþ of 1 (enlargement in Figure 1); its
isotopic pattern agrees well with the calculated one (see
Experimental Section). It should be noted that the ion at
465.3 u is not part of the isotopic pattern of 1þ, but results
from the loss of one H atom (1þ - H). When the oven
temperature was increased to 260 °C (Figure 1 top
spectrum), 1þ could no longer be detected,15 and the
highest mass appeared at 171.2 u followed by a mass of
higher intensity at 169.2 u. The mass of þ171.2 u matches
with the empirical formula of C9H21N3 (calc. mass of
171.17 u) and can be interpreted as the cation of the
protonated ligand 2 (Chart 1). The peak at 169.2 u fits
Results and Discussions
Since Pimentel’s pioneering work on matrix-isolation
spectroscopy,13,14 this technique matured into a well-known
method to trap and identify reactive intermediates. Mass
spectrometry is a highly sensitive method to investigate gas-
phase species, and it has been used to elucidate thermal
fragmentations of precursors. Recently, we combined both
techniques in a unique apparatus in which the thermolysis
of a CVD precursor can be investigated using MI FTIR
spectroscopy and TOF MS.12 The facility was designed in
such a way that a series of experiments can be performed
using one method, which can then be followed by a similar
series of experiments on the same precursor sample and
carrier gas supply using the second method. Consequently,
MI FTIR and TOF MS data are collected consecutively,
but the conditions in both series of experiments are kept
as similar as possible so that their data complement one
another.
þ
well to C9H19N3 (calc. mass of 169.16 u) and can be
interpreted as dehydrogenated 2. From the numerous
possibilities to assign a structural formula to the mass
of 169.2 u, we assign structure 3 (Chart 1) for reasons that
will become apparent in the following discussion.
Figure 2 shows how signal intensities of selected ions
changed in the temperature range of 130-200 °C. The
signal intensity of Mþ (466.3 u) of the precursor 1 decreases
slightly from 160 to 170 °C, followed by a steeper decrease at
temperatures above 170 °C. A comparable change of
(11) Wu, J. P.; Li, J. Y.; Zhou, C. G.; Lei, X. J.; Gaffney, T.; Norman,
J. A. T.; Li, Z. W.; Gordon, R.; Cheng, H. S. Organometallics 2007, 26, 2803–
2805.
€
(12) Ziffle, L. C.; Kenney, A. P.; Barry, S. T.; Muller, J. Polyhedron 2008,
27, 1832–1840.
(13) Whittle, E.; Dows, D. A.; Pimentel, G. C. J. Chem. Phys. 1954, 22,
1943.
(14) Norman, I.; Pimentel, G. C. Nature 1954, 174, 508–509.
(15) Only traces of 1þ were detectable; intensity of 466.3 u was 25 counts
out of 10 million sweeps.

2846 Inorganic Chemistry, Vol. 49, No. 6, 2010
Coyle et al.
up to 800 °C in the presence of a copper metal surface,
species 2 does not eliminate H2 to give 3. In all mass
spectra of 2 the signal at 169.2 u appeared only with
negligible intensities compared to that at 171.2 u (2þ).
When considering Scheme 2, it might seem possible for
the reaction to proceed by disproportion of Cu(I) to
Cu(II) and Cu(0), giving an undetected (and perhaps
unstable and transient) copper(II) species. This has been
suggested for the amidinate depositing copper on silicon
dioxide.17 However, our DFT calculations (at the
B3LYP/6-31þG(d,p) level of theory with DCPs and
making the same assumptions as previous) show this to
be unlikely for a gas phase species: it would have a ΔG of
10.9 kcal/mol (compared to -44.4 kcal/mol for β-hydro-
gen elimination from a copper(I) species). As well, the
FTIR spectra throughout the series of thermolysis experi-
ments did not show any peaks that could not be assigned
specifically to the reactants and products of Scheme 2.
Specifically, the region from 1500-1800 cm-1 (where a
CdN stretching mode for a Cu(II) guanidinate species
would appear) contained no unassigned peaks. Finally,
similar mechanistic work for surface-bound copper species
on a nickel surface resulting from the thermolysis of a copper
amidinate similarly found no experimental evidence for a
copper(II) species.18 The sum of the evidence suggests that we
do not see a copper(II) species in these experiments.
Figure 2. Temperature dependence of intensities of selected ions arising
from the thermolysis of 1 [m/z = 466.3 (1þ), 171.2 (2þ), 169.2 (3þ),
2.0 (H2þ) u].
Scheme 2
Considering Scheme 2, it also seems possible that the
decomposition might proceed by a process involving
homolytic scission of the Cu-N bonds to produce free
radical intermediates, which may then disproportionate
to form 2 and 3. DFT calculations at the B3LYP/
6-31þG(d,p) level of theory with DCPs indicate that this
process is endergonic by 33.6 kcal/mol, and therefore also
unlikely. The thermolysis of copper(I) alkyls has been
studied by Whitesides et al.19-21 Their study of n-butyl-
copper(I) compounds revealed a β-hydride elimination
pathway.19 The decomposition of 1-propenylcopper(I)
was also found not to proceed by a pathway involving
free radical intermediates despite the stability of the vinyl
radical.20 However, neophyl(tri-n-butylphosphine)copper-
(I) was found to produce the neophyl radical, which is
known to rearrange to an exceptionally stable radical.21
These studies have shown that free radicals are unlikely
intermediates in the thermolysis of copper(I) alkyls unless
the free radical itself is very stable. Copper(I) amides have
been found to decompose via β-hydride elimination and
to be more stable than their corresponding alkyls.22 On
this basis, it seems unlikely that a mechanism involving
free radical intermediates is involved in the decomposi-
tion of 1.
intensities was observed for the signal of 2þ (171.2 u). At the
same time, the signal of 3þ (169.2 u) showed a significant
increase in intensity, which indicates that 3þ is an ion of a
thermolysis fragment of precursor 1.
In addition to the identification of the peak at 169.2 u,
which we assign to the structural formula 3 (Chart 1),
there is a significant increase of the signal intensity at
2.0 u, which is due to the presence of H2þ. Copper atoms
were not detected in the gas phase; however, after a long
series of thermolyses, the thermolysis oven was inspected
and had a copper color at the oven orifice. This is not
surprising, as it was shown that 1 is a precursor for the
thermal deposition of Cu.10 On the basis of these data, we
propose that the thermolysis of precursor 1 can be
described as shown in Scheme 2. We performed DFT
calculations at the B3LYP/6-31þG(d,p) level of theory
employing dispersion-correcting potentials (DCPs),
which showed that the production of 3 through β-hydro-
gen elimination is thermodynamically favorable (ΔG =
-44.4 kcal/mol).16
Compound 3 is the formal dehydrogenated product of
the protonated ligand 2, and it is feasible that 2, instead of
3, is the primary thermolysis product of the precursor 1.
Species 2 could further decompose to give the detected
species 3. Therefore, an independent thermolysis series
with the protonated ligand 2 as a starting compound was
undertaken. This study revealed that, for temperatures of
Alternatively, one might expect the thermolysis reac-
tion as illustrated in Scheme 2 to be initiated by the
(17) Dai, M.; Kwon, J.; Halls, M. D.; Gordon, R. G.; Chabal, Y. J.
Langmuir 2010, ASAP, DOI: 10.1021/la903212c.
(18) Qiang Ma, Q.; Hansheng Guo, G.; Gordon, R. G.; Francisco Zaera,
F. Chem. Mater. 2010, 22, 352–359.
(19) Whitesides, G. M.; Stedronsky, E. R.; Casey, C. P.; San Filippo, J.
J. Am. Chem. Soc. 2002, 92, 1426–1427.
(20) Whitesides, G. M.; Casey, C. P.; Krieger, J. K. J. Am. Chem. Soc.
2002, 93, 1379–1389.
(21) Whitesides, G. M.; Panek, E. J.; Stedronsky, E. R. J. Am. Chem. Soc.
2002, 94, 232–239.
(22) Tsuda, T.; Watanabe, K.; Miyata, K.; Yamamoto, H.; Saegusa, T.
Inorg. Chem. 1981, 20, 2728–2730.
(16) This value was calculated using ΔG = -127.3 kcal/mol for the
process 2Cu(g) f 2Cu(s) calculated from data tabulated in: Lide, D. R.
Handbook of Chemistry and Physics 79th ed.; CRC Press: Boca Raton, FL,
1998. Using the same level of theory, ΔG is computed to be -11.5 kcal/mol for the
CDI deinsertion process shown in Scheme 1.

Article
Inorganic Chemistry, Vol. 49, No. 6, 2010 2847
Figure 4. Optimized structure for trans-3 obtained with B3LYP/
6-31þG(d,p)-DCP.
which appears as a shoulder of the intense features between
1600 and 1700 cm-1 in spectrum B (see dotted line in
Figure 3). The TOF MS experiments clearly indicated that
3 was one of the key fragments of the thermolysis of the
copper precursor 1 (Scheme 2). Similarly to the identifica-
tion of 2 as a thermolysis product, we intended to identify 3
through comparison with FTIR spectra. To our surprise,
species 3 was unknown in the literature, and consequently
needed to be synthesized.
On the basis of a known method to synthesize imines
from secondary amines,23 2 was oxidized using 2 equiv of
CuBr2 and 2 equiv of LiOtBu to give species 3. Species 3
was first characterized by 1H NMR spectroscopy, which
showed that in addition to the target compound 3, small
amounts of the starting compound 2 were present. In
particular, the two singlets of equal intensity at 1.43 and
1.72 ppm (in C6D6) clearly revealed the two diastereo-
topic Me groups of the N=CMe2 moiety of 3 (Chart 1).
This compound proved very difficult to isolate from the
Figure 3. Measured IR spectra of copper guanidinate 1 and its thermo-
lysis products (x axis, 1500-1800 cm-1; y axis, absorbance). Spectrum A:
2 in solid argon (thermolysis ovenat 260 °C); see Supporting Information,
Figure S3 for a complete IR spectrum. Spectrum B: results of a thermo-
lysis of complex 1 at 260 °C; see Supporting Information, Figure S2 for a
complete IR spectrum. Spectrum C: 3 in solid argon (thermolysis oven at
260 °C); residual amounts of 2 have been electronically eliminated; see
Supporting Information, Figure S4 for a complete IR spectrum.
t
parent guanidine and BuOH, all of which have very
decomposition of the dimer 1 to form two copper guanidi-
nate monomers, which subsequently decompose. Using
DFT calculations, we predict that this pathway is very
unlikely, namely, ΔG = 66.7 kcal/mol for the formation
of two copper(I) guanidinate monomers. In addition, there
was no evidence of the monomer [Me2NC(NiPr)2Cu] in the
thermolysis series of the dimer 1; the measured intensities of
signals showing fragments with one Cu atom indicated that
they were caused by the impact ionization of 1 alone.
MI FTIR Spectroscopy. The thermal decomposition of
precursor 1 was also investigated with MI FTIR spectro-
scopy, using similar experimental conditions as for the TOF
MS experiments discussed above. At oven temperatures of
up to 150 °C, IR spectra did not change, thus indicating that
precursor 1 did not decompose (see Supporting Informa-
tion, Figure S1). At temperatures of 160 °C, new IR peaks
indicated the onset of the thermolysis. The most indicative
feature of the resulting spectrum was a set of intense and
structured peaks between 1600 and 1700 cm-1, an area
where precursor 1does not have any absorbances. Spectrum
B of Figure 3 shows the result of a thermolysis of 1 at 260 °C.
The investigation of 1 by TOF MS revealed that the
protonated ligand 2 was a product in the gas phase. To find
out if 2 was also present in the matrixes, a MI FTIR
spectrum of the synthesized sample of 2 was measured
(Figure 3, spectrum A). The comparison of spectrum A
with B indicates that 2 was one of the thermolysis products
similar boiling points, but the IR was captured in matrix
isolation, and matched the calculated vibrational fre-
quencies such that we are satisfied of its formulation
(Supporting Information, Figure S4, Table S1). We are
continuing to try to isolate this material to further our
thermolysis studies of this system.
A so-prepared sample was vaporized and trapped in
solid argon; spectrum C of Figure 3 shows the resulting
IR spectrum. All measured IR bands of 3 (spectrum C)
are present in spectra of the thermolysis of precursor 1
(spectrum B), giving strong evidence that 3 is a thermolysis
product. The structured IR band in the range of 1600 and
1700 cm-1 with four maxima at 1687.9, 1668.9, 1635.1, and
1626.6 cm-1 must be caused by CdN stretching modes. As
compound 3 exhibits two CdN bonds, two CdN stretching
modes are expected in the IR range. The fact that the IR
absorbance between 1600 and 1700 cm-1 shows more than
two bands, suggests that perhaps different isomers of 3 are
present. To better understand the structure of the IR band in
the CdN stretching region, DFT calculations at the BLYP/
6-31G(d) level of theory were performed. The lowest energy
structure of 3 (Figure 4) has a trans orientation of the Me2N
and the iPr group, and was the only isomer that appeared in
an appreciable amount (Figure 4).
The calculated frequencies of the CdN stretches of trans-
3 fit very well to the experimental data: the calculated
of 1. Species 2 has its most intense IR band at 1646.0 cm-1
,
(23) Yamaguchi, J.; Takeda, T. Chem. Lett. 1992, 1933–1936.

2848 Inorganic Chemistry, Vol. 49, No. 6, 2010
Coyle et al.
Scheme 3. Calculated Fragmentation Pathways for Precursor 1a
a Free energies relative to A1 for the reaction steps are shown (values in parentheses are relative to species 1).
(BLYP/6-31G(d))24/experimental frequencies for ν(C6dN1)
were 1652.9/1668.9 and for ν(C1dN2) they were 1598.8/
1635.1 cm-1 (Figure 4). The calculated intensity pattern
(ca. 3:1) also fits well to the measured peak ratios: the two
IR bands at lower frequencies are more intense than those at
higher frequencies. However, these assignments are only
tentative. For an unambiguous assignment of experimental
IR bands, a complete vibrational analysis of 3 would be
necessary. Such an analysis would necessarily be based on
experimental IR data of different isotopomers of 3, which
would require extended synthetic work first. However, such
an investigation is beyond the scope of our present work.
It is interesting that the thermolysis at these temperatures
found no indication of CDI, as was previously found in
sealed NMR tube thermolysis experiments.11 An extensive
DFT study of the decomposition of compound 1 was under-
taken, and this helps rationalize these two different thermo-
lysis pathways. These calculations were undertaken at the
B3LYP/6-31þG(d,p) level of theory employing DCPs.
DFT Calculations. The calculated mechanism showed
that 1 undergoes a shift of a nitrogen atom of one of the
guanidinate ligands from an μ1 to a μ2 bonding arrangement,
as well as a rotation of one of this ligand’s isopropyl groups
such that the hydrogen on this group is oriented toward one
copper center (Scheme 3, path A). This step is calculated to be
uphill in free energy by 22.5 kcal/mol. The resulting species
A1 is the branching point between β-hydrogen elimination
(path B) and CDI elimination (path C).
(24) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb,
M. A.; Cheeseman, J. R.; J. A. Montgomery, J.; Vreven, T.; Kudin, K. N.;
Burant, J. C.; Millam, J. M.; Iyengar, S. S.; Tomasi, J.; Barone, V.;
Mennucci, B.; Cossi, M.; Scalmani, G.; Rega, N.; Petersson, G. A.;
Nakatsuji, H.; Hada, M.; Ehara, M.; Toyota, K.; Fukuda, R.; Hasegawa,
J.; Ishida, M.; Nakajima, T.; Honda, Y.; Kitao, O.; Nakai, H.; Klene, M.; Li,
X.; Knox, J. E.; Hratchian, H. P.; Cross, J. B.; Bakken, V.; Adamo, C.;
Jaramillo, J.; Gomperts, R.; Stratmann, R. E.; Yazyev, O.; Austin, A. J.;
Cammi, R.; Pomelli, C.; Ochterski, J. W.; Ayala, P. Y.; Morokuma, K.;
Voth, G. A.; Salvador, P.; Dannenberg, J. J.; Zakrzewski, V. G.; Dapprich,
S.; Daniels, A. D.; Strain, M. C.; Farkas, O.; Malick, D. K.; Rabuck, A. D.;
Raghavachari, K.; Foresman, J. B.; Ortiz, J. V.; Cui, Q.; Baboul, A. G.;
Clifford, S.; Cioslowski, J.; Stefanov, B. B.; Liu, G.; Liashenko, A.; Piskorz,
P.; Komaromi, I.; Martin, R. L.; Fox, D. J.; Keith, T.; Al-Laham, M. A.;
Peng, C. Y.; Nanayakkara, A.; Challacombe, M.; Gill, P. M. W.; Johnson,
B.; Chen, W.; Wong, M. W.; Gonzalez, C.; Pople, J. A. Gaussian 03, Revision
D.02 ed.; Gaussian, Inc.: Wallingford, CT, 2004.
As shown in path B of Scheme 3, compound 3 can be
formed through elimination of the isopropyl methine proton
to form a copper hydride (B1), which is an uphill process by
13.1 kcal/mol (free energy). The hydride B1 eliminates 3 in a
barrierless process. The path from A1 to the products B2 and
3 is endergonic by 10.5 kcal/mol (33.0 kcal/mol from 1).

Article
Instead of undergoing β-hydrogen elimination, A1 can
Inorganic Chemistry, Vol. 49, No. 6, 2010 2849
undergo a shift of the nitrogen back to μ1 coordination, with a
concurrent rotation about the Cu-N bond to allow the
exocyclic dimethylamido to coordinate to the other copper
atom to form species C1. This step is thermodynamically
downhill by 6.4 kcal/mol. Compound C1 can then eliminate
the dimethylamido moiety to give the copper compound C2,
which further reacts to form C3 and free CDI. Energetically,
the process from A1 to products C3 and CDI is exergonic by
0.8 kcal/mol (but endergonic by 21.6 kcal/mol from 1).
These calculations are a good model for the vapor phase
decomposition, and show that β-hydrogen abstraction is
the more thermodynamically favored for decomposition of
1 to metallic copper (as suggested in Scheme 2). However, the
previously reported thermolysis to eliminate carbodiimide
by deinsertion was performed at lower temperatures and in
the solution phase.11 These calculations also show that
carbodiimide deinsertion is reasonable as a decomposi-
tion mechanism, as demonstrated by the convincing
divergence at species A1.
Fragmentation at Higher Temperatures. We investi-
gated the fragmentation of the copper precursor 1 and
the protonated ligand 2 at temperatures of up to 800 °C,
and new fragments appeared in the thermolyses of both 1
and 2. Figure 5 shows the results of two series of thermo-
lysis, spectrum A for 2 and spectrum B for 1. As shown in
Figure 5, the amount of species 3 (171.2 u), which is a key
fragment of the thermolysis of precursor 1, decreased
with increasing temperature, while only two signals, at
124.1 and 45.0 u, showed a significant increase of intensity.
The lower mass at 45.0 u fits to Me2NH (calculated mass of
45.06 u) and the one at higher mass fits to the carbodiimide 4
(calculated mass of 124.10 u, Chart 2). These data indicate
that the thermolysis product 3 eliminates Me2NH to give the
carbodiimide 4. This interpretation is consistent with the
decomposition of the protonated ligand 2 (Figure 5A). The
signal for 2þ dropped off, whereas signals at 126.1 and 45.0 u
rose with increasing temperatures, showing that 2 eliminates
Me2NH to give the well-known CDI (Chart 2). As species 3
is also produced in the thermolysis of 2, its signal at 171.2 u
decreased (Figure 5). The intensity for the signal of CDI
(126.1 u) varies over the temperature range but does not
show a trend, which might be due to its overall low intensity.
The TOF MS data and their interpretation are consistent
with MI FTIR data obtained from high temperature ther-
molyses. For example, in the thermolysis of 2 at 700 °C we
unambiguously identified HNMe2 and CDI, trapped in
solid argon (Supporting Information, Figure S6). In the
thermolysis of 1, HNMe2 was identified, but the unknown
carbodiimide 4 could not be unambiguously identified, as a
required IR spectrum for comparison is unknown in the
literature. However, an intense and structured IR band with
a maximum at 2127.7 cm-1 clearly indicates the presence of
a carbodiimide. Moreover, an IR band in the typical range
of CdC valence vibrations at 1649.2 cm-1 supports the
proposed structure of 4 (see Chart 2). Frequency calcula-
tions on 4 at the BLYP/6-31G(d) level of theory give a CdC
stretch at 1639.0 cm-1 and a NdCdN asymmetric stretch at
Figure 5. Temperature dependences of selected ions of the thermolysis
at high temperatures of 1 (spectrum B) and 2 (spectrum A) [m/z = 171.2
(2þ), 169.2 (3þ), 126.2 (4þ), 124.1 (CDIþ) 45.0 (HNMe2þ) u]. The
intensities of masses at 169.2 and 124.1 u were negligible in spectrum A
(counts between 2 and 22 at sweep rates of 8 ꢀ 106).
Chart 2. Carbodiimides
MS, MI FTIR spectroscopy, and DFT calculations. The
experimental results showed that precursor 1 fragments
above oven temperatures of 150 °C to release the oxidized
guanidine Me2NC(=NiPr)(N=CMe2) (3). In addition to 3,
the formation of H2 was detected by TOF MS. The key
intermediate 3 was synthesized independently in this study
and through comparison of FTIR spectra of matrix-isolated
3 with those from thermolysis experiment with copper pre-
cursor 1, the formation of the oxidized guanidine 3 as a
product of the thermolysis was proven. The structure of the
major IR bands could be rationalized with the help of DFT
calculations, which showed that species 3 exists as a pre-
domination of the trans isomer. These findings can be
explained by assuming a β-hydrogen elimination of the
ligands in precursor 1 takes place. DFT calculations indi-
cated that the two decomposition mechanisms result from the
same parent species (A1), which rationalizes the fact that
carbodiimide deinsertion is seen at lower temperatures in the
2164.7 cm-1
.
Summary and Conclusions
The fragmentation of the dimeric copper(I) precursor
[Me2NC(NiPr)2Cu]2 (1) has been investigated with TOF

2850 Inorganic Chemistry, Vol. 49, No. 6, 2010
Coyle et al.
solution phase, but β-hydrogen elimination is favored in
higher temperature, gas-phase reactions.
equipped with one roughing vacuum pump (Varian DS 602)
and three turbo vacuum pumps, and a reflectron TOF spectro-
€
meter with electron impact ionization (Stefan Kaesdorf, Gerate
fur Forschung and Industrie; time resolution under favorable
When higher temperatures were explored, it became evi-
dent that both species 2 and species 3 underwent deinsertion
to form free amine and either CDI or 4, respectively. This
important observation indicates that CDI detected at high
temperatures can result from deinsertion from a thermolysis
product, and not necessarily from the precursor itself.
€
conditions T/ΔT = 10000). For more details please see reference
12 (Note that the distance between the end of the heated oven
tube to the orifice of the skimmer was shortened from 12 to
8 mm.). In a typical experiment, the sample container was kept
at constant temperature (1: 90-92 °C; 2: 0 °C) while a flow of
argon was conducted over the sample. To get acceptable signal-
to-noise ratios, the container temperature for 1 needed for the
TOF MS measurements was slightly higher than those for MI
FTIR spectroscopy (see above). For all measurements electron
impact of 70 eV and 1 to 20 million sweeps per oven temperature
were applied. The signal of the carrier gas argon at 40 u was
reduced in intensity by using a mass filter at 4.35 kV (delay:
3.1000 μs; duration: 0.9000 μs). This prevented the MCP
detector from exceeding its detection limit. Calibration of flight
times was done within every new series of measurements using
Xe as an additional standard. Calibrations were done with the
following masses: 129Xeþ (128.90478 u), 129Xe2þ (64.45239 u),
40Arþ (39.96230 u), and H216Oþ (18.01056 u). As the calibration
masses did not span the entire range of measured ions from 1, the
experimental masses were usually within the range of (0.1 u of
the calculated ones.
Calculated masses of selected ions (intensities in parentheses):
1þ (C18H40Cu2N6þ): m/z = 466.19 (100.0%), 467.19 (19.5%),
467.19 (2.2%), 468.19 (89.1%), 468.20 (1.8%), 469.19 (17.4%),
469.19 (2.0%), 470.19 (19.9%), 470.20 (1.6%), 471.19 (3.9%).
DFT Calculations. All calculations were done using the
GAUSSIAN 03 program package.24,25 The mechanism25,26 of
thermolysis of 1 was studied with the B327LYP28 functional and
the 6-31þG(d,p) basis set augmented with dispersion correcting
potentials (DCPs).29 DCPs are atom-centered effective core
potentials which are employed to better account for long-range
dynamical correlation effects (i.e., dispersion interactions).
Coefficients for DCPs have been optimized30 for a wide variety
of functionals and basis sets and can be implemented into
quantum chemical software simply by appending the input file.
All thermodynamic values discussed in the text are calculated
free energies at 260 °C. To help interpret the IR spectra acquired,
molecular vibrations were studied at the BLYP/6-31G(d) level
of theory. The BLYP functional was used rather than B3LYP as
it is known that unscaled frequencies obtained from BLYP are
more accurate than those from B3LYP.31
Experimental Section
General Remarks. Complex 1 was synthesized via known
procedures11 and purified by sublimation prior to use (1: 70-
77 °C, approximately 10-2-10-3 mbar). Compound 2 and was
synthesized via known procedures22 and purified via flask-to-
flask condensation at ambient temperatures (approximately
10-2-10-3 mbar).
The matrix isolation and the TOF MS apparatus were
separate instruments each equipped with its own thermolysis
oven, both identical in construction. Both instruments used the
same carrier gas supply and were in close proximity so that the
heatable container, loaded with the compound of interest, could
be moved from one to the other instrument without exposing the
sample to air. Both instruments used the same thermolysis oven
temperature controller. The facility was constructed in this way
so that a series of experiments with one sample batch could be
performed using one method, for example, MI FTIR spectro-
scopy, followed by a series of experiments using the other
method, for example, TOF MS, and still being able to keep
the experimental conditions similar in both consecutive series of
experiments.
Each thermolysis oven was equipped with an Al2O3 tube
(OD = 4 mm) with two parallel, inner canals (ID = 1 mm). The
overall length of the Al2O3 tube was 35 mm with the last 10 mm
heated with a tungsten wire, coiled around the outside, and
coated with ZrO2 cement. One of the inner canals was equipped
with a thermocouple (Thermocoax: NiCrSi/NiSi), while the
second canal was used to let a carrier gas/sample mixture pass
through. With this setup, reliable thermolysis temperatures at
the inside of the thermolysis tube could be measured without
exposing the thermocouple to the gaseous sample. A schematic
drawing of the thermolysis oven can be found in reference 26.
A thermolysis series with compound 1 was done in the
temperature range between 100 and 800 °C for both MI FTIR
spectroscopy and TOF MS. The thermolysis of compound 2 was
investigated in the temperatures range 200 to 800 °C for both MI
FTIR spectroscopy and TOF MS.
Acknowledgment. We thank the Natural Sciences and
Engineering Research Council of Canada (NSERC Discovery
Grants, J.M. and S.T.B.) for their generous support. We thank
the Canada Foundation for Innovation (CFI) and the govern-
ment of Saskatchewan for funding of the MI FTIR spectrometer
and the TOF MS facility.
MI FTIR Spectroscopy. The matrix apparatus consisted of a
vacuum line (Varian TV 551 Navigator; Varian DS 402) and a
Sumitomo CSW-71 cryogenic closed cycle system (Sumitomo
Heavy Industries. Ltd.) including a Sumitomo RDK-415D 4K
cold head fitted with CsI windows (vacuum shroud design and
fabrication by Janis Research Company, Inc.). The sample was
loaded into a heatable container connected to the thermolysis
oven. In a typical experiment, the evacuated container, loaded
with the compound of interest, was kept at constant temperature
(1: 86 °C; 2: 0 °C) while a flow of argon (1.25 sccm; Praxair 6.0) was
conducted over the sample (mass-flow controller: MKS Type
1179A). The resulting gaseous mixture was passed through a
thermolysis oven for 30-90 min and deposited onto a CsI window
cooled to 20 K. The distance between the end of the heated oven
tube and the CsI window was 54 mm. The so-prepared matrixes
were cooled to 4 K for FTIR measurements (Bruker Tensor 27
FTIR; 200 to 5000 cm-1; resolution of 1 cm-1).
Supporting Information Available: FTIR spectra of matrix-
isolated species 1, 2, and 3, and FTIR spectra of thermolysis of 1
and 2 (Figures S1-S6; range 4000-220 cm-1); Cartesian co-
ordinates for the trans-3 species, as well as the species in Figure 4
(Tables S1-S11). This material is available free of charge via the
Internet at http://pubs.acs.org.
(25) Farrugia, L. J. J. Appl. Crystallogr. 1997, 30, 565.
€
(26) Muller, J.; Wittig, B.; Sternkicker, H.; Bendix, S. J. Phys. IV 2001,
11, 17–22.
(27) Becke, A. D. J. Chem. Phys. 1993, 98, 5648–5652.
(28) Lee, C.; Yang, W.; Parr, R. G. Phys. Rev. B 1988, 37, 785–789.
(29) DiLabio, G. A. Chem. Phys. Lett. 2008, 455, 348–353.
(30) Mackie, I. D.; DiLabio, G. A. J. Phys. Chem. A 2008, 112, 10968–
10976.
TOF MS. The apparatus consisted of a custom designed
€
vacuum chamber (designed by Stefan Kaesdorf, Gerate fur
Forschung and Industrie; built by Johnsen Ultravac Inc.),
€
(31) Rauhut, G.; Pulay, P. J. Phys. Chem. 1995, 99, 3093–3100.