Controlling the Thermal Stability and Volatility of Organogold(I) Compounds for Vapor Deposition with Complementary Ligand Design

Article abstract of DOI:10.1002/ejic.201901087

Atomic layer deposition (ALD) of gold is being studied by multiple research groups, but to date no process using non-energetic co-reactants has been demonstrated. In order to access milder co-reactants, precursors with higher thermal stability are required. We set out to uncover how structure and bonding affect the stability and volatility of a family of twelve organogold(I) compounds using a combination of techniques: X-ray diffraction (XRD), thermogravimetric analysis (TGA), differential scanning calorimetry (DSC), and density functional theory (DFT). Small, unsubstituted phosphonium ylide ligands bind more strongly to Au(I) than their silyl-substituted analogues, but the utility of both these ligands suffers due to their poor volatility and substantial thermal decomposition. Pentafluorophenyl (C₆F₅) is introduced as a new, very electronegative ligand for gold vapor deposition precursors, and it was found that the disadvantage to volatility due to π-stacking and other intermolecular interactions in the solid state was overshadowed by dramatic improvements to kinetic and thermodynamic stability. We introduce a new figure of merit to compare and rank the suitability of these and other complexes as precursors for vapor deposition. Finally, DFT calculations on four compounds that have high figures of merit show a linear correlation between the gold-coordinative ligand bond dissociation energies and the observed decomposition temperatures, highlighting and justifying this design strategy.

Full text of DOI:10.1002/ejic.201901087

A Journal of
Accepted Article
Title: Controlling thermal stability and volatility of organogold(I)
compounds for vapor deposition with complementary ligand
design
Authors: Matthew B E Griffiths, Zachary S. Dubrawski, Goran Bačić,
Achini Japahuge, Jason D. Masuda, Tao Zeng, and Seán T.
Barry
This manuscript has been accepted after peer review and appears as an
Accepted Article online prior to editing, proofing, and formal publication
of the final Version of Record (VoR). This work is currently citable by
using the Digital Object Identifier (DOI) given below. The VoR will be
published online in Early View as soon as possible and may be different
to this Accepted Article as a result of editing. Readers should obtain
the VoR from the journal website shown below when it is published
to ensure accuracy of information. The authors are responsible for the
content of this Accepted Article.
To be cited as: Eur. J. Inorg. Chem. 10.1002/ejic.201901087
Link to VoR: http://dx.doi.org/10.1002/ejic.201901087

10.1002/ejic.201901087
European Journal of Inorganic Chemistry
Controlling thermal stability and volatility of organogold(I)
compounds for vapor deposition with complementary ligand
design
Matthew B. E. Griffiths*[a], Zachary S. Dubrawski [a], Goran Bačić [a], Achini Japahuge [a], Prof.
Jason D. Masuda [b], Prof. Tao Zeng [a], and Prof. Seán T. Barry [a]
[a] Department of Chemistry, Carleton University, 1125 Colonel By Drive, Ottawa, Ontario, K1S
5B6, Canada
E-mail: mbgriffi016@gmail.com
Homepage: carleton.ca/barrylab
Twitter: Carleton Chemistry @Carletonchem, MBEG @mattbegriffiths, ZSD @DubrawskiZ, STB
@sean_t_barry
[b] Department of Chemistry, Saint Mary’s University, 923 Robie Street, Halifax, Nova Scotia,
B3H 3C3, Canada
Twitter: @jasonmasuda
TOC Graphic
TOC Text
Ligand-structure relationships in organogold(I) compounds strongly influence the resulting
thermal stability and volatility of these potential atomic layer deposition precursors.
Key topic: ALD precursors
Keywords: atomic layer deposition, precursor design, gold, crystallography, thermogravimetric
analysis, figure of merit
Abstract
Atomic layer deposition (ALD) of gold is being studied by multiple research groups, but to date no
process using non-energetic co-reactants has been demonstrated. In order to access milder co-
reactants, precursors with higher thermal stability are required. We set out to uncover how
structure and bonding affect the stability and volatility of a family of twelve organogold(I)
This article is protected by copyright. All rights reserved.

10.1002/ejic.201901087
European Journal of Inorganic Chemistry
compounds using a combination of techniques: X-ray diffraction (XRD), thermogravimetric
analysis (TGA), differential scanning calorimetry (DSC), and density functional theory (DFT).
Small, unsubstituted phosphonium ylide ligands bind more strongly to Au(I) than their silyl-
substituted analogues, but the utility of both these ligands suffers due to their poor volatility and
substantial thermal decomposition. Pentafluorophenyl (C6F5) is introduced as a new, very
electronegative ligand for gold vapor deposition precursors, and it was found that the
disadvantage to volatility due to π-stacking and other intermolecular interactions in the solid state
was overshadowed by dramatic improvements to kinetic and thermodynamic stability. We
introduce a new figure of merit to compare and rank the suitability of these and other complexes
as precursors for vapor deposition. Finally, DFT calculations on four compounds that have high
figures of merit show a linear correlation between the gold-coordinative ligand bond dissociation
energies and the observed decomposition temperatures, highlighting and justifying this design
strategy.
Introduction
Atomic layer deposition (ALD) is a chemical vapor deposition (CVD) technique that relies on
specific surface chemistries to force film growth into a self-limiting layer-by-layer regime.[1] Films
deposited by ALD are highly conformal to substrate geometry, and sub-nanometer thickness
control can be easily achieved by changing the number of sequential reactant exposures (i.e.,
cycles). Recently, two different ALD processes to deposit gold metal were reported using
organogold(III) precursors: one by our group using oxygen-plasma and water as co-reagents, and
the other by Mäkelä and co-workers using ozone.[2,3] While both processes were effective at
depositing gold metal with high growth rates at low temperatures, they inherently suffer similar
limitations in their scope. Firstly, thermal instability due to recombination of energetic oxygen
species prevents high-degree of thickness uniformity of the films down high-aspect ratio trenches,
vias, or tortuous geometries.[4,5] Secondly, many substrates are incompatible with oxygen plasma
or ozone and are therefore incompatible with these processes.
The strategy most often used for selecting viable compounds to use as new potential ALD
precursors is to turn to previous literature; pick the best candidates based on volatility, thermal
stability, and reactivity; and try to adapt them for use in an ALD process. While this was fruitful for
gold, as shown by both reported processes using previously reported CVD precursors,[6–8] this
selection process only returns a handful of molecules that are often quite different from one
another in their structure and chemistry. ALD precursor design for other materials is quite mature
in some cases and has been proven to be a useful tool in accelerating process development for
Cu, Ru, Co, Ni, and many other metallic films.[9,10] As such, we felt that a thorough study on the
fundamental factors that govern thermal stability and volatility would be useful in the development
of new gold ALD precursors.
Many Au(I) CVD precursors decompose below 200°C.[11] By comparison, the current two ALD
precursors undergo thermolysis: at 140°C for trimethylgold(III) trimethylphosphine (A) and at
220°C for dimethyl(diethyldithiocarbamato-κ2-S,Sʹ)gold(III) (B) (Scheme 1), which manifested in
upper process temperatures being limited to 120 °C and 180 °C, respectively. We hypothesize
that milder co-reagents than plasma or ozone could be used if higher deposition temperatures
could be accessed, therefore requiring precursors with higher thermal stability.
This article is protected by copyright. All rights reserved.

10.1002/ejic.201901087
European Journal of Inorganic Chemistry
Scheme 1. Two previously reported precursors for gold ALD.
This work reports a study of a family of potential gold(I) vapor deposition precursors based on a
framework with interchangeable coordinative and anionic ligands to allow a systematic structure-
function understanding not only of the steric and electronic effects of these ligands, but also of
their cooperative effects on the volatility and thermal stability of the compound.
Alkylgold(I) phosphine compounds are known to decompose through a bimolecular reductive
elimination pathway (Scheme 2). Trimethylphosphine methylgold(I) (1a) is reported to be stable
as a neat liquid until 150°C, at which point dissociation of trimethylphosphine occurs followed by
bimolecular reductive elimination of ethane gas.[12] However, on active metal surfaces this
decomposition is known to occur as low as room temperature: the Au-PMe3 bond dissociates on
the surface, allowing bimolecular reductive elimination to take place.[13,14]
Scheme 2. Decomposition pathways for (PMe3)AuMe as a neat liquid and in the presence of an
active metal surface.
Given this, we focused on three complementary synthetic strategies to increase the thermal
stability of Au(I) complexes. First, increasing the steric bulk of the anionic ligand should hinder
the bimolecular reductive elimination pathway by preventing the association of adjacent surface-
bound Au(I) species. Second, since the rate-limiting step of this reaction is dissociation of the
coordinative phosphine ligand, a more electron-withdrawing anionic ligand should make the Au(I)
center more acidic, thus strengthening the bond of the coordinative ligand. Lastly, using
coordinative ligands of other known thermally stable Au(I) complexes may yield novel and useful
ligand combinations. Thus, we envisioned a family of compounds that would allow us to discover
which factors conferred the most thermal stability to organogold(I) compounds (Scheme 3).
This article is protected by copyright. All rights reserved.

10.1002/ejic.201901087
European Journal of Inorganic Chemistry
Scheme 3. Family of Au(I) compounds considered in this study.
Starting from the known methylgold(I) CVD precursor 1a, substituting trimethylsilylmethyl
(CH2(SiMe3)) and pentafluorophenyl (C6F5) ligands should impart a stepwise increase in the
overall thermal stability of the compound. CH2(SiMe3) is sterically bulkier than Me and is known
to stabilize Au(I) species more so than non-silylated ligands.[15,16] Both factors increase the
thermal stability of transition metal complexes bearing this ligand, and have been used previously
for manganese ALD precursor design.[17] The C6F5 is much more σ-electron withdrawing due to
perfluorination, which should result coordinative ligands forming a stronger bond to the gold center
and increase the activation energy required for decomposition via bimolecular reductive coupling.
Some C6F5 complexes of gold with and without phosphine coordinative ligands have previously
been shown to have high thermal stability, but have not been used for CVD or ALD.[18,19]
Phosphines are strongly binding coordinative ligands for organogold(I) compounds. Their
chemistry is well known and predictable, the gold(I) compounds are stable to air and moisture,
and they are generally quite volatile. We chose trimethylphosphine (PMe3) as one of four neutral
coordinative ligands (1a, 2a, 3a) due to its low molecular weight, ease of use in synthesis, and
good σ-donation capabilities.
Schmidbaur and Franke reported “remarkably stable” organometallic gold(I) complexes bearing
trimethylmethylene phosphorane (ylide) and trimethyl(trimethylsilylmethylene) phosphorane
(TMS-ylide) phosphonium ylides as neutral coordinative ligands.[20] These phosphonium ylides
readily displace PMe3 from 1a and 2a giving 1b and 1c, and 2b and 2c which were reported to
be more thermally stable than the parent PMe3 complexes (decomposing above 150 °C in the
case of 1b). Since phosphonium ylide compounds of Au(I) have to our knowledge not yet been
This article is protected by copyright. All rights reserved.

10.1002/ejic.201901087
European Journal of Inorganic Chemistry
tested for vapor deposition applications, and their volatilities have not been assessed, we included
them and their C6F5 analogues in this study.
N-Heterocyclic carbenes (NHCs) have been successfully used in the design of ALD precursors
for copper metal, a CVD precursor for gold metal, and an ALD precursor for silver metal.[21–24] The
compound N,N’-di-tert-butylimidazolidin-2-ylidene[25] (NHC) is a monomeric, thermally stable, and
sterically bulky saturated NHC with better σ-donor and π- acceptor properties than the more
common imidazol-2-ylidene ligand class. This NHC should enhance the thermal stability of
organogold(I) compounds due to its strong electron donating ability as well as its ability to
sterically protect the Au(I) center.
Results and Discussion
Synthesis and Characterization
In literature examples of alkylgold(I) compounds, the desired species is often synthesized by salt
metathesis from the corresponding gold(I) halide. However, given the propensity of phosphonium
ylide- and NHC-gold(I) halides to form salts of the type [Au(L)2]+ [AuX2]– [26,27] and to reduce the
required number of compounds for this study, we opted for a divergent synthetic strategy where
the phosphonium ylide and NHC compounds would be synthesized from their respective parent
alkylgold(I) phosphine complexes (Scheme 4).
This article is protected by copyright. All rights reserved.

10.1002/ejic.201901087
European Journal of Inorganic Chemistry
Scheme 4. Synthesis of all compounds. Reagents: a, MeLi (1a, 77%), (SiMe3)CH2Li (2a, 85%);
b, CH2PMe3 (1b, 87%), (2b, 82%), (3b, 65%); c, CH(PMe3)(SiMe3) (1c, 84%), (2c, 56%), (3c,
51%); d, N,N’-di-tert-butylimidazolidin-2-ylidene (1d, 94%), (2d, 87%), (3d, 89%), e, PMe3 (3a,
94%).
The ligand exchange reaction produced volatile PMe3 and no other by-products which allowed
for facile workup of these reactions. As detailed in the experimental section, an excess of the
desired ligand was often used in order to completely consume the starting gold(I) phosphines
which are difficult to remove by recrystallization or sublimation. Synthesis of 3b was initially
attempted from the known compound (THT)AuC6F5 since this saved one step in its synthesis
(bypassing (PMe3)AuC6F5). Unfortunately, this reaction produced a mixture of PMe3-,
phosphonium ylide-, and C6F5-containing compounds. However, when we attempted the same
exchange reaction with 3a instead of (THT)AuC6F5, 3b and 3c were produced as the majority
products and were isolated by recrystallization. Compound 3d could be synthesized
successfully from both (THT)AuC6F5 and 3a by ligand exchange, but we deem the former
procedure superior due to a reduction in synthetic steps.
This article is protected by copyright. All rights reserved.

10.1002/ejic.201901087
European Journal of Inorganic Chemistry
We observed during workup that the ylide-containing complexes 1b and 1c decomposed upon
attempted sublimation, and so we chose to purify phosphonium ylide species exclusively by
recrystallization. All PMe3 and NHC complexes were purified by sublimation (or by distillation for
for 2a).
All compounds were characterized by 1H-, 31P-, 19F-, and 13C-NMR spectroscopy, and were
matched to literature reports where appropriate (1a,[28] 1b,[29] 1c,[30] 2a,[30] 2b,[30] 2c,[26] 3a[31] are
known, compounds 1d, 2d, 3b, 3c, 3d are novel). Our analysis of 3a differed slightly by 31P-
NMR analysis where the literature gives a chemical shift of 3.28 ppm and we instead observed
the resonance at -5.00 ppm after purifying the compound by vacuum sublimation. The NHC
compounds 1d, 2d, and 3d display characteristic NHC carbon resonances at 221.74, 219.94,
and 211.05 ppm in the 13C-NMR spectra which are similar to the free NHC (218.7 ppm)[32]. The
compounds 3b and 3c were characterized by their 31P-NMR resonances which were observed
at 26.50 and 23.65 ppm respectively, whose chemical shifts are very similar to 1b and 1c, and
2b and 2c. The CH(SiMe3)PMe3 resonances (1c, 2c, 3c) were shielded by the increased
electron density from the SiMe3 group compared to the non-substituted ylide compounds.
Compounds 3a, 2b, 2c, and 2d could not be differentiated from ionic salts of the type [Au(L)2]+
[AuX2]– using NMR spectroscopy alone if the sample contains only one species, so the following
X-ray crystallographic analyses served to confirm their neutral molecular arrangement.
Crystallography
Volatility is arguably the most important property of an ALD precursor as it sets the lower limit of
useable process temperatures. It is primarily dictated by intermolecular forces and, to a lesser
extent, by the molecular weight of the compound. Since a compound’s volatility can be readily
determined by TGA, we wanted to understand which intermolecular interactions most affect
volatility. Single crystal X-ray crystallography was used to determine the solid-state packing of 1c
and 1d, 2b and 2d, and 3a, 3b, 3c, and 3d (Figure 1). Selected bond lengths, angles, and other
data are displayed for comparison in Table 1.
Figure 1. Solid-state structures of 1c, 1d, 2b, 2d, 3a, 3b, 3c, and 3d. In the cases of 1c and 3c
only the S enantiomers are shown. Ellipsoids are drawn at 50 % probability in all diagrams.
This article is protected by copyright. All rights reserved.

10.1002/ejic.201901087
European Journal of Inorganic Chemistry
Compound 1c, like its TMS-ylide derivative 3c, crystallizes as a 1:1 mixture of R and S
enantiomers at the chiral ylidic carbon center coordinated to Au. The P-Cylide bond length is shorter
than the mean P-Me bond length (1.742(3) Å vs. 1.793(9) Å respectively), and the P-Cylide-Si angle
(121.6(2)°) is much larger than expected for a tetrahedral geometry. These factors indicate that
the bonding of this ligand lies somewhere between its two extreme canonical structures (Scheme
5). Steric repulsion between the PMe3 and SiMe3 groups of the ylide may influence the large angle
and therefore hinder the σ-donating ability of this ligand. Likewise, compound 3c displays a P-
Cylide bond length of 1.759(3) Å, an average P-Me bond length of 1.798(11) Å, and a P-Cylide-Si
bond angle of 121.7(2)°, all of which implies a significant ylene character. More ylene character
translated to higher volatility in TMS-ylide compounds (vide infra).
In the structure of 2b, the acceptance of electron density from the ylidic carbon by the Au center
is much more pronounced as shown by the shorter Au-L bond. The P-Cylide bond length in 2b is
elongated compared to that of trimethylmethylene phosphorane (1.77(1) Å and 1.640(6) Å
respectively)[33] and the P-Cylide bond length is not significantly different from the mean P-Me bond
length (1.77(1) Å vs. 1.80(4) Å, respectively). This is similarly observed in 3b and distinguishes
2b and 3b from 1c and 3c as being primarily ylidic in nature, which led to a suppression of
volatility.
Scheme 5. Resonance structures depicting the extreme canonical resonance structures and
ligand geometries of compounds 1,2,3c.
Table 1. Selected bond distances and angles for the analyzed compounds
P-Cylide
(Å)
Average
P-Me (Å) angle (°)
NCN bond R-Au-L
R-
L
Au-R (Å) Au-L (Å)
(°)
TMS-
ylide
1c
1d
2b
Me
Me
2.057(3) 2.136(3) 1.742(3) 1.793(9)
-
178.1(1)
180
NHC
2.069(8) 2.055(6) -
2.10(1) 2.09(1) 1.77(1)
2.059(7) 2.045(7) -
-
108.5(5)
-
CH2(SiMe3) Ylide
CH2(SiMe3) NHC
1.80(4)
178.5(3)
2d
-
108.5(6)
178.1(3)
This article is protected by copyright. All rights reserved.

10.1002/ejic.201901087
European Journal of Inorganic Chemistry
3a
3b
C6F5
C6F5
PMe3
2.053(5) 2.280(1) -
-
-
172.8(1)
177.9(1)
Ylide
TMS-
ylide
2.046(3) 2.081(4) 1.763(3) 1.789(10) -
2.041(3) 2.104(3) 1.759(3) 1.789(11) -
3c
3d
C6F5
C6F5
178.4(1)
NHC
2.039(4) 2.031(4) -
-
109.4(3)
176.7(1)
When comparing R-Au bond lengths, the C6F5 compounds all exhibit shorter bonds than other
compounds with the same coordinative ligand: 1d ≈ 2d > 3d, 1c > 3c, and 2b > 3b. This is primarily
due to the covalent radius of the sp2-hybridized carbon in the C6F5 ring being slightly smaller than
that of the sp3-hybridized carbon atoms in the Me and CH2(SiMe3) ligands. Furthermore, C6F5
complexes 3a, 3b, 3c, and 3d display shorter Au-L bond lengths than their aliphatic derivatives,
which is due to the strong electron-withdrawing ability of the C6F5 ligand. From these results, it
appeared that C6F5 was an especially good ligand for improving the thermal stability of Au(I)
compounds for two reasons. Firstly, the rate-limiting step for decomposition must proceed through
cleavage of a stronger coordinative bond, and secondly because the final reductive elimination
step must occur between two strongly bound, electron-deficient carbon atoms.
In the solid state, each molecule of 3a is associated to its neighbors by polymeric Au-Au aurophilic
interactions (Au-Au distance 3.3703(6) Å). This was the only compound structurally characterized
in this work that displayed aurophilicity (Figure 2). This is an intermediate-to-long range aurophilic
interaction,[34] and is likely only observed for 3a because PMe3 is the least sterically ligand in this
study. By comparison the previously reported PPh3 analogue does not display aurophilic
interactions, but instead undergoes a π-stacking interaction between the C6F5 ligands, possibly
due to the steric bulk of the PPh3 ligand interfering with the Au-Au interaction.[35]
Figure 2. Extended structure of 3a showing the aurophilic polymer chain. Thermal ellipsoids are
shown at 50 % probability. Selected intermolecular interaction distances (Å): Au(1)-Au(2)
3.3703(6), H(1a) – F(5**) 2.492, H(2c) – F(5**) 2.527.
This article is protected by copyright. All rights reserved.

10.1002/ejic.201901087
European Journal of Inorganic Chemistry
Some studies have previously linked the volatility of Au(I) compounds to solid-state aurophilic
interactions.[36,37] However a recent computational paper by Mata et al[38] cautioned that aurophilic
interactions aren’t necessarily the dominant intermolecular force holding the molecules together,
and Au-ligand interactions or inter-ligand interactions dictate solid-state packing arrangements to
a greater degree. Thus, we have attempted to describe the number and types of interactions that
occur throughout the system which overall affect the compound’s volatility.
In Figure 2 other in-chain intermolecular interactions are observed between the PMe3 ligands.
Each phosphine engages in two H – F interactions to an adjacent molecule (H(1a) - F(5**) 2.492
Å, H(2c) – F(5**) 2.527 Å). These strong hydrogen bonding interactions translated to a
suppression of the volatility of 3a compared to 1a and 2a.
Compound 3b is involved in a π-stacking interaction with a centroid to centroid distance of 3.497
Å. These π-stacked dimers are involved in secondary interactions with other π-stacked dimers
along the crystallographic b x (a+c) plane through a set of F-F, C-F, and C-C interactions (Figure
3). The extensive intermolecular network formed by 3b shows that the addition of a C6F5 group
introduces many more intermolecular interactions when coupled with a small ligand like PMe3 or
CH2PMe3 than with larger ligands like CH2PPh3 and resulted in a reduced volatility of this
compound. This compound displayed the largest number of intermolecular interactions of the
family (Table 2).
Figure 3. C6F5 – C6F5 intermolecular interactions in 3b. The diagram displays pairs that occur
along the crystallographic b x (a+c) plane. Methyl moieties of the ylide and hydrogen atoms have
been omitted for clarity. Thermal ellipsoids are shown at 50 % probability. Selected intermolecular
interaction distances (Å): centroid-to-centroid 3.497, F(2**) – C(10) 3.210(3), F(2**) – C(9)
3.166(3), C(7**) – F(4) 3.231(6), F(1**) – F(4) 2.928(2), F(1**) – C(9) 3.127(3), F(1**) – C(8)
3.145(3).
Compound 3c does not π-stack in the solid state, instead the C6F5 ligands interact to form chains
via (C-F)-F interactions, although there are considerably fewer than observed in 3b (Figure 4).
This demonstrates the ability of sterically bulky coordinative ligands to suppress intermolecular
interactions in C6F5-Au complexes.
This article is protected by copyright. All rights reserved.

10.1002/ejic.201901087
European Journal of Inorganic Chemistry
Figure 4. C6F5 – C6F5 intermolecular interactions in 3c. Hydrogen atoms have been omitted for
clarity and thermal ellipsoids have been drawn at the 50% probability level. The top two molecules
are the (S) enantiomer while the bottom two are the (R) enantiomer. Selected intermolecular
interaction distances (Å): F(3) – F(4*) 2.912(3), F(3) – C(2*) 3.151(4), F(3) – C(3*) 3.214(3).
Compound 3d displays fewer overall intermolecular contacts than other C6F5 compounds which
we attribute to the large steric bulk of the NHC. The C6F5 inter-ligand interactions are suppressed
compared to 3c, where here only a single pair of (C-C) – F p-π interactions (Figure 5) are allowed.
Figure 5. C6F5 – C6F5 intermolecular interactions in 3d. Thermal ellipsoids have been drawn at
the 50% probability level. Selected intermolecular interaction distances (Å): C(14) – F(3*)
3.095(4), C(15) – F(3*) 3.043(4).
This article is protected by copyright. All rights reserved.

10.1002/ejic.201901087
European Journal of Inorganic Chemistry
Complexes bearing Me (1c and 1d) or CH2(SiMe3) (2b and 2d) ligands display no intermolecular
contacts. In contrast, C6F5 complexes (3a, 3b, 3c, and 3d) display far more intermolecular
interactions in the solid state, and we surmise that this is the main reason for their decreased
volatility (see below).
Thermolysis
For a compound to be a potential ALD candidate, high thermal stability and volatility are required.
Thermogravimetric analysis (TGA) of a thermally stable and volatile compound shows an
exponential mass loss with constantly increasing temperature as described by the Clausius-
Clapeyron equation. Thermally stable compounds also leave a very low final residual mass
indicative of evaporation without decomposition to non-volatile by-products. The vapor pressure
of a compound can be calculated from TGA data using a previously developed method (Figures
S38-S45).[39] The temperature at which a compound achieves a vapor pressure of 1 Torr (i.e., its
1 Torr temperature, TV) may be used as a benchmark for comparison since commercial ALD
reactors typically operate close to this pressure. While TGA can also give clues about
decomposition of the sample compound, differential scanning calorimetry (DSC) is better at
analyzing the onset of decomposition of these compounds. By graphically determining the point
at which the decomposition exotherm has reached 5 % of it’s maximum height, we obtained a
decomposition temperature (TD) for the compound in question (Figures S46-S57). The results of
these analyses are tabulated below (Table 2).
Table 2. Important metrics obtained from TGA and DSC experiments
Useful
Fractional
Gold
remaining (%)
Residual mass
(%)
Figure of Merit
TV (°C) TD (°C)
temperature
(σ)b
range (°C)a
1a 68
130
107
129
189
62.5
-
34.3
22.9
36.9
34.9
50.2
35.1
70.1
69.9
31.1
-
1b
1c
-
-
-
-
1d 157
32.5
9.8
2a 87
146
152
118
184
59.1
-
-16.8
33.1
0.8
2.5
5.9
11.9
1.5
58.2
-
-14.6
23.8
2b
-
4.8
13.4
28.2
2c 134
2d 150
3a 152
185
220
253
300
33.1
-
1.1
2.5
32.3
-
3b
-
41.3
33.3
1.6
95.2
88.8
4.4
3c 148
3d 200
104.7
99.7
11.7
95.3
a
b
Difference between TV and TD. See Equation 1. Dashes indicate that insufficient mass loss data could be obtained before the
measured self-decomposition temperature of that compound.
In the TGA ramp experiments of 1a-d (Figure 6), none of the compounds were thermally stable
enough to evaporate completely, however in each case some sample did evaporate intact which
This article is protected by copyright. All rights reserved.

10.1002/ejic.201901087
European Journal of Inorganic Chemistry
we can infer from the percent atomic mass of Au in the compound compared to the final residual
mass of the experiment (see inset). All sample pans were left with a visibly golden coating after
the experiments.
In the case of 1a it is interesting that such a high residual mass is observed, given that the self-
decomposition point (130 °C) is close to the end of the mass loss curve. This can be explained
by the known autocatalytic decomposition of alkylgold(I) phosphines that occurs as low as room
temperature on Cr, Cu and Au surfaces.[13] Some decomposition likely occurs on the Pt surface
of the pan at temperatures below the onset of self-decomposition results in an increased residual
mass. This occurs more readily for the phosphine than for the NHC which implies a strong NHC–
Au bond. No inflection is observed in the exponential mass loss curves of 1a and 1d suggesting
that the decomposition products are gaseous (ethane, PMe3, NHC) or non-volatile (metallic gold),
and therefore their production does not affect the rate of mass loss of the sample.
In the case of compounds 1b and 1c, concurrent evaporation and decomposition were observed
by the presence of multiple inflection points in their respective mass loss curves. Here the
products formed by thermolysis are semi-volatile or are themselves thermally unstable. This
makes the by-products observable as changes to the mass loss rate, implying low-volatility
organic molecules or other gold complexes are the primary decomposition products of 1b and 1c.
Figure 6. TGA of 1a-d at a ramp rate of 10 °C/min. The inset vertical ranges show the % mass of
Au in each compound (top line) and the final residual mass of the respective analysis (bottom
line).
The CH2(SiMe3)gold(I) compounds 2a and 2d performed much better in TGA analyses (Figure 7)
than 1a and 1d. Compound 2a appears to evaporate very cleanly, the only indication of slight
decomposition being the small residual mass of 0.8 %. Compound 2d evaporates with
decomposition in a single step to 11.9 %.
Compounds 2b and 2c undergo multiple decomposition events like 1b and 1c. However, it is
obvious that the CH2(SiMe3) ligand improves thermal performance by TGA because the residual
masses of 2b and 2c are much lower, meaning more of the gold has ultimately evaporated from
the system.
This article is protected by copyright. All rights reserved.

10.1002/ejic.201901087
European Journal of Inorganic Chemistry
Figure 7. TGA ramp experiments of 2a-d at a ramp rate of 10 °C/min. The inset vertical ranges
show the % mass of Au in each compound (top line) and the final residual mass of the respective
analysis (bottom line).
The C6F5 compounds 3a and 3d performed very well by TGA (Figure 8), each evaporating in a
single step and leaving a very low residual mass (3a: 1.1%; 3d: 1.6%). Volatilization occurs at
higher temperatures for these species, and 3d shows a marked improvement in its thermal
stability compared to 1d and 2d. Clearly C6F5 imparts a strong stabilization to gold(I) and improves
the thermal stability of the system, although at the expense of volatility.
Compounds 3b and 3c performed worse than their Me and CH2(SiMe3) counterparts,
decomposing extensively and leaving the highest gold residues of the family (>85%). The
introduction of the C6F5 ligand does improve thermal stability when compared with the aliphatic
derivatives, but in the case of 3b and 3c, the reduction in volatility is too great to result in a useful
precursor.
Figure 8. TGA ramp experiments of 3a-d at a ramp rate of 10 °C/min. The inset vertical ranges
show the % mass of Au in each compound (top line) and the final residual mass of the respective
analysis (bottom line).
Using the thermal data collected, we defined a figure of merit (σ) to compare the viability of these
compounds for use in vapor deposition processes (Equation 1). This figure of merit is made up of
thermodynamic and kinetic terms that together describe the suitability of the vapor deposition
precursor. Not only does it include decomposition temperature and vapor pressure, but if two
This article is protected by copyright. All rights reserved.

10.1002/ejic.201901087
European Journal of Inorganic Chemistry
compounds that both undergo the same decomposition mechanism are compared, the one which
resists decomposition longer will receive a higher figure of merit. This highlights that some of the
compounds are more kinetically stable (e.g. 2a, 3a) than others (e.g. 1a). Particularly, 3a persists
in the TGA experiment far past its onset of self-decomposition measured by DSC (185 °C). Thus,
weighting the figure of merit by the residual mass demonstrates the importance of kinetic and
thermodynamic stability of a potential precursor compound.
Equation 1. Figure of merit equation for vapor deposition precursor usefulness.
The calculated σ values for the family of compounds are included in Table 3. Certain compounds
(1b, 1c, 2b, 3b) were not given a σ value because insufficient mass loss data could be obtained
before the TGA experiment reached the self-decomposition temperature (DSC) of the analyte in
question. The merits of each compound in this family can be visualized using a “minefield”
diagram (Figure 9). Since the delivery temperature of a precursor is an unavoidable consideration
in process design, each circle is centered at the compound’s respective TV.
If delivery temperature is not an important requirement, then clearly compound 3d is the most
promising precursor candidate for vapor deposition. It has a TV of 200 °C however, so if only lower
delivery temperatures are desired 2a is likely the preferred candidate. These are two excellent
candidates for the development of ALD processes that include gold, and they are currently under
investigation in our group.
Figure 9. Figure of merit “minefield” plot. Green circles indicate a positive figure of merit while red
circles indicate a negative figure of merit. The σ value of each compound is shown as the radius
of each circle in arbitrary units.
DFT study
To better understand the reason behind the high figures of merit obtained for phosphine and NHC
systems, we undertook a density functional theory (DFT) study of compounds 2a, 2d, 3a, and 3d.
This article is protected by copyright. All rights reserved.

10.1002/ejic.201901087
European Journal of Inorganic Chemistry
The calculations were first performed using model compounds 2a*, 2d*, 3a*, and 3d* where the
PMe3 and NHC ligands were replaced with proto analogues (PH3 and N,N’-dihydroimidazolidin-
2-ylidene (H2-NHC), to examine the energy required to break the Au-R and Au-L bonds. All
calculations were carried out using the 휔B97X-D functional[40,41] and the iMCP-SR2 model core
potentials and basis sets.[42] Scalar-relativistic model core potentials are known to provide
satisfactory description for heavy elements like Au.[43] The dissociation energies of the R– and L
ligand fragments were determined by calculating the difference in energy between the free ligand
in question and the remaining Au-L or Au-R fragment. Three scenarios were considered:
dissociation of the coordinative ligand, forming a neutral R-Au fragment and a neutral L fragment;
homolytic cleavage of the R-Au bond resulting in R• and •Au-L fragments; and heterolytic
+
cleavage of the R-Au bond resulting in R- and Au-L fragments (Table 4). Heterolytic cleavage
was discounted after considering the results of 3a* and 3d*, because it was consistently the
highest energy case and was therefore unlikely to contribute to the first thermolysis events of the
molecules. For 2a, 2d, 3a, and 3d, only dissociation of the neutral L ligand was considered since
it is known to be the primary decomposition pathway for alkylgold(I) compounds. It was also the
lowest energy case, implying it was the largest contributor to thermal decomposition.
Table 4. Calculated energies and measured decomposition temperatures of the three
ligand dissociation cases for model compounds and synthesized compounds.
Dissociation of Homolytic Heterolytic
coordinative
ligand (kJ/mol) (kJ/mol)
cleavage
cleavage
(kJ/mol)
Compound
Td (°C)a
CH2(SiMe3)-Au-PH3
(2a*)
CH2(SiMe3)-Au-PMe3
(2a)
119
171
215
293
-
-
-
-
-
145
-
CH2(SiMe3)-Au-H2NHC
(2d*)
452
CH2(SiMe3)-Au-tBuNHC
(2d)
221
153
212
263
277
-
-
184
-
C6F5-Au-PH3 (3a*)
422
727
C6F5-Au-PMe3 (3a)
C6F5-Au-H2NHC (3d*)
C6F5-Au-tBuNHC (3d)
-
-
185
-
593
-
665
-
300
a Onset of self-decomposition as measured by DSC.
The model compounds 2a*, 2d*, 3a*, and 3d* mirrored the general trends of thermal stability that
were observed experimentally. The energy required for homolytic R-Au bond cleavage increased
from CH2(SiMe3) to C6F5 which reflects the tendency for C6F5 ligands to resist reductive
elimination. Exchanging PH3 for H2-NHC resulted in an increase in both Au-L dissociation energy
and R-Au homolytic bond cleavage energy, which is due to the relatively higher electron donating
ability of the NHC ligand. The Au-L bond dissociation energies increased from CH2(SiMe3) to C6F5
and from PMe3 to NHC which is reflective of the σ-withdrawing capability of the C6F5 ligand as
well as the increased σ-donor ability of the NHC compared to PMe3.
Compounds 2a and 3a show significantly higher Au-L bond dissociation energy than 2a* and 3a*
since PH3 is a much worse σ-donor than PMe3 due to the electron donating Me groups being
This article is protected by copyright. All rights reserved.

10.1002/ejic.201901087
European Journal of Inorganic Chemistry
directly bound to the ligating atom (P). The change in N-substituents in the NHC compounds has
comparatively little effect on the Au-L bond strength since the alkyl group is relatively remote from
the ligating C atom. Compounds 2d* and 3d* were calculated to have Au-L bond dissociation
energies that were comparable to 2d and 3d.
Thermal stability observed by DSC correlates well to the calculated Au-L bond dissociation
energies (Figure 10). Since the decomposition of these compounds is known to proceed via the
rate-limiting step of Au-L bond dissociation, a linear correlation between the observed
decomposition temperature and the calculated Au-L bond dissociation energy is expected and is
observed for these compounds. The calculational and experimental thermal results corroborate
each other.
Figure 10. Correlation between ab initio-calculated Au-L bond dissociation energies and the
observed self-decomposition temperatures for 2a,d and 3a,d.
Being able to predict a compound’s thermal stability from ligand choice is a useful tool that allows
for rational design and theoretical evaluation of different ligand classes before undertaking
synthesis and testing. Furthermore, for a single type of ligand, if the nature of alkyl groups has
little effect on the overall thermal stability of the compound (such as 3d* ≈ 3d) then these alkyl
groups can be repurposed to introduce volatility or other desirable properties for the potential
precursor. The implementation of this strategy is of ongoing interest to us in the further
development of vapor deposition precursors for Au and other metals.
Conclusions
A family of 12 gold(I) compounds were synthesized and analyzed for their use as potential
precursors for gold metal vapor deposition applications. Single crystal x-ray diffraction was used
to structurally characterize 8 of these compounds to assess the effect of the anionic and
coordinative ligands on intermolecular interactions. Compounds with Me and CH2(SiMe3) ligands
displayed fewer intermolecular interactions than those with C6F5 ligands due to the ability of the
latter to undergo π-stacking and p-π interactions. These strong intermolecular interacctions
resulted in a substantial decrease in the volatility of C6F5-containing compounds. Ylide compound
3b displays more intermolecular interactions than TMS-ylide compound 3c, which translated to
This article is protected by copyright. All rights reserved.

10.1002/ejic.201901087
European Journal of Inorganic Chemistry
an increase in volatility of TMS-ylide containing compounds vs. their ylide counterparts. TGA and
DSC were used to assess the volatility and thermal stability of the family of compounds, and it
was found that CH2(SiMe3) provides higher kinetic stability than Me. NHC and C6F5 ligands
provide substantially higher thermodynamic stability to the gold(I) center, but adversely affected
volatility. To deconvolute these effects we derived a figure of merit with which the viability of all
gold(I) compounds may be compared for vapor deposition. 3d has the highest merit, but is only
useful at high delivery temperatures, while 2a is most suitable for lower delivery temperature
applications. We are currently studying these two compounds as precursors in ALD process
development. Using DFT we found a linear correlation between the calculated Au-L bond strength
and the measured onset of self-decomposition as measured by DSC. In future studies, using this
computational method for the design of vapor deposition precursors should inform the process of
potential candidates prior to undertaking chemical synthesis.
Experimental Section
Crystallography
A crystal of the desired compound was mounted from Paratone-N oil on an appropriately sized
MiTeGen MicroMount. The data were collected on a Bruker APEX II charge-coupled-device
(CCD) diffractometer, with an Oxford 700 Cryocool sample cooling device. The instrument was
equipped with graphite-monochromated Mo Kα radiation (λ = 0.71073 ꢀ; 30 mA, 50 mV), with
MonoCap X-ray source optics. For data collection, four ω-scan frame series were collected with
0.5° wide scans, 5-30 second frames and 366 frames per series at varying φ angles (φ = 0°, 90°,
180°, 270°). Data collection, unit cell refinement, data processing and multi-scan absorption
correction were applied using the APEX211 or APEX312 software packages. The structures were
solved using SHELXT13 and all non-hydrogen atoms were refined anisotropically with SHELXL14
using a combination of shelXle15 and OLEX216 graphical user interfaces. Unless otherwise noted,
all hydrogen atom positions were idealized and ride on the atom to which they were attached.
The final refinement included anisotropic temperature factors on all non- hydrogen atoms. Details
of crystal data, data collection, and structure refinement are listed in Table 1. Figures were made
using UCSF-Chimera.17 Crystal structures were deposited in the (reference numbers 1915544,
1915545, 1915546, 1915547, 1915548, 1915549, 1915550, and 1915551)
Thermogravimetric analysis (TGA) and differential scanning calorimetry (DSC)
TGA was performed on a TA Instruments Q500 instrument which was housed in a nitrogen-filled
MBraun glovebox. The purge gas during TGA and DSC experiments was nitrogen gas of (99.999
%, 5.0) and (99.998 %, 4.8) respectively. In a typical TGA experiment 10.000 ± 2 mg of analyte
was placed in a platinum pan whose mass was monitored by the instrument during the analysis.
For DSC experiments, due to the volatile nature of the analytes and their decomposition products,
mass loadings greater than 1.000 mg often lead to a rupturing of the hermetically sealed aluminum
pans. Thus, small mass loadings of 0.300 ± 0.200 mg were used for typical DSC experiment.
DSC samples were hermetically sealed in aluminum pans inside the glovebox before analysis.
Both TGA and DSC temperature ramp rates were 10.0 °C / min. Langmuir vapour pressure
equations were derived from TGA data using a previously reported method with Cu(tmhd)2 as the
calibrant.[39,44]
Synthetic procedures
This article is protected by copyright. All rights reserved.

10.1002/ejic.201901087
European Journal of Inorganic Chemistry
CAUTION: After synthesizing the compounds in this manuscript, we learned that
suspensions of pentafluorophenyllithium[45] have been reported to explode violently upon
standing even when prepared at -78 °C under inert atmosphere. People performing this
reaction have been seriously injured by such explosions in other laboratories. We must
therefore recommend that the reaction to prepare pentafluorophenyllithium and its
subsequent use to synthesize (THT)AuC6F5 be performed behind a blast shield in small
amounts, or not at all.
General details
All manipulations were performed under an inert atmosphere of dry nitrogen gas using an MBraun
Labmaster 130 glovebox or standard Schlenk techniques unless otherwise stated. All synthesized
compounds were treated as light sensitive materials during synthetic procedures, and were stored
at -35 °C in the freezer of a glovebox. NMR spectra were collected at room temperature on a
Bruker 300 MHz or a Jeol 400 MHz spectrometer and were referenced to an internal standard of
tetramethylsilane (TMS) in the case of CDCl3 or residual protio solvent signal in the case of C6D6
(7.16 ppm relative to TMS). C6D6 was purchased from Aldrich and was degassed using freeze-
pump-thaw cycles and subsequently stored over activated 4 Å molecular sieves under inert gas.
Trimethylmethylene phosphorane (Ylide), trimethyl(trimethylsilylmethylene) phosphorane (TMS-
ylide),[46] (THT)AuCl, (PMe3)AuCl,[47] and N,N’-di-tert-butylimidazolidinium chloride[48] were
prepared according to literature procedures. Trimethylphosphine (PMe3) was prepared by the
literature procedure using iodomethane instead of bromomethane.[49] Tetrahydrothiophene
pentafluorophenylgold(I), (THT)AuC6F5, was prepared using chloropentafluorophenylbenzene
instead of the bromo- derivative.[45] Methyllithium (1.6
M
in diethyl ether),
(trimethylsilyl)methyllithium (1.0 M in pentane), n-butyllithium (1.6 M in hexanes), potassium
hydride (30 weight % dispersion in mineral oil), sodium tert-butoxide, tetrahydrofuran (anhydrous),
pentane (anhydrous), and dichloromethane (anhydrous) were purchased from Aldrich and used
as received. Diethyl ether, hexanes, and toluene were purchased from VWR chemicals and
purified using an MBraun solvent purification system prior to use. Gold metal (99.99 %) was
purchased at market price and was used as received. HAuCl4 • x H2O was obtained from the
digestion of gold metal by chlorine gas as previously described, with the modification of using up
to 1 Troy ounce of (32.2 g) gold metal in 250 mL of H2O.
Elemental analysis was performed by the elemental analysis laboratory at the Université de
Montréal.
N,N’-di-tert-butylimidazolidin-2-ylidene
A modification of a previously reported procedure was used.[32] To a 500 mL Schlenk flask
containing 150 mL of tetrahydrofuran was added 10.01 g (45.76 mmol) of dry N,N-di-tert-
butylimidazolidinium chloride, 4.366 g (181.9 mmol) of sodium hydride, 0.262 g (2.7 mmol) of
sodium tert-butoxide, and a teflon-coated stir bar. This white suspension was sealed and stirred
overnight, at which time it was filtered through a plug of celite in a medium frit into a 250 mL
Schlenk flask which contained a Teflon-coated stir bar. Approximately two-thirds of the THF was
removed by evaporation, then the flask was cooled in a -20 °C bath and the residual
tetrahydrofuran was stripped under high vacuum leaving an off-white solid in the flask which is
the frozen free carbene product. This solid was then purified by static high-vacuum (10 mTorr)
distillation using a warm water pot flask (approx. 40 °C) and a -78 °C receiving flask. The clear
crystalline solid caught in the receiving flask was allowed to warm to room temperature and liquify
This article is protected by copyright. All rights reserved.

10.1002/ejic.201901087
European Journal of Inorganic Chemistry
under nitrogen, and was then transferred into a glovebox for storage in a -35 °C freezer. Yield =
7.501 g, 90%. NMR analysis was consistent with the reported literature values.
Trimethylphosphine methylgold(I), (PMe3)AuMe (1a)
4.151 g (13.46 mmol) of (PMe3)AuCl was suspended in 150 mL of diethyl ether and cooled to -78
°C. Then, 8.41 mL (1.6 M in diethyl ether, 13.50 mol) of methyllithium solution was then added
dropwise over 5 minutes. The resulting yellow suspension was stirred at -78 °C for 1 hour, and
was then allowed to warm to room temperature, at which point the colour of the suspension had
returned to white. Then the reaction mixture was cooled to 0 °C and quenched by adding 20 mL
of distilled and degassed water dropwise. The resulting biphasic mixture was separated, and the
ethereal layer was dried over MgSO4. After filtration the solvent was removed using a rotary
evaporator leaving a slightly purple crystalline solid. This material was purified by sublimation (50
°C, 10 mTorr) which gave a white crystalline solid and left a purple-brown coloured residue in the
pot. Yield 3.007 g (77 %).
Tm (DSC) = 62 °C.1H-NMR (300 MHz, C6D6):  = 1.22 (d, 3JH-P = 8.7 Hz, 3H, Au-CH3), 0.62 (d, 2JH-
13
1
P = 8.7 Hz, 9H, P-CH3). C-NMR (75.5 MHz, C6D6):  = 15.56 (d, JC-P = 29.1 Hz, P-CH3), 8.73
(d, 2JC-P = 102.5 Hz, Au-CH3). 31P-NMR (121.5 MHz, C6D6):  = 13.84.
Methyl(trimethylphosphoniummethylide)gold(I), (Ylide)AuMe (1b)
Modifying the previously reported procedure,[46] 0.144 g (0.501 mmol) of (PMe3)AuMe was
dissolved in 10 mL of diethyl ether. To this was added 0.050 mL (ρ = 0.90 g / mL. 0.499 mmol) of
neat trimethylphosphoniummethylide (ylide) which immediately produced a white precipitate. This
suspension was stirred for one hour, filtered through a medium frit, washed with 3 x 1 mL of cold
pentane and dried under high vacuum. Yield = 0.131 g, 87% of a white powder.
Tm (DSC) = N/A. Decomposition begins at 107 °C. 1H-NMR (300 MHz, C6D6):  = 1.16 (d, 4JH-P
=
2
1.2 Hz, 3H, Au-CH3), 0.61 (d, 2JH-P = 12.9 Hz, 9H, P-CH3), 0.45 (d, JH-P = 12.6 Hz, 2H, P-CH2).
13
1
1
C-NMR (75.5 MHz, C6D6):  = 18.02 (d, JC-P = 36.8 Hz, P-CH2), 15.11 (d, JC-P = 56.2 Hz, P-
CH3), 2.52 (d, 3JC-P = 3.0 Hz, Au-CH3). 31P-NMR (121.5 MHz, C6D6):  = 25.83.
Methyl(trimethylphosphonium(trimethylsilyl)methylide)gold(I), (TMS-ylide)AuMe (1c)
Modifying the previously reported procedure,[30] 0.106 g (0.368 mmol) of (PMe3)AuMe was
dissolved in 10 mL of diethyl ether. To this was added 0.114 g (0.702 mmol)
trimethylphosphonium(trimethylsilyl)methylide (TMS-ylide) and the solution was stirred overnight.
The solvent was removed under high vacuum resulting leaving an off-white solid. Yield = 0.117 g
(84 %). Crystals of this compound suitable for X-ray crystallography were grown from hot toluene.
1
4
Tm (DSC) = 129 °C (dec.). H-NMR (300 MHz, C6D6):  = 1.16 (d, JH-P = 0.9 Hz, 3H, Au-CH3),
0.72 (d, 2JH-P = 12.6 Hz, 9H, P-CH3), 0.25 (s, 9H, Si-CH3), -0.17 (d, 2JH-P = 18.3 Hz, 1H, P-CH3).
13
1
1
C-NMR (75.5 MHz, C6D6):  = 17.20 (d, JC-P = 54.1 Hz, P-CH3), 16.58 (d, JC-P = 33.4 Hz, P-
CH2), 4.46 (d, JC-P = 4.7 Hz, Si-CH3), 0.77 (d, JC-P = 4.0 Hz, Au-CH3). 31P-NMR (121.5 MHz,
C6D6):  = 20.00
3
3
*Note: The excess of TMS-ylide was necessary to push the reaction to completion. 1:1
stoichiometry only converted 75 % of 1a to 1c.
This article is protected by copyright. All rights reserved.

10.1002/ejic.201901087
European Journal of Inorganic Chemistry
N,N’-di-tert-butylimidazolidin-2-ylidene methylgold(I), (NHC)AuMe (1d)
1.554 g (5.40 mmol) of (PMe3)AuMe and 1.588 g (8.71 mmol) of N,N’-di-tert-butylimidazolidin-2-
ylidene were added to 30 mL of toluene and the solution was stirred overnight. The next day, a
very small amount of purple precipitate was present in the otherwise clear solution. The solution
was filtered through a medium frit, and the solvent was removed under high vacuum. Then the
material was dried for a further 3 hours at 40 °C under high vacuum to remove excess free NHC.
The resulting slightly purple white solid was then purified by sublimation (100 °C, 10 mTorr) giving
a white crystalline solid. Yield = 2.002 g (94%). Crystals of this compound suitable for X-ray
crystallography were grown from hot toluene.
Tm (DSC) = 111 °C. Elemental analysis: Calcd. for C12H25N2Au: C, 36.51; H, 6.34; N, 7.10.
Found: C, 36.60; H, 6.26; N, 6.97. 1H-NMR (300 MHz, C6D6):  = 2.65 (s, 4H, CH2-CH2), 1.52 (s,
18H, C(CH3)3), 1.20 (s, 3H, Au-CH3). 13C-NMR (75.5 MHz, C6D6):  = 221.74 (s, NCN), 56.15 (s,
C(CH3)3), 46.31 (s, CH2-CH2), 31.35 (s, C(CH3)3), -3.26 (s, Au-CH3).
*Note: The excess of free NHC was used in order to push the reaction to completion. At 1:1
stoichiometry, the starting material 1a persisted in ~25 mol %, and it is difficult to purify 1d when
1a is present.
Trimethylphosphine trimethylsilylmethylgold(I), (PMe3)AuNeoSi (2a)
7.80 g (25.3 mmol) of (PMe3)AuCl was suspended in 150 mL of diethyl ether and cooled to -78
°C. (Trimethylsilyl)methyllithium solution (1.0 M, 27.8 mL) was added dropwise over 20 minutes.
The mixture was then allowed to warm to room temperature. The resulting white suspension was
then cooled to 0 °C and quenched by adding 25 mL of distilled and degassed water dropwise
over 10 minutes. The resulting biphasic mixture was separated, the brown-purple coloured water
layer was washed with 2 x 20 mL of diethyl ether, and the organic layers were combined and
dried over MgSO4. After filtration, the solvent was removed using a rotary evaporator and the
resulting dark purple liquid was purified by vacuum distillation (90 °C, 10 mTorr) giving a clear
liquid. Yield 7.735 g (85 %). This compound freezes upon refrigeration at 5 °C.
1H-NMR (300 MHz, C6D6):  = 0.84 (d, 3JH-P = 10.5 Hz, 2H, Au-CH2), 0.54 (d, 2JH-P = 8.7 Hz, 9H,
P-CH3), 0.51 (s, 9H, Si-CH3). 13C-NMR (75.5 MHz, C6D6):  = 15.58 (d, 2JC-P = 85.7 Hz, Au-CH2),
15.39 (d, 1JC-P = 29.6 Hz, P-CH3), 4.78 (d, 4JC-P = 2.2 Hz, Si-CH3). 31P-NMR (121.5 MHz, C6D6): 
= 12.70
Trimethylsilylmethyl(trimethylphosphoniummethylide)gold(I), (Ylide)AuNeoSi (2b)
Modifying the previously reported procedure,[30] 0.481 g (1.33 mmol) of (PMe3)AuNeoSi was
dissolved in 10 mL of diethyl ether in a glovebox. To this was added 0.233 mL (ρ = 0.90 g / mL.
2.33 mmol) of neat trimethylphosphoniummethylide (ylide). The resulting solution was stirred
overnight and then dried under high vacuum leaving a white solid. Yield 0.411 g (82 %). Crystals
of this compound suitable for X-ray crystallography were grown from hot toluene.
Tm (DSC) = 134 °C. 1H-NMR (300 MHz, C6D6):  = 0.62 (br, 2H, Au-CH2), 0.61 (d, 2JH-P = 12.9 Hz,
9H, P-CH3), 0.57 (s, 9H, Si-CH3), 0.42 (d, 2JH-P = 12.6 Hz, 2H, P-CH2). 13C-NMR (75.5 MHz, C6D6):
1
1
3
 = 17.36 (d, JC-P = 37.3 Hz, P-CH2), 15.04 (d, JC-P = 56.2 Hz, P-CH3), 8.47 (d, JC-P = 2.6 Hz,
Au-CH2), 5.01 (s, Si-CH3). 31P-NMR (121.5 MHz, C6D6):  = 25.51
This article is protected by copyright. All rights reserved.

10.1002/ejic.201901087
European Journal of Inorganic Chemistry
Trimethylsilylmethyl(trimethylphosphonium(trimethylsilyl)methylide)gold(I), (TMS-ylide)AuNeoSi
(2c)
Modifying the previously reported procedure,[30] 0.121 g (0.335 mmol) of (PMe3)AuNeoSi and
0.062 g (0.382 mmol) of trimethylphosphonium(trimethylsilyl)methylide (TMS-ylide) were
dissolved in 4 mL of pentane. The resulting solution was stirred overnight, and then the solvent
was removed leaving a white crystalline solid which was purified by recrystallization from a
minimum of pentane. Yield 0.084 g (56 %).
Tm (DSC) = 66 °C. 1H-NMR (300 MHz, C6D6):  = 0.72 (d, 2JH-P = 12.6 Hz, 9H, P-CH3), 0.65 (br,
2H, Au-CH2), 0.54 (s, 9H, NeoSi-CH3), 0.23 (s, 9H, TMS-ylide-Si-CH3), -0.21 (d, 2JH-P = 18.3 Hz,
13
1
1
1H, P-CH). C-NMR (75.5 MHz, C6D6):  = 16.72 (d, JC-P = 42.7 Hz, P-CH3), 15.97 (d, JC-P
=
3
27.6 Hz, P-CH2), 6.94 (s, Au-CH2), 4.37 (s, NeoSi-CH3), 4.03 (d, JC-P = 3.6 Hz, TMS-ylide-Si-
CH3). 31P-NMR (121.5 MHz, C6D6):  = 19.80
N,N’-di-tert-butylimidazolidin-2-ylidene trimethylsilylmethylgold(I), (NHC)AuNeoSi (2d)
1.551 g (4.31 mmol) of (PMe3)AuNeoSi and 0.794 g (4.36 mmol) of N,N’-di-tert-butylimidazolidin-
2-ylidene were added to 30 mL of diethyl ether and the solution was stirred overnight. The next
day, the solution was evaporated to dryness leaving a white crystalline material. Then the material
was dried for a further 3 hours at 40 °C under high vacuum to remove excess free NHC. The
resulting white solid was then purified by sublimation (100 to 120 °C, 10 mTorr) giving a white
crystalline solid. Yield = 1.754 g (87 %). Crystals of this compound suitable for X-ray
crystallography were grown from hot toluene.
Tm (DSC) = 105 °C. Elemental analysis: Calcd for C15H33N2AuSi: C, 38.58; H, 7.07; N, 6.00.
Found: C, 37.95; H, 6.97; N, 5.67. 1H-NMR (300 MHz, C6D6):  = 2.58 (s, 4H, CH2-CH2), 1.49 (s,
18H, C(CH3)3), 0.75 (s, 2H, Au-CH2), 0.54 (s, 9H, Si-CH3). 13C-NMR (75.5 MHz, C6D6):  = 219.94
(NCN), 56.15 (C(CH3)3), 46.23 (CH2-CH2), 31.26 (C(CH3)3), 4.30 (Si-CH3), 3.30 (Au-CH2).
Trimethylphosphine pentafluorophenylgold(I), (PMe3)AuC6F5 (3a)
0.550 g (1.22 mmol) of (THT)AuC6F5 and 0.142 g (1.82 mmol) of PMe3 were dissolved in 15 mL
of diethyl ether and the solution was stirred overnight. The next day the solution was filtered
through a medium frit and evaporated to dryness leaving a white solid. This solid was then purified
by sublimation (100 to 120 °C, 10 mTorr). Yield = 0.503 g (94 %). Crystals of this compound
suitable for x-ray crystallography were grown from hot toluene.
Tm (DSC) = 151 °C. 1H-NMR (300 MHz, CDCl3):  = 1.51 (d, 2JH-P = 8.1 Hz, 9H, P-CH3). 13C-NMR
(75.5 MHz, CDCl3):  = 15.64 (d, 1JC-P = 35.2 Hz, P-CH3). 31P-NMR (121.5 MHz, CDCl3):  = -5.00.
19
3
F-NMR (282.4 MHz, CDCl3):  = -118.51 (m, ortho-F), -160.63 (t, JF-F = 19.8 Hz, para-F), -
164.49 (m, meta-F).
Pentafluorophenyl(trimethylphosphoniummethylide)gold(I), (Ylide)AuC6F5 (3b)
0.120 g (0.273 mmol) of (PMe3)AuC6F5 and 0.031 mL (ρ = 0.90 g / mL. 0.310 mmol) of
trimethylphosphoniummethylide (ylide) were added to 5 mL of diethyl ether which caused a white
precipitate to form. The mixture was stirred for one hour and then filtered, washes with diethyl
ether and dried under high vacuum giving a white crystalline solid. This material was purified by
This article is protected by copyright. All rights reserved.

10.1002/ejic.201901087
European Journal of Inorganic Chemistry
recrystallization from hot toluene, which also produced crystals that were suitable for x-ray
crystallography. Yield = 0.080 g (65 %).
Tm (DSC) = 157 °C. Elemental analysis: Calcd for C10H11AuF5P: C, 26.42; H, 2.42. Found: C,
26.59; H, 2.50. 1H-NMR (300 MHz, C6D6):  = 0.56 (d, 2JH-P = 9.6 Hz, 2H, P-CH2), 0.42 (d, 2JH-P
9.6 Hz, 9H, P-CH3. C-NMR (75.5 MHz, C6D6):  = 14.26 (d, JC-P = 42.7 Hz, P-CH3), 12.52 (d,
1JC-P = 28.9 Hz, P-CH2). 31P-NMR (121.5 MHz, C6D6):  = 26.50. 19F-NMR (282.4 MHz, C6D6):  =
-116.23 (m, ortho-F), -161.29 (t, 3JF-F = 15.2 Hz, para-F), -162.86 (m, meta-F).
=
13
1
Pentafluorophenyl(trimethylphosphonium(trimethylsilyl)methylide)gold(I),
(TMS-ylide)AuC6F5
(3c)
0.478 g (2.94 mmol) of trimethylphosphonium(trimethylsilyl)methylide (TMS-ylide) was added to
a solution of 1.070 g (2.45 mmol) (THT)AuC6F5 in 10 mL of diethyl ether causing a white
precipitate to form immediately. The suspension was allowed to stir overnight and was then
stripped of its solvent, re-dissolved in a minimum of toluene, filtered, and left in a -35 °C freezer
overnight to recrystallize. Clear platelike crystals were observed in a mother liquor of brown-red
solution. These crystals were of suitable quality for crystallography (0.211 g, 33%). A second
batch of crystals was grown by concentrating the mother liquor by 2/3, filtering and storing at -35
°C overnight (0.114 g,18%). Yield = 0.325 g (51 %).
Tm (DSC) = 150 °C. Elemental analysis: Calcd for C13H19AuF5PSi: C, 29.64; H, 3.61. Found: C,
1
29.37; H, 3.76. H-NMR (300 MHz, C6D6):  = 0.62 (d, 2JH-P = 12.6 Hz, 9H, P-CH3), 0.17 (s, Si-
CH3), 0.03 (d, 2JH-P = 18.3 Hz, 1H, P-CH). 13C-NMR (75.5 MHz, C6D6):  = 16.61 (d, 1JC-P = 57.2
Hz, P-CH3), 13.20 (d, JC-P = 36.4 Hz, P-CH), 4.14 (d, 3JC-P = 4.6 Hz, Si-CH3). 31P-NMR (121.5
1
19
MHz, C6D6):  = 23.65. F-NMR (282.4 MHz, C6D6):  = -118.67 (m, ortho-F), -163.36 (t, 3JF-F
20.3 Hz, para-F), -165.07 (m, meta-F).
=
N,N’-di-tert-butylimidazolidin-2-ylidene pentafluorophenylgold(I), (NHC)AuC6F5 (3d)
0.278 g (0.615 mmol) of (THT)AuC6F5 and 0.168 g (0.922 mmol) of N,N’-di-tert-butylimidazolidin-
2-ylidene were dissolved in 5 mL of diethyl ether and the solution was stirred overnight. The
solvent was removed, and the residue was extracted with 3 x 3 mL of toluene, filtered, and dried
under high vacuum at 40 °C for at least 3 hours to remove excess free NHC, leaving a slightly
purple solid. This material was purified by sublimation (110 to 130 °C, 10 mTorr) which gave a
white crystalline solid on the cold finger and a purple residue in the pot. Crystals suitable for x-ray
crystallography were grown from hot toluene. Yield = 0.292 g (89%).
Tm (DSC) = 177 °C. Elemental analysis: Calcd for C17H22AuF5N2: C, 37.33; H, 4.026; N, 5.13.
1
Found: C, 37.42; H, 4.15; N, 4.99. H-NMR (300 MHz, C6D6):  = 2.55 (s, 4H, CH2-CH2), 1.42 (s,
18H, C(CH3)3). 13C-NMR (75.5 MHz, C6D6):  = 211.05 (NCN), 56.44 (C(CH3)3), 46.21 (CH2-CH2),
31.25 (C(CH3)3. 19F-NMR (282.4 MHz, C6D6):  = -117.46 (m, ortho-F), -162.16 (t, 3JF-F = 20.1 Hz,
para-F), -165.12 (m, meta-F).
Associated Content
Author Information
Corresponding Author
*E-mail: mbgriffi016@gmail.com
This article is protected by copyright. All rights reserved.

10.1002/ejic.201901087
European Journal of Inorganic Chemistry
ORCID
Matthew B. E. Griffiths: 0000-0002-3187-4064
Zachary S. Dubrawski: 0000-0003-3999-3874
Goran Bačić: 0000-0003-2437-2418
Jason D. Masuda: 0000-0002-6195-9691
Tao Zeng: 0000-0002-1553-7850
Seán T. Barry: 0000-0001-5515-4734
Notes
The authors declare no competing financial interest
Acknowledgements
M. B. E. G thanks the Natural Sciences and Engineering Research Council (NSERC) of Canada
for funding through the Alexander Graham Bell CGS-D scholarship. T. Z. thanks Carleton
University for the start-up grant (186853) and NSERC for funding (RGPIN-2016-06276). J. D. M.
thanks NSERC for funding in the form of a discovery grant and acknowledges the Canada
Foundation for Innovation (CFI) and Nova Scotia Research and Innovation Trust Fund for funding.
S. T. B. acknowledges the Natural Sciences and Engineering Research Council of Canada
(NSERC grant RGPIN-2014-06250) for funding.
References
[1]
V. Miikkulainen, M. Leskelä, M. Ritala, R. L. Puurunen, J. Appl. Phys. 2013, 113, DOI
10.1063/1.4757907.
[2]
[3]
M. B. E. Griffiths, P. J. Pallister, D. J. Mandia, S. T. Barry, Chem. Mater. 2016, 28, 44–46.
M. Mäkelä, T. Hatanpää, K. Mizohata, J. Räisänen, M. Ritala, M. Leskelä, Chem. Mater.
2017, 29, 6130–6136.
[4]
[5]
H. B. Profijt, S. E. Potts, M. C. M. van de Sanden, W. M. M. Kessels, J. Vac. Sci.
Technol. A Vacuum, Surfaces, Film. 2011, 29, 050801.
H. C. M. Knoops, J. W. Elam, J. A. Libera, W. M. M. Kessels, Chem. Mater. 2011, 23,
2381–2387.
[6]
[7]
[8]
R. J. Puddephatt, I. Treurnicht, J. Organomet. Chem. 1987, 319, 129–137.
J. Messelhäuser, E. B. Flint, H. Suhr, Appl. Surf. Sci. 1992, 54, 64–68.
R. G. Parkhomenko, S. V. Trubin, A. E. Turgambaeva, I. К. Igumenov, J. Cryst. Growth
2015, 414, 143–150.
[9]
K. Bernal Ramos, M. J. Saly, Y. J. Chabal, Coord. Chem. Rev. 2013, 257, 3271–3281.
[10] P. G. Gordon, A. Kurek, S. T. Barry, ECS J. Solid State Sci. Technol. 2014, 4, N3188–
N3197.
[11] T. H. Baum, P. B. Comita, in Chem. Met. CVD (Eds.: T.T. Kodas, M.J. Hampden-Smith),
This article is protected by copyright. All rights reserved.

10.1002/ejic.201901087
European Journal of Inorganic Chemistry
Wiley-VCH Verlag GmbH, Weinheim, Germany, 1994, pp. 303–327.
[12] A. Tamaki, J. K. Kochi, J. Organomet. Chem. 1973, 61, 441–450.
[13] M. M. B. Holl, P. F. Seidler, S. P. Kowalczyk, F. R. McFeely, Inorg. Chem. 1994, 33, 510–
517.
[14] F. Jansen, T. Kruck, Adv. Mater. 1995, 7, 297–300.
[15] W. Mowat, A. Shortland, G. Yagupsky, N. J. Hill, M. Yagupsky, G. Wilkinson, J. Chem.
Soc. Dalt. Trans. 1972, 533.
[16] B. Wozniak, J. D. Ruddick, G. Wilkinson, J. Chem. Soc. A Inorganic, Phys. Theor. Chem.
1971, 3116–3120.
[17] J. S. Price, P. Chadha, D. J. H. Emslie, Organometallics 2016, 35, 168–180.
[18] L. G. Vaughan, W. A. Sheppard, J. Organomet. Chem. 1970, 22, 739–742.
[19] R. Usón, A. Laguna, M. Laguna, A. Usón, Inorganica Chim. Acta 1983, 73, 63–66.
[20] H. Schmidbaur, R. Franke, Chem. Ber. 1975, 108, 1321–1328.
[21] J. P. Coyle, G. Dey, E. R. Sirianni, M. L. Kemell, G. P. A. Yap, M. Ritala, M. Leskelä, S.
D. Elliott, S. T. Barry, Chem. Mater. 2013, 25, 1132–1138.
[22] J. P. Coyle, E. R. Sirianni, I. Korobkov, G. P. A. Yap, G. Dey, S. T. Barry,
Organometallics 2017, 35, 2800–2810.
[23] M. B. E. Griffiths, S. E. Koponen, D. J. Mandia, J. F. McLeod, J. P. Coyle, J. J. Sims, J. B.
Giorgi, E. R. Sirianni, G. P. A. Yap, S. T. Barry, Chem. Mater. 2015, 27, 6116–6124.
[24] N. Boysen, T. Hasselmann, S. Karle, D. Rogalla, D. Theirich, M. Winter, T. Riedl, A. Devi,
Angew. Chemie Int. Ed. 2018, 57, 16224–16227.
[25] M. K. Denk, A. Thadani, K. Hatano, A. J. Lough, Angew. Chemie Int. Ed. English 1997,
36, 2607–2609.
[26] H. Schmidbaur, R. Franke, Angew. Chemie 1973, 85, 449–450.
[27] S. Gaillard, P. Nun, A. M. Z. Slawin, S. P. Nolan, Organometallics 2010, 29, 5402–5408.
[28] A. Battisti, O. Bellina, P. Diversi, S. Losi, F. Marchetti, P. Zanello, Eur. J. Inorg. Chem.
2007, 865–875.
[29] H. Schmidbaur, K. H. Raethlein, Inorganic Syntheses, Volume 18, 1978.
[30] H. Schmidbaur, R. Franke, Chem. Ber. 1975, 108, 1321–1328.
[31] J. M. López-de-Luzuriaga, M. Monge, M. E. Olmos, D. Pascual, T. Lasanta, Chem.
Commun. 2011, 47, 6795.
[32] K. Arentsen, S. Caddick, F. G. N. Cloke, Tetrahedron 2005, 61, 9710–9715.
[33] E. A. V. Ebsworth, T. E. Fraser, D. W. H. Rankin, Chem. Ber. 1977, 110, 3494–3500.
[34] H. Schmidbaur, A. Schier, Chem. Soc. Rev. 2012, 41, 370–412.
[35] R. W. Baker, P. J. Pauling, J. Chem. Soc. Dalt. Trans. 1972, 2264.
This article is protected by copyright. All rights reserved.

10.1002/ejic.201901087
European Journal of Inorganic Chemistry
[36] W. G. Carden, J. Pedziwiatr, K. A. Abboud, L. McElwee-White, ACS Appl. Mater.
Interfaces 2017, 9, 40998–41005.
[37] W. G. Carden, R. M. Thorman, I. Unlu, K. A. Abboud, D. H. Fairbrother, L. McElwee-
White, ACS Appl. Mater. Interfaces 2019, 11, 11976–11987.
[38] A. Wuttke, M. Feldt, R. A. Mata, J. Phys. Chem. A 2018, 122, 6918–6925.
[39] G. V. Kunte, S. A. Shivashankar, A. M. Umarji, Meas. Sci. Technol. 2008, 19, 025704.
[40] J.-D. Chai, M. Head-Gordon, J. Chem. Phys. 2008, 128, 084106.
[41] J.-D. Chai, M. Head-Gordon, Phys. Chem. Chem. Phys. 2008, 10, 6615.
[42] C. C. Lovallo, M. Klobukowski, J. Comput. Chem. 2003, 24, 1009–1015.
[43] T. Zeng, M. Klobukowski, J. Chem. Phys. 2009, 130, 204107.
[44] C. Colominas, K. H. Lau, D. L. Hildenbrand, S. Crouch-Baker, A. Sanjurjo, J. Chem. Eng.
Data 2001, 46, 446–450.
[45] R. Uson, A. Laguna, M. Laguna, D. A. Briggs, H. H. Murray, J. P. Fackler, Inorg. Synth.
2007, 26, 85–91.
[46] H. Schmidbaur, Inorg. Synth. 1978, 18, 135–140.
[47] M. B. E. Griffiths, P. J. Pallister, D. J. Mandia, S. T. Barry, Chem. Mater. 2016, 28, 44–46.
[48] S. Barry, J. P. Coyle, T. J. Clark, J. J. M. Hastie, Group 11 Mono-Metallic Precursor
Compounds and Use Thereof in Metal Deposition, 2011, WO2012155264.
[49] M. L. J. Leutkens, A. P. Sattelberger, H. H. Murray, J. D. Basil, J. P. J. Fackler, Inorg.
Synth. 1990, 28, 305.
This article is protected by copyright. All rights reserved.