United States

Motivation To reduce the device cost, body diode is expected as the freewheeling diode for commercial trench SiC power MOSFETs in the high frequency applications. Therefore, it is necessary to investigate the reliability of body diodes. This study compares the body diode degradation of commercial 1.2 kV planar and trench SiC power MOSFETs from several vendors listed in Table 1. Besides, 1.2 kV planar SiC power MOSFETs are fabricated with the nominal and deep JFET. The deep JFET extends down from the nominal JFET into the drift layer, and part of the additional n+ layer is in contact with p-body. We assume that the deep JFET enhances the recombination of the injected minority carriers from p-body. This leads to the suppression of body degradation by reducing the recombination energy provided to expand SFs. These two different JFET structures are presented in Fig. 1. Results Fig. 2 compares the average degradation ratio of VF, RON, and forward leakage current in the blocking mode (IL) on the commercial 1.2 kV planar SiC power MOSFETs from Vendors E, I, and C. Decline in device performance from body diode degradation is at negligible or acceptable levels for all these three vendors. It can be concluded that body diode degradation is not a big issue in most commercial 1.2 kV planar SiC power MOSFETs. In Fig. 3, commercial trench DUTs from Vendor K have more significant average body diode degradation than planar DUTs. Trench DUTs from Vendor D have no obvious degradation. However, the influence of body diode degradation on IL is not reflected as expected. This may be due to the fact that devices are breaking down in the termination area where no SFs exist, as opposed to the active area. As shown in Fig. 4, 7 of 10 DUTs from Vendor K have reached the degradation criterion while there’s no significant body diode degradation in 8 DUTs, apart from the 2 dead devices, for Vendor D. This can be explained in Fig. 5 that the deep p+ contact by the Al implantation with high dose and energy is required to shield the electric-field-crowded corner of the trench gate, which creates many BPDs around the p+ contact in the trench structure from Vendor K. Although the p+ contact with the heavy dose is still needed for Vendor D, it is implanted after the dummy trench formed by RIE. That means the p+ contact can be shallow which significantly reduces the new BPDs. In the planar structure, high density of new BPDs around the implanted regions is hard to create because of light-dose and shallow p-type contact, as well as many improved designs and processes that have already minimized the BPDs arising from the SiC substrate wafer and epitaxial growth. In addition, Vendor D not only has the Gen 3 trench structure mentioned above, but also has the Gen 4 trench structure as shown in Fig. 6. In the Gen 4 trench structure, the Al implantation with the high energy and dose for the deep p+ contact is performed before the RIE for the trench dummy. Therefore, this new trench structure is expected to have significant body diode degradation as that of Vendor K. The body diode degradation test on the Gen 4 DUTs from Vendor D is still in progress. Furthermore, Fig. 7 compares the body diode degradation of the fabricated 1.2 kV SiC power MOSFETs with the nominal and deep JFET. It can be observed that there is increasing average body diode degradation in the DUTs with the nominal JFET during the stress. However, for the DUTs with the deep JFET under the same process, there is no average body diode degradation during the stress. It is also reflected that the breakdown voltage of the DUTs with the nominal JFET decreases due to the increasing leakage current while there is no variation for the DUTs with the deep JFET. This confirms the assumption that body diode degradation in commercial 1.2 kV SiC power MOSFETs, especially in the trench devices, is mainly caused by the new BPDs around the p-type contacts introduced by the deep Al implantation with heavy dose.

EMC 2023

A Comparison of Body Diode Degradation in Commercial 1.2 kV SiC Power MOSFETs with the Planar and Trench Structure

trench

body diode degradation

implantation

mosfet

technical paper

The Electronic Materials Conference (EMC) is the premier annual forum on the preparation, characterization and use of electronic materials. This year’s event will feature a plenary session, topical sessions, a poster session and more.

EMC is coordinated with the Device Research Conference (DRC). This recognizes the strong interaction between device and electronic materials research and provides fruitful exchange of information between attendees of both Conferences.

The Electronic Materials Conference (EMC) is the premier annual forum on the preparation, characterization and use of electronic materials. 

Generation of basal plane dislocations (BPDs) in 4H-silicon carbide (SiC) epitaxial layers due to ion implantation was reported by Stahlbush et. al. [1]. These BPDs formed loops that were generated at the epitaxial layer surface within the implanted layer and propagated towards the substrate interface. Such BPD generation was observed when high dosage Al ion implantation was carried out at room temperature (RT) for typical contact formation. With recent advances in device processing, several of these newly generated BPDs have been mitigated either by high temperature (HT) ion implantation or by other processing techniques such as post implant oxidation [2]. It is, however, desirable to perform ion implantation at RT to reduce high temperature fabrication steps, which could affect wafer yield due to less thermal induced stress in the wafer. It is also unclear in device layers with differently doped regions if the BPDs would result in stacking fault (SF) formation and if the typical carrier injection conditions would cause SF expansion. In this work, we investigate SF expansion from potential BPDs generated by ion implantation in SiC epitaxial layers that had both RT and HT ion implantation and compare their robustness using high carrier injection using high intensity ultraviolet (UV) laser excitation. For this work, MOSFET device structure with 30 mm drift layers were fully fabricated on 4 commercial 150 mm diameter SiC wafers at a commercial foundry. Two of the wafers had a RT implant for the high Al dose contact layer and two remaining wafers had a HT implant performed at approximately 600 oC. Several of the device dies were removed from each of the four wafers and their metal layers were stripped to observe the generation of defects. UV photoluminescence (UVPL) imaging using a 355nm laser and a microscopy setup was performed with typical (10-20 W/cm2) excitation power on all the 4 samples. This was followed by high intensity (500 W/cm2) UV carrier injection. Sequence of images were collected to study progression of SF formation. The UVPL images obtained using typical excitation power did not introduce any BPDs in either the RT or the HT implanted samples even after exposure for several minutes. Upon UV excitation at 500 W/cm2, SF expansion from BPDs were observed in the RT implanted samples near the edge of the termination regions and not in the active areas within 8 secs. UV excitation generates carriers with an exponential depth decay uniformly across the device region, which is different from current injection that depends on background doping. Hence, in highly doped regions, UV excitation simulates significantly higher minority carrier generation responsible for SF generation compared to current injection. A comparison of UV excitation power with current injection conditions will be presented. The active device region did not show SF expansion, which is likely due to fast recombination of electrons from the implanted region due to depletion in the p-well. Details of the mechanisms will be presented. In the HT implanted samples, no SFs were observed even after 96 secs of high intensity UV exposure, indicated no generation of new BPDs. Even though SF expansion was not observed in active regions of the RT samples, BPD are likely formed in the contact layers and pose a reliability issue for long term device operation, especially if surge current conditions are expected during switching. [1] R. E. Stahlbush, N. A. Mahadik, Q. Zhang, et. al., Mater. Sci. Forum 821-823, 389 (2015) [2] Y. Bu, H. Yoshimoto, N. Watanabe, et. al., J. Appl. Phys., 122, 244504 (2017) Acknowledgement: Author Kallon’s research was sponsored by the Office of Naval Research NREIP program.

Reliability of Room Temperature vs High Temperature Implantation in 3.3kV SiC MOSFETs

Molecular electronics are being developed as materials for light-harvesters and -emitters, sensors, and quantum technology. The themes for optimizing these systems are miniaturization and improved precision, which increasingly moves these materials into the quantum-mechanical regime. Natural systems like photosynthetic pigment-protein complexes have shown that vast improvements to functional bottlenecks like energy-transport can be achieved, if the structural and energetic properties are finely controlled. In that spirit, we are developing modular molecular networks whose structures are controlled by DNA scaffolds. This approach controls the network’s structural, energetic, and dynamic properties. However, this increased control presents new needs and challenges. We must know what the ideal parameters are for optimal function. In turn, that optimization requires a better understanding of the molecular-scale characteristics of the systems. Another challenge is that these characteristics often include many degrees of freedom, especially for large chemical networks, so an approach must be developed to treat large amounts of data efficiently, yet thoroughly. Here, we describe approaches to address these challenges. By combining machine-learning tools with experimental spectroscopies and computational models (Figure 1), we can simultaneously understand and map the sample characteristics on the molecular scale, the energy transport characteristics on the mesoscale, and the material performance on the bulk scale. We will discuss how we have used these tools to predicted optimal system configurations for energy-transport, and how they reveal the interchromophore effects in chromophore arrays.

Understanding Electronic Energy Transport in DNA-Scaffolded Molecular Networks Using Machine-Learning Methods

Organic molecules, such as pentacene, rubrene and phthalocyanine are at the forefront of materials research, due to their desired electric, excitonic and mechanical properties. They represent the feasible alternative to traditional silicon-based semiconductor applications for flexible electronics platforms, where materials in flexible devices will experience complex strain environment. Octabutoxy phthalocyanine (H2-OBPc) is a soluble derivative of the Phthalocyanine (Pc). It has π-conjugated systems and large directional intermolecular interactions, which allow the formation of molecular crystal with well-defined structure like J-aggregate, where molecules are stacked in a head-to-tail fashion (Fig.1a). 1, 2 In this quasi one-dimension system, the Frenkel exciton can couple with lattice and intramolecular vibrations to form exciton-polarons which are delocalized along the molecular aggregates. The formation of these one-dimensional delocalized exciton-polaron states grants H2-OBPc the potential for efficient excitation energy transport in the coherent exciton regime. 3-6 In this work, H2-OBPc polycrystalline thin films were prepared using a solution-based pen-writing deposition technique. These ordered thin films with macroscopic grain size (Fig.1b, millimeter scale) uniquely offer the opportunity to spatially explore optical and excitonic properties within a single crystalline grain, free from the influence of grain boundaries or static disorder (Fig.1b). Temperature-dependent absorption and photoluminescence (PL) experiment points to the existence of red-shifted bandgap when temperature T < 175 K (Fig.2c and d). This new bandgap at low temperature is the result of the formation of exciton-polaron states, where the molecular Frenkel excitons couple with intermolecular low energy phonon modes (i.e. dynamic disorder fluctuations ), leading to the wavefunction spreading over the π-stacked molecular chain (Fig.1a). These delocalized exciton-polaron states have robust emission properties and they are potentially coherent. 6 The critical role played by low-energy phonons (25-300 cm-1) 7 in the formation of delocalized, 1D coherent exciton-polaron states is explored using an externally applied uniaxial mechanical strain (example of the tensile strain geometry is shown in Fig.2a). Strain-dependent absorption (Fig.2b), PL microscopy (Fig.2c), and confocal Raman measurements (Fig2.e) reveal the strain-induced softening (red shift) of the phonon modes favors the survival of the coherent exciton-polaron states at room temperature (A2 peak in Fig.2b and P2 peak in Fig.2c). Remarkably, an 80% enhancement in PL intensity is observed with a uniaxial tensile strain of 4.7% applied at room temperature (Fig.2c), which is equivalent to lowering the temperature to 200K for the unstrained film. 8 Time-resolved transient absorption (TA) spectroscopy confirms the behavior of the exciton-polaron state as a function of strain. Reference (1) Rawat, N.; Pan, Z.; Manning, L. W.; Lamarche, C. J.; Cour, I.; Headrick, R. L.; Waterman, R.; Woll, A. R.; Furis, M. I. J. Phys. Chem. lett. 2015, 6, (10), 1834-1840. (2) Manning, L. W.; Rawat, N.; Lamarche, C.; Waterman, R.; Headrick, R. L.; Furis, M. J. Phys. Chem. C 2016, 120, (22), 11966-11976. (3) Ghosh, R.; Spano, F. C. Acc. Chem. Res. 2020, 53, (10), 2201-2211. (4) Spano, F. C. Acc. Chem. Res. 2010, 43, (3), 429-439. (5) Yamagata, H.; Spano, F. C. J. Phys. Chem. lett. 2014, 5, (3), 622-632. (6) Liang, L.; Burrill, K.; Furis, M. I. J. Phys. Chem. C 2021, 125, (51), 27966-27974. (7) Fornari, R. P.; Aragó, J.; Troisi, A. J. Phys. Chem. C 2016, 120, (15), 7987-7996. (8) Liang, L.; Czar, K.; Furis, M. I. J. Phys. Chem. C 2022, 126, (20), 8889-8896.

Strain-Enhanced Formation of 1D Coherent Exciton-Polaron States in Small Molecule Semiconductors

There have been studies of understanding and controlling the charge transport through molecular junction wince it gives fundamental grounds that helps to gain a novel idea to realize numerous potential molecular functional devices. [1,2] There are some key factors considered in molecular charge transport, such as molecular barrier which is directly related to molecular orbital levels, coupling strength, asymmetric factor, and molecular configuration and conformation. Much research has been reported to obtain deeper insights of such key factors under certain controlled conductions with the effect of the external stimuli applied on the molecular junction such as light irradiation [3], electrochemical gating [4], electrical field [5], and mechanical stress such by tip loading force [6]. Molecular heterojunction with two-dimensional (2D) semiconductor provides various key factors for the charge transport through the heterojunction; however, the effect of molecular tilt configuration on the molecular heterojunction have never been studied. Here, we present the effect of molecular tilt configuration on the molecule/mono-layer (1L) 2D semiconductor heterojunction by tip loading force with conductive atomic force microscopy (C-AFM) technique. With various tip loading force (1-30 nN), the molecular tilt configuration affects differently for the three different interfaces in Au/molecule/1L-2D semiconductor/Au heterojunction that the transition voltage spectroscopy (TVS), which is a conventional tool to analyze charge transport through molecular junction, rectification ratio (RR), non-linearity (NL) and possible array size are demonstrated. Molecule/2D semiconductor heterojunction system has been reported as molecular diode (with RR) and selector (with NL) and it is a novel platform owning so much various potential in, that understanding the charge transport by the effect of external stress on the heterojunction will help developing such future molecular scale functional devices. [1] L. A. Bumm, J. J. Arnold, M. T. Cygan, T. D. Dunbar, T. P. Burgin, L. Jones, D. L. Allara, J. M. Tour, P. S. Weiss, Science 271, 1705 (1996),. [2] H. Song, Y. Kim, Y. H. Jang, H. Jeong, M. A. Reed, T. Lee, Nature 462, 1039 (2009). [3] G. Reecgt, C. Lotze, D. Sysoiev, T. Huhn, K. J. Franke, ACS Nano 10, 10555 (2016). [4] R. A. Wong, Y. Yokta, M. Wakisaka, J. Inukai, Y. Kim, Nat. Commun. 11, 4194 (2020). [5] J. Shin, S. Yang, Y. Jang, J. S. Eo, T.-W. Kim, T. Lee, C.-H. Lee, G. Wang, Nat. Commun. 11, 1 (2020). [6] G. Wang, T. -W. Kim, G. Jo, T. Lee, J. Am. Chem. Soc. 131, 5980 (2009).

Effect of Molecular Tilt Configuration in Molecular Heterojunction with Two-Dimensional Semiconductor

Many semiconductor devices (e.g., light emitters and photovoltaics) utilize heterojunctions of doped and undoped layers or depend on gradients of electronic doping to control charge transport. Doping of conjugated polymers requires the introduction of small molecules dopants which undergo charge transfer reactions with the polymer. The rate of the doping reaction and diffusion of the dopant controls the overall doping level. Conjugated polymers can be electrically doped by strong acids, but the details of the reaction mechanism and subsequent stability are not understood. Here, we show a clear kinetic isotope effect in the doping of thin films of poly(3-hexylthiophene) (P3HT) by bis(trifluoromethane)sulfonimide (HTFSI) from solution indicating that this doping process is limited by proton transfer to the polymer. The subsequent charge transfer reaction occurs rapidly between polymer segments in the solid state, thus circumventing the need to match redox potentials of dopants and conjugated polymers. Dynamic secondary ion mass spectrometry (DSIMS) of doped films show that H/D is retained in doped films after the doping process. Complementary X-ray photoelectron spectroscopy and DSIMS depth profiling of dopant concentrations show definitive evidence of enrichment at the P3HT surface. These surface-limited concentration profiles suggest that diffusivity of dopants vary inversely with dopant concentration due to doping-induced changes to the structure of the conjugated polymer.

Surface-Confined Brønsted Acidic Doping of Conjugated Polymers Thin Films

The paper reports on a novel process flow to manufacture conductive organic electrodes from highly conductive doped PEDOT:PSS polymer films that can be patterned and display good adhesion to oxidized Si wafers as well as to flexible substrates, such as Mylar. It is shown that multiple depositions of PEDOT:PSS increase the electrical conductivity by more than two orders of magnitude without increasing the film thickness of PEDOT:PSS significantly. An exponential dependence between sheet resistance and the number of PEDOT:PSS coatings is found. Additional coatings, in excess of 6 coatings provide diminishing returns to the lowering of the sheet resistance. The dependence of the sheet resistance Rsq(n) as function of the number of PEDOT:PSS coatings, n, can be described by the formula Rsq(n)=230×exp[-(n-1)×1.28] kΩ. The multiple coating methods can be further optimized by selecting a low spinning speed (e.g. 1500 rpm) for the initial coatings and a high spinning speed (e.g. 3000 rpm) for the final coatings. The reason for the decrease in the sheet resistance is the densification of the conductive PEDOT ribbons at the bottom part of the PEDOT:PSS layer. The electrical conductivity of PEDOT:PSS can be increased by another two orders of magnitude by doping it with Cu nanoparticles (Cu NP) when coated on the surface of a soft-baked PEDOT:PSS film. However, if Cu NP are dispersed in the aqueous solution of PEDOT:PSS then the impact on the sheet resistance is nil. It is found that both kinds of conductivity enhancement are not additive. The doping with Cu nanoparticles is desirable, as it provides the surface of an organic electrode with readily ionized copper ions useful for several applications including Cu filament formation in organic switching layers of a resistive RAM (ReRAM) memory cell. Adhesion of PEDOT:PSS to oxidized Si wafers and to BoPET (Mylar) has been ensured by applying an oxygen plasma cleaning step prior to spin coating. The manufactured high conductivity PEDOT:PSS film can be patterned using a sacrificial metal layer with subsequent etching of PEDOT:PSS in oxygen plasma, followed by the removal of the patterned segments of the sacrificial metal layer in an aqueous acid solution on both Si wafers and flexible Mylar substrates. A single PEDOT:PSS layer has a thickness of around 60 nm depending on the spinning speed and has sheet resistance (Rsq) of 230 kΩ. The same deposition applied 6 times results in a PEDOT:PSS layer of 90 nm thickness (instead of 6×60 nm=360 nm) and Rsq=980 Ω. A surface doping of 1xPEDOT:PSS with Cu NP lowers the sheet resistance from 230 kΩ to 60 kΩ, whereas in the case of multiple coatings, the 6xPEDOT:PSS with the same Cu doping lowers the Rsq from 980 Ω to 6.2 Ω. Thus, with reference to one PEDOT:PSS layer, the sheet resistance has been reduced by a factor of 4×104. The methods described above can be further combined with the methods of in situ doping of PEDOT:PSS with inert metal NP such as Ag and Au NP, that are largely immune to the oxidation reactions in the aqueous solutions, and with acid treatment such as nitric or phosphoric acid. It has been found that all the various methods to increase the conductivity mentioned above are not additive and, in some cases, work at cross-purposes with one another. Therefore, a tailored optimum combination of the available methods requires a detailed understanding of their respective mechanisms. Careful morphological analysis performed by dint of optical microscope picture and surface roughness plots by AFM reveals a close interplay between the thickness of the PEDOT:PSS film, the density of the PEDOT:PSS particles - here controlled by the multiple coatings -, and the size of the Cu NP. The size of Cu NP should be smaller than the film thickness, and the density of PEDOT:PSS particle should be neither too small nor too high to accommodate the Cu NP in the PSS spaces between the conductive PEDOT ribbons to achieve the lowest sheet resistance.

Manufacturing of Highly Conductive Organic PEDOT:PSS Films for Electronic Devices

Electrocorticography (ECoG) is a critically important neuroscientific tool for recording neural activity in the cerebral cortex. Advances in recent years have produced high-resolution ECoG arrays, which have enabled brain-machine interface (BMI) applications such as speech and motor intention decoding systems. Alternatively, cortical activity can be studied using optical tools such as genetically encoded calcium indicators (GECIs) and optogenetics. Optical access to the brain is typically achieved with a small craniotomy covered by a transparent window. Electrophysiology and optical imaging provide complementary neural sensing capabilities: electrophysiological recordings have high temporal resolution, while optical imaging allows recording of genetically-defined populations at high spatial resolution. Combining these two modalities for simultaneous large-scale, multimodal sensing of neural activity across multiple brain regions can be very powerful. To this end, many research groups have developed transparent ECoG arrays that can be used simultaneously with optical recording methods in small rodents and nonhuman primates. These devices are made possible by using transparent conductive materials such as graphene or indium tin oxide (ITO). Alternatively, poly(3,4-ethylenedioxythiophene) polystyrene sulfonate (PEDOT:PSS) is a biocompatible transparent organic conductor which is routinely used to coat chronic neural interfaces in order reduce the interfacial impedance of the contact site. We have developed inkjet-printed transparent PEDOT:PSS microelectrodes which use a novel ring-shaped geometry to reduce interfacial impedance and improve charge injection. [1] This is accomplished by taking advantage of the volumetric capacitance in PEDOT:PSS. Ring-shaped microelectrodes with diameters ranging from 300 µm to 550 µm were fabricated by inkjet-printing PEDOT:PSS, encapsulating with Parylene C, and then exposing a contact site that is much smaller than the microelectrode outer diameter. This unique structure allowed the encapsulated portion of the microelectrode volume surrounding the contact site to participate in signal transduction, which reduced impedance and enhanced charge storage capacity. For the same size contact site, increasing the outer diameter of the encapsulated electrode from 300 to 550 μm reduced the impedance from 294 ± 21 to 98 ± 2kΩ, respectively, at 1 Hz. Similarly, the charge storage capacity was enhanced from 6 to 21 mC cm-2. The PEDOT:PSS microelectrodes provide a low-haze, high-transmittance optical interface, demonstrating their suitability for optical neuroscience applications. They remained functional after a million 1 V stimulation cycles, up to 600 μA of stimulation current, and more than 1000 mechanical bending cycles. We have integrated these inkjet-printed PEDOT:PSS electrodes into transparent cranial windows for multi-modal recording of activity from the cortex.[2] Implanted on transgenic mice expressing the Calcium (Ca2+) indicator GCaMP6f in excitatory neurons, these “eSee-Shells” provide a robust opto-electrophysiological interface for over 100 days. eSee-Shells enable simultaneous mesoscale Ca2+ imaging and electrocorticography acquisition from multiple brain regions covering 45 mm2 of cortex under anesthesia and in awake animals. These eSee-Shells allow either mesoscale imaging across the entire field of view (FOV) or random-access cellular level imaging of small FOVs throughout ≈45 mm2 of the cerebral cortex, an area that is an order of magnitude larger than previous transparent ECoG electrode arrays. [1] P. D. Donaldson and S. L. Swisher, Physica Status Solidi (a) 2022. [2] P. D. Donaldson et al., Adv. Healthcare Mater. 2022.

Transparent, Inkjet-Printed PEDOT:PSS Electrotrode Arrays for Large-Area Multimodal Neural Interfaces

We present vanadium (V) based ohmic contacts to high composition n-AlGaN grown on single crystal bulk aluminium nitride substrates by plasma-assisted molecular beam epitaxy (PAMBE). Metal-semiconductor contacts are integral for the performance of electronic devices such as transistors, light-emitting diodes (LEDs), lasers and photodetectors [1,2]. Low-resistance Ohmic contacts have proven difficult to achieve on n-AlGaN epilayers grown on single crystal bulk AlN substrates as compared to the ones grown on AlN-on-sapphire templates [3]. This is because the large density of threading dislocations allow for conduction pathways [4]. While the low dislocation density of the films grown on bulk substrates reduce optical losses, it also makes low resistance ohmic contacts difficult due to lack of such conduction channels. Since the bandgap of AlGaN increases with increasing Al composition, it is difficult to find suitable metals with low work functions for a low energy barrier in the conduction band. Therefore, an alternate way to make ohmic contacts to wide bandgap semiconductors is to heavily dope the semiconductor layers to thin down the depletion enough to enable tunneling of electrons. A Ti-based metal stack has been shown to work for low resistance ohmic contacts to low composition n-AlGaN [5]. More recently, vanadium based ohmic contacts have been achieved on entire AlGaN composition grown on sapphire wafers [6]. It is hypothesised, that through annealing, vanadium nitride (VN) forms at the junction and has a work function of 3.56 eV [7]. In this work we demonstrate V-based Ohmic contacts on Si-doped Al0.7Ga0.3N grown by PAMBE on bulk AlN. Degenerately doped AlGaN layers with Silicon doping concentrations [Si] = 3e19/cm3-9e19/cm3 were used. The metal-polar single-crystal AlN substrates were cleaned ex-situ using solvents and acids, as well as in-situ by Al-assisted cleaning [8,9]. A 500 nm thick buffer layer was grown in Al-rich conditions to separate the AlGaN layer from the substrate. Then a 400-440 nm thick AlGaN layer was grown in metal-rich conditions. Atomic steps were observed in the coherently strained epilayers that have an energy bandgap of 4.8 eV as measured by room temperature photoluminescence. The samples were blanket etched 70 nm using chlorine-based reactive ion etching to emulate the n-AlGaN contacts as required for a light-emitting diode or laser diode. Two metal stacks were explored. Bottom-up, they consist of V(20n m)/Al(80 nm)/V(40 nm)/Au(100 nm) and V(20 nm)/Al(80 nm)/Ni(40 nm)/Au(100 nm). They were deposited using an e-beam evaporator and annealed by rapid thermal annealing (RTA) for 30 seconds at 800oC in N2 ambience. Linear transfer length method (TLM) analysis was done to determine the specific contact resistance and sheet resistance for which the results are shown in Figure 2. Successive annealing steps were performed to explore the effect of annealing time on the contact resistance. We found that a 30 second anneal is enough to make a low resistance ohmic contact. For both samples V/Al/V/Au contact gave a lower specific contact resistance than the V/Al/Ni/Au contact. We find that only a very high doping (9e19/cm3) results in low resistance ohmic contacts in the 10-4 ohm.cm2 range. These optimised doping conditions and metal stacks are now available to be used for ultrawide bandgap photonic and electronic devices on homoepitaxial AlGaN. References: [1] AIP Advances 12, 035023 (2022) [2] Appl. Phys. Lett. 121, 222103 (2022) [3] Appl. Phys. Lett. 114, 113501 (2019) [4] Journal of Applied Physics 117, 245702 (2015) [5] Sampath, A., et al. "Metal Contacts to n-AlxGa1-xN." MRS Online Proceedings Library (OPL) 482 (1997) [6] Appl. Phys. Lett. 90, 062115 (2007) [7] Fomenko, V. S. (2012). Handbook of thermionic properties. Springer Science & Business Media [8] Applied Physics Letters 116.17 (2020): 17210 [9] Applied Physics Letters 116.26 (2020): 262102

Ohmic Contacts to Homoepitaxial Ultrawide Bandgap n-AlGaN Grown on Bulk AlN Substrates

Low specific contact resistivity ohmic contacts (ρc ~ 10-6 Ω-cm2) with thermal and chemical stability play an important role in III-nitrides based devices. A Ti-Al-Ni-Au multilayer metallization is the most commonly used scheme and requires high temperatures (>800oC) to form reliable ohmic contacts. There are a few reports on the surface microstructures and chemical composition of the Ti-Al based contacts [1-2]. The low contact resistivity has been explained via the formation of TiN and TiAl3 phases along with Al-Ni agglomerates surrounded by alloys of Al-Au, mainly AlAu2, Al2Au5 and AlAu4. We have investigated low ρc ohmic contacts (ρc ~ 10-7 Ω-cm2) formed on n-GaN using a Re-Al-Ni-Au multilayer stack annealed at 650-700oC [3]. In this work, we report the effect of Re layer thickness and annealing temperature on the microstructure of these contacts. Several samples with Re thickness (10, 30, 60 nm) were prepared and annealed at 450oC—800oC temperatures. The ρc was estimated using C-TLM measurements while AFM, GIXRD and SEM equipped with EDS were used to analyze the chemical composition and microstructures of the contacts. AFM images show surface microstructures as a function of annealing temperatures in the range of 500—800oC in steps of 50oC for 30 nm Re samples. The metals form almost a continuous film with signs of segregation into small clusters at an annealing temperature up to 500oC. The films show more extensive segregations into separate islands, forming larger agglomerates and the density of these islands keeps on reducing, with a concomitant increase in height with increasing temperature. The RMS roughness increases linearly with annealing temperature. AFM images of the surface microstructures as a function of Re thickness for the annealing temperature where lowest ρc is observed; (a) 10 nm at 650oC and (b) 30 and 60 nm at 700oC; shows that as the Re thickness increases, the size of the agglomerates increases, with the corresponding reduction in the density. The RMS surface roughness also increases with Re content. For contacts with 10 nm Re, the RMS roughness starts increasing slowly with increasing temperature from 450oC with a faster rate of increase beyond 600oC. For contacts with 30 nm Re, the surface roughness increases linearly over the entire temperature range 500—800oC. However, in case of contacts with 60 nm Re, the surface roughness starts increasing initally and then saturates around at ~170 nm. EDS elemental scans for Re, Al, Ni and Au for samples with Re thickness 10, 30 and 60 nm, annealed at temperatures for which ρc is minimum, shows that the Ni layer disintegrates to form isolated islands and also Al-Ni agglomerates. There is a thin continuous layer of Al over the entire surface with a few isolated patches devoid of aluminum. A continuous layer of Re underneath is observed across the surface. The dark spots observed in the Re scan are due to the thick agglomerates which absorb the signal from the underlying Re layer. The gold layer seems to get accumulated around the agglomerates. Glancing angle SEM image of the surface of 60 nm Re sample annealed at 650oC shows hills formed due to the Al-Ni agglomerates covered by Al-Au films spread across entire surface. A cross section view of one of the agglomerates shows a thick lump of Al-Ni in the centre of the hill, while rest of the surface is covered by an Al-Au layer. The Re layer is delineated from the top structure. The Re layer may form intermetalic phases of Re-Ni-Al in the center, while the different phases of Al-Re are formed on the entire surface. The combinations of phases of Al-Re, Re-Al-Ni, Al-Au along with a thin layer of Re-N formed between the Re and GaN interface gives the low ρc of this material system. [1] S. Ruvimov, et al., Appl. Phys. Lett., 69, 1556 (1996). [2] W. Macherzynski et al., Optica Applicata, XLIII, 67 (2013). [3] A. P. Shah, et al., J. Appl. Phys., 132, 075701 (2022).

Evolution of Surface Microstructure of Re-Al-Ni-Au Based Ohmic Contacts on N-Type GaN

Quantum well infrared photodetectors (QWIP) based on III-Nitrides are a subject of increasing interest due to their potential for applications in telecommunications, optical gas sensors, military, and imaging. These detectors rely on intersubband transitions (ISBTs) of an electron (hole) between the ground state and the first excited state in the conduction (valance) band. Large conduction band offset (CBO) of nearly 2 eV [1], [2] and valance band offset (VBO) of nearly 0.85 eV [3] in AlN/GaN QWIPs make this material system suitable for a wide range of ISBTs depending on Al composition and specifically in the spectral regions where GaAs-based QWIPs are limited due to their smaller bandgap offsets. Also, the longitudinal-optical (LO) phonon energy of GaN (92 meV) is more than two times higher than GaAs [4], which makes this material an ideal choice for room temperature detection application. By employing GaN-based QWIPs, ISBT energies can be tuned from a few meV to 2 eV based on QWIP design, allowing for full NIR to IR band tunability within a single material system. While III-N materials can be grown in cubic zincblende structure, only hexagonal wurtzite structure is thermodynamically stable. In this structure, lack of inversion symmetry along the (growth) c-axis, generates a spontaneous polarization (Psp) along this axis [5]. Additionally, at AlGaN/GaN heterointerface, the lattice mismatch causes strain leading to piezoelectric polarization [6]. The discontinuity of polarization at heterointerfaces results in an uncompensated sheet charge at the interface. Further due to the polarization selection rule, normal incidence absorption is not allowed in n-QWIPs [7], requiring a light coupler and adding to design complexity. To circumvent this issue, p-QWIPs are being considered. In this work, we report on the growth of 5 periods Mg-doped Al0.4Ga0.6N/GaN QWIP, by metal organic chemical vapor deposition (MOCVD) on c-plane sapphire for a target 1.55 µm absorption. Critical to the performance of the QWIP are high controllability of layer thicknesses and composition in superlattice and interface quality. Sub-nanometer control of layer thicknesses is challenging in MOCVD growth due to the high temperature and the relatively poorer control of precursor flux on the sample surface, compared to other methods such as MBE. To overcome this, a pulsed growth method was used to achieve higher control of growth rate [8]. A TCAD-based physics model was implemented in the commercial Synopsys Sentaurus to model the energy band diagram, electric field and calculate the band offsets between AlxGa1-xN/GaN heterointerfaces by considering polarization charges. Based on the CBO and VBO in the QW, layers thicknesses and compositions can be adjusted to achieve desired absorption wavelength. X-ray diffraction and transmission electron microscope (TEM) are used to assess strain, interface quality, layer thicknesses, and aluminum composition. Transmission/reflection measurement is used to determine the absorption spectrum of the QWIP. [1]D. Cociorva et al., ‘Quasiparticle calculations of band offsets at AlN–GaN interfaces’, Solid State Commun, 2002 [2]A. N. Westmeyer et al., ‘Determination of energy-band offsets between GaN and AlN using excitonic luminescence transition in AlGaN alloys’, J Appl Phys, 2006 [3]G. Martin et al., ‘Valence-band discontinuities of wurtzite GaN, AlN, and InN heterojunctions measured by x-ray photoemission spectroscopy’, Appl Phys Lett, 1996 [4]K. Ahi, ‘Review of GaN-based devices for terahertz operation’, Optical Engineering, 2017 [5]H. Ünlü et al., ‘Band offsets in III-nitride heterostructures’, J Phys D Appl Phys, 2002 [6]S. Nicolay et al., ‘Strain-induced interface instability in GaN/AlN multiple quantum wells’, Appl Phys Lett, 2007 [7]B. F. Levine, ‘Quantum-well infrared photodetectors’, J Appl Phys, 1993 [8]P. Suvarna et al., ‘Design and growth of visible-blind and solar-blind III-N APDs on sapphire substrates’, J Electron Mater, 2013

Premium content

A Comparison of Body Diode Degradation in Commercial 1.2 kV SiC Power MOSFETs with the Planar and Trench Structure

Downloads

Next from EMC 2023

Reliability of Room Temperature vs High Temperature Implantation in 3.3kV SiC MOSFETs

Stay up to date with the latest Underline news!

PRESENTATIONS

CONFERENCES

COMPANY

RESOURCES