Skip to main content
NCBI home page
ACS Materials Au logoLink to ACS Materials Au
. 2023 Jan 11;3(3):206–214. doi: 10.1021/acsmaterialsau.2c00075

Reductive Thermal Atomic Layer Deposition Process for Gold

Anton Vihervaara †,*, Timo Hatanpää †, Heta-Elisa Nieminen †, Kenichiro Mizohata ‡, Mykhailo Chundak †, Mikko Ritala †,*
PMCID: PMC10176613  PMID: 38089133

Abstract

graphic file with name mg2c00075_0015.jpg

In this work, we developed an atomic layer deposition (ALD) process for gold metal thin films from chloro(triethylphosphine)gold(I) [AuCl(PEt3)] and 1,4-bis(trimethylgermyl)-1,4-dihydropyrazine [(Me3Ge)2DHP]. High purity gold films were deposited on different substrate materials at 180 °C for the first time with thermal reductive ALD. The growth rate is 1.7 Å/cycle after the film reaches full coverage. The films have a very low resistivity close to the bulk value, and a minimal amount of impurities could be detected. The reaction mechanism of the process is studied in situ with a quartz crystal microbalance and a quadrupole mass spectrometer.

Keywords: atomic layer deposition, gold, thin films, reducing agent, gold halide adducts, germyl dihydropyrazine

Introduction

Throughout the human history, gold has held a unique place among the elements, a precious metal that does not tarnish is used as a currency and in ornamentation. However, its beauty is not the only desirable trait. Gold has unprecedented chemical stability, unique catalytic properties, and the third highest conductivity of all the elements, exceeded only by silver and copper. These properties indicate high potential for gold thin films in applications such as interconnects, contact platings, biosensors, and microelectromechanical system devices.1−4 Furthermore, gold nanorods have been used to make localized surface plasmon resonance biosensors.5

For the majority of these applications, especially in microelectronics, having high-quality thin films deposited uniformly on complex three-dimensional (3D) structures is an essential requirement. Atomic layer deposition (ALD) stands as the prominent method to meet the requirements. ALD relies on saturative alternate surface reactions. Excellent thickness control and superior conformality provide uniform coatings even on complex 3D structures.6,7

Although several chemical vapor deposition (CVD) processes for gold exist,8 only three ALD processes have been reported for gold thin films. Two of the processes require hydrogen and oxygen plasma, respectively,9,10 and the only thermal ALD process11 uses ozone to combust away the ligands of the gold precursor. Thermal ALD processes for gold nanoparticles have also been reported but are not considered here in detail.12,13

The first gold ALD process9 was plasma-enhanced ALD (PEALD) and used trimethyl(trimethylphosphine)gold(III) [Me3Au(PMe3)], oxygen plasma, and water at 120 °C. The resulting gold films had some impurities, most notably carbon (6.7 at. %). The growth rate was 0.5 Å/cycle. The other PEALD process for gold10 used Me3Au(PMe3) with hydrogen plasma at 50–120 °C and yielded films of considerably higher purity, with the total impurity content of less than 1 at. %. The resistivity was 6 μΩ cm for the thickest films. The growth rate was 0.3 Å/cycle.

PEALD processes typically yield smooth films with little impurities, and the processes operate at low temperatures. However, possible damage to the substrates needs to be considered. Furthermore, repeatability and limited film conformality present challenges for PEALD.14 Thermal ALD processes do not have these problems with plasma damage and film conformality. Thermal processes are also repeatable and can be used in batch processing, and the reactor design is simpler.

The first and so far the only thermal ALD process for gold thin films used Me2Au(S2CNEt2) as the gold precursor.11 Combined with ozone, self-limiting growth was observed at a deposition temperature of 180 °C. This process had a relatively high growth rate of 0.9 Å/cycle and yielded films with resistivities of 4 μΩ cm at best. The impurity contents were also low, with 2.9 at. % of oxygen and less than 1 at. % of other elements. However, this process is highly oxidative, and thus, there is a need for reductive alternative. The best attempt to this direction so far is the combination of Au(N(SiMe3)2)(PEt3) and dimethylamine borane [BH3(NHMe2)] that resulted in gold deposition, but the growth rate was less than 0.10 Å/cycle, and the growth was not fully self-limiting as required in ALD.13

In our recent research,15 we developed a new ALD process for nickel from novel precursors. 1,4-bis(trimethylgermyl)-1,4-dihydropyrazine ((Me3Ge)2DHP), a new reducing agent in ALD combined with dichlorobis(triethylphosphine)nickel(II) (NiCl2(PEt3)2), afforded high quality nickel thin films at a very low temperature of 110 °C. (Me3Ge)2DHP is analogous with the previously presented 1,4-bis(trimethylsilyl)-1,4-dihydropyrazine [(Me3Si)2DHP] reducing agent.16,17 The reducing capability of (Me3Si)2DHP is based on the nonaromatic dihydropyrazine being prone to oxidize and turn into aromatic pyrazine giving two electrons for reduction of, for example, metals. The reaction apparently involves a removal of the ligands L on the metal as L-SiMe3. Metal chlorides are especially reactive due to the thermodynamically very favorable release of alkylsilylchlorides. Our presumption is that the Me3Ge groups in place of the Me3Si groups make the compound more reactive because of the weaker Me3Ge–N bonds and still quite strong Me3Ge–Cl bonds formed in reactions with metal halides. In Ni ALD, (Me3Ge)2DHP indeed turned out to be a better reducing agent than (Me3Si)2DHP. Therefore, we proposed that (Me3Ge)2DHP could have good reactivity with other metal halides as well. In this paper, we demonstrate the first ever reductive thermal ALD process for gold thin films. In this process, a commercially available gold compound, chloro(triethylphosphine)gold(I) [AuCl(PEt3)], is combined with (Me3Ge)2DHP (Figure 1).

Figure 1.

Figure 1

Open in a new tab

Structures of AuCl(PEt3) and (Me3Ge)2DHP.

Similar to NiCl2(PEt3)2, also AuCl(PEt3) is a phosphine adduct of a metal chloride and hence expected to be reactive with (Me3Ge)2DHP. Indeed, high purity films were deposited at low temperatures of 160–180 °C. Furthermore, we investigate the reaction mechanism of the process.

Experimental Section

Film Deposition

Both precursor compounds, AuCl(PEt3) and (Me3Ge)2DHP, were synthesized in-house. Details about the reducing agent have been reported elsewhere.15 Details of the synthesis and thermal properties studies of AuCl(PEt3) are reported in the Supporting Information.

A F-120 hot-wall flow-type ALD reactor (ASM Microchemistry), with a pressure of approximately 10 mbar, was used in the experiments. The reactor uses inert gas valving to pulse the precursors. Pulse times varied from 1.0 to 6.0 s, and a purge time of 3.0 s was used for most of the depositions. N2 (AGA 5.0) was used as a carrier and purging gas. AuCl(PEt3) and (Me3Ge)2DHP were evaporated inside the reactor from open glass boats held at 160 and 45 °C, respectively. The depositions were carried out on 5 × 5 cm2 native oxide terminated Si substrates that were cut from wafers and on soda lime glass substrates of the same size. In addition, nucleation and film growth were studied on Ru, TiN, and Al2O3 films.

Film Characterization

Crystal structures were studied with X-ray diffraction using a PANalytical X’Pert Pro MPD X-ray diffractometer. A grazing incidence geometry was used with the Cu Kα (λ = 1.54 Å) radiation having an incidence angle of 1°. PANalytical Highscore Plus 4.1 software was used to analyze the results.

Energy-dispersive X-ray spectrometry (EDS) was used to measure film thicknesses. An oxford INCA 350 connected to a Hitachi scanning electron microscope was used for the measurements with an electron beam voltage of 20 kV. Film thicknesses were calculated from Au M X-ray lines with the GMRFILM program using a bulk density of 19.3 g/cm3 for gold.18 These thicknesses are nominal thicknesses and do not match with island heights of noncontinuous films. However, they do indicate the amount of the deposited gold and are therefore a good measure for the overall growth. An uncertainty of 5% was estimated for the thickness measurements due to a minor fluctuation of the electron beam current and possible deviations from the bulk density.

Elemental analysis of the film composition was carried out with time-of-flight elastic recoil detection analysis (ToF-ERDA) using a 40 MeV 127I7+ ion beam. Further elemental characterization was carried out with X-ray photoelectron spectroscopy (XPS). The XPS measurements were performed with a PREVAC system having an EA-15 hemispherical electrostatic energy analyzer and a RMC50 monochromatic X-ray source with Al Kα anode (1486.7 eV). The pressure during all the measurements was 10–10 mbar. The analyzer pass energy was 50 eV, and the slit was C 0.3 × 25. Before the measurements, the films were rastered for 1 h with 5 keV Ar+ ions over an area of 5 × 2 mm to remove oxygen and other impurities from the surface of the films. The peak fitting was carried out with CasaXPS by GL line shape.

Nucleation and morphology of the films were studied with a field emission scanning electron microscope (Hitachi S-4800). Morphology and roughness of the films were studied also with atomic force microscopy (AFM, Veeco Multimode V instrument). The Si probe with a nominal tip radius of 10 nm and a spring constant of 5 N/m was used to capture images in air from films deposited on Si substrates. Images were flattened to remove artefacts from sample tilt and scanner bow. Roughness was calculated as a root-mean-square value (Rq) from flattened images.

A four-point probe (CPS Probe Station, Cascade Microtech combined with a Keithley 2400 SourceMeter) was used to measure sheet resistances of films deposited on soda lime glass substrates. Sheet resistance was obtained as an average from nine measurement points uniformly distributed across the square-shaped substrate. The resistivity was calculated by multiplying the sheet resistance with the film thickness.

Reaction Mechanism Studies

Reaction mechanism studies were carried out in a F120 SAT flow-type ALD reactor (ASM Microchemistry) to which an in situ quartz crystal microbalance (QCM) and a quadrupole mass spectrometer (QMS) are attached. Details of the specially modified ALD reactor are described elsewhere.19 The pressure drop from the ALD reactor (1–5 mbar) to the required QMS pressure (10–5 mbar) is carried out by differential pumping through a 150 μm orifice that is at the reaction temperature. The deposition area in the reactor is enlarged to produce high enough amounts of reaction byproducts for detection with QMS. The total area of the glass substrates and walls is 3500 cm2. The process parameters were modified to fit the special ALD reactor, and the pulse lengths of AuCl(PEt3) and (Me3Ge)2DHP were 10 and 12 s, respectively. The purge times after both precursors were 30 s. The sublimation temperature was 166 °C for AuCl(PEt3) and 47 °C for (Me3Ge)2DHP. The deposition temperature was 180 °C.

QMS measurements were carried out with a Hiden Analytical HAL/3F 510RC mass spectrometer with an ionization energy of 70 eV. QCM measurements were carried out on the gold-coated quartz crystal with Maxtek TM 400 with a sampling rate of 20 Hz. The QCM head is placed in the deposition zone between the substrates and the QMS inlet.

Results and Discussion

Properties of the Gold Precursor

Although the gold precursor used in this study, chloro(trimethylphosphine)gold(I) [AuCl(PEt3)], is a known compound, its suitability for ALD remained unexplored. AuCl(PEt3) melts at 86.5–89.9 °C and can be sublimed with a quantitative yield under vacuum at 140–160 °C/0.1 mbar. The thermogravimetric (TG) curve measured in the flowing N2 atmosphere at atmospheric pressure (Figure 2) shows a one-step weight loss around 200–325 °C and a residue of 31.4% while the theoretical Au content of the compound is 56.2%. This indicates that a large fraction of the compound evaporates even at atmospheric pressure. The TG curve measured in the flowing N2 atmosphere at 10 mbar pressure (Figure 2) shows a single-step weight loss around 100–200 °C and a residue of 1.3%, proving almost complete evaporation of the compound. Thus, it seems that in condensed form, the compound is sufficiently stable at least up to 190 °C. Vapor pressures of the AuCl(PEt3) compound at different temperatures were estimated using the TG method.20 Vapor pressures of 0.1, 0.3, 0.5, and 1 mbar are achieved at 150, 163, 169, and 178 °C, respectively. The Antoine equation derived takes the form log P(mbar) = 6.47 ± 0.04 + 0.0364 ± 0.0003 × T(°C).

Figure 2.

Figure 2

Open in a new tab

Thermogravimetric analysis and SDTA curves measured for AuCl(PEt3) at atmospheric pressure (black) and 10 mbar pressure (red). Measurements were carried out under a flowing N2 atmosphere, and heating rates of 10 and 5 °C/min were used for 1 atm and 10 mbar measurements, respectively.

Gold Film Deposition and Growth Characteristics

AuCl(PEt3) was combined with (Me3Ge)2DHP at deposition temperatures of 160–180 °C. Highly reflective gold films were obtained within this temperature range (Figure 3). The lower limit was dictated by the sublimation temperature of the gold precursor (160 °C) and the upper limit by the decomposition of the same. The decomposition of AuCl(PEt3) was observed visually in several experiments as formation of AuCl in the hot end of the precursor source tube where the precursor molecules have long residence time. However, despite this minor decomposition, saturation with both precursors was achieved and ALD-type growth verified. The growth characteristics of the process are presented in Figure 4. Some results are presented as film thickness instead of the more commonly used growth rate. The reason for this choice is that the process has two growth regimes with significantly different growth rates (Figure 4d–f). This is a quite typical observation in ALD of metals and will be discussed below. Calculating the growth rate by dividing the final film thickness with the total number of deposition cycles would therefore provide a value that does not correspond to either of the actual growth rates. It is emphasized that we use nominal film thicknesses that are calculated from EDS results and thereby represent the total amount of gold deposited rather than, for example, island height.

Figure 3.

Figure 3

Open in a new tab

Reflective gold film deposited on soda lime glass at 180 °C.

Figure 4.

Figure 4

Open in a new tab

Growth characteristics of the AuCl(PEt3) + (Me3Ge)2DHP process. The film thickness as a function of deposition temperature (a), AuCl(PEt3) pulse length (b), and (Me3Ge)2DHP pulse length (c). Film thickness as a function of deposition cycles on Si (d), TiN (e), and Ru (f).

The film growth depended heavily on the deposition temperature. The film thickness almost doubled when the deposition temperature was increased from 160 to 180 °C (Figure 4a). Saturation tests were carried out at 180 °C using 500 cycles and 3.0 s purges. Self-limiting film growth was observed with 4.0 s pulses for both precursors (Figure 4b,c). The films were uniform in thickness across the substrate (Figure S1).

Film thickness as a function of number of deposition cycles was not linear, but two growth regimes were observed instead. At 180 °C, at the start of the deposition, the growth rate was approximately 4.1 Å/cycle, then decreased to 1.7 Å/cycle, and stayed constant from this point onward. As will be seen below, the change in the growth rate occurs when the films become continuous. The high growth rate at the beginning of the process can be attributed to the roughness of the films. Films that consist of islands have much larger surface area than films with smooth surfaces and thus have more adsorption sites for the precursors. When the films become continuous, the roughness decreases and so does the growth rate. However, the growth rate of 1.7 Å/cycle is still very high for a metal ALD process, being approximately 60% of a full gold monolayer thickness (2.88 Å, calculated from the lattice constants21). Finally, halving the purges from 3.0 to 1.5 s did not affect the growth rate.

AuCl(PEt3) was also tested with (Me3Si)2DHP as a reducing agent. At 180 °C, only gold islands (Figure S2) were obtained with (Me3Si)2DHP, whereas (Me3Ge)2DHP yielded a fully continuous film when the other process parameters were similar. Similarly, in our previous study15 on ALD of Ni from NiCl2(PEt3)2, only some scattered islands were achieved with (Me3Si)2DHP, whereas the combination of NiCl2(PEt3)2 and (Me3Ge)2DHP yielded a fully continuous nickel film.

Film Characterization

XRD measurements were carried out from films deposited at 160, 170, and 180 °C with the optimized parameters on silicon substrates. Regardless of the deposition temperature, all the diffractograms were similar and exhibited reflections matching with the cubic gold metal (Figure 5).

Figure 5.

Figure 5

Open in a new tab

Grazing-incidence X-ray diffractogram of Au films deposited at 160–180 °C. The red bars correspond to the ICDD PDF card 4–784 for gold.

Nucleation and morphology of the films on the different substrates were studied with scanning electron microscopy (SEM) (Figures 6 and 7). Films were deposited on Si, TiN, Al2O3, and Ru surfaces at 180 °C with the saturative pulses. On Si, a slight nucleation delay was observed during the first 10–20 cycles (Figure 4d). However, after 30 cycles, the growth was rapid, and the process had a growth rate of approximately 4.1 Å/cycle. Up to 175 cycles, the films consisted of islands of increasing size (Figure 6). When the films became continuous after 250 cycles, the growth rate decreased to approximately 1.7 Å/cycle. While the nucleation and overall growth were similar on Si and TiN (not shown), there was a clear difference with the other two surfaces. On Ru, the nucleation density was much higher, and the films were continuous after 175 cycles corresponding to a thickness of approximately 70 nm (Figure 7). This can be attributed to better wetting characteristics of gold on another noble metal. On Al2O3, there was a clear delay in nucleation. Even though the differences between the substrates were clear, there is no difference in the type of the growth itself as the two regimes were seen on all the substrates (Figure 4d–f).

Figure 6.

Figure 6

Open in a new tab

SEM images of gold films deposited with varying number of cycles at 180 °C on silicon.

Figure 7.

Figure 7

Open in a new tab

SEM images of gold films deposited with varying number of cycles at 180 °C on Ru and Al2O3 surfaces.

Morphology and roughness were also studied with AFM from the films deposited on Si at 180 °C with the saturative pulses. The root-mean-square roughness (Rq) of the films was around 6.3 nm after the first 30 cycles, then sharply rose to 18–20 nm, and finally settled around 16 nm when the films became continuous (Figure 8). This dramatic increase in roughness is attributed to the strong agglomeration, as evident in the SEM and AFM images (Figures 6 and 8). Between 30 and 70 cycles, approximately 80% of the islands coalesced into larger ones, greatly increasing the surface roughness. These films are very rough when compared to other gold ALD films. Gold films deposited with hydrogen plasma had Rq of 6.5 nm.10 Also, the gold films deposited with the Me2Au(S2CNEt2)–O3 thermal ALD process11 seemed smoother in SEM images (no Rq reported) but were closer to our films than the PEALD gold films in roughness.

Figure 8.

Figure 8

Open in a new tab

AFM images of Au films deposited with varying number of cycles at 180 °C on Si with their respective Rq values.

Elemental composition of the films was studied with ToF-ERDA measurements (Figure 9). Films were very pure with the total amount of impurities being less than 0.5 at. %. Furthermore, majority of the impurities was found at the film surface. The high film purity indicates fast and complete reactions between the precursors. XPS measurements further confirmed the film purity as nothing else than gold could be seen after removing the contamination layer from the ambient exposure by sputtering. The Au 4f5/2 and Au 4f7/2 peaks (Figure 10) were found at 87.74 and 84.02 eV, respectively.22 Wide-scan spectra can be found in the Supporting Information (Figure S3).

Figure 9.

Figure 9

Open in a new tab

Elemental depth profiles and atomic percentages from a 145 nm Au film deposited at 180 °C.

Figure 10.

Figure 10

Open in a new tab

Au 4f XPS spectrum of a 145 nm thick Au film deposited at 180 °C.

Resistivity of the films on glass substrates was measured with the four-point probe. A fully continuous film with a thickness of 166 nm had a resistivity of approximately 2.4 μΩ cm, which is very close to the bulk resistivity of gold (2.2 μΩcm).21 The low resistivity was expected as the films had a large grain size and high purity.

Mechanistic Studies

The growth mechanism was studied with a QCM and a QMS. The overall reaction is expected to proceed, as shown in reactions 1–2. A similar reaction mechanism was found for an ALD process of tin utilizing (Me3Si)2DHP and SnCl4.17 AuCl(PEt3) adsorbs on the surface, and the triethylphosphine adduct ligand detaches and desorbs (eq 1).

graphic file with name mg2c00075_m001.jpg 1

Next, (Me3Ge)2DHP reacts with AuCl forming gaseous byproducts (CH3)3GeCl and pyrazine and depositing Au (eq 2).

graphic file with name mg2c00075_m002.jpg 2

The QCM measurements show that the mass increase is rapid during the AuCl(PEt3) pulse, and a slight decrease occurs during the (Me3Ge)2DHP pulse (Figure 11a). The mass change m1 is the change after the AuCl(PEt3) pulse and purge. The mass change m0 is the change after a complete ALD cycle. The ratio of the mass changes m1/m0 is 1.15 ± 0.02. When the precursors were pulsed alone for reference before and after 10 ALD cycles, no mass increase was observed (Figure 11b).

Figure 11.

Figure 11

Open in a new tab

QCM measurements during (a) three ALD cycles and (b) seven (Me3Ge)2DHP reference pulses, 10 ALD cycles, and seven AuCl(PEt3) pulses. QMS measurement on m/z = 62 (c) during three ALD cycles and (d) during seven (Me3Ge)2DHP reference pulses, 10 ALD cycles, and seven AuCl(PEt3) reference pulses.

With QMS, the release of the triethylphosphine adduct ligand was studied first. Figure 11c shows a close up on the QMS measurement monitoring m/z = 62, [H2PC2H5]+ originating from the P(CH2CH3)3 adduct ligand, during three ALD cycles. Based on all the measured m/z values (Table S1), the adduct ligand was identified to release as triethylphosphine23 and mostly (90%) during the AuCl(PEt3) pulse. When AuCl(PEt3) was pulsed alone, only a very slight release of PEt3 was detected (Figures 11d and S4). It is expected that no AuCl(PEt3) adsorbs on a AuCl(PEt3)1–x saturated surface; hence, no PEt3 cleavage occurs during the reference pulses. All in all, very tiny peaks at m/z = 315 [AuPEt3]+, 118 [PEt3]+, 90 [HPEt2]+, and 62 [H2PEt]+ from AuCl(PEt3) were visible in QMS when the precursor was pulsed alone (Figure S4). The low intensities observed from the precursor molecule might be because of its possible condensation on the cone of the sampling orifice and because effusion fluxes of heavier molecules through the orifice to the QMS are lower than those of lighter molecules.

The behavior of the triethylphosphine adduct ligand was studied further by pulsing triethylphosphine alone on a freshly deposited Au surface (Figure 12). During the triethylphosphine pulses, a clear increase of the m/z = 62 was visible in QMS. As no mass increase was detected with QCM, it is clear that triethylphosphine does not adsorb on Au at this temperature (180 °C). AuCl(PEt3), on the other hand, seems to first adsorb on the surface, which is seen as a mass increase, and then desorb if the surface is already saturated with AuCl(PEt3)1–x (Figure 11b). The adsorption and desorption seem to occur without a cleavage of the triethylphosphine ligand since almost no increase of the m/z = 62 is visible in QMS.

Figure 12.

Figure 12

Open in a new tab

QCM (top row) and QMS (bottom row) measurements on reference pulses of triethylphosphine on the Au surface.

QMS proves clearly the formation and release of the suggested reaction byproduct Me3GeCl. Figure 13a shows the QMS measurement on m/z = 139, [ (CH3)274Ge35Cl]+, a fragment ion which has the highest intensity in the Me3GeCl mass spectrum,24 during three ALD cycles. Almost all (around 90%) of the compound is formed during the (Me3Ge)2DHP pulse. Figure 13b shows a QMS measurement on m/z = 139 during the full measurement pattern. A clear signal was visible also during the (Me3Ge)2DHP reference pulses. Presumably, the precursor or its fragments react with residual chlorine in the measurement system and create a background signal of m/z = 139 (marked also as a red curve in Figure 13a). Hence, the real intensity associated with the reaction with AuCl on the surface during the ALD cycles is the difference between the m/z = 139 intensities during the ALD process and the (Me3Ge)2DHP reference pulses. The highest peak of the mass spectrum of pyrazine25 (m/z = 80) was observed too (Figure 13c). Pyrazine is the byproduct of the redox reaction that produces metallic gold (eq 2), and it was seen only during the (Me3Ge)2DHP precursor pulses during the ALD cycles. A comparable intensity is observed during the (Me3Ge)2DHP reference pulses, however.

Figure 13.

Figure 13

Open in a new tab

QMS measurement on (a) m/z = 139 during three ALD cycles, (b) during the whole measurement pattern of seven (Me3Ge)2DHP reference pulses, 10 ALD cycles, and seven AuCl(PEt3) reference pulses, and (c) m/z = 80 during the same measurement pattern. The m/z = 139 signal during (Me3Ge)2DHP reference pulses is shown as a red curve in Figure 13a.

Based on the QMS measurements, during the AuCl(PEt3) most of the PEt3 desorbs, and a small amount of (CH3)3GeCl is formed in the reaction between adsorbed Me3Ge and chloride ligands (eqs 3 and 4).

graphic file with name mg2c00075_m003.jpg 3
graphic file with name mg2c00075_m004.jpg 4

During the (Me3Ge)2DHP pulse, most of the chlorides react, and the rest of the PEt3 desorbs. Gold in AuCl is reduced by (Me3Ge)2DHP, and pyrazine and Me3GeCl are released as byproducts (eq 5). A small fraction of Me3Ge groups adsorbs onto the Au surface.

graphic file with name mg2c00075_m005.jpg 5

In reactions 3–5, y and x are approximately 0.9 and 0.1, respectively. The overall reaction is the sum of the reactions that take place during each pulse (eq 6).

graphic file with name mg2c00075_m006.jpg 6

The QMS and QCM results can be compared by calculating the m1/m0 ratio based on the QMS results and comparing that with the m1/m0 measured with QCM. After the AuCl(PEt3) pulse, the overall mass change on the surface is m1 = M(AuCl(PEt3)) – x × M(Me3GeCl) – y × M(PEt3). After the (Me3Ge)2DHP pulse, that is, full ALD cycle, metallic Au is deposited and m0 = M(Au). The ratio m1/m0 calculated using the molar masses of the compounds and elemental gold with the x and y obtained from QMS is approximately 1.16 (eq 7). The result agrees well with the m1/m0 of 1.15 ± 0.02 measured with the QCM.

graphic file with name mg2c00075_m007.jpg 7

However, these calculations are only to support the qualitative observations of the reaction mechanism. In fact, the m1/m0 ratio is heavily dominated by the gold containing species because the M(AuCl)/M(Au) = 1.18.

Conclusions

We successfully developed the first reductive thermal ALD process for elemental gold using AuCl(PEt3) and (Me3Ge)2DHP as precursors. Highly conductive and pure gold films could be deposited at moderate temperatures of 160–180 °C. The process was proven to work on multiple substrates, although with a clear difference in nucleation that was the most favorable on a Ru surface and the least favorable on Al2O3. Furthermore, the reaction mechanism was studied and found to proceed stepwise, as expected based on the literature. The combination of high growth rate and purity of the films shows potential for many applications and furthermore proves the capabilities of the recently discovered reducing agent, (Me3Ge)2DHP.

Acknowledgments

ASM Microchemistry Oy and Academy of Finland (decision number 309552) and the Doctoral Programme in Materials Research and Nanoscience (MATRENA) are gratefully acknowledged for funding this research. ALD center of Finland is acknowledged for the infrastructure.

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsmaterialsau.2c00075.

  • Synthesis of the gold precursor, film uniformity as a function of distance from the inlet area, SEM from the gold film deposited with AuCl(PEt3) and (Me3Si)2DHP, wide-scan XPS spectra of a gold film, all measured m/z values, and QMS measurements of AuCl(PEt3) (PDF)

Author Contributions

The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript.

The authors declare no competing financial interest.

Supplementary Material

mg2c00075_si_001.pdf (342.2KB, pdf)

References

  1. Holliday R.; Goodman P. Going for Gold [Gold in Electronics Industry]. IEE Rev. 2002, 48, 15–19. 10.1049/ir:20020302. [DOI] [Google Scholar]
  2. Handoyo T.; Kondoh J. Development of Gold Thin-Film for Optical-Based Biosensor. AIP Conf. Proc. 2020, 2230, 020008. 10.1063/5.0006687. [DOI] [Google Scholar]
  3. Ravi Sankar A.; Lahiri S. K.; Das S. Performance Enhancement of a Silicon MEMS Piezoresistive Single Axis Accelerometer with Electroplated Gold on a Proof Mass. J. Micromech. Microeng. 2009, 19, 025008. 10.1088/0960-1317/19/2/025008. [DOI] [Google Scholar]
  4. Chang C.; Chang P. Innovative Micromachined Microwave Switch with Very Low Insertion Loss. Sens. Actuators, A 2000, 79, 71–75. 10.1016/S0924-4247(99)00218-6. [DOI] [Google Scholar]
  5. Cao J.; Sun T.; Grattan K. T. V. Gold Nanorod-Based Localized Surface Plasmon Resonance Biosensors: A Review. Sens. Actuators, B 2014, 195, 332–351. 10.1016/j.snb.2014.01.056. [DOI] [Google Scholar]
  6. Miikkulainen V.; Leskelä M.; Ritala M.; Puurunen R. L. Crystallinity of Inorganic Films Grown by Atomic Layer Deposition: Overview and General Trends. J. Appl. Phys. 2013, 113, 021301. 10.1063/1.4757907. [DOI] [Google Scholar]
  7. Hagen D. J.; Pemble M. E.; Karppinen M. Atomic Layer Deposition of Metals: Precursors and Film Growth. Appl. Phys. Rev. 2019, 6, 041309. 10.1063/1.5087759. [DOI] [Google Scholar]
  8. Baum T. H.; Comita P. B.. Chemical Vapor Deposition of Gold and Silver. The Chemistry of Metal CVD August 25, 1994; pp 303–327. [Google Scholar]
  9. Griffiths M. B. E.; Pallister P. J.; Mandia D. J.; Barry S. T. Atomic Layer Deposition of Gold Metal. Chem. Mater. 2016, 28, 44–46. 10.1021/acs.chemmater.5b04562. [DOI] [Google Scholar]
  10. Van Daele M.; Griffiths M. B. E.; Raza A.; Minjauw M. M.; Solano E.; Feng J.-Y.; Ramachandran R. K.; Clemmen S.; Baets R.; Barry S. T.; Detavernier C.; Dendooven J. Plasma-Enhanced Atomic Layer Deposition of Nanostructured Gold Near Room Temperature. ACS Appl. Mater. Interfaces 2019, 11, 37229–37238. 10.1021/acsami.9b10848. [DOI] [PubMed] [Google Scholar]
  11. Mäkelä M.; Hatanpää T.; Mizohata K.; Räisänen J.; Ritala M.; Leskelä M. Thermal Atomic Layer Deposition of Continuous and Highly Conducting Gold Thin Films. Chem. Mater. 2017, 29, 6130–6136. 10.1021/acs.chemmater.7b02167. [DOI] [Google Scholar]
  12. Hashemi F. S. M.; Grillo F.; Ravikumar V. R.; Benz D.; Shekhar A.; Griffiths M. B. E.; Barry S. T.; van Ommen J. R. Thermal Atomic Layer Deposition of Gold Nanoparticles: Controlled Growth and Size Selection for Photocatalysis. Nanoscale 2020, 12, 9005–9013. 10.1039/D0NR01092H. [DOI] [PubMed] [Google Scholar]
  13. Mäkelä M.; Hatanpää T.; Ritala M.; Leskelä M.; Mizohata K.; Meinander K.; Räisänen J. Potential Gold(I) Precursors Evaluated for Atomic Layer Deposition. J. Vac. Sci. Technol., A 2017, 35, 01B112. 10.1116/1.4968193. [DOI] [Google Scholar]
  14. Profijt H. B.; Potts S. E.; van de Sanden M. C. M.; Kessels W. M. M. Plasma-Assisted Atomic Layer Deposition: Basics, Opportunities, and Challenges. J. Vac. Sci. Technol., A 2011, 29, 050801. 10.1116/1.3609974. [DOI] [Google Scholar]
  15. Vihervaara A.; Hatanpää T.; Mizohata K.; Chundak M.; Popov G.; Ritala M. A Low-Temperature Thermal ALD Process for Nickel Utilizing Dichlorobis(Triethylphosphine)Nickel(Ii) and 1,4-Bis(Trimethylgermyl)-1,4-Dihydropyrazine. Dalt. Trans. 2022, 51, 10898–10908. 10.1039/D2DT01347A. [DOI] [PubMed] [Google Scholar]
  16. Klesko J. P.; Thrush C. M.; Winter C. H. Thermal Atomic Layer Deposition of Titanium Films Using Titanium Tetrachloride and 2-Methyl-1,4-bis(trimethylsilyl)-2,5-cyclohexadiene or 1,4-Bis(trimethylsilyl)-1,4-dihydropyrazine. Chem. Mater. 2015, 27, 4918–4921. 10.1021/acs.chemmater.5b01707. [DOI] [Google Scholar]
  17. Stevens E. C.; Mousa M. B. M.; Parsons G. N. Thermal Atomic Layer Deposition of Sn Metal Using SnCl4 and a Vapor Phase Silyl Dihydropyrazine Reducing Agent. J. Vac. Sci. Technol., A 2018, 36, 06A106. 10.1116/1.5055212. [DOI] [Google Scholar]
  18. Waldo R. A.An iteration procedure to calculate film compositions and thicknesses in electron-probe microanalysis. In Microbeam Analysis; Newbury D. E., Ed.; San Francisco Press: San Francisco, 1988; pp 310−314.
  19. Rahtu A.; Ritala M. Integration of a Quadrupole Mass Spectrometer and a Quartz Crystal Microbalance for In Situ Characterization of Atomic Layer Deposition Processes in Flow Type Reactors. Electrochem. Soc. Proc. 2000, 2000-13, 105. [Google Scholar]
  20. Price D. Vapor Pressure Determination by Thermogravimetry. Thermochim. Acta 2001, 367–368, 253–262. 10.1016/S0040-6031(00)00676-6. [DOI] [Google Scholar]
  21. PeriodicTable.com. Technical data for Gold. Mathematica’s ElementData function from Wolfram Research, Inc. http://periodictable.com/Elements/079/data.html (accessed Oct 24, 2022).
  22. Darrah Thomas D. T.; Weightman P. Valence Electronic Structure of AuZn and AuMg Alloys Derived from a New Way of Analyzing Auger-Parameter Shifts. Phys. Rev. B Condens. Matter 1986, 33, 5406–5413. 10.1103/physrevb.33.5406. [DOI] [PubMed] [Google Scholar]
  23. Wada Y.; Kiser R. W. A Mass Spectrometric Study of Some Alkyl-Substituted Phosphines1. J. Phys. Chem. 1964, 68, 2290–2295. 10.1021/j100790a044. [DOI] [Google Scholar]
  24. Bittremieux W.; Chen C.; Dorrestein P. C.; Schymanski E. L.; Schulze T.; Neumann S.; Meier R.; Rogers S.; Wang M.. Universal MS/MS Visualization and Retrieval with the Metabolomics Spectrum Resolver Web Service, 2020. bioRxiv:2020.05.09.086066, 10.1101/2020.05.09.086066, (accessed July 06, 2022). [DOI] [Google Scholar]
  25. Wallace W. E.‘Mass Spectra’ by NIST Mass Spectrometry Data Center. In NIST Chemistry WebBook, NIST Standard Reference Database Number 69; Linstrom P. J., Mallard W. G., Eds.; National Institute of Standards and Technology.

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

mg2c00075_si_001.pdf (342.2KB, pdf)

Articles from ACS Materials Au are provided here courtesy of American Chemical Society

RESOURCES