Synthesis and characterization of copper(l) amidinates as precursors for atomic layer deposition (ALD) of copper metal

Article abstract of DOI:10.1021/ic048492u

A series of copper(I) amidinates of the general type [(R NC(R)NR JCu] ₂ (R and R = n-propyl, isopropyl, n-butyl, isobutyl, sec-butyl, tert-butyl; R = methyl, n-butyl) have been synthesized and characterized. These compounds are planar dimers, bridged by nearly linear N-Cu-N bonds. Their properties (volatility, low melting point, high thermal stability, and self-limited surface reactivity) are well-suited for atomic layer deposition (ALD) of copper metal films that are pure, highly conductive, conformal, and strongly adherent to substrates.

Full text of DOI:10.1021/ic048492u

Inorg. Chem. 2005, 44, 1728−1735
Synthesis and Characterization of Copper(I) Amidinates as Precursors
for Atomic Layer Deposition (ALD) of Copper Metal
Zhengwen Li,† Sea´n T. Barry,‡ and Roy G. Gordon*,†
Department of Chemistry and Chemical Biology, HarVard UniVersity,
Cambridge, Massachusetts 02138, and Department of Chemistry,
Carleton UniVersity, 1125 Colonel By DriVe, Ottawa, Ontario, Canada K1S 5B6
Received October 27, 2004
A series of copper(I) amidinates of the general type [(R
isobutyl, sec-butyl, tert-butyl; R methyl, n-butyl) have been synthesized and characterized. These compounds
are planar dimers, bridged by nearly linear N Cu N bonds. Their properties (volatility, low melting point, high
′NC(R)NR′′)Cu]2 (R′ and R′′ ) n-propyl, isopropyl, n-butyl,
)
−
−
thermal stability, and self-limited surface reactivity) are well-suited for atomic layer deposition (ALD) of copper
metal films that are pure, highly conductive, conformal, and strongly adherent to substrates.
Introduction
formality, and high aspect ratio, thin films made by ALD
might be able to satisfy the future requirements in micro-
electronics metallization.9 ALD has been demonstrated to
make barrier layers such as WN,10 Co adhesion layers,11 and
thin copper films as seed layers for subsequent electrochemi-
cal deposition of copper.5
Atomic layer deposition (ALD) is a method currently used
for making highly conformal thin films in computer chips,
displays, magnetic devices for information storage, and
optical communication systems.1,2 ALD processes have been
investigated intensively in recent years, particularly for metal
oxides,3 metal nitrides, and pure metals.4 Metal amidinates
have been found to be suitable as precursors for the ALD of
transition metals and metal oxides.5 Certain amidinates of
Ti, V, Mn, Fe, Co, Ni, Cu, and La have the required
volatility, high thermal stability, and high reactivity required
for successful use in ALD.6
Copper is replacing aluminum as the interconnect metal
in microelectronics because of its relatively low resistivity
and excellent electromigration properties. Much attention has
focused on the deposition of copper in increasingly narrow
trenches.7,8 Because of their excellent step coverage, con-
Despite the promising properties of ALD films, a lack of
suitable copper precursors has hindered the use of ALD for
copper metallization or other applications. All of the copper
compounds previously proposed as precursors (summarized
in Table 1) have one or more shortcomings. Copper halides,
such as copper(I) chloride, Cu3Cl3, can be reduced with
molecular hydrogen,12 but their high vaporization and
deposition temperatures (around 400 °C) resulted in ag-
glomerated, discontinuous copper films. Also, the hydro-
chloric acid byproduct from this process can etch substrates
and corrode deposition equipment. ALD of copper from
copper(II) 2,2,6,6-tetramethylheptanedionate, Cu(thd)2, suc-
ceeded only on platinum/palladium substrates.13,14 Fluorina-
tion of â-diketonates [e.g., copper(II) hexafluoroacetyl-
* To whom correspondence should be addressed. E-mail: Gordon@
chemistry.harvard.edu.
† Harvard University.
‡ Carleton University.
(1) Rosenberg, R.; Edelstein, D. C.; Hu, C.-K.; Rodbell, K. P. Annu. ReV.
Mater. Sci. 2000, 30, 229-262.
(9) International Technology Roadmap for Semiconductors 2003 Up-
date: http://public.itrs.net/Files/2003ITRS/Home2003.htm.
(10) Becker, J. S.; Gordon, R. G. Appl. Phys. Lett. 2003, 82, 2239-2241.
(11) Gordon, R. G.; de Rouffignac, P.; Li, Z. Presented at the AVS Topical
Conference on Atomic Layer Deposition-ALD2004, Helsinki, Finland,
Aug 16-18, 2004.
(2) Ritala, M.; Leskela¨, M. In Handbook of Thin Film Materials; Nalwa,
H. S., Ed.; Academic Press: New York, 2001; Vol. 1, pp 103-159.
(3) Gordon, R. G.; Becker, J.; Hausmann, D.; Suh, S. Chem. Mater. 2001,
13, 2463-2464.
(4) Kim, H. J. Vac. Sci. Technol. B 2003, 21, 2231-2261.
(5) Lim, B. S.; Rahtu, A.; Gordon, R. G. Nat. Mater. 2003, 2, 749-754.
(6) Lim, B. S.; Rahtu, A.; Park, J.; Gordon, R. G. Inorg. Chem. 2003, 42,
7951-7958.
(7) Niskanen, A.; Rahtu, A.; Sajavaara, T.; Arstila, K.; Ritala, M.; Leskela,
M. J. Electrochem. Soc. 2005, 152, G25-G28.
(8) Leskela¨, M.; Ritala, M. Angew. Chem., Int. Ed. 2003, 42, 5548-5554.
(12) To¨rndahl, T.; Ottosson, M.; Carlsson, J.-O. Thin Solid Films 2004,
458, 129-136.
(13) Martensson, P.; Carlsson, J.-O. J. Electrochem. Soc. 1998, 145, 2926-
2931.
(14) Martensson, P.; Juppo, M.; Ritala, M.; Leskela, M.; Carlsson, J.-O. J.
Vac. Sci. Technol. B 1999, 17, 2122-2128.
1728 Inorganic Chemistry, Vol. 44, No. 6, 2005
10.1021/ic048492u CCC: $30.25
© 2005 American Chemical Society
Published on Web 02/03/2005

Cu(I) Amidinates as Precursors for ALD of Copper Metal
Table 1. Melting Points and Vapor Pressures of Various Copper
Precursors Used in ALD of Copper
and hexanes were purified by being passed through a drying column
and a deoxygenating column using an Innovative Technology
solvent purification system. Solvents were then stored over 4-Å
molecular sieves. Alkylamines (RNH2, R ) n-propyl, isopropyl,
n-butyl, isobutyl, sec-butyl, diethyl) were distilled from sodium or
KOH prior to use. Methyllithium, butyllithium, valeronitrile, 1,3-
diisopropylcarbodiimide, 1,3-di-tert-butylcarbodiimide, N-bromo-
succinimide (NBS), tert-butyl isothiocyanate, and CuCl were
purchased and used as received from Aldrich Chemical Co. sec-
Butyl isothiocyanate was used as received from Alfa Aesar Co.
Dichloromethane was used as received from EMD Chemicals Inc.
Physical Measurements. 1H and 13C NMR spectra were obtained
on a Bruker AM-500 spectrometer. Elemental analysis was
performed by Desert Analytics Laboratory, Tucson, AZ. Thermal
gravimetric analysis (TGA) and differential scanning calorimetry
(DSC) data were obtained on a Netzsch STA 449C instrument in
nitrogen atmosphere inside a drybox. The measurements were done
in open aluminum crucibles with 10-20 mg of sample; the
temperature ramp rate was 10 K/min. Melting points were measured
in sealed capillaries using a Mel-Temp II apparatus and are
compared with the results from TGA/DSC. Molecular weight
determinations by the freezing point depression method in solution
were done by dissolving 0.3-0.4 g of compound in 3.0-4.0 g of
recrystallized p-xylene. The temperature was recorded electronically
from a thermocouple every 10 s for at least 2 min, and the
measurements were repeated at least three times to ensure reproduc-
ibility. Vapor pressures were estimated using a thermocouple
pressure gauge during sublimation or distillation of purified
compounds. In the case of compound 9, the vapor pressure was
also corroborated by transport measurements of a known mass from
a source bubbler.
vapor pressure/
melting
temperature
copper precursor
Cu3Cl3
point (°C)
(Torr/°C)
ref(s)
12
13, 14
16, 17
5, 6
this work
this work
430
-
-
0.25/60
0.05/70
0.1/85
0.05/55
Cu(thd)2
198 (dec)
130-134
147
77
65
Cu(hfac)2.xH2O
[Cu(iPr-Me-amd)]2, 3
[Cu(sBu-Me-amd)]2, 9
[Cu(nPr-Me-amd)]2, 1
acetonate, Cu(hfac)2] increases their volatility;15 ALD of
copper was demonstrated with Cu(hfac)2 using formaldehyde
as a reducing agent.16,17 However, fluorine impurities in these
copper films cause poor adhesion to substrates such as
tantalum and reliability problems in microelectronic devices.
Copper(I) N,N′-diisopropylacetamidinate, [Cu(iPr-amd)]2,
contains no halogens, and is volatile, thermally stable, and
reactive to molecular hydrogen at low temperatures
(<200 °C).5
All of these copper precursors previously used for ALD
have melting points above the sublimation temperatures
typically used for operating ALD processes. Thus, their solid
sublimation rates are generally nonreproducible, because they
depend on the particle size, which changes during the
sublimation process. The purification and handling of solid
precursors is also inconvenient. One of the advantages of
copper(I) amidinates is that one can vary the six pendant
alkyl groups to adjust the properties of the precursors. In
this paper, we report a family of 11 previously unknown
copper(I) amidinate compounds with various different pen-
dant alkyl groups. These new amidinato copper precursors
provide at least four benefits compared with other types of
ALD precursors: Many of these amidinate complexes have
lower melting points, such as copper(I) N,N′-di-sec-butyl-
acetamidinate ([Cu(sBu-Me-amd)]2, 9) and copper(I) N,N′-
di-n-propylacetamidinate ([Cu(nPr-Me-amd)]2, 1). These
compounds (melting points around 60-80 °C) evaporate
from their thermally stable liquid states (around 100 °C),
thereby providing very reproducible sources of vapor for
ALD. Their second advantage is that the deposition tem-
peratures are lower (around 180 °C) than has been achieved
by any of the previous precursors, providing smoother film
morphology because less agglomeration occurs at lower
temperatures. A third advantage is that the copper films are
pure [impurities below the detection limit of Rutherford
backscattering spectroscopy (RBS), <1 atom %]; thus low
electrical resistivity can be achieved. Finally, unlike copper
halides, the byproduct vapors do not corrode substrates or
deposition equipment and do not cause poor adhesion of the
films.
Synthesis of Symmetric Amidine Ligands. N,N′-Dialkyl-
acetamidines were synthesized as illustrated in Scheme 1, in a
manner similar to that reported in the literature.18 A solution of
50:50:1 mole ratio of alkylamine/acetonitrile/La(CF3SO3)3 was
refluxed for 18 h at around 80 °C under nitrogen atmosphere. The
unreacted starting materials (mainly acetonitrile) were then removed
under partial vacuum at room temperature. The byproduct 2,4,6-
trimethyl-1,3,5-triazine [C6N3H9, white crystals, melting point 55
°C, 1H NMR (C6D6, 23 °C, 500 MHz) 2.37 (s)] was removed from
the reaction mixture by sublimation at 40 °C at around 0.18 Torr.
The colorless liquid amidine was then isolated in 50-80% yield
by distilling the remaining liquid at approximately 90 °C and 0.06-
0.18 Torr. The N,N′-dialkylvaleroamidines were prepared similarly
to the N,N′-dialkylacetamidines, except that the original solution
was refluxed for a longer time (3 days) to increase the yield.
Preparation of Symmetric Copper(I) Amidinate Compounds.
For symmetric compounds (those having the same alkyl group on
all nitrogens), the free symmetric amidines were synthesized by
the above procedure and then deprotonated with methyllithium. The
resulting lithium amidinates were then reacted with copper chloride.
Copper(I) N,N′-Di-n-propylacetamidinate [Cu(nPr-Me-amd)]2
(1). 1,3-Di-n-propylacetamidine was synthesized by lanthanide-
catalyzed reaction of acetonitrile and propylamine with a yield of
60%, according to Scheme 1. A solution of methyllithium (1.6 M
in Et2O, 12.5 mL, 0.02 mol) was added dropwise by syringe into
a solution of 1,3-di-n-propylacetamidine in 30 mL of Et2O at -30
°C. The solution was stirred for 2 h at room temperature. The
resulting solution was then transferred to a copper monochloride
Experimental Section
General Considerations. All reactions and manipulations were
done under a pure N2 atmosphere using either standard Schlenk
techniques or a glovebox. Ether, tetrahydrofuran (THF), acetonitrile,
(15) Dopplet, P. Coord. Chem. ReV. 1998, 178, 1785-1809.
(16) Solanki, R.; Pathangey, B. Electrochem. Solid State Lett. 2000, 3, 479-
480.
(18) Forsberg, J. H.; Spaziano, V. T.; Balasubramanian, T. M.; Liu, G. K.;
Kinsley, S. A.; Duckworth, C. A.; Poteruca, J. J.; Brown, P. S.; Miller,
J. L. J. Org. Chem. 1987, 52, 1017-1021.
(17) Huo, J.; Solanki, R.; McAndrew, J. J. Mater. Res. 2002, 17, 2394-
2398.
Inorganic Chemistry, Vol. 44, No. 6, 2005 1729

Li et al.
Scheme 1. Synthesis of Symmetric Amidines and the Corresponding Symmetric Copper(I) Amidinate Compounds
(1.98 g, 0.02 mol) solution in 30 mL of Et2O. The solution was
stirred for 18 h under nitrogen atmosphere. The solvent was
removed under reduced pressure, and the resulting solid was
extracted with hexanes (30 mL). The hexane extract was filtered
through a glass frit to afford a pale yellow solution. A saturated
solution was formed by removing part of the hexane under reduced
pressure. The saturated solution was cooled to -30 °C and afforded
colorless crystals (3.62 g, 83%). Pure white compound was obtained
through sublimation at 50 °C (0.05 mmHg). mp: 65 °C. 1H NMR
(C6D6, 25 °C, ppm): 3.06 (t, 4H), 1.70 (sextet, 4H), 1.60 (s, 3H),
0.96 (t, 6H). 13C NMR (C6D6, 25 °C, ppm): 173.6, 53.2, 27.49,
13.98, 12.05. Anal. Calcd for C16H34N4Cu2: C, 46.92; H, 8.37; N,
13.68. Found: C, 46.70; H, 8.20; N, 14.02.
C20H42N4Cu2: C, 51.59; H, 9.09; N, 12.03. Found: C, 51.81; H,
8.82; N, 12.28.
Copper N,N′-Di-n-butyl-2-butylamidinate [Cu(nBu-Bu-amd)]2
(6): Yellow viscous supercooled liquid at room temperature. After
being quenched in liquid nitrogen, it remained as a solid glass at
room temperature. It decomposed at 190 °C, 0.1 mmHg. 1H NMR
(C6D6, 25 °C, ppm): 3.25 (t, 4H), 2.28 (m, 2H), 1.77 (m, 4H),
1.46 (m, 6H, overlapping signals), 1.28 (m, 2H), 0.94 (t, 6H), 0.86
(t, 3H). 13C NMR (C6D6, 25 °C, ppm): 176.5, 50.2, 37.4, 29.7,
27.5, 23.4, 20.9, 14.5, 14.0. Anal. Calcd for C26H54N4Cu2: C, 56.80;
H, 9.90; N, 10.19. Found: C, 56.72; H, 9.69; N, 9.97.
Copper N,N′-Di-isobutylacetamidinate [Cu(iBu-Me-amd)]2
(7): Colorless crystals (94%). Sublimation: 90 °C, 0.2 mmHg.
mp: 114 °C. 1H NMR (C6D6, 25 °C, ppm): 2.98 (d, 4H), 1.94 (m,
2H), 1.62 (s, 3H), 1.02 (d, 12H). 13C NMR (C6D6, 25 °C, ppm):
173.7, 59.63, 32.71, 21.1, 14.4. Anal. Calcd for C20H42N4Cu2: C,
51.59; H, 9.09; N, 12.03. Found: C, 51.66; H, 8.82; N, 12.09.
Copper(I) N,N′-Di-n-propyl-2-butylamidinate [Cu(nPr-Bu-
amd)]2 (2). This compound was obtained in a fashion similar to
[Cu(nPr-Me-AMD)]2, 1. 1,3-Di-n-propyl-2-butylamidine was pre-
pared according to Scheme 1, similarly to 1,3-di-n-propylacet-
amidine, in 58% yield. Treatment of (nPr-butylamidinato)lithium
Copper N,N′-di-isobutyl-2-butylamidinate [Cu(iBu-Bu-amd)]2
n
[obtained from Pr-butylamidine (3.68 g, 0.02 mol) and a 1.6 M
1
(8): Pale yellow crystals (81%). mp: 111 °C. H NMR (C6D6, 25
solution of methyllithium in diethyl ether (12.5 mL, 0.02 mol)]
with CuCl (1.98 g, 0.02 mol) in ether solution (30 mL). Crystal-
lization afforded colorless crystals (3.75 g, 76%). An analytical
sample was obtained by distillation at 75 °C (0.05 mmHg). mp:
°C, ppm): 3.10 (d, 4H), 2.49 (m, 2H), 2.05 (m, 2H), 1.50 (m, 2H),
1.27 (m, 2H), 1.05 (d, 12H), 0.841 (t, 3H). 13C NMR (C6D6, 25
°C, ppm): 176.9, 58.9, 33.2, 29.6, 27.6, 23.4, 21.1, 14.0. Anal.
Calcd for C26H54N4Cu2: C, 56.80; H, 9.90; N, 10.19. Found: C,
56.60; H, 10.08; N, 10.09.
Copper N,N′-Di-sec-butylacetamidinate [Cu(sBu-Me-amd)]2
(9): Colorless crystals (90%). Vapor pressure: 85 °C, 0.1 mmHg
(by thermocouple vacuum gauge during distillation); 95 °C, 0.23
mmHg (by mass loss from a bubbler). mp: 77 °C. 1H NMR (C6D6,
25 °C, ppm): 3.16 (m, 2H), 1.71 (d, 3H), 1.69 (m, 2H), 1.52 (m,
2H), 1.19 (dd, 6H), 0.94 (m, 6H). 13C NMR (C6D6, 25 °C, ppm):
170.8, 55.0, 34.7, 25.2, 14.5, 11.6. (Note: equilibrium or isomeri-
zation.) Anal. Calcd for C20H42N4Cu2: C, 51.59; H, 9.09; N, 12.03.
Found: C, 51.66; H, 8.82; N, 12.09.
Copper N,N′-Di-sec-butyl-2-butylamidinate [Cu(sBu-Bu-
amd)]2 (10): Colorless crystals (14% after several recrystallizations).
Sublimation: 60 °C, 0.01 mmHg. mp: 71 °C. 1H NMR (C6D6, 25
°C, ppm): 3.25 (m, 2H), 2.28 (t, 2H), 1.73 (m, 2H), 1.54 (dm,
4H), 1.28 (m, 2H), 1.26, (dd, 6H), 0.98 (dt, 6H), 0.88 (dt, 3H).13C
NMR (C6D6, 25 °C, ppm): 174.3, 54.5, 34.9, 30.7, 28.0, 25.9, 23.4,
14.1, 11.9. Anal. Calcd for C26H54N4Cu2: C, 56.80; H, 9.90; N,
10.19. Found: C, 56.96; H, 9.65; N, 9.92.
1
60 °C. H NMR (C6D6, 25 °C, ppm): 3.19 (t, 4H), 2.26 (t, 2H),
1.77 (m, 4H), 1.49 (m, 2H), 1.26 (m, 2H), 0.98 (t, 6H), 0.84 (t,
3H). 13C NMR (C6D6, 25 °C, ppm): 186.5, 176.7, 52.4, 29.7, 28.0,
27.5, 23.4, 14.0, 12.0. Anal. Calcd for C22H46N4Cu2: C, 53.52; H,
9.39; N, 11.35. Found: C, 53.13; H, 9.07; N, 11.18.
All compounds 3-10 were obtained in the same way as
[Cu(nPr-Me-amd)]2, 1 unless described otherwise.
Copper N,N′-Di-isopropylacetamidinate [Cu(iPr-Me-amd)]2
(3): Colorless crystals (70%). Sublimation: 70 °C, 0.05 mmHg.
mp: 147 °C. 1H NMR (C6D6, 25 °C, ppm): 1.16 (d, 12H), 1.65 (s,
3H), 3.40 (m, 2H). 13C NMR (C6D6, 25 °C, ppm): 170.1, 48.8,
27.2, 13.6. Anal. Calcd for C16H34N4Cu2: C, 46.92; H, 8.37; N,
13.68. Found: C, 46.80; H, 8.29; N, 13.35.
Copper N,N′-Di-isopropyl-2-butylamidinate [Cu(iPr-Bu-
amd)]2 (4): Colorless crystals (74%). Sublimation: 75 °C, 0.04
mmHg. mp: 97 °C.1H NMR (C6D6, 25 °C, ppm): 3.19 (t, 4H),
2.26 (t, 2H), 1.77 (m, 4H), 1.49 (m, 2H), 1.26 (m, 2H), 0.98
(t, 6H), 0.84 (t, 3H). 13C NMR (C6D6, 25 °C, ppm): 186.5,
176.7, 52.4, 29.7, 28.0, 27.5, 23.4, 14.0, 12.0. Anal. Calcd for
C22H46N4Cu2: C, 53.52; H, 9.39; N, 11.35. Found: C, 55.80; H,
9.33; N, 12.06.
Synthesis of Asymmetric Carbodiimides and Copper Amid-
inates. As illustrated in Scheme 2, a thiourea solution was made
by the addition of 1.46 g (0.02 mol) of isobutylamine dissolved in
20 mL of dichloromethane to 2.30 g (0.02 mol) of sec-butyl
isothiocyanate (for compound 11) or tert-butyl isothiocyanate (for
compound 12) in 200 mL of dichloromethane at room temperature;
then 4.40 g (0.06 mol) of N,N-dimethylethylamine was added to
Copper N,N′-Di-n-butylacetamidinate [Cu(nBu-Me-amd)]2
(5): Dark brown crystals (84%). Distillation: 70 °C, 0.1 mmHg.
1
mp: 65 °C. H NMR (C6D6, 25 °C, ppm): 3.12 (t, 4H), 1.69 (m,
4H), 1.63 (s, 3H), 1.44 (m, 4H), 0.93 (t, 6H). 13C NMR (C6D6, 25
°C, ppm): 173.5, 51.0, 36.9, 21.0, 14.4, 14.0. Anal. Calcd for
1730 Inorganic Chemistry, Vol. 44, No. 6, 2005

Cu(I) Amidinates as Precursors for ALD of Copper Metal
Scheme 2. Synthesis of Unsymmetric Carbodiimides and the Corresponding Unsymmetric Copper(I) Amidinate Compounds
the solution, which had been cooled to -30 °C.19 N-Bromo-
succinimide (NBS; 3.54 g, 0.02 mol) was dissolved in 100 mL of
dichloromethane and stirred for 12 h to promote dissolution. Then
the NBS solution was added slowly to the thiourea/NMe2Et
solution.20 Undissolved NBS was washed into the reaction vessel
using minimal additional CH2Cl2. The mixture solution turned
orange after being stirred overnight. The pale yellowish powder
precipitate (sulfur) was filtered off and the volatiles were removed
from the filtrate under partial pressure. The carbodiimide was
isolated by distillation of the remaining liquid.
by least-squares method on F2 by SHELXL-97, incorporated in
SHELXTL-PC V 6.10.
The structure was solved in the space group C2/c (No. 15). All
non-hydrogen atoms were refined anisotropically. Hydrogens were
calculated by geometrical methods and refined as a riding model.
The ligand was modeled as 50% disordered at the chiral carbons.
The crystal used for the diffraction study showed no decomposition
during data collection. The drawing was done at 50% ellipsoids.
NMR Studies of Thermal Stability. Decomposition rates of
four of these copper amidinates were determined in deuterated
benzene solution. Each of the four compounds [Cu(iPr-Me-
amd)2]2, 3; [Cu(iPr-nBu-amd)2]2, 4; [Cu(sBu-Me-amd)2]2, 9; and [Cu-
(sBu-nBu-amd)2]2, 10, was dissolved in C6D6 NMR solvent. These
solutions were sealed in thick-walled glass NMR tubes and heated
to 190 °C in an oven. Periodically, the tubes were removed from
the oven and their proton NMR spectra recorded. The intensity of
the relatively isolated peak near chemical shift δ ) 3.25 ppm was
recorded relative to the proton impurity peak (δ ) 7.15) in the
C6D6 solvent taken as a reference.
The following two asymmetric copper amidinates were synthe-
sized according to Scheme 2, using asymmetric carbodiimides
prepared as described above. Methyllithium (1.6 M in Et2O, 6.25
mL, 0.02 mol) was added to a solution of the carbodiimide (3.76
g, 0.02 mol) in 30 mL of diethyl ether. The resulting solution of
lithium amidinate was added to CuCl (1.98 g, 0.02 mol) in 30 mL
of ether solution, and the product was isolated as described above
for 1.
Copper N-Isopropyl-N′-sec-butylacetamidinate [Cu(iPr-Me-
sBu-amd)]2 (11): Pale yellow crystals (80%). Sublimation: 72 °C,
1
Results and Discussion
0.05 mmHg. mp: 86 °C. H NMR (C6D6, 25 °C, ppm): 3.46 (m,
2H), 3.15 (m, 2H), 1.70 (s, 3H), 1.69 (m, 4H), 1.52 (m, 4H), 1.20
(m, 18H, overlapping signals), 0.93 (dt, 6H). 13C NMR (C6D6, 25
°C, ppm): 170.5, 55.0, 48.9, 34.6, 27.3, 25.3, 14.1, 11.8. Anal.
Calcd for C18H38N4Cu2: C, 49.40; H, 8.75; N, 12.80. Found: C,
49.42; H, 8.52; N, 12.61.
Synthesis and Structures. All Cu(I) amidinate compounds
listed in Table 2 below were synthesized by the metathesis
reactions of anhydrous CuCl with the corresponding equiva-
lents of lithium amidinates (N,N′-dialkylacetamidinate or
N,N′-dialkylvaleroamidinate) in diethyl ether at ambient
temperature to afford the final products, according to reac-
tion 1.
Copper N-Isopropyl-N′-tert-butylacetamidinate [Cu(iPr-Me-
tBu-amd)]2 (12): Dark brown crystals (70%). Sublimation: 75 °C,
0.05 mmHg. mp: 121 °C. 1H NMR (C6D6, 25 °C, ppm): 3.43 (m,
2H), 1.89 (s, 3H), 1.39 (ds, 18H), 1.20 (dd, 12H). 13C NMR (C6D6,
25 °C, ppm): 172.1, 53.8, 48.4, 34.1, 33.9, 27.1, 26.8, 18.0, 17.9.
Anal. Calcd for C18H38N4Cu2: C, 49.40; H, 8.75; N, 12.80.
Found: C, 49.68; H, 8.62; N, 12.64.
2CuCl + 2Li(R’NC(R)NR′′) f [Cu(R’NC(R)NR′′)]2 +
2LiCl (1)
The overall process is illustrated in Schemes 1 and 2.
Symmetric amidinates (R′ ) R′′) were formed by Scheme
1, in which the lithium amidinate is formed by first coupling
two amines with a nitrile to form an amidine that is
subsequently deprotonated with an alkyllithium. Unsym-
metric amidines (R′ not the same as R′′) were synthesized
by Scheme 2, in which an unsymmetric lithium amidinate
is made by first coupling an amine and an isothiocyanate to
form a thiourea, followed by sulfur removal with
N-bromosuccinimide and treatment of the resulting carbo-
diimide with an alkyllithium. Scheme 1 has the advantage
of avoiding the highly toxic carbodiimide intermediate used
in Scheme 2. Scheme 1 fails, however, for tertiary amines
such as tert-butylamine, because the coupling reaction
becomes too slow.
X-ray Structure Determination. The crystal structure of
[Cu(sBu-Me-amd)]2, 9, was found by X-ray crystallography.
Crystals were grown by sublimation at 75 °C under vacuum.
Diffraction data were collected using a Bruker SMART CCD
(charge coupled device) based diffractometer equipped with an
Oxford Cryostream low-temperature apparatus operating at 193 K.
A suitable crystal was chosen and mounted on a glass fiber using
grease. Data were measured using Ω scans of 0.3° per frame for
30 s, such that a hemisphere was collected. A total of 1271 frames
were collected with a maximum resolution of 0.76 Å. The first 50
frames were recollected at the end of data collection to monitor
for decay. Cell parameters were retrieved using SMART software
and refined using SAINT on all observed reflections. Data reduction
was performed using the SAINT software, which corrects for
Lorenz polarization and decay. Absorption corrections were applied
using SADABS supplied by Sheldrick. The structures were solved
by the direct method using the SHELXS-97 program and refined
All of the compounds are easily soluble in hexane. They
were purified by crystallization of the crude materials from
hexane at -30 °C and by sublimation. All of the pure
compounds are colorless or light yellow crystalline solids
(19) Benniston, A. C.; Harriman, A. J. Chem. Soc., Faraday Trans. 1994,
90, 2627-2634.
(20) Furumoto, S. Nippon Kagaku Zasshi 1970, 91, 359-361.
Inorganic Chemistry, Vol. 44, No. 6, 2005 1731

Li et al.
Table 2. Properties of Amidinato Cu(I) Compounds 1-12
vapor pressure/
temperature
(Torr/°C)
residue after
TGA
melting
point (°C)
yield
(%)
half-lifea
(days)
compd
color
white
white
white
white
dark brown
yellow
white
pale yellow
white
(%)
[Cu(nPr-Me-amd)2]2, 1
[Cu(nPr-nBu-amd)2]2, 2
[Cu(iPr-Me-amd)2]2, 35,6
[Cu(iPr-nBu-amd)2]2, 4
[Cu(nBu-Me-amd)2]2, 5
[Cu(nBu-nBu-amd)2]2, 6
[Cu(iBu-Me-amd)2]2, 7
[Cu(iBu-nBu-amd)2]2, 8
[Cu(sBu-Me-amd)2]2, 9
[Cu(sBu-nBu-amd)2]2, 10
[Cu(iPr-Me-sBu-amd)2]2, 11
[Cu(iPr-Me-tBu-amd)2]2, 12
65
60
147
97
65
38
114
111
77
71
86
0.05/50
0.05/75
0.05/70
0.04/75
0.1/70
0.1/190
0.2/90
-
0.1/85
0.01/60
0.05/72
0.05/75
83
76
70
74
84
-
84
81
90
-
80
70
9.7
31.9
2.8
11.7
32.3
22.0
5.4
18.3
0.7
3.5
45
20
33
21
white
pale yellow
dark brown
17.4
14.2
121
a Measured at 190 °C in C6D6 solution.
Table 3. Crystal Data and Data Collection Parameters for
[Cu(sBu-Me-amd)2]2, 9
empirical formula
formula weight
temperature
wavelength
crystal system
space group
C20H42Cu2N4
465.66
193(2) K
0.71073 Å
monoclinic
C2/c
unit cell dimensions
a ) 18.884(2) Å
b ) 10.950(2) Å
c ) 12.9230(19) Å
2405.1(7) Å3
4
R ) 90°
â ) 115.842(4)°
γ ) 90°
volume
Z
density (calcd)
absorption coefficient
θ range for data collection
completeness to θ ) 28.26°
GOF on F2
1.286 Mg/m3
1.779 mm-1
2.21 to 28.26°.
99.8%
1.046
final R indices [I > 2σ(I)]
R indices (all data)
largest diff peak and hole
R1 ) 0.0302
R1 ) 0.0497
0.300 and -0.574 e.Å-3
wR2 ) 0.0700
wR2 ) 0.0762
Table 4. Bond Distances (Å) and Angles (deg) for
[Cu(sBu-Me-amd)2]2, 9
Cu-Na
1.877(2)
2.4031(6)
1.326(3)
1.505(15)
176.62(10)
119.62(17)
Figure 1. Structure of [Cu(sBu-Me-amd)2]2, 9, showing 50% probability
ellipsoids and partial atom-labeling schemes. Atoms labeled A are related
to other atoms by an inversion center.
Cu-Cu
N-Ca,b
N-Ca
N-Cu-Na
N-C-Nb
except [Cu(nBu-nBu-amd)]2, 6, which is waxy and becomes
a glass at room temperature after being quenched in liquid
nitrogen. Dark impurities were not successfully removed
from compounds 5 and 12.
a Mean value. b Amidinate backbone.
is shown in Figure 1 for clarity. Acetamidinate ligands bridge
copper atoms in a µ, η1:η1 fashion, which is similar to the
structure of [Cu(I) (N,N′-diisopropylacetamidinate)]2.6 The
average distance of Cu-N is 1.877(2) Å, and this distance
is similar to that obtained for the analogous compounds 3
and [Cu(p-tolyl-amd)]2.21 The distance between copper atoms
is 2.4031(6) Å,22 which might be short enough to indicate
some bonding interaction between the Cu-Cu atoms. On
the other hand, the N-Cu-N bonds are nearly linear with
an average angle of 176.62(10)°, suggesting little or no
bonding between the copper atoms. The dihedral angle for
N-Cu-Cu-N is 178.96(8)°, almost planar with imposed
centrosymmetry.
The structures of two of these copper complexes were
found to be dimeric by X-ray crystal structure determinations
([Cu(sBu-Me-amd)2]2, 9; [Cu(iPr-Me-amd)2]2, 36). Melting
point depression of p-xylene solutions showed that
[Cu(nPr-Me-amd)2]2, 1, is a dimer and confirmed that [Cu-
(sBu-Me-amd)2]2, 9, remains dimeric in solution [complexity
2.0(2)]. Other copper compounds were found to be dimers
in the vapor phase by mass spectroscopy: [Cu(nPr-Me-
amd)2]2, 1; [Cu(iPr-Me-amd)2]2, 3; [Cu(nBu-Me-amd)2]2, 5;
[Cu(iBu-Me-amd)2]2, 7; [Cu(sBu-Me-amd)2]2, 9; [Cu(sBu-Bu-
amd)2]2, 10. The physical properties are summarized in
Table 2.
The X-ray structure of [Cu(sBu-Me-amd)2]2, 9, is shown
in Figure 1 and selected metric data are collected in Tables
3 and 4. Its structure was disordered at the chiral centers in
the sec-butyl groups on the ligand, but only one configuration
(21) Cotton, F. A.; Daniels, L. M.; Feng, X.; Maloney, D. J.; Matonic, J.
H.; Murilio, C. A. Inorg. Chim. Acta 1997, 256, 291-301.
(22) Franzen, S.; Miskowski, V. M.; Shreve, A. P.; Wallace-Williams, S.
E.; Woodruff, W. H.; Ondrias, M. R.; Barr, M. E.; Moore, L.; Boxer,
S. G.; Inorg. Chem. 2001, 40, 6375-6382.
1732 Inorganic Chemistry, Vol. 44, No. 6, 2005

Cu(I) Amidinates as Precursors for ALD of Copper Metal
Figure 2. Melting points plotted as a function of the number of disruptive
degrees of freedom of the alkyl groups (see text).
Figure 3. NMR decomposition study of four compounds at 190 °C. The
lines represent the best fits to a first-order decomposition rate law with the
following rate constants: (9) [Cu(iPr-Me-amd)2]2, 3, rate ) 1.8 × 10-7
s-1; (b) [Cu(iPr-nBu-amd)2]2, 4, rate ) 4.0 × 10-7 s-1; (2) [Cu(sBu-Me-
amd)2]2, 9, pyramids, rate ) 2.4 × 10-7 s-1; (1) [Cu(sBu-nBu-amd)2]2, 10,
Melting Points. Melting points, listed in Table 2, are
below 100 °C for most of the new compounds. Compound
6, having all n-butyl groups, has the lowest melting point,
whereas compound 3 (isopropyl and methyl) has the highest.
Although there is no rigorous theory of melting points that
can be applied to these compounds, it seems reasonable that
the large number of conformations accessible to n-butyl
groups would hinder crystallization of compounds and lower
their melting points. The various alkyl groups can be ranked
in terms of the number of accessible configurations by
counting the number of single bonds about which nearly free
rotation can take place. In this scheme for counting the
degrees of freedom, we ignore a single bond that only rotates
a methyl group around its 3-fold axis, because the resulting
changes in energy might not be sufficient to disrupt crystal-
lization. Similarly, rotating a whole tert-butyl group might
not have much influence on crystal packing, so we do not
count this rotation as an effective degree of freedom. The
sec-butyl group introduces an additional degree of disorder
because of its chiral carbon; thus we increase its degrees of
freedom from 2 (if we counted only its disruptive single
bonds) to 3. This somewhat arbitrary increase in degrees of
freedom for sec-butyl might seem excessive, but the two
enantiomers cannot interconvert at normal temperatures; in
contrast, the rotations about single bonds can readily adjust
to the conditions within the crystals, thereby reducing the
disorder that they might produce. In summary, we estimated
the effective number of disruptive degrees of freedom for
the alkyl groups as follows: methyl ) 0, tert-butyl ) 0,
isopropyl ) 1, n-propyl ) 2, isobutyl ) 2, sec-butyl ) 3,
n-butyl ) 3. Then the total number of disruptive degrees of
freedom was calculated for the six alkyl groups on each of
the compounds. In Figure 2, we plot the melting points as a
function of this total number of disruptive degrees of
freedom. A reasonable qualitative correlation is shown
between these quantities, although there is certainly a degree
of scatter. One conclusion that we can draw from these trends
is that reduction of melting points can be achieved most
readily by synthesizing compounds having more sec-butyl
and n-butyl substitutents.
rate ) 3.8 × 10-7 s-1
.
Stability of the Compounds. All of the compounds listed
in Table 2 are much more stable in air than other first-row
transition metal amidinates, such as those of Ti, V, Mn, Fe,
Co, and Ni.6 Over a period of days, ambient air does cause
gradual darkening, presumably by oxidation to copper(II)
compounds. The compounds were synthesized with exclusion
of ambient air to avoid this oxidation and because the
intermediate lithium amidinates are very moisture-sensitive.
Initial NMR decomposition measurements showed that
compounds [Cu(iPr-Me-amd)]2, 3; [Cu(iPr-nBu-amd)]2, 4;
[Cu(sBu-Me-amd)]2, 9; and [Cu(sBu-nBu-amd)]2, 10, showed
very little decomposition after the NMR tubes containing
these compounds had been kept for 1 week at 175 °C. Thus
the oven temperature was raised to 190 °C to induce thermal
decomposition at a more readily measurable rate. Figure 3
shows the relative intensity of the peak (C) as a function of
time t (×105 s), where the vertical axis is on a logarithmic
scale. We found approximately linear relationships between
ln(C) and t at 190 °C, indicating that the decomposition is a
first-order reaction. The first-order rate constants for decom-
position were extracted from the slope as kd ) 1.8 × 10-7
s-1 for [Cu(iPr-Me-amd)2]2, 3; 4.0 × 10-7 s-1 for [Cu(iPr-
nBu-amd)2]2, 4; 2.4 × 10-7 s-1 for [Cu(sBu-Me-amd)2]2, 9;
and 3.8 × 10-7 s-1 for [Cu(sBu-nBu-amd)2]2, 10, correspond-
ing to the half-lives around 1 month (see Table 2). These
NMR results show that these compounds are very stable even
at 190 °C, which is a typical substrate temperature for ALD
of copper using these precursors. Although the vacuum
environment of these precursors during ALD is different from
the solvent environment in the NMR tubes, the NMR results
suggest that these copper amidinates should not thermally
decompose during the short time (seconds) that they are
inside the deposition chamber during an ALD cycle. These
results also demonstrate that the compounds are completely
stable at typical vaporization temperatures of around 100 °C.
Inorganic Chemistry, Vol. 44, No. 6, 2005 1733

Li et al.
Isothermal TGA runs with precursor 9 showed that, at
100 °C, its liquid vaporizes at a rate of 4.1 × 10-4 mg cm-2
s-1, which is sufficiently high to replenish the vapor needed
for an ALD cycle within a typical cycle time of a few
seconds.
Investigation of These Copper Compounds as ALD
Precursors. To satisfy the requirements of ALD, the
precursors must be sufficiently volatile, thermally stable, and
highly reactive to the surface produced by a complementary
ALD reactant. Lim et al.5,6 successfully used [Cu(iPr-Me-
amd)2]2, 3, as a precursor to make Cu films, using molecular
hydrogen as the complementary reactant. However, the
melting point of 3 (147 °C) is higher than the typical
temperatures (∼100 °C) used for its vaporization, so the
precursor 3 remains in the solid state as it sublimes into the
vapor. Thus the rate of vaporization of 3 depends on its
surface area, which is not reproducible from run to run.
Many of the copper(I) compounds listed in Table 3 have
much lower melting points (60-80 °C) and vapor pressures
similar to that of 3. Thus we can easily keep these new
precursors in the liquid phase while vaporizing them, which
makes the vapor delivery more reproducible than sublimation
from solid sources. As an example, we tried the two
compounds [Cu(nPr-Me-amd)2]2, 1, and Cu(sBu-Me-amd)2]2,
9, for the ALD of copper. Keeping the precursor reservoir
at 95 °C, we obtained ALD Cu films from both of these
precursors at rates up to 1 Å per cycle. The nitrogen and
carbon impurities are below the detection levels of RBS
(<1 atom %), showing that the ligands were completely
removed during the ALD process. The byproducts of these
reactions (most probably hydrogenated amidine ligands) are
not corrosive toward the reactor chamber and substrates,
providing a further advantage over copper halide precursors.
More detailed ALD results will be reported elsewhere.
Conclusions. Most of the copper(I) amidinates listed in
Table 2 are suitable as ALD precursors. Bis(copper(I) N,N′-
di-sec-butylacetamidinate) [Cu(sBu-Me-amd)2]2, 9, has a
particularly useful combination of sufficiently low melting
point and excellent thermal stability. These new compounds
provide several advantages over traditional precursors, such
as copper halides and â-diketonates. The copper(I) amidinates
are sufficiently volatile, low melting, thermally stable, highly
reactive, and noncorrosive. The weak intermolecular attrac-
tion between the hydrocarbon outer parts of the complexes
contributes to their high volatility. The high thermal stability
is due to the chelating effects from the amidinate ligands
that connect the dimer together. The copper-nitrogen bond
is weaker than the copper-oxygen bond, which leads to
higher reactivity compared to copper â-diketonates. We
obtained ALD copper films from precursor 9 at 160 °C, a
much lower temperature than those required for CuCl
(400 °C) and Cu(hfac)2‚H2O (230-300 °C). Finally, these
ALD copper films show strong adhesion to cobalt films as
substrates, so that robust microelectronic interconnections
can be based on ALD copper seed layers from these new
precursors.24
Figure 4. TGA/DSC data for [Cu(sBu-Me-amd)2]2, 9. The experiment
was done at N2 carrier gas (∼50 sccm). The ramp rate for TGA/DSC is
10 K/min.
Comparison of compounds 3 and 9 (methyl) with 4 and
10 (n-butyl) shows that n-butyl groups are less stable than
methyl groups on the amidinate ligand. This approximate
doubling of the decomposition rate might be due to additional
pathways for â-hydrogen elimination from the n-butyl groups
of 4 and 10. A compound with sec-butyl groups (9) showed
better stability than a similar compound with n-butyl groups
(5), which was difficult to purify because of decomposition.
Thus sec-butyl groups are the best choice for compounds
combining low melting points with high thermal stability.
The NMR spectra also showed formation of unsaturated
hydrocarbon decomposition products, although assigning all
of the new peaks is difficult because of their relatively low
intensities. One major decomposition product was identified
from the NMR spectra as a dimer of two amidinate ligands
joined by a nitrogen-nitrogen bond.23 Copper metal mirrors
were also deposited on the inside walls of the NMR tubes
after the long thermal decomposition runs. Taken together,
these two byproducts indicate the presence of a reductive
elimination pathway for the thermal decomposition of these
compounds.
Vaporization of the Compounds. Table 2 lists the TGA/
DSC data (melting point, sublimation temperature, residual
mass) for most of these copper compounds. Most of the
copper amidinates left little residue after thermal evaporation.
Figure 4 gives the TGA/DSC graphs for 9; the Supporting
Information contains TGA/DSC curves for compounds 1,
6-8, 10, and 12. Figure 4 shows the TGA curve for complex
[Cu(sBu-Me-amd)2]2, 9, which left very little residual mass
(0.7%), showing that negligible decomposition happened
during its melting and evaporation. The DSC curves have
only endothermic peaks due to the melting and evaporation
of the compounds, confirming their thermal stabilities. Many
of the compounds have similar vapor pressures, except that
the ones with n-butyl or tert-butyl substituents tended to have
lower vapor pressures.
(23) Full characterization of the amidine ligand dimer will be reported by
Huazhi Li and Roy G. Gordon.
(24) de Rouffignac, P.; Li, Z.; Gordon, R. G. Electrochem. Solid State Lett.
2004, 7, G306.
1734 Inorganic Chemistry, Vol. 44, No. 6, 2005

Cu(I) Amidinates as Precursors for ALD of Copper Metal
Supporting Information Available: TGA and DSC curves for
compounds 1, 6-8, 10, and 12. X-ray crystallographic file in CIF
format for the structure determination of compound 9. This material
is available free of charge via the Internet at http://pubs.acs.org.
Acknowledgment. We thank Dr. Xinjian Lei of Schu-
macher Chemical Company for the TGA/DSC analyses and
Dr. Richard Staples for X-ray structure analysis. We ap-
preciate helpful discussions with Dr. Booyong Lim. This
work was supported in part by the National Science Founda-
tion (Grant ECS-0354213)
IC048492U
Inorganic Chemistry, Vol. 44, No. 6, 2005 1735