Ge-ion implantation and activation in (100) β-Ga₂O₃ for ohmic contact improvement using pulsed rapid thermal annealing

Beta gallium oxide (β-Ga₂O₃) with its ultrawide bandgap of ∼4.8 eV has emerged as a promising semiconducting material for next-generation power electronic devices.^1,2 The estimated dielectric strength of 8 MV/cm along with the expected Baliga’s figure of merit which is more than 3000 times higher than that of silicon provide the ideal conditions to realize power devices with even higher breakdown voltages and efficiencies than their SiC and GaN counterparts.³ Moreover, β-Ga₂O₃ bears the advantage of large-area bulk substrates being available from melt growth techniques such as Czochralski,^4,5 vertical Bridgeman,^6,7 and edge-defined film-fed growth,⁸ which is very beneficial for low-cost wafer mass production. Up to now, promising advances have been made with the realization of β-Ga₂O₃ power devices such as metal-oxide-semiconductor field-effect transistors,^9–12 metal-semiconductor field-effect transistors,^13–15 modulation-doped field-effect transistors,^16–18 or heterojunction field-effect transistors^19–21 reaching a record average breakdown strength of 5.5 MV/cm,¹⁷ maximum breakdown voltage of 8.56 kV,²² and a power figure of merit as high as 0.9 GW/cm².²³ Furthermore, the development of high-performance unipolar Schottky barrier diodes^24–26 or heterojunction diodes^27–29 based on β-Ga₂O₃ is rapidly progressing showing a power figure of merit up to 13.2 GW/cm².³⁰ 

One major key requirement for high performance transistors based on wide bandgap semiconductors such as β-Ga₂O₃ is the formation of low ohmic contact resistances in order to reduce conduction losses inside the devices. This becomes even more important for RF devices, where the channel resistance is much lower compared to high-voltage devices. The specific contact resistance of the standard contact metallization system Ti/Au to β-Ga₂O₃ is typically in the range of 10⁻⁶–10⁻⁵ Ω cm².³¹ Here, it has to be noted that, apart from the metallization stack, the ohmic contact properties also strongly depend on the crystal orientation of β-Ga₂O₃ due to the anisotropic nature of this material. A comparative study revealed significantly lower contact resistances of Ti/Au contacts to β-Ga₂O₃ in the (100) orientation in contrast to the (010) orientation.³² This can be explained by the fact that the in situ-formed Ti-TiO_x layer on (100) β-Ga₂O₃ is much thinner and more homogenous than on (010) β-Ga₂O₃ which is caused by the anisotropic surface energy and mass diffusity. Nevertheless, improvement of the contact resistance can be effectively achieved by high n-type doping of the β-Ga₂O₃ contact region prior to metal deposition either using ion implantation^16,20 or epitaxial regrowth techniques.^15,17 So far, the lowest specific contact resistance with a value of 8.3 × 10⁻⁷ Ω cm² was achieved by using the latter approach on (010) β-Ga₂O₃.³³ However, the advantage of ion implantation is the fact that the process itself is less complex as no hardmask is needed for local n-type doping. For a long time, Si as a shallow donor to β-Ga₂O₃ was the only candidate being investigated for local n-type doping by implantation reaching decent activation efficiencies beyond 60%.³⁴ Here, the lowest specific contact resistance with a value of 4.6 × 10⁻⁶ Ω cm² was achieved after annealing at 950 °C. Just recently, a comparative study on the activation of other implanted shallow donors in (001) β-Ga₂O₃ was reported showing efficiencies of 28.2% for Sn, 40.3% for Ge, and 64.7% for Si after annealing at 925 °C for 30 min in N₂.³⁵ The resulting specific contact resistance for all species was still found to be well above 1 × 10⁻⁵ Ω cm². However, this could be due to the fact that the activation conditions for Ge and Sn were probably not optimal as no temperature-dependent analysis of the activation efficiency for these dopants in β-Ga₂O₃ has been carried out yet. Moreover, all these investigations so far focused on annealing temperatures below 1000 °C. In this regard, it has been stated that the implant activation temperature in general follows a two-thirds rule with respect to the melting point for most semiconductors.³⁶ This would mean that the optimum activation temperature of ion implanted β-Ga₂O₃ should roughly be 1100 °C considering a melting temperature of around 1800 °C. Therefore, it remains unclear whether the usually applied annealing conditions below 1000 °C for implanted shallow donors including Si are ideal for reaching low ohmic contact resistances.

In this work, a temperature-dependent analysis on the activation efficiency of Ge in (100) β-Ga₂O₃ is carried out in order to determine the optimum annealing conditions for improving ohmic contact properties. The annealing process involves the use of pulsed rapid thermal annealing (RTA) at temperatures between 900 and 1200 °C. Our investigations reveal a significant reduction of the specific contact resistance after annealing at 1100 °C using 40 pulses to a record value as low as 4.8 × 10⁻⁷ Ω cm² which is an important step forward to the realization of highly efficient β-Ga₂O₃ power devices.

The experiments were carried out on epitaxial (100) β-Ga₂O₃ wafers consisting of Mg-doped semi-insulating substrates prepared from Czochralski growth with a miscut angle of 4°,^4,37 on which a 200 nm thick Si-doped β-Ga₂O₃ layer was homoepitaxially grown by metal-organic chemical vapor deposition. In this regard, Mg is used as a very efficient compensating acceptor which allows the realization of semi-insulating substrates.³⁸ The details of the optimal growth parameters used to grow the films were previously reported elsewhere.^39,40 The layers featured a charge carrier concentration and mobility of 3 × 10¹⁷ cm⁻³ and 123 cm²/(V s), respectively, as extracted from Hall effect measurements. The samples were then subjected to a multiple energy Ge⁺ ion implantation process to obtain a 100 nm box-like profile with an ion concentration of 5 × 10¹⁹ cm⁻³ using implantation energies of 300, 130, and 60 keV and doses of 5.6 × 10¹⁴, 1.4 × 10¹⁴, and 6.7 × 10¹³ cm⁻², respectively. The depth profiles of the Ge⁺ ions were calculated by using the transport-of-ions-in-matter (trim) simulation software⁴¹ and are shown in Fig. 1. It should be noted that the Mg concentration in the semi-insulating substrate is roughly around 2 × 10¹⁸ cm⁻³ which is according to the simulation sufficient to compensate Ge below 200 nm so that the effective thickness of the Ge-implanted region is 200 nm. The implantation was carried out on a High Voltage Engineering Europe HVEE implanter from a germanium oven operated at 1200 °C. Argon was used as supporting gas and the ionization potential was 40 keV. The most abundant ⁷⁴Ge isotope (36%) was chosen. Afterward, Ge⁺ activation was carried out using a pulsed rapid thermal annealing (RTA) system (HQ-Dielectrics Hemera HT). The annealing process involved for all experiments using the pulsed RTA treatment, a preheating step at a base temperature of 900 °C for 10 min. Then, this was followed by pulses up to 1000, 1100, or 1200 °C with a total number of 10 or 40 pulses without a holding phase. The temperature gradient was set to 25 K/s in the heating phase and 12 K/s in the cooling phase. A representative temperature profile for the pulsed RTA treatment of the Ge⁺-implanted β-Ga₂O₃ samples up to 1100 °C is presented in Fig. 2. Subsequently, ohmic contacts for the transfer length method (TLM) and van der Pauw structures for the Hall measurements are realized by starting with a surface preparation of the β-Ga₂O₃ samples in BCl₃/Ar-plasma (20 SCCM/10 SCCM) with a power of 50 W at 1 Pa for 2 min. yielding an etch depth of 10 nm. Then, lift-off metallization of evaporated Ti/Au (20 nm/130 nm) was done followed by an RTA step in nitrogen at 470 °C for 60 s. Finally, interdevice isolation of the TLM and Hall structures was carried out by multiple energy nitrogen implantation.⁴² A schematic cross section of the test structures used for electrical characterization is shown in Fig. 3.

The surface morphology of the β-Ga₂O₃ layers before and after Ge⁺ implantation and subsequent pulsed RTA treatment were analyzed using atomic force microscopy (AFM) (Bruker Dimension Icon, USA). The electrical characterization started with TLM measurements to extract the contact resistivity, sheet resistance, and specific contact resistance. A total number of at least 30 TLM structures were measured per each wafer to ensure statistical certainty. Afterward, Hall effect measurements at room temperature were carried out, which allowed the calculation of the activation efficiency. Finally, the implantation profile of the Ge⁺ ions was determined by secondary ion mass spectrometry performed by RTG Mikroanalyse GmbH Berlin.

Figure 4 shows the AFM surface topograms of Ge⁺-implanted β-Ga₂O₃ samples which have been annealed at 900 °C for 10 min. without any pulsing, and after 10 and 40 pulses to 1000, 1100, and 1200 °C as well as an as-grown reference sample without implantation and thermal treatment. It can be seen that the initial as-grown β-Ga₂O₃ layers show very smooth surfaces with the desired step-flow morphology featuring a root-mean-square (rms) surface roughness of around 0.3 nm. Although this step-flow morphology vanishes after ion implantation and subsequent annealing at 900 °C for 10 min in N₂ atmosphere without any additional pulsed RTA treatment, the value for the surface roughness remains unchanged. In addition, no significant change in the surface roughness is observed after pulsed RTA treatment at 1000 and 1100 °C using 10 pulses. However, increasing the number of pulses to 40 results in an increase of the rms value for the surface roughness of the 1000 and 1100 °C annealed samples to 1.5 and 1.8 nm, respectively. Moreover, pulsed RTA treatment at 1200 °C causes a drastic change in the surface morphology even after only 10 pulses which gets more pronounced after 40 pulses reaching an rms roughness up to 7.3 nm. This emphasizes that high-temperature pulsing beyond 1100 °C as well as the use of more than 10 pulses leads to significant changes in the surface morphology of Ge⁺-implanted β-Ga₂O₃. Here, the fast ramping and the involved thermal stress which is repeated for several pulses seems to lead to microcleaving of the (100) surface, which is the easiest cleavage plane. However, further investigations are necessary to validate this assumption.

The results of the TLM measurements are presented in Fig. 5, which shows the extracted contact resistivity, sheet resistance, and specific contact resistance for the as-grown β-Ga₂O₃ reference sample as well as the Ge⁺-implanted and -annealed samples. Here, it should be noted that the sample annealed with 40 pulses at 1200 °C was highly resistive and not useful for the evaluation. This might be related to the drastic morphological change observed in the AFM measurements. The nonimplanted reference sample features a contact resistivity, sheet resistance, and specific contact resistance of around 2 Ω mm, 8.4 kΩ/sq, and 4.9 × 10⁻⁶ Ω cm², respectively, which is consistent with our previous experiments on epitaxial β-Ga₂O₃ wafers with comparable layer properties.⁴³ Up to now, these already very low contact resistances at such a low doping concentration without any additional n⁺⁺-doping using ion implantation or epitaxial regrowth have only been demonstrated so far on (100) β-Ga₂O₃.³¹ This again emphasizes the strong influence of the crystal orientation of β-Ga₂O₃ on the ohmic contact properties. After Ge⁺ implantation and annealing at 900 °C for 10 min, already a significant reduction of the contact resistivity to 1.4 Ω mm and sheet resistance to 2.3 kΩ/sq is observed. Further increase of the annealing temperature and the number of pulses results in a steady decrease of the contact resistivity, which reaches its minimum with 0.2 Ω mm using 40 pulses at 1100 °C. Interestingly, the contact resistivity rises again using 10 pulses at 1200 °C which could be related to the drastic morphological changes causing the degradation of the ohmic contact interface. Likewise, the sheet resistance steadily decreases with increasing annealing temperature and the number of pulses reaching a minimum of 590 Ω /sq at 1200 °C using 10 pulses. Finally, the specific contact resistance reaches the lowest value of 4.8 × 10⁻⁷ Ω cm² at 1100 °C using 40 pulses, revealing the optimum annealing condition in our experiment for low resistance ohmic contact formation. Besides the fact that this specific contact resistance is one order of magnitude lower than what is measured on the reference sample, it is indeed a new record value. We speculate that this is the result of applying the optimum annealing conditions for Ge⁺-implanted β-Ga₂O₃ in combination with the fact that (100) β-Ga₂O₃ shows as already mentioned better properties regarding the ohmic contact formation compared to (010) β-Ga₂O₃.³² It should be noted that an increase of the number of pulses beyond 40 did not show any further improvement of the electrical properties for all annealing temperatures.

In order to calculate the activation efficiency as a function of the annealing conditions in our experiments, Hall effect measurements were carried out, which allowed the extraction of the charge carrier concentration and mobility. A summary of the measurement data is listed in Table I. Here, it should be noted that we assumed a homogeneous distribution of the charge carriers within the 200 nm epitaxial layer down to the interface to the Mg-doped semi-insulating substrate. However, as can be seen in Fig. 1, the Ge concentration shows a drastic nonlinear decrease below a depth of 100 nm. Therefore, the extracted values for charge carrier concentration and mobility might be slightly underestimated in our analysis. As an outcome of the Hall effect measurements, we can conclude that the extracted data are in agreement with the results obtained from TLM measurements. With increasing temperature and number of pulses, the activated charge carrier concentration increases and thus the activation efficiency. The highest value for the activation efficiency of 19.2% was achieved on the Ge⁺-implanted (100) β-Ga₂O₃ sample which was annealed using 10 pulses at 1200 °C. Compared to previous investigations on the Ge⁺ implantation and activation in (001) β-Ga₂O₃,³⁵ this value is relatively low, which could either be due to the underestimated calculations in our Hall effect analysis or the implantation and activation processes strongly dependent on the crystal orientation of β-Ga₂O₃. The latter would mean that the (001) orientation is much more suitable for implantation purposes than the (100) orientation which should be further analyzed in comparative studies in the future. Moreover, it was not possible to measure the sample annealed at 1200 °C using 40 pulses due to the high resistivity of the material as observed in the TLM measurements. The sample annealed at 1100 °C using 40 pulses showing the lowest specific contact resistivity in our experiments featured a charge carrier concentration of 7.1 × 10¹⁸ cm⁻³ which yields an activation efficiency of 14.2%.

Finally, SIMS measurements of Ge⁺-implanted (100) β-Ga₂O₃ samples without annealing and with annealing at 1100 °C using 40 pulses were carried out, as shown in Fig. 6 in order to analyze the influence of the annealing process on the diffusion of the implanted Ge⁺ ions. Here, it should be noted that the measurement is based on a qualitative analysis since no standard for Ge⁺ implantation into β-Ga₂O₃ was available. The diagram shows that the shape of the implantation profile of the as-implanted sample matches the calculated profile shown in Fig. 1 with a reasonably constant Ge concentration down to 100 nm followed by an instant decrease with a comparable slope. However, after pulsed RTA treatment, the profile within the first 100 nm is not constant anymore and shows a more distorted shape. This phenomenon has been also observed during the high-temperature annealing of Si⁺-implanted β-Ga₂O₃ at 1100 °C and could be explained by a thermal diffusion process of the implanted species to highly implantation damaged regions upon annealing.³⁴ Nevertheless, it is important to note that this redistribution of Ge has no negative side effect on the ohmic contact properties but quite the opposite, as confirmed by the electrical characterizations. Furthermore, the SIMS measurements verify no vertical diffusion of Ge into the bulk after annealing, which is most probably suppressed due to the annealing in N₂ atmosphere.⁴⁴ 

In this work, we analyzed the optimum annealing conditions for the activation of Ge-implanted β-Ga₂O₃ in order to improve ohmic contact formation by using annealing temperatures between 900 and 1200 °C. Our investigations revealed reduced ohmic contact resistances with a record value of 4.8 × 10⁻⁷ Ω cm² after pulsed rapid thermal annealing at 1100 °C using 40 pulses yielding an activation efficiency of 14.2%. Annealing at 1200 °C using 10 pulses results in an activation efficiency of 19.2%, which unfortunately is accompanied by an increase of the contact resistance due to significant changes in the surface morphology. Our results verify the high potential of applying high-temperature annealing processes above 1000 °C after Ge implantation for reaching low ohmic contact resistances to β-Ga₂O₃, which might be also suitable for other implanted shallow donors.

The authors would like to thank the colleagues in the process department at FBH for process technological support, in particular, Alexander Külberg and Nico Thiele. This work was performed in the framework of GraFOx, a Leibniz-ScienceCampus partially funded by the Leibniz association. Furthermore, this work was funded by the Federal Ministry of Education and Research in Germany within the Research Project GoNext, Funding No. 16ES1084K and within the “Forschungsfabrik Mikroelektronik Deutschland (FMD)” framework under Ref. No. 16FMD02.

The authors have no conflicts to disclose.

Kornelius Tetzner: Conceptualization (lead); Formal analysis (lead); Investigation (lead); Methodology (lead); Validation (lead); Writing – original draft (lead). Andreas Thies: Conceptualization (supporting); Formal analysis (supporting); Investigation (supporting); Methodology (supporting); Writing – original draft (supporting). Palvan Seyidov: Data curation (equal); Formal analysis (equal); Investigation (equal); Writing – original draft (supporting). Ta-Shun Chou: Data curation (equal); Formal analysis (equal); Investigation (equal); Writing – original draft (supporting). Jana Rehm: Data curation (equal); Investigation (equal). Ina Ostermay: Investigation (supporting); Methodology (equal); Writing – original draft (supporting). Zbigniew Galazka: Methodology (equal); Resources (equal); Writing – original draft (supporting). Andreas Fiedler: Conceptualization (supporting); Data curation (supporting); Formal analysis (supporting); Investigation (equal); Writing – original draft (supporting). Andreas Popp: Investigation (supporting); Methodology (supporting); Resources (supporting); Writing – original draft (supporting). Joachim Würfl: Conceptualization (equal); Formal analysis (equal); Funding acquisition (lead); Investigation (equal); Methodology (equal); Project administration (equal); Writing – original draft (supporting). Oliver Hilt: Formal analysis (equal); Funding acquisition (equal); Investigation (equal); Project administration (lead); Supervision (lead); Writing – original draft (supporting).

The data that support the findings of this study are available from the corresponding author upon reasonable request.