Solution-processed aluminum metals using liquid-phase aluminum-hydrides

Takashi Masuda *abc and Hideyuki Takagishi a
aSchool of Materials Science, Japan Advanced Institute of Science and Technology, 1-1 Asahidai, Nomi, Ishikawa 923-1292, Japan
bCucullus Incorporation, 1-18, Chuo-dori, Kanazawa, Ishikawa 920-0866, Japan
cVerein, Skodagasse, 1080, Wien, Austria

Received 3rd May 2020 , Accepted 12th July 2020

First published on 14th July 2020

Amine-alane (AlH3) adducts have been extensively studied for numerous years due to their ability to function as excellent precursors for aluminum metals in chemical vapor deposition (CVD). This study highlights the potential versatility of amine-alane adducts in the field of coating processes, rather than in CVD processes which have been traditionally studied. The purpose of this study is to demonstrate a solution-processed Al metal film using amine-alane adducts. We prepared a precursor solution for the production of Al metal films named “liquid-Al”, which consisted of triethylamine-alane, bis(cyclopentadienyl) titanium dichloride (Cp2TiCl2), and 4-methylanisole. Moreover, “precoat-ink” was synthesized using a diisopropoxy-bisethylacetoacetatotitanate (DBEAT) solution. We clarified that by combining two types of Ti catalyst (Cp2TiCl2 and DBEAT), which perform different roles, a solution-processed Al metal film can be realized. The DBEAT in the precoat-ink provides adsorption/nucleation sites for Al on the substrate surface, while Cp2TiCl2 in the liquid-Al facilitates the elimination of non-Al elements (ligand and hydrogen) from the amine-alane adducts, leading to an effective liquid-to-metal Al conversion at temperatures >100 °C. Consequently, the Al films produced using liquid-Al demonstrated a resistivity of 5.8 μΩ cm (annealed at 250 °C), which is comparable to that of the bulk resistivity (3 μΩ cm). This result is of significant importance because it facilitates the utility of Al metals in the field of printed electronics.

1. Introduction

Since the first report of trimethylamine aluminum hydride (AlH3·TMA) by Stecher et al. in 1942,1 tertiary amine-alane adducts AlH3·NR3 have been studied in detail until now due to their versatility in material science.2–4 Alane, (AlH3) is a sterically and electronically unsaturated part that reacts with a range of Lewis donors resulting in 1[thin space (1/6-em)]:[thin space (1/6-em)]1 and 1[thin space (1/6-em)]:[thin space (1/6-em)]2 adducts at the Al center that are four- and five-coordinate, respectively.5,6 Of these, the alane adducts of TMA, triethylamine (TEA), dimethylethylamine (DMEA), and bistrimethylamine (BTMA) have received attention as high-quality Al sources in chemical vapor deposition (CVD).7

More recently, their potential to function as liquid precursors for solid-state metal materials was demonstrated, thereby restimulating fundamental research on this class of compounds. In particular, the alane adducts of dibutyl ether (DBE) and TMA were identified as promising Al sources in solution processes.8,9 The pathway for obtaining metal Al by coating using AlH3·DBE and AlH3·TMA is to expose the coated films to the catalytic surroundings using a vapor method.10 If the catalyst feeding could be replaced the vapor method to a coating one, the utility of liquid-phase amine-alane adducts would be further enhanced. The main advantages of the coating process are the facile handling and operation and, more importantly, the ability to directly supply the material to the substrate surface with inexpensive solution processes.

AlH3·TEA compounds captured our interest, not only because of their ability to be gas-phase precursors for high-quality Al films under high vacuum (e.g. CVD processes) but also due to the fact that they are liquids at atmospheric pressure.11,12 However, AlH3·TEA has only been employed as the gas source in vacuum processes. To the best of our knowledge, the direct deposition of Al metal from liquid-phase AlH3·TEA by a coating process has not been reported. One of the main reasons for this dearth of research is the difficulty associated with obtaining coating films due to the lack of adsorption/nucleation sites on the substrate. Furthermore, the elimination of ligands under atmospheric pressure forces the formation of an intractable polymeric solid with higher resistivity than that of the Al metal.13–15

The purpose of this study is to demonstrate the potential versatility of liquid-phase amine-alane adducts in the field of coating processes by demonstrating solution-processed Al metal films. We prepared two types of inks; “liquid-Al” that is a precursor solution (AlH3·TEA + Ti catalyst + solvent) for metal Al, and a “precoat-ink” that provides adsorption/nucleation sites for Al on the substrate. Both inks include a Ti catalyst that performs different roles, an Al film is not formed in the absence of either of the Ti catalysts. The combination of the liquid-Al and the precoat-ink successfully accomplished liquid-to-metal Al conversion with high efficiency, even during the coating process. This result is of fundamental importance for the potential applications of such liquid-Alane materials in printed electronics.

2. Experimental section

General procedure

All air-sensitive procedures were carried out in a N2-filled glovebox (O2 < 0.5 ppm and dew point <−75 °C). Lithium aluminum hydride (LiAlH4), triethylammonium chloride (Et3N·HCl), diethyl ether, and dimethylsulfoxide (DMSO) all obtained from Fujifilm Wako Pure Chemical Industries, Ltd, and bis(cyclopentadienyl)titanium dichloride (Cp2TiCl2, Kanto Chemical), and diisopropoxy bisethylacetoacetatotitanate (DBEAT, Gelest) were used as purchased without further purification. 4-Methylanisole (Tokyo Chemical Industry) was dried over 4 Å molecular sieves and stored under N2.

1H and 13C NMR spectra were obtained on a Bruker AVANCE III spectrometer (400 MHz, benzene-d6, 25 °C, TMS). The abbreviations: br (broad), t (triplet), q (quartet), and J (spin–spin coupling constant) were used. FTIR spectra were obtained using a Bruker ALPHA spectrometer with a resolution of 4 cm−1 over the spectral range ν = 650–4000 cm−1, typically using an average of 32 scans. Gas chromatographic analysis (GC) was carried out using a GC-2014 (Shimadzu) with a Shincarbon ST 50/80 packed column (Shinwa Chemical). The temperature of the column, vaporizing chamber, and detector was maintained at 40, 150, and 200 °C, respectively. The flow rate of the carrier gas (Ar) was 50 mL min−1. Thermogravimetric analysis (TG) was conducted using a Thermo plus EVO2 (Rigaku). The flow rate of the N2 purge gas was 400 mL min−1, and the rate of the temperature increase was 5 °C min−1. Quantitative analysis was performed using a triple quadrupole inductively-coupled plasma mass spectrometer (ICP-MS: Agilent 8900, Agilent Technologies) and a secondary ion mass spectroscopy (SIMS: PHI ADEPT-1010, Ulvac-PHI) with a 2.0 kV Cs+ primary ion beam. The XPS analyses were performed using a Kratos Axis-Ultra DLD (Shimidzu) with a monochromatic Al Kα source (1486.6 eV). The measurement chamber was in the 10−8 Pa range. The spectrometer was calibrated by using the photoemission lines of Au 4f7/2 (83.9 eV) and Cu 2p3/2 (932.6 eV).

Preparation of liquid-Al

The “liquid-Al” was a solution of AlH3·TEA and Cp2TiCl2 dissolved in 4-methylanisole. Cp2TiCl2 and 4-methylanisole enhanced the resistivity of the Al films and the wettability of the liquid-Al, respectively. The synthesis of AlH3·TEA followed the protocol (eqn (1)) of Ruff and Hawthorne,5 which involved a direct reaction between LiAlH4 and the hydrochloride salt of the ligand.
LiAlH4 + Et3N·HCl → AlH3·NEt3 + LiCl + H2(1)

The Et3N·HCl (17.8 g; 130 mmol) was slowly added onto a suspension of the LiAlH4 (3.8 g; 100 mmol) in diethyl ether (100 mL) at 25 °C; the resulting mixture was stirred for 2 h. All byproducts were removed by filtration (0.2 μm). The filtrate was concentrated under reduced pressure to yield pure AlH3·TEA as a colorless liquid (4.0 g, 30% yield with respect to LiAlH4). The AlH3·TEA (250 μL), 4-methylanisole (250 μL), and Cp2TiCl2 (240 μg) were mixed ([Ti]/[Al] = 0.05 mol%).

Physical data for AlH3·TEA, 1H NMR: δ = 4.196 (br, 3H, AlH3), 2.359 (q, J = 7.3 Hz, 6H, N–CH2), 0.853 (t, J = 7.3 Hz, 9H, CH3). The signal integration of the alane Al–H peak in reference to that of N–CH2 and CH3 was almost 1[thin space (1/6-em)]:[thin space (1/6-em)]2[thin space (1/6-em)]:[thin space (1/6-em)]3, as predicted by the molecular formula. 13C NMR: δ = 47.683 (CH2), 9.784 (CH3). 27Al NMR; δ = 139.5 (Al–H). IR (neat): ν (cm−1) = 2977 (s), 2942 (s), 2883 (m), 2801 (w), (νC–H); 1770 (s), (νAl–H); 1455 (m), 1392 (m), (δC–H); 1297 (w), 1164 (w), 1086 (w), (γC–H); 1041 (w), 894 (w), (νC–N); 780 (s), (γC–H); 713 (s), (δsAl–H).

Film preparation

Fig. 1 shows the procedure used to obtain the Al film. A DBEAT solution (1.5 wt% in DMSO) was prepared as the precoat-ink, which introduces adsorption/nucleation sites for Al on the substrate surface. The precoat-ink was spin-coated (2000 rpm, 10 s) onto a glass substrate with the size of 2 × 2 cm2, and dried on a hotplate (150 °C, 30 min). There was no change in the appearance of the substrate surface after precoating due to the extremely low quantity of Ti adsorbed. Next, liquid-Al was spin-coated (1200 rpm, 10 s) onto the precoated substrate. The resultant film was annealed at 250 °C for 30 min. The annealing changed the appearance of the film from transparent to metallic.
image file: d0tc02162h-f1.tif
Fig. 1 Schematic diagram showing the process of Al metal film production using two inks (precoat-ink and liquid-Al). The precoat-ink provides adsorption/nucleation sites for effective liquid-to-metal Al conversion on the substrate surface.

The film thickness, microstructure, and resistivity was measured by using a contact stylus profiler Dektak-3030ST (Veeco Instruments), a scanning electron microscope S-4500 (SEM, Hitachi), and a 4-point probe resistivity meter Loresta-EP (Mitsubishi Chemical). The binding energy of Al 2p core level was recorded with a step of 0.1 eV and a pass energy of 80 eV, which was corrected to give the adventitious C 1s spectral component (C–C, C–H) a binding energy of 284.8 eV. The H, C, O, and Ti content in the films were quantified by SIMS using standard sample (Al). Measurement accuracy is ±40% (2σ). The quantity of precoated Ti adsorbed on the substrate was obtained by ICP-MS. Hydrofluoric acid (0.2 mL) was dropped onto the DBEAT-coated substrate (2 × 2 cm2), and then the adsorbed substance was collected together with the hydrofluoric acid and measured. Since the detection limit of Ti (6 × 1012 atom per cm2) is less than 6% of the measured value, the obtained value is sufficiently quantitative.

Pyrolysis behavior of the TEA ligand and hydrogen

The desorption temperature and weight reduction of the TEA ligand was measured by TG. The liquid sample (10 mg) was added to a pan, and was heated to 250 °C. Sample pans with or without the Ti coating were prepared to verify the function of the precoat-ink. Precoat-ink droplets were placed on the pans, followed by drying at 150 °C for 30 min on a hotplate to obtain Ti-coated pans. The volume of hydrogen gas released from AlH3·TEA by annealing was traced using GC. An accurately weighted sample was added to a cylindrical reactor, and was heated at different temperatures (50–250 °C) for 15 min. Next the volume of hydrogen gas released from the sample was determined. The correlation between peak area and hydrogen content was calibrated with 99.99% H2 gas (standard gas, GL sciences). To verify the function of precoating, the interior of several reactors was coated with the precoat-ink prior to use.

3. Results

Description of the Al film

Fig. 2(a) shows the specific resistivity of the coated films as a function of annealing temperature. Photographs of the films are inserted on the upper side. All the substrates were precoated. The films prepared below 75 °C were gray and non-conductive. As the temperature was increased from 75 to 100 °C, the insulating films changed dramatically to conductive films (19 μΩ cm) with a metallic appearance. This change in appearance is attributed to a liquid-to-metal Al conversion at 100 °C. The resistivity of the films increased with temperature, and eventually attained 5.8 μΩ cm at 250 °C, which is comparable to the resistivity of bulk Al (3 μΩ cm).16 We defined this film as “Film A” in Fig. 2(b) and (c). Considering that the commonly used conductive ink (Ag nano-colloidal ink) has a resistivity of 200–400 μΩ cm (100 °C) to several μΩ cm (250 °C),17,18 the liquid-Al provides advantages in the resistivity at low annealing temperatures (100 °C).
image file: d0tc02162h-f2.tif
Fig. 2 Photographs of the liquid-Al films coated on a 2 × 2 cm2 glass substrate. (a) Specific resistivities (and standard error, n = 10 samples) and photographs of the obtained films using different annealing temperatures. The black object on the film surface annealed at 100, 125, and 250 °C is a reflection of the camera. (b) Films prepared under three different preparation conditions to confirm the DBEAT catalytic effect, all the films were annealed at 250 °C. The substrate in Films A and B was precoated with DBEAT, whereas the substrate in Film C was not. Liquid-Al with Cp2TiCl2 was used for Film A (and C), whereas the liquid-Al without Cp2TiCl2 was used for Film B. (c) Surface SEM images of Films A and B obtained in (b).

Fig. 2(b) shows the resulting films (all annealed at 250 °C) obtained using three different preparation conditions. The substrate in Films A and B was precoated, whereas the one in Film C was not. In the absence of precoating, no film was obtained on the substrate. The difference between Films A and B is that the former used liquid-Al with Cp2TiCl2, whereas the latter used liquid-Al without Cp2TiCl2. We obtained a mirror-like Al film with a thickness of 120 nm and a specific resistivity of 5.8 μΩ cm in Film A. The liquid-Al without Cp2TiCl2 produced a white cloudy film with a thickness of 160 nm and a resistivity of 15 μΩ cm, as shown in Film B. The surface SEM images of Films A and B are shown in Fig. 2(b). The mirror-like Film A exhibited a flat surface composed of smaller grains, whereas the cloudy Film B revealed a rough surface composed of grains with larger aggregation. Addition of Cp2TiCl2 enhanced the resistivity, uniformity, and smoothness of the resultant film.

Fig. 3 shows the Al 2p spectrum in the binding energy range from 68 to 80 eV for the Film A in Fig. 2. The reproducibility of the measurements was confirmed in three samples. The representative spectrum is shown here. Intense peaks are observed at 71.7 and 74.8 eV. The peak at 74.8 eV is the oxide component derived from the surface oxidation, and the peak at 71.7 is attributed metal component.19 The spectrum supports that Film A is metal Al.

image file: d0tc02162h-f3.tif
Fig. 3 Al 2p XPS spectrum of Film A in Fig. 2. Dotted and thin lines are experimental and fitted data, respectively. The spectrum is separated into two peaks, metal (71.7 eV) and oxide (74.8 eV).

The precoating process assisted the Al adsorption/growth on the substrate surface. Herein, we focused on the quantity of Ti adsorbed on the substrate. The optimal concentration of DBEAT in the precoat-ink was 1–2 wt%. Precoat-inks containing DBEAT in concentrations less than 1 wt% resulted in a lower number of adsorption/nucleation sites, and therefore a lower functionality. However, there was no difference in the uniformity of the Al film obtained using DBEAT with concentrations above 1 wt%. The ICP-MS reveals the quantity of Ti adsorbed on the substrate after the precoating process. Fig. 4 depicts the relationship between the quantity of Ti adsorbed and the concentration of DBEAT used in the precoat-ink. The quantity of Ti on the substrate surface was saturated at 5.0–5.5 × 1014 atoms cm−2 when we used DBEAT at concentrations above 1 wt%. A lower concentration of DBEAT resulted in lower Ti coverage. The functionality of the adsorption/nucleation sites correlated with the quantity of adsorbed Ti.

image file: d0tc02162h-f4.tif
Fig. 4 The relationship between the quantity of Ti adsorbed on the glass substrate and the concentration of DBEAT used in the precoat-ink (and standard error, n = 3 samples).

Table 1 shows the concentration of impurities, measured by SIMS, of Films A and B in Fig. 2. Although liquid-Al contains large amounts of carbon, pyrolysis released almost all of the carbon and left Al metal with a small amount of H. The carbon content was less than 0.1 at%, indicating the elimination of solvent and the TEA ligand. The oxygen incorporation was 1–5 at%, which may be due to the oxidation during annealing. Cp2TiCl2 facilitated the pyrolysis of hydrogen even at low concentrations. The hydrogen content in Film A was 11.4 at%, whereas that in Film B (without Cp2TiCl2) was 33.9 at%. The residual Ti content in the film was less than 0.01 at%, as expected from the low amount of Cp2TiCl2 in the liquid-Al. Table 1 also suggests that further pyrolysis of the residual hydrogen enhances the resistivity.

Table 1 Concentration of impurities for Films A and B obtained by SIMS measurement
Element Concentration (at%)
Film A Film B
H 11.4 33.9
C 0.05 0.08
O 1.1 5.2
Ti <0.01 <0.01

Pyrolysis behavior of the ligand and hydrogen

As shown in Fig. 5(a), the pyrolysis behavior of AlH3·TEA was investigated using four different sample conditions: sample A, B, C, and D. Sample D is pure AlH3·TEA, while the other samples include the catalysis of Cp2TiCl2 and/or precoated DBEAT. The measurements were repeated five times in each sample to ensure their reproducibility.
image file: d0tc02162h-f5.tif
Fig. 5 Measurements for the pyrolysis behavior of the ligand and hydrogen. (a) TG, DTG, and GC measurements for samples A, B, C, and D. (b) TG curves, weight reduction as a function of the heating temperature. (c) DTG curves. (d) Dehydriding curves (and standard error, n = 5 samples), hydrogen release results for each sample with various decomposition temperatures.

Fig. 5(b) depicts the TG curves as a function of the heating temperature. Assuming that all the components except for Al are desorbed from the AlH3·TEA, the weight should be 20%. The experimental TG curves for all samples indicate a weight loss of 80–82% following heating to 250 °C and consequently, 18–20% of the material remained. A temperature of 250 °C completed the desorption of almost all the carbon in the film.

According to the differential thermogravimetry (DTG) curve in Fig. 5(c), samples A, B, C, and D exhibit a desorption peak at 91, 113, 97, and 130 °C, respectively. The peak at 130 °C in sample D is interpreted as the elimination of TEA ligands, based on the study of Al–N dissociation at 130–150 °C using a differential scanning calorimeter.20 The comparison of samples C and D reveals that the addition of the Cp2TiCl2 lowered the desorption temperature of the TEA ligand, and the sharp peak related to the Al–N dissociation at 130 °C shifted to 90–100 °C. This temperature is consistent with the boiling point of TEA (90 °C), suggesting that Cp2TiCl2 weakens the Al–N bond in AlH3·TEA and releases TEA into the system. The comparison of samples B with D reveals that the precoating with DBEAT lowered the desorption temperature of TEA from 130 to 113 °C. The main role of DBEAT is to provide adsorption/nucleation sites, and the catalysis of DBEAT as a pyrolysis accelerator is not as effective as that of Cp2TiCl2.

The GC measurement in Fig. 5(d) depicts the dehydriding curves for samples A, B, C, and D; 100% refers to the state where all hydrogens on the AlH3 are released. Each plot is average of five measurements. The dehydriding curve at 250 °C for samples A and B demonstrates an 84 and 67% hydrogen release, respectively, which agrees with the SIMS results in Table 1. The curves indicate two distinct stages—a rapid first stage, followed by a considerably slower second stage, which may reflect two different hydrogen-release mechanisms (e.g., nucleation and growth). The break between the two stages for the curves of samples A, B, C, and D occurs at 100, 125, 100, and 150 °C, respectively. The break point coincided with the position of the TG peak, indicating the coordinated elimination of the hydrogen and ligand. The slopes of all curves in the second stage are similar, suggesting a similar pyrolysis mechanism for the hydrogen release in this stage. In contrast, significant differences appeared in the first stage of the curves. The hydrogen release from sample D (pure AlH3·TEA) was negligible at 125 °C, whereas those from samples A, B, and C were substantial. Both Cp2TiCl2 and the precoated DBEAT facilitated the dehydrogenation in the first stage. The curves of samples A and C overlap, regardless of the presence or absence of precoating. The Cp2TiCl2, which is contained in both samples A and C, influences the dehydriding of AlH3·TEA to a greater extent than the precoated DBEAT (sample B).

4. Discussion

Liquid-Al (mixture of AlH3·TEA, Cp2TiCl2, and 4-methylanisole) was an effective liquid precursor for high-quality Al films in coating processes. The coating ability and pyrolysis temperature of the film was extremely sensitive to the presence of Ti catalysts (Cp2TiCl2 and DBEAT). In this section, we discuss the possible mechanism for the Al deposition, and offer several suggestions on the role of the Ti compounds.


The growth of Al on an Al surface differs from the initial nucleation of Al on a substrate. The kinetic barrier towards initial nucleation is usually higher than that towards steady state growth.21 The chemical reaction on the substrate surface can play a dominant role in initiating nucleation. Hence, we consider the role of precoating in the nucleation process in this section. The main purpose of precoating is to fix active Ti on the substrate surface, as a nucleation site. As shown in Fig. 4, the coverage of Ti on the glass substrate is saturated at 5.0–5.5 × 1014 atoms cm−2, which is almost consistent with the OH surface coverage (4.6 × 1014 OH atoms cm−2) on SiO2.22,23 Furthermore, the functionality of the adsorption/nucleation sites was found to be almost unchanged by further wet cleaning of the precoated substrate, suggesting that Ti was immobilized on the glass via a hydrolysis reaction24 (eqn (2)) between DBEAT and the surface hydroxyl groups. In the CVD process, pretreatment using TiCl4 gas demonstrated excellent agreement between the OH surface coverage on the glass substrate and the quantity of Ti adsorbed,24–27 which is similar to the phenomenon that occurred in this study.
Si–OH + Ti(OR)4 → Si–O–Ti(OR)3 + ROH(2)

It was found that not all of the Ti compound solutions performed as suitable precoat-inks. In addition to DBEAT, we also evaluated Ti compounds such as titanium tetraisopropoxide (TTIP), TiCl4, tetrakis(dimethylamido)titanium, Cp2TiCl2, and CpTiCl3, as well as sputtered Ti and TiO2 films. However, DBEAT was found to provide the most effective adsorption/nucleation sites, as an Al metal film could only be obtained on a substrate precoated with DBEAT. Slight structural differences in TTIP (alkoxide) and DBEAT (coordination compound) resulted in significant differences in their performance as precoat-inks. TTIP has four alkoxide ligands, which is advantageous in that the hydrolysis reaction shown in eqn (2) proceeds quickly; however, it also leads to catalysis deactivation. By contrast, DBEAT has two alkoxide ligands that are readily hydrolyzed, and also has two chelating ligands that provide structural stability. Therefore, catalysis could be maintained, even after chemisorption onto the substrates. Although the exact identity of the catalyst and the mechanism of catalysis are still unknown, the precoated DBEAT could lower the kinetic barrier of nucleation on glass surfaces.

Many researchers postulate that the catalytic dehydrogenation of AlH3 by Ti is important in the reaction of NaAlH4 (a hydrogen storage material), and a wide variety of catalytic mechanisms have been proposed. For instance, Sun et al.28 and Kang et al.29 proposed a model in which Ti–Al and Ti–H interactions, respectively, facilitate the dehydrogenation of Al hydride and the nucleation of the Al phase nuclei. Assuming that Ti(IV) compounds adsorb on the substrate in our system, the catalysis of the Ti–Al alloy or Ti0 appears challenging. Given the strong reducibility of AlH3, a catalytic reaction path through the formation of Ti–H is feasible. The catalytic dehydrogenation of the carbon system with a Ti complex containing chelating ligands is also reported.30 Further investigation is required to elucidate the catalytic mechanism.

Nuclear growth

AlH3 must be reduced from an oxidation state of +3 to obtain an Al0 metal film. The reduction is achieved by the elimination of molecular hydrogen. Although this reductive elimination is not feasible for isolated AlH3, detailed studies of surface reaction mechanisms have reported that the reduction readily occurs on metallic Al as an autocatalytic reaction.31 It is interesting to note that the activation energy for the elimination of H2 from gaseous AlH3 is 43 kcal mol−1, whereas the energetic barrier for the desorption of H2 from a metallic Al surface is 16 kcal mol−1.13,32,33

In the temperature region above 150 °C, all the dehydriding curves exhibit a similar slope, as shown in Fig. 5(c), thereby suggesting that the dehydrogenation follows the same decomposition mechanism. The Ti catalyst has minimal effect on the slope. Since the catalysis of the precoated Ti is a reaction limited to the substrate surface, its catalytic effect should disappear when the surface Ti is covered with the generated Al. As the droplet/film loses fluidity following drying, Cp2TiCl2 is no longer catalytically active after reducing the surrounding AlH3. In the second stage, therefore, dehydrogenation and nuclear growth proceed on the Al metal via an autocatalytic reaction.

Given the structural similarity between AlH3·TEA and AlH3·TMA,34 we expect that the thermal reaction mechanism should be comparable. We therefore propose a simplified description of film growth on a metal Al surface occurring during annealing, as originally indicated for AlH3·TMA in the CVD process.35

AlH3·TEA ⇌ AlH3 + TEA(3)
AlH3 + 2Al (surface) → 3AlH(4)
3AlH → 3Al + 1.5H2(5)

First, pyrolysis releases the AlH3 (eqn (3)). Next, the hydrogen atoms disperse themselves over the Al surface (eqn (4)), based on the observed surface reaction of TIBA (triisobutylaluminum).36 With the adsorption of TIBA on Al surfaces, the three isobutyl units behaved identically, and molecular TIBA desorption was not observed in the TDS study. These results suggested that the alkyl ligand becomes equivalent by migration to neighboring surface Al atoms. Then, in eqn (5), it decomposes via an autocatalytic process, releasing hydrogen and forming an Al film.5,37 The Ti catalysts in the system effectively deposited only Al metals on the substrate. Thus, no film contamination resulted from the presence of carbon elements in liquid-Al, as shown in Table 1. These facts indicate that the proposed model in eqn (3)–(5) is more dominant than other possible processes related to carbon incorporation, even in liquids.

Role of Cp2TiCl2

The addition of Cp2TiCl2 to liquid-Al dramatically reduced the desorption temperature of hydrogen and TEA. Moreover, this, coupled with the fact that the obtained film was formed with finer grains, suggests that Cp2TiCl2 causes homogeneous nucleation at lower temperature. Therefore, here, we discuss some possibilities about the dehydrocoupling of AlH3 by Cp2TiCl2. However, a feasible chemical reaction mechanism is still under consideration due to insufficient evidence.

Before discussing the dehydrocoupling, we briefly examine the desorption of TEA. Many alane complexes form 1[thin space (1/6-em)]:[thin space (1/6-em)]1 and 1[thin space (1/6-em)]:[thin space (1/6-em)]2 adducts, but AlH3·TEA form only 1[thin space (1/6-em)]:[thin space (1/6-em)]1 adducts because of the structural effects.14,38 Therefore, if a five-coordinated intermediate is obtained as a result of the interaction between Cp2TiCl2 and the vacant orbitals of AlH3·TEA, it causes structural instability and releases TEA. Indeed, the lower limit of the DTG peak coincided with the boiling point of free TEA (90 °C).

The typical reactions of Cp2TiCl2 include transmetalation, σ-bond metathesis, and oxidative-addition/reductive-elimination. In particular, there are numerous reports on dehydrocoupling by the σ-bond metathesis and oxidative-addition/reductive-elimination.

The σ-bond metathesis is well studied in the dehydrocoupling of Si hydride. Fig. 6 shows the reaction path of σ-bond metathesis drawn with reference to Si. It proceeds via two possible paths, one in which Al is in the α-position and a second in which Al is in the β-position.39 Dehydrocoupling proceeds when both pathways are energetically feasible. In the Al hydride, it is presumed that the α-position is very disadvantageous in terms of the activation energy because of the instability of Ti–Al. We calculated the four-center transition state of the σ-bond metathesis using DFT calculations, but did not elucidate the structure of the α-position Ti–Al with a feasible activation energy. Hence, the reaction appears unsuitable for the dehydrocoupling of Al hydride.

image file: d0tc02162h-f6.tif
Fig. 6 The two possible pathways for the σ-bond metathesis reaction. Al in the α-position yields a dehydrogenated Ti–Al compound when R = H, while the Al in the β-position undergoes ligand exchange with the metallocene. Dehydrocoupling proceeds when both pathways are energetically feasible.

Oxidative-addition/reductive-elimination is well reported in the dehydrocoupling of borane complexes. Luo et al.40 reported the dehydrocoupling mechanism based on the computational analysis, in which dehydrogenation initiates at the Ti(II) catalyst, Cp2Ti. Sloan et al. published a detailed kinetic study of the catalytic dehydrocoupling of amine-boranes by titanocene, which demonstrated that the catalysis likely proceeds through Ti(III) intermediates.41 In addition to the catalysis reaction, although the reaction path is unknown, Bel'sky et al.42 and Harrod et al.43 reported a direct dehydrogenation from metallocene complexes. On the basis of the aforementioned reports, the regioselectivity of the intermediate structure in σ-bond metathesis, the valence number of active Ti, and the possibility of direct dehydrogenation from Ti complexes will be reported in a future publication.

5. Conclusion

In this work, we prepared “liquid-Al” consisting of an appropriate ratio of AlH3·TEA, Cp2TiCl2, and 4-methylanisole, as well as a precoat-ink containing a Ti catalyst (DBEAT solution). DBEAT in the precoat-ink provided adsorption/nucleation sites (active Ti) on the substrate surface thorough a hydrolysis reaction between DBEAT and the surface OH groups on the glass substrate. Cp2TiCl2 in the liquid-Al decreased the pyrolysis temperature of AlH3·TEA from 130 °C to 90–100 °C; furthermore, it enhanced the film uniformity and resistivity via homogeneous nucleation. In the presence of the two Ti catalysis, AlH3·TEA within the coated liquid-Al rapidly initiated pyrolysis and was deposited on the substrate surface at lower temperatures. Therefore, the coated film potentially follows the common decomposition steps that have been previously proposed for other gas-phase amine-alane adducts. The pyrolysis first releases the ligand to yield intermediate alane structures, which further decompose to release hydrogen gas, thus producing the Al metal. Mirror-like Al films with a specific resistivity of 19 and 5.8 μΩ cm were obtained by annealing the liquid-Al film at 100 and 250 °C, respectively.

Liquid-Al demonstrated the utility of liquid amine-alane adducts as a conductive ink in the field of printed electronics. Traditionally, Al films have been deposited by physical vapor deposition or CVD techniques, which limits the efficiency in energy and materials applications due to the requirement for large deposition instruments and vacuum systems. In an attempt to overcome these challenges, metal Al obtained using the liquid-Al coating process presented here offers a functional method for environmentally friendly film deposition.

Conflicts of interest

There are no conflicts to declare.


The authors would like to thank K. Fukudome for technical assistance. Financial support by The Noguchi Institute, Japan Aluminium Association, The Mitani Foundation for Research and Development, and JSPS KAKENHI (Grant Numbers 19K21969) is gratefully acknowledged.


  1. O. Stecher and E. Wiberg, Ber. Dtsch. Chem. Ges. A, 1942, 75, 2003 CrossRef.
  2. I. P. Jain, P. Jain and A. Jain, J. Alloys Compd., 2010, 503, 303 CrossRef CAS.
  3. H. Li, M. J. Meziani, A. Kitaygorodskiy, F. Lu, C. E. Bunker, K. A. S. Fernando, E. A. Guliants and Y. P. Sun, J. Phys. Chem. C, 2010, 114, 3318 CrossRef CAS.
  4. I. G. Aviziotis, T. Duguet, K. Soussi, G. Kokkoris, N. Cheimarios, C. Vahlas and A. G. Boudouvis, Phys. Status Solidi C, 2015, 12, 923 CrossRef CAS.
  5. J. K. Ruff and M. F. Hawthorne, J. Am. Chem. Soc., 1960, 82, 2141 CrossRef CAS.
  6. J. K. Ruff and M. F. Hawthorne, J. Am. Chem. Soc., 1961, 83, 535 CrossRef CAS.
  7. M. G. Simmonds and W. L. Gladfelter, in The Chemistry of Metal CVD, ed. T. T. Kodas and M. J. M. J. Hampden-Smith, Wiley-VCH, 1994, pp. 45–103 Search PubMed.
  8. H. M. Lee, H. B. Lee, D. S. Jung, J. Y. Yun, S. H. Ko and S. B. Park, Langmuir, 2012, 28, 13127 CrossRef CAS PubMed.
  9. H. M. Lee, J. Y. Seo, A. Jung, S. Y. Choi, S. H. Ko, J. Jo, S. B. Park and D. Park, ACS Appl. Mater. Interfaces, 2014, 6, 15480 CrossRef CAS PubMed.
  10. H. M. Lee, S. Y. Choi and A. Jung, ACS Appl. Mater. Interfaces, 2013, 5, 4581 CrossRef CAS PubMed.
  11. O. Lehmann and M. Stuke, Appl. Phys. Lett., 1992, 61, 2027 CrossRef CAS.
  12. M. E. Gross, C. G. Fleming, K. P. Cheung and L. A. Heimbrook, J. Appl. Phys., 1991, 69, 2589 CrossRef CAS.
  13. G. N. Nechiporenko, L. B. Petukhova and A. S. Rozenberg, Bull. Acad. Med. Sci. USSR, 1975, 24, 1584 CrossRef.
  14. E. Wiberg and E. Amberger, Hydrides of the Elements of Main Groups I–V, Elsevier Science Ltd, 1971 Search PubMed.
  15. M. J. van Setten, V. A. Popa, G. A. de Wijs and G. Brocks, Phys. Rev. B: Condens. Matter Mater. Phys., 2007, 75, 035204 CrossRef.
  16. J. R. Rumble, CRC Handbook of Chemistry and Physics, 99th edn, CRC Press, 2018 Search PubMed.
  17. J. R. Greer and R. A. Street, Acta Mater., 2007, 55, 6345 CrossRef CAS.
  18. D. Kim, S. Jeong, H. Shin, Y. Xia and J. Moon, Adv. Mater., 2008, 20, 3084 CAS.
  19. J. van den Brand, W. G. Sloof, H. Terryn and J. H. W. de Wit, Surf. Interface Anal., 2004, 36, 81 CrossRef CAS.
  20. D. M. Frigo and G. J. M. Eijden, Chem. Mater., 1994, 6, 190 CrossRef CAS.
  21. J. A. Venables, G. D. T. Spiller and M. Hanbucken, Rep. Prog. Phys., 1984, 47, 399 CrossRef.
  22. B. M. Wright, P. A. Peaden and M. L. Lee, Chromatographia, 1982, 15, 584 CrossRef CAS.
  23. R. Mueller, H. K. Kammler, K. Wegner and S. E. Pratsinis, Langmuir, 2003, 19, 160 CrossRef CAS.
  24. J. B. Kinney and R. H. Staley, J. Phys. Chem., 1983, 87, 3735 CrossRef CAS.
  25. I. Bakardjiev, M. Majdraganova and G. Bliznakov, J. Non-Cryst. Solids, 1976, 20, 349 CrossRef.
  26. B. A. Morrow and A. H. Hardin, J. Phys. Chem., 1979, 83, 3135 CrossRef CAS.
  27. M. L. Hair and W. Hertl, J. Phys. Chem., 1973, 77, 2070 CrossRef CAS.
  28. D. Sun, S. S. Srinivasan, G. Chen and C. M. Jensen, J. Alloys Compd., 2004, 373, 265 CrossRef CAS.
  29. X. D. Kang, P. Wang and H. M. Cheng, J. Appl. Phys., 2006, 100, 034914 CrossRef.
  30. S. Taubmann and H. G. Alt, J. Mol. Catal. A: Chem., 2008, 289, 49 CrossRef CAS.
  31. A. Wnkler, C. Resch and K. D. Rendulic, J. Chem. Phys., 1991, 95, 7682 CrossRef.
  32. H. Hara, K. Domen, T. Onishi, H. Nozoye, C. Nishihara, Y. Kaise and H. Shindo, Surf. Sci., 1991, 242, 459 CrossRef.
  33. J. M. Mundenar, R. Murphy, K. D. Tsuei and E. W. Plummer, Chem. Phys. Lett., 1988, 143, 593 CrossRef CAS.
  34. G. N. Nechiporenko, L. B. Petukhova and A. S. Rozenberg, Izv. Akad. Nauk SSSR, Ser. Khim., 1975, 8, 1697 Search PubMed.
  35. L. H. Dubois, B. R. Zegarski, C. T. Kao and R. G. Nuzzo, Surf. Sci., 1990, 236, 77 CrossRef CAS.
  36. B. E. Bent, R. G. Nuzzo and L. H. Dubois, J. Am. Chem. Soc., 1989, 111, 1634 CrossRef CAS.
  37. C. W. Heitsch, C. E. Nordman and R. W. Parry, Inorg. Chem., 1963, 2, 508 CrossRef CAS.
  38. T. D. Humphries, K. T. Munroe, A. Decken and G. S. McGrady, Dalton Trans., 2013, 42, 6953 RSC.
  39. L. Perrin, O. Eisenstein and L. Maron, New J. Chem., 2007, 31, 549 RSC.
  40. Y. Luo and K. Ohno, Organometallics, 2007, 26, 3597 CrossRef CAS.
  41. M. E. Sloan, A. Staubitz, T. J. Clark, C. A. Russell, G. C. Lloyd-Jones and I. Manners, J. Am. Chem. Soc., 2010, 132, 3831 CrossRef CAS PubMed.
  42. V. K. Bel'sky, A. I. Sizov, B. M. Bulychev and G. L. Soloveichik, J. Organomet. Chem., 1985, 280, 67 CrossRef.
  43. J. F. Harrod and S. S. Yun, Organometallics, 1987, 6, 1381 CrossRef CAS.

This journal is © The Royal Society of Chemistry 2020