Preventing thermolysis: Precursor design for volatile copper compounds

Article abstract of DOI:10.1039/c2cc35415b

A copper(i) iminopyrrolidinate was synthesized and evaluated by thermal gravimetric analysis (TGA), solution based ¹H NMR studies and surface chemistry to determine its thermal stability and decomposition mechanism. Copper(i) tert-butyl-imino-2

Full text of DOI:10.1039/c2cc35415b

View Article Online / Journal Homepage / Table of Contents for this issue
Dynamic Article Links
ChemComm
Cite this: Chem. Commun., 2012, 48, 10440–10442
www.rsc.org/chemcomm
COMMUNICATION
Preventing thermolysis: precursor design for volatile copper compoundsw
Jason P. Coyle,a Agnieszka Kurek,a Peter J. Pallister,a Eric R. Sirianni,b Glenn P. A. Yapb
a
´
and Sean T. Barry*
Received 26th July 2012, Accepted 1st September 2012
DOI: 10.1039/c2cc35415b
A copper(I) iminopyrrolidinate was synthesized and evaluated by
thermal gravimetric analysis (TGA), solution based 1H NMR
studies and surface chemistry to determine its thermal stability
and decomposition mechanism. Copper(I) tert-butyl-imino-2,2-
dimethylpyrrolidinate (1) demonstrated superior thermal stability
and showed negligible decomposition in TGA experiments up to
300 8C as well as no decomposition in solutions at 165 8C over
3 weeks.
Atomic layer deposition (ALD) is a deposition process that relies
on successive self-limiting reactions between an entrained precursor
gas and a solid substrate. Progress in ALD has relied on novel
precursor chemistry to allow processes for thin film deposition to
be defined for a variety of target films and process conditions.1 In
many cases, the ‘‘design’’ of the precursor entails selecting a new
ligand and modifying it to provide volatility and thermal stability
to a precursor. However, such ligands are often not tested and
redesigned to improve on their original performance. More
often, a new ligand is employed to overcome the failings of a
previous ligand.
Fig. 1 Thermal decomposition pathways for copper(I) amidinates.
Where R can be an alkyl or an amine.
There have been several designs for precursors for the
deposition of copper metal, a key element in the fabrication
of microelectronic circuits.2 One influential ligand used in
copper ALD is the amidinate family, which includes both
amidinates and guanidinates,3 including copper(I) alkylamidinates4
and copper(I) guanidinates5 (Fig. 1). These ligands suffer from two
main thermal decomposition pathways: loss of carbodiimide
(herein called ‘‘deinsertion’’) and b-hydrogen abstraction.6
These low temperature decompositions lead to the formation
of copper metal, with the ligand acting as a reducing agent.
The redesign of the amidinate ligand was undertaken to prevent
these thermolyses. By using a judiciously chosen pyrrolidinate,
both thermal pathways could be eliminated (Fig. 2).
Fig. 2 The synthetic procedure for producing the tert-butyl-imino-
2,2-dimethylpyrrolidine ligand.
position as well as replacing the b-hydrogens on the ring with
methyl groups. The ligand was made in high yield, and the
copper(I) pyrrolidinate (1) was formed through simple salt meta-
thesis. Recently, trimeric heteroleptic copper(I) guanidinates were
isolated from the attempted synthesis of [Me2NC(NtBu)2Cu]2.7
Compared to this trimer, the ring of the iminopyrrolidinate
reduced the steric bulk to permit the isolation of 1 as the
more typical dimer, as seen by single crystal X-ray diffraction.
The structure of 1 is consistent with previously known amidinate4
and guanidinate5 systems (Fig. 3, Table 1). The compound has
a Cu–Cu distance of B2.5 A, with an approximately linear
coordination geometry around each copper atom. The metallo-
cycle core is planar with equivalent N–C bonds (1.34 A)
suggesting delocalization.
The deinsertion of carbodiimide (CDI) was prevented by
tethering the alkyl group of the chelate position to the exocyclic
group through use of a pyrrole ring. b-hydrogen abstraction
was eliminated by using a tert-butyl group in the chelate
This compound showed excellent volatility: thermogravi-
metric analysis (TGA) showed it to be volatile as low as 165 1C
at atmospheric pressure, with a residual mass of o2% of
the sample mass. Compound 1 could also be quantitatively
sublimed at 130 1C at a pressure of 20 mtorr. This indicated
that the compound cleanly volatilizes at these temperatures
with no thermal decomposition.
a Department of Chemistry, Carleton University, 1125 Colonel By
Drive, Ottawa, K1S 5B6, Canada. E-mail: sean_barry@carleton.ca;
Tel: +1 613-520-2600
b Department of Chemistry & Biochemistry, University of Delaware,
Newark, DE 19716, USA
w Electronic supplementary information (ESI) available: Experimental
and CIF for 1. CCDC 893889. For ESI and crystallographic data in
CIF or other electronic format see DOI: 10.1039/c2cc35415b
c
10440 Chem. Commun., 2012, 48, 10440–10442
This journal is The Royal Society of Chemistry 2012

View Article Online
residual mass of o2% to 9.0% of the initial mass as the
sample mass was increased from 10 mg to 63 mg. It has been
previously demonstrated that 2 undergoes CDI deinsertion at
lower temperatures, a process which has been prevented
through this ligand design.4
This contrast in thermal behaviour was highlighted by the
decomposition kinetics of these two compounds at 165 1C.
Compound 2 was found to decompose following first order
decomposition kinetics with a calculated half-life of 33.8 hours
while 1 showed no appreciable decomposition over 20 days at
165 1C.
To test compound 1 for vapour phase stability, it was heated
in a bubbler at 160 1C and the evolved vapour was entrained
over high surface area silica for 17 hours to saturate the silica
with a chemisorbed monolayer. Characterization was under-
taken using 1H/13C cross-polarized magic-angle spinning solid
state NMR, and there was only physisorbed species detected
up to 200 1C. At a deposition temperature of 275 1C, solid
state 13C NMR showed signals attributable to the copper
precursor (Fig. 5a, top trace). These signals were similar to
the high resolution 13C NMR of 1 in deuterated benzene
(Fig. 5a, bottom trace). There was excellent corroboration
between the solution NMR and the solid-state NMR. Although
this suggested that the precursor chemisorbed ‘‘whole’’ at the
surface, literature suggests that the alkyl group at the chelate
position might be lost at high temperatures.7 The line width of
the solid-state NMR did not permit unambiguous identification
of the tert-butyl moiety, so another experiment was needed.
The silica with the deposited monolayer was rinsed with
D2O to etch off the surface species, and a high resolution
1H NMR was collected (Fig. 5b, top trace), showing a singular
surface species. Although the chemical shifts were different
because of the necessity of using different solvents, the inte-
gration of the peaks for the surface species showed loss of the
tert-butyl group: integrations of 2.1, 2.0, and 5.7 for the peaks
at 2.94 ppm, 2.07 ppm, and 1.36 ppm respectively. The
1H NMR of the ligand in deuterated chloroform clearly
showed a singlet for the tert-butyl group (Fig. 5b, bottom
trace) with integration ratios of 2.0, 2.0, 9.4, and 6.1 for the
peaks at 2.42 ppm, 1.57 ppm, 1.32 ppm, and 1.18 ppm
respectively. As well, 29Si SS-NMR of the silica showed no
additional signals for silicon beyond what was detected for
pure silica, suggesting nucleation did not occur at a silicon
atom. This is reasonable, considering that the most likely
nucleation point for this precursor on silica surface is at the
oxygen of a silanol group. Given these data, nucleation likely
occurs through the copper atom to a surface oxygen, and
subsequently undergoes butene elimination (Fig. 5c). This
elimination occurs at 300 1C higher in temperature than shown
for the copper amidinate,8 and highlights the excellent thermal
stability designed into this ligand.
Fig. 3 The structure of 1, with hydrogens removed for clarity, and
the thermal ellipsoids shown at 30%.
Table 1 The selected bond lengths and angles for 1
Selected bond
Length (A)
Cu–Cu
Cu–N1
Cu–N2
N1–C4
N2–C4
2.47
1.87
1.89
1.33
1.33
Selected angle
Angle (1)
N1–Cu–N2
N1–C4–N2
175.6
122.5
More importantly, a thermal stress test designed for this
study showed 1 to be very resistant to thermal decomposition
at higher temperatures. In this stress test, the compound is
measured by TGA using the same temperature ramp rate
(10 1C minꢀ1) but with different sample masses (Fig. 4).
Due to the kinetics of volatilization, as sample mass is
increased, more sample becomes exposed to higher temperatures.
This valuable test can be used to gauge the behaviour of a
compound with respect to thermal handling during a deposition
process. As seen in Fig. 4, varying the sample weights of 1 gave
no deviation in residual mass, whereas performing the same
stress test on the similar N0,N00-diisopropyl-N-dimethylguani-
dinatocopper5 (2) compound showed a marked increase of
In summary, a new volatile precursor for copper metal deposi-
tion was synthesized with its ligand designed to prevent two
known, low temperature decomposition pathways. The precursor
1 showed excellent stability both in the solid state, in solution, and
as a vapour species. Surface analysis suggested that the precursor
undergoes an alkene elimination from the chelating nitrogen atom
during chemisorption on silica at 275 1C. This novel precursor
suggests that a family of iminopyrrolidinates of copper might be
Fig. 4 Thermal stress test of compounds (a) 1 and (b) 2, showing
effect on residual mass as starting mass was increased.
c
This journal is The Royal Society of Chemistry 2012
Chem. Commun., 2012, 48, 10440–10442 10441

View Article Online
Fig. 5 Reactivity of 1 with silica: (a) shows the SS-13C NMR of silica after deposition of 1 at 275 1C (top trace) compared to the HR-13C NMR of
1 in deuterated benzene (bottom trace). (b) Shows the HR-1H NMR of the surface species etched from the silica with D2O (in D2O; top trace)
compared to the HR-1H NMR of 1 in CDCl3 (bottom trace). (c) Shows the proposed nucleation and chemisorption of 1 at a silanol group on a
silica surface.
synthesized with various b-hydrogen atoms, which would allow
tuning of the precursors thermal reactivity.
4 Z. Li, A. Rahtu and A. R. G. Gordon, J. Electrochem. Soc., 2006,
153, C787.
5 J. P. Coyle, W. H. Monillas, G. P. A. Yap and S. T. Barry, Inorg.
Chem., 2008, 47, 683.
6 J. P. Coyle, P. A. Johnson, G. A. DiLabio, S. T. Barry and
J. Muller, Inorg. Chem., 2010, 49, 2844.
7 A. M. Willcocks, T. P. Robinson, C. Roche, T. Pugh,
S. P. Richards, A. J. Kingsley, J. P. Lowe and A. L. Johnson, Inorg.
Chem., 2012, 51, 246.
Notes and references
1 (a) M. Ritala and M. Leskela, in Handbook of Thin Film Materials,
¨
ed. H. S. Nalwa, Vol. 1, Academic Press, San Diego, CA, 2001,
pp. 103–156; (b) M. Ritala and M. Leskela, Thin Solid Films, 2002,
409, 138.
¨
8 Q. Ma, H. Guo, R. G. Gordon and F. Zaera, Chem. Mater., 2011,
23, 3325.
9 U. Beckmann, E. Eichberger, M. Lindner, M. Bongartz and
P. C. Kunz, Eur. J. Org. Chem., 2008, 4139.
2 (a) H. Kim, J. Vac. Sci. Technol. B, 2003, 21, 2231; (b) P. Dopplet,
Coord. Chem. Rev., 1998, 178–180, 1785.
3 F. T. Edelmann, Chem. Soc. Rev., 2012, DOI: 10.1039/
C2CS35180C.
c
10442 Chem. Commun., 2012, 48, 10440–10442
This journal is The Royal Society of Chemistry 2012