GOX 2023 Session HM-TuP: Heterogeneous Material Integration Poster Session II

Making Power Semiconductor Devices on starting heterogeneous engineered substrates gives numerical advantages over making them on bulk blanket wafers.

For β-Ga₂O₃ it allows to mitigate its critical disadvantage – low thermal conductivity that heavily limits its applications to power semiconductor devices.

In our process flow, we bond 100 mm cmmercially available polycrystalline SiC wafer to 100 mm commercially available β-Ga₂O₃ wafer using room temperarature surface activation bonding process. Then we thin the initial β-Ga₂O₃ to a mnimum thickness needed for desired device voltage.

As each micron of β-Ga₂O₃ withstand up to 800V, the typical final thickness is several microns.

Using polycrystalline SiC substate is for 2 reasons - it is more than 10X cheaper than single crystalline one, and it can have 100X lower electricall resistance compared to aregular nitrogen doped SiC. Indeed, the substrate here is just mechanical support and electrical contact, no semiconductor properties needed. For processes based on epitaxial growth - crystalline lattice is needed, while our process - wafer bonding - is independent on crystal structure.

Next we etch the continuous β-Ga₂O₃ layer into islands equal in shape and area to future power chips to be made on them. Reason is, the continuous layer will not withstand ~1000C processing steps for making MOSFETS and even Schottky diodes. The continuous layer breaks due to difference in thermal expansion. While the islands withstand the thermal processing. The process is being patented.

A P-type diamond and N-type Ga₂O₃ PN heterostructure could address these issues. Diamond could form an ideal heterojunction with β-Ga₂O₃ due to their ultra-wide bandgaps. Diamond is relatively easy to make P-type and has high thermal conductivity; a simulated PN junction is shown in the supplemental document [3,4]. This structure can simultaneously address both problems. However, significant work in device design and integration of epitaxial growth is needed to realize this concept.

In this study, we established a model for the PN heterojunction. We investigated the energy band diagram for the P-diamond/I-Ga₂O₃/N-Ga₂O₃ structure, edge termination to mitigate electric field crowding, drift layer design (I-Ga₂O₃) to increase the breakdown voltage and reduce the on-resistance, and temperature dependence. We successfully designed 10 kV P-diamond/I-Ga₂O₃/N-Ga₂O₃ power PN diodes, and the results are very promising for this type of ultra-wide bandgap PN heterojunction. Effects of interface states on device performance were also investigated due to the importance of epitaxial growth.

[1] A. J. Green et al., "β-Gallium oxide power electronics," APL Materials, vol. 10, no. 2, p. 029201, 2022.

[2] Y. Yuan et al., "Toward emerging gallium oxide semiconductors: A roadmap," Fundamental Research, vol. 1, no. 6, pp. 697-716, Nov. 2021

[3] S. Pearton et al., "A review of Ga₂O₃ materials, processing, and devices," Applied Physics Reviews, vol. 5, no. 1, p. 011301, 2018.

[4] P. Sittimart, S. Ohmagari, T. Matsumae, H. Umezawa, and T. Yoshitake, "Diamond/β-Ga₂O₃ pn heterojunction diodes fabricated by low-temperature direct-bonding," AIP Advances, vol. 11, no. 10, p. 105114, 2021.

The use of UWBGs (Ultra-Wide Bandgap Semiconductors) based power converters is an emerging technology that will revolutionize power electronics industries.Space-rated DC-DC converters' performance and power density are primarily limited by high-power Metal Oxide Semiconductor Field Effect Transistors (MOSFETs). Power MOSFETs are very susceptible to damage and degradation from the irradiation found in space, especially ionizing radiation. As a response to the current technology gap, Syrnatec in collaboration with University at Buffalo has developed a revolutionary Ga2O3 technology-based UWBGs. These Ga2O3 MOSFETs are capable of demonstrating more robustness to single event effects than their rad-hard power MOSFET counterparts. Because Ga2O3 MOSFETs do not have a metal oxide layer, they are very robust to ionizing radiation, which prevents charge entrapment from TID in high radiation environments. After exposure to 500 krad (Si) ionizing doses, early radiation tests on first generation Ga2O3 MOSFETs showed less than 4% threshold voltage variation (VTH) and less than 3% RDSON change. When the devices were in the OFF state, higher variation was reported (18% VTH and 8% RDSON). In second generation Ga2O3 MOSFETs, no performance degradation has been observed from TID to 1.0 Mrad (Si).Syrnatec’s Gallium Oxide MOSFET, an upcoming wide bandgap material that is not only inherently radiation tolerant, but is also suitable for operating in environments with extreme temperatures such as lunar night, where the temperature changes from -153 degrees Celsius to 123 degrees Celsius , and -125 degrees Celsius to 80 degrees Celsius.

Syrnatec will incorporated its Ga2O3 technology into DC-DC converters with a bulk voltage of 20% to 80% and a trickle voltage of above 80%. With a maximum and minimum bulk charge timer (validated as the charge parameters), a Trickle voltage per cell (to be 2V), Boost and trickle voltage settings (Boost is 120% of the rated voltage, Trickle is 2V) and a Device switch off setting (tested on a battery under 20%). With no errors in over voltage and over current test conditions at 150% rated input for 1 sec out of every 10 seconds while maintaining an average of 100% overall rated values for the other 9 seconds. In addition, we have evaluated the success of fault detection across the entire Military Grade Temperature flow. In both buck and boost modes, power conversion efficiency exceeded 96% over the entire temperature range.

Overall, Ga2O3 based power converters can bring several novel features to the US commercial market, including high breakdown voltage, high thermal conductivity, wide bandgap, and low cost.

Ga₂O₃, an ultrawide-bandgap semiconductor, has attracted substantial attention in recent years due to its exceptional electronic properties and its vast potential in power electronics and solar-blind optoelectronics [1]. Despite these attractive properties of Ga₂O₃, there are some challenges to be addressed. For instance, the long-standing issue of lack of p-type doping in Ga₂O₃ has persisted [2]. The inefficiency stems from high ionization energy of acceptors when using the common dopants in Ga₂O₃. As a result, the design and fabrication of high-performance bipolar Ga₂O₃ devices, such as p-n diodes, and HBTs, are still in the research and development stage.

Semiconductor grafting [3], which enables the formation of heterostructures between two arbitrary monocrystalline semiconductors, could be the approach to overcoming the current constraints through the creation of Ga₂O₃ heterostructures, wherein a foreign semiconductor with good p-type doping to integrate with Ga₂O₃ at the atomic level. In this approach, an ultrathin oxide (UO) layer at sub-nanometer scale serves both as the interfacial passivation layer and an effective quantum tunneling layer. In the present case, the surface Ga₂O₃ layer and the possible native oxide of Si should have played the role of the UO layer in the grafting approach.

Employing the semiconductor grafting technology, two types of Ga₂O₃ heterojunctions are created, including Si/Ga₂O₃ and GaAsP/Ga₂O₃, to address the current challenges of ineffective p-type doping in Ga₂O₃ and lack of bipolar devices. In these two structures, p-type Si and p-type GaAsP nanomembranes (NM) are released from their respective epi substrates, transfer-printed and then subsequently chemically bonded to the n-type Ga₂O₃ substrates, forming PN abrupt heterojunctions, and the grafted heterostructures were subsequently fabricated into PN diodes. Their respective diode schematics are shown in Figs. 1 (a) and (b), with preliminary I-V curves for both diodes displayed in Figs. 1 (c) and (d). Both Si/Ga₂O₃ and GaAsP/Ga₂O₃ exhibit excellent rectifying behaviors with rectification ratios of 10⁷ and 10³ at ±2V, respectively. Meanwhile, their ideality factors are characterized to be 1.13 for Si/Ga₂O₃ diode and 1.35 for GaAsP/Ga₂O₃ diode.

In conclusion, we have demonstrated the feasibility to fabricate Ga₂O₃ bipolar devices via the semiconductor grafting approach. The demonstration of the high-performance Si/Ga₂O₃ and GaAsP/Ga₂O₃ PN diodes could lead to functional Ga₂O₃ HBTs in the near future.

References:

[1] M. Higashiwaki et al. (2016). Semi. Sci. and Tech.

[2] E. Chikoidze et al. (2017). Materials Today Physics

[3] Liu et al. (2018). arXiv:1812.10225.