Atmospheric pressure metal organic chemical vapor deposition of thin germanium films

The chlorogermanes GeCl₂Cp^4M₂ and GeCl₂Cp*₂ were prepared by salt metathesis reactions starting from GeCl₄ with two equivalents of the corresponding lithium salts LiC₅Me₅ and LiC₅HMe₄, respectively (Scheme 1).

Hydrogenation of the chlorogermanes was carried out with LiAlH₄ and gave the corresponding dihydrides GeH₂Cp^4M₂ (1) and GeH₂Cp*₂ (2) in nearly quantitative yields (1: 95%; 2: 98%). Compared with the syntheses described in the literature of other molecular precursors such as GeH₂tBu₂ (reported yield: 6%) [63] and GeBzlH₃ (reported yield: 31%) [47], which were shown to be suitable for the deposition of germanium, the yields obtained for the dihydridogermanes 1 and 2 are significantly higher. The hydrogenation of the dichlorogermane GeCl₂Cp^4M₂ represents a significant improvement with regard to the synthesis as described in the literature, which is based on a disproportionation reaction of GeH₃Cp^4M [47]. Noteworthy, the chlorogermanes are highly sensitive to moisture, while the hydrides 1 and 2 are only slightly sensitive to oxygen and water. They can be stored at 20 °C under argon for months without decomposition using normal glassware, while chlorogermanes need to be stored under inert atmosphere in sealed tubes.

In order to elucidate the potential of the precursors 1 and 2 for the deposition of germanium, both germanes were investigated with regard to their thermal behavior by means of Thermogravimetric Analysis (TGA), decomposition temperature measurements, and temperature-dependent PXRD measurements. The decomposition temperatures of 1 and 2 were determined to be 225 °C and 200 °C, respectively, which is significantly lower as compared to other tetravalent organogermanes such as GeVinyl₄ (dec. 450 °C), GeEt₄ (dec. 400 °C), or GeHEt₃ (dec. 350 °C) and thus are in the range as reported for divalent germanium precursors such as GeI₂ (dec. 210 °C to 300 °C) [20] and 1,3 di-tert-butyl-1,3,2-diazagermolidin-2-ylidines (dec. 140 °C) [64]. When heating the germanes 1 and 2 to 550 °C under inert conditions using a heating rate of 50 K/min, evaporation was observed and minor residues of 8.5% and 12.6% were obtained in case of 1 and 2, respectively. The TGA residues were examined by means of EDX, and high germanium contents of 86.2% and 89.3% in case of 1 and 2, respectively, were detected, while carbon (9.4 – 11.0%) and oxygen (1.3 – 2.8%) represent only minor impurities. In temperature-dependent PXRD measurements in sealed quartz glass capillaries starting from the pure germanes 1 and 2 reflections for crystalline germanium are observed at approximately 500 °C. It is thus assumed that the decomposition of the precursor provides amorphous germanium first, which then crystallizes at 500 °C. It should be noted that the crystallization temperature of germanium significantly depends on purity and annealing conditions [65, 66], however, a significantly lower crystallization temperature of the amorphous germanium in the range 300 °C to 400 °C was expected [67].

In conclusion, both germanes 1 and 2 are easily accessible, easy to handle, show high volatility and low decomposition temperatures. For these reasons, the two germanes were chosen as promising candidates for the development of a simple AP MOCVD process for the production of thin and flexible germanium films.

The deposition setup (Fig. 1) is kept simple and does not require vacuum, plasma, gas flow, or cooling traps which are usually required for the CVD process. The apparatus consists of three parts: i) the evaporator vessel, which here is a silver crucible containing the precursor; ii) a holder made of copper wire, which carries the evaporation vessel; iii) a glass vessel with a flat opening, which serves as the reactor and is placed on the substrate. The experimental setup is placed on a ceramic heater, which at best but not necessarily is located in a glove box under inert gas. The maximum temperature used for the deposition on float glass, and silicon wafers was set to 550 °C, whereas the maximum temperature for the deposition on Kapton® was set to 400 °C.

For the deposition experiments silicon wafers with a natural SiO₂ film, float glass and Kapton® were chosen. These substrates provide adequate thermal stability for the deposition, a smooth surface with a roughness < 0.5 nm, and are chemically resistant to the solvents used for cleaning the surfaces (acetone, ethanol, and Et₂O).

In an initial step, several experiments were carried out in order to determine the minimum substrate temperature necessary for the deposition starting from both precursors. The most important requirement at this stage was that the film adheres to the substrate, which was tested using a standardized Scotch tape test of the American Society for Testing and Materials (ASTM D3359-17). The substrate temperature was successively increased in steps of 5 K beginning at 250 °C. A temperature of 270 °C for 1 and 260 °C for 2 was determined as the lowest applicable deposition temperature. At deposition temperatures higher than 350 °C for 2 and 400 °C for 1 inhomogeneous films on all selected substrates, which partially peeled off from the substrate, were observed. Thus, the optimum temperature window for the deposition process ranges from 275 °C up to 350 °C for 2 and from 280 °C to 400 °C for 1. The closed setup is advantageous in that a high concentration of the precursor is present in the gas phase, as previously described by Harrison et al. for tetravinylgermane [46]. Figure 2 shows two deposited germanium films on float glass and Kapton® deposited by our AP MOCVD method. Noteworthy, germanium films deposited on Kapton® are flexible and permit to apply a bending radius below 2 mm without the germanium films chipping off from the substrate and without cracking (Fig. 3).

In the simple setup a roll rim glass was used, but scaling-up is easily possible. Germanium films with a diameter of up to 11 cm were fabricated using glass reactors of larger diameter. However, in the experiments as described in the following commercially available 10 mL roll rim glasses with a neck diameter of 18 mm were used.

For a more detailed characterization of the thin germanium films scanning electron microscopy (SEM), energy dispersive X-ray spectroscopy (EDX), X-ray photoelectron spectroscopy (XPS), and Raman spectroscopy were used. Deposition on silicon wafers was carried out for 3 min at 275 °C, 300 °C, and 325 °C with 2, respectively, at 300 °C, 325 °C, and 350 °C with 1. The as-deposited films were characterized using Raman spectroscopy. All samples revealed similar and characteristic spectra showing broad bands with peak maxima at 265—273 cm⁻¹, which is characteristic for amorphous germanium (Fig. 4). Despite the presence of carbon distinct bands for carbon were not observed, and only the humps at 1500–1600 cm⁻¹ might indicate carbon residues. Noteworthy, the films might be easily crystallized using laser-induced crystallization. We did observe that a symmetric Raman band at 294 cm⁻¹ evolved upon measuring the Raman spectra for more than 30 s using an Argon Laser with a wavelength of 514.5 nm, which is indicative for the beginning of crystallization. However, as the phonon frequency of crystalline Ge is 300 cm⁻¹ the shift to 294 cm⁻¹ shows the formation of small crystallites. Moreover, the remaining broad feature at lower frequency indicates that the crystallization is not complete, which is in line with previous studies on hydrogenatead carbon germanium films (Fig. 5) [68].

Figure 6 (left) shows the SEM cross-section of the sample, that was prepared by deposition of 1 at 350 °C on a silicon wafer. Here, the germanium film (red frame) deposited on the silicon wafer is well visible, and EDX proofs the deposition a germanium-rich film on the silicon wafer.

In addition to surface analysis, the deposited film was analyzed using XPS before and after sputtering, whereby approximately 30 nm of the films was removed (Fig. S1-S12). At the surface, all samples exhibited oxygen and carbon in addition to germanium. Regardless of the precursor and the deposition temperature carbon contents in the range of 20–30% were observed. The high oxygen content of 25–31% most likely results from a native germanium dioxide layer. This assumption is confirmed by the fact that the sputtered samples did not reveal any oxygen. However, even the sputtered samples and thus the bulk material contain significant amounts of carbon (10 – 15%) [69]. In conclusion, our deposition experiments revealed two trends: (i) at higher deposition temperatures the amount of incorporated carbon is lower and (ii) the Cp* substituted germane 2 gives slightly lower carbon contents at the same deposition temperature as used for Cp^4M substituted 1 despite an increased carbon content in the precursor. The lowest carbon contents were found for deposition of 2 at 325 °C (10.6%) and for deposition of 1 at 350 °C (11.5%).

Atomic force microscopy (AFM) was used to determine the film thickness and the surface texture of the deposited germanium films on solid substrates such as float glass and silicon. Therefore, a stainless steel needle is used to scratch the germanium film on the substrate, and the measurement is carried out at the thus produced edge. The thickness as well as the surface roughness is determined in a single measurement at the edge (method A, for details see ESI Fig. S13). For both precursors, it is observed that the film thickness does not significantly depend on the precursor used and increases with increasing temperature while using a constant deposition time of three minutes. The final deposited films are quite similar in thickness for both precursors (113 nm-120 nm) when a temperature of 300 °C is applied. However, at 275 °C film deposition was not observed using precursor 1 and at 350 °C homogeneous films were only obtained from precursor 2. In the following, we concentrated our work on compound 1, which offers a better synthetic availability.

In order to check the homogeneity of the deposited films using different time intervals, germanium films starting from 1 were deposited at 325 °C in time intervals of one minute up to four minutes. The film thickness determination was carried out at three positions of the substrate using method A, [distance from the edge: X1: 5 mm; X2: 2.5 mm; X3: 1 mm from the edge (Table 1)]. All samples show only a small variation of the film thickness (2.2% to 7.8%) when comparing the thickness at the center with the film thickness at outer parts of the Si wafer.

In addition to the film thickness, the morphology and roughness R_RMS (RMS—root-mean-squared) of the as-deposited films (deposition time: 3 min) were investigated using AFM. With respect to the morphology, the AFM images of films show a characteristic grain size (Fig. 7), which reaches from a few to 200 nm depending on the deposition temperature and is similar to that observed for germanium films deposited by low pressure CVD or epitaxy [27, 54, 70]. As the deposition temperature increases, an increase in the diameter of the individual grains at the surface is observed. The specific roughness of the films is between (1 ± 0.15) nm and (4.5 ± 1.5) nm for films deposited at 275 °C and 400 °C, respectively. The roughness increases with increasing deposition temperature, deposition rate, and film thickness. With regard to the morphology and roughness, the germanium films deposited at temperatures below 300 °C are comparable with germanium films deposited by means of CVD using GeH₄ or magnetron sputtering using Ge targets [27, 54, 70].

Due to the optical properties of thin germanium films, it is possible to determine the film thickness by vis–NIR measurements (Figs. 8–10) as was demonstrated recently [71]. This method is particularly interesting, since in contrast to AFM studies the film is not destroyed for or by the measurement (method B). For the film thickness determination, the interference effect is used as described for a Fabry-Pérot interferometer. The reflection of the irradiated light within the film at the interfaces germanium-glass and germanium-air is used. Within the film, constructive and destructive interference occur depending on the irradiated wavelength and the film thickness. Thus, the position of the transmission maxima depends on the film thickness, which then is calculated by 2·n·d = m·λ_min. For this purpose, the refractive index of germanium is used for variable n at λ_min (minimum absorption) [72], d is the film thickness to be determined, m is the order of the interference (integer), and λ_min corresponds to the wavelength with minimal absorption.

The film thicknesses of Ge on float glas determined by means of vis–NIR spectroscopy are plotted as a function of deposition temperature (Figs. 11 and 12). The comparison shows a very good agreement of the values determined by means of vis–NIR spectroscopy (non-destructive method B) with the values determined from the AFM measurements (destructive method A).

The film thickness of deposited films on float glass was additionally determined using method A: center at 0 mm, at ± 4.5 mm distance from the center, and at ± 7.5 mm distance from the center. By this, we get an idea of the uniformity of the film deposition at different temperatures (deposition time: 3 min) on the example of precursor 1 (Fig. 13). At temperatures below 330 °C, quite uniform films were deposited and a linear correlation of deposition temperature and film thickness is observed. In a second temperature regime from 335 °C to 360 °C, a film thickness of approximately 300 nm is obtained, regardless of the temperature, and the films appear to be less uniform. At higher temperatures the film thickness again increases but also the deviation increases significantly and thus non-uniform films are obtained. Methods A and B provide similar results.

The observed deposition characteristics are well in line with common CVD process parameters. A characteristic relationship between film growth and substrate temperature is observed, which can be divided into three processes: i) kinetically controlled film growth at low temperatures; ii) stagnation of the film growth in a certain temperature range as a result of saturation of the surface with the precursor; iii) an increase of the deposition rate by premature decomposition of the precursor in the gas phase. With regard to this, we analyzed the deposition process of our setup using an Arrhenius plot (Fig. 14).

In the temperature range between 280 °C to 330 °C (Fig. 14 green line), an increase in the deposition rate is observed with increasing temperature. Extrapolation would give an intersection with the y-axis at about 222 °C, which marks the beginning of the film deposition. This value correlates well with the experimentally determined decomposition temperature of 225 °C of precursor 1. The deposited films show a uniform film thickness. The film growth is kinetically controlled in this temperature regime, and the activation energy of the thermal decomposition of the precursor 1 was determined from the Arrhenius Plot to amount to 101 kJ/mol (Fig. 14). This value is significantly lower as reported for other organogermanium(IV) precursors (GeEt₄ 142 kJ/mol [73], GeH₃tBu 160 kJ/mol [45], and GeVinyl₄ 185 kJ/mol [46]). The digermane Ge₂H₆ also shows a higher activation energy of 125 kJ/mol at 0.0039 mbar and 163 kJ/mol at 0.295 mbar [74]. The only report for a lower activation energy is given by Veprek et al. for 3-di-tert-butyl-1,3,2-diazagermolidin-2-ylidine, which shows an activation energy of 90 kJ/mol. However, this precursor follows a disproportionation reaction, and thus a maximum of 50% of the germanium might be deposited [64]. The low activation energy for the decomposition of 1 confirms that Cp^4M is easily cleaved off, and the corresponding diorganogermane is suitable for the deposition of germanium at low temperatures. In the second temperature range between 330 °C to 360 °C (Fig. 14 pink line), there is no further increase in the deposition rate, and the latter remains constant, which is explained by a diffusion controlled process and surface saturation with precursor 1. Films deposited in this temperature regime were roughly (290 ± 15) nm thick. In the third temperature range, 365 °C to 400 °C (Fig. 14 blue line), an irregular increase in the deposition rate occurs. The deposition is inhomogeneous starting from 365 °C. A film thickness of (455 ± 95) nm was determined at 370 °C and films deposited at 400 °C showed a thickness of (529 ± 224) nm. Here the precursor decomposes prematurely in the gas phase, but still contributes to film growth (Fig. 14). In a classical CVD system, a continuous gas stream removes the precursor that is decomposed in the gas phase, and thus, they do not contribute to the growth of the film.

A series of samples of germanium films was produced using the diorganogermanium precursors on Kapton® and float glass as substrates. The deposition of the germanium films was carried out at a substrate temperature of 280 °C and was increased by 5 °C to 400 °C for each new sample. The deposition of germanium on a silicon substrate was carried out at 350 °C using the precursor GeH₂Cp^4M₂. A deposition time of 3 min was chosen unless otherwise described.

All experiments were carried out at atmospheric pressure in a glove box under a nitrogen atmosphere (oxygen and water content < 0.5 ppm) using a computer-controlled ceramic heating plate (Bath Resistor Ceramics). The temperature of the heating plate can be varied between 25 and 550 °C. The substrates used were float glass (25 × 80 mm) from VWR Collection, silicon wafer with a natural oxide film (10 × 10 mm), or Kapton® 100 QR 2 films (25 × 25 × 0.025 mm) from DuPont. All substrates were treated with acetone, ethanol, and diethylether for 3 min in an ultrasonic bath before the deposition experiments and then dried at 120 °C. As evaporation vessels, home-build silver crucibles made of silver foil (∅ approx. 5 mm, height-7-mm, thickness-0.1-mm) were used. Unless otherwise described, all deposition experiments were performed with 10.0 ± 0.3 mg of the precursors GeH₂Cp^4M₂ (1) [75] and GeH₂Cp*₂ (2), which we prepared according to literature procedures [62, 75].