Symposium EN02—III-V Semiconductors for Energy Conversion Technologies

With an increase of control over crystalline defects, metallorganic vapor phase epitaxy (MOVPE)-grown III-V-on-Si multijunction solar cells have seen rapid increases in efficiency in recent years [1, 2], pointing to a promising path to lower cost III-V solar cells. However, the cost of chemo-mechanical polishing the Si wafers to prepare them for epitaxy is high. The use of V-groove nanopatterns enables similar defect reduction to that achieved on planar wafers [3–5], but the nanopatterns can be fabricated with a low-cost process [6]. While V-grooves offer advantages over planar Si, they add complexity to the growth process. In particular, coalescence can cause the formation of threading dislocations [7], and the highly-directional growth conditions required for coalescence are unusual for MOVPE.

We observed that for optimized growth conditions (V/III=5,000 and T=800 C) two growth modes were possible, and the resulting morphology depended on the exact geometry of the SiNx cap used to cover the (001)-oriented Si at the tops of the grooves. For caps with a width >100 nm, non-coalescing, nano ridge-like growth terminating in {111} facets was observed. For thinner caps, coalescence with an RMS roughness of 0.2 nm as measured by atomic force microscopy was observed. We will discuss possible mechanisms for this phenomenon, including the role of Si from the substrate surface.
The dislocation dynamics of this system were studied with electron chan- neling contrast imaging (ECCI) and transmission electron microscopy (TEM). V-grooves do not block dislocation glide; ECCI measurements show misfit dis- locations greater than 10 μm long observed to continue perpendicularly across neighboring V-grooves. In addition, the TEM micrographs did not show evi- dence of sessile dislocations expected to result from coalescence-related dislo- cation formation [7]; instead, the elevated threading dislocation density (TDD) of 5 × 10⁷ cm^-2 likely is driven inefficient strain relaxation in the GaP due to non-optimized growth conditions rather than coalescence-related problems. The dislocation dynamics and morphological evolution of the coalescence of GaP on Si, possible mechanisms behind the observed phenomenons, and further dislo- cation mitigation strategies for these materials will be presented.
References
(1) Feifel, M. et al. Sol. RRL 2021, 5, 2000763.
(2) Lepkowski, D. L. et al. Sol. Energy Mater. Sol. Cells 2021, 230, 111299.
(3) Kunert, B. et al. Semicond. Sci. Technol. 2018, 33, 093002.
(4) Li, Q.; Lau, K. M. Prog. Crys. Growth Char. Mater. 2017, 63, 105–120.
(5) Saenz, T. E. et al. Crys. Growth Des. 2020, 20, 67456751.
(6) Warren, E. L. et al. In Proc. WCPEC-7, 2018.
(7) McMahon, W. E. et al. APL Mater. 2018, 6, 120903.

Numerous optoelectronic devices use III-V materials, and the materials and devices produced by growth technologies like organometallic vapor phase epitaxy and molecular beam epitaxy have been refined to an exquisite level of purity and performance. There are times, however, when cost comes into play, for example in large area/volume application such as solar cells, thermophotovoltaics, or thermoelectrics that require scaling to vast quantities to realize the economies of scale and achieve acceptable costs. This is where the dynamic hydride vapor phase epitaxy (D-HVPE) growth technique can play a role. D-HVPE is a variant of traditional HVPE growth that makes use of the technique’s low-cost precursors, high precursor utilization, and extremely high growth rates, but uses substrate motion to remove previous difficulties with the formation of the abrupt heterojunctions required in most device structures. Our recent roadmap shows that solar cell costs, for instance, can decrease from well in excess of $100/W to < $2/W, with D-HVPE playing a major role in the reduction. Lower costs of these area-intensive devices could enable a bright future for high-efficiency III-V photovoltaics.

Traditionally, there were materials that proved challenging during HVPE growth, e.g. aluminum-containing materials like AlGaAs and AlInP, while other materials and structures could not be attained with sufficient quality due to the high growth rates and the chemical inertia present in the process, such as compositionally-graded buffer layers, LEDs, quantum wells, or distributed Bragg reflectors. Recent work showed that greatly accelerating the surface kinetics through hydride-enhanced chlorine reduction is a key to unlocking even higher growth rates, as well as the formation of controllable compositions of Al-containing materials through use of the AlCl₃ precursor. It is simple to modify the parameter space during HVPE growth to be controlled predominantly by kinetics or mass flow, which allows for the HVPE practitioner to tune for a desired effect, including maximizing the growth rate for thick layers or dialing back for the deposition of very thin layers, accentuating the growth on different crystal planes, or in situ etching for purposes of texturing or regrowth.

In this presentation, we will discuss some of the recent materials science advances using D-HVPE and how they were achieved, including GaAs growth rates in excess of 0.5 mm/h (that’s millimeter, not a typo!) and GaInP growth rates of 0.2 mm/h, the development of AlGaAs and AlInP and their incorporation into high-efficiency solar cell devices, and atomic ordering and the dynamics of dislocation glide in GaInP grown at high growth rates. We will also cover additional topics that relate to lowering the cost of III-V devices, including the reuse of substrates by epitaxial planarization and thinning solar cells with the incorporation of high-throughput backside texturing. We will close with a short discussion of remaining opportunities in HVPE growth and the future of D-HVPE toward the production of lower-cost III-V materials and devices.

III-V semiconductors are desireable for many device applications due to their favorable electronic properties and superior device performance. The high cost of III-V device production, however, has prevented their widespread use in cost-sensitive markets such as solar energy conversion. One method of reducing III-V device costs is through the use of lower cost substrates. Rough substrates, such as those that have been cut from the boule and not polished, avoid costs of chemomechanical polishing (CMP), which is estimated to be $10-$25 per wafer. Rough substrates produced by layer transfer techniques further reduce the substrate cost per device through reuse. However, the morphological features on rough substrates may affect the quality of overgrowth and be detrimental to device performance if not removed. Crystallographic features can potentially be removed through epitaxial planarization by varying the growth conditions to favor lateral growth that facilitates the formation of the original surface plane. Understanding the growth behavior on different planes and morphologies can help move toward successful device growth on rough substrates.
In this work, we focus on planarization growth by hydride vapor phase epitaxy (HVPE) over facets produced by controlled spalling of (100)-oriented GaAs. HVPE is a potentially low-cost III-V growth technique, and the growth rate on different planes can be varied through parameters such as gas velocity, V/III ratio, and gas-phase supersaturation. Controlled spalling is a layer transfer technique that uses a stressor layer to help initiate and guide a lateral crack that allows rapid exfoliation of device layers from a parent substrate. (100)-oriented GaAs substrates do not have favorable cleavage planes oriented parallel to the wafer surface, so controlled spalling fractures follow low energy planes oriented at a high angle to the surface, resulting in the formation of triangular facets that expose low energy planes. The identity of the exposed planes and presence of surface steps on the facet faces can be controlled through the spalling process, allowing us to study the effect of specific morphologies on the growth evolution.
We present the observed HVPE overgrowth behavior on {110} facets with surface steps due to the substrate off-cut [6^○ towards (111)A], {112} facets without significant surface steps, and {112} facets with surface steps. The growth conditions used to compare the overgrowth behavior were determined from a statistical analysis of the effect of the flows used in HVPE growth on favoring planarizing growth over as-cut substrates. The growth evolution over the facets was tracked using AlGaAs marker layers. Our results show different growth evolution over the three facet types using the same growth conditions, and only growth over the stepped {112} facets achieved planarization. The overgrowth on the {110} facets was largely conformal over the approximately 20 µm of growth resulting in less than 10% areal fraction of (100) surface at the end of growth. The growth on the non-stepped {112} facets showed evidence of a lack of step flow growth indicated by instability of the growth evolution and formation of surface pits. Comparing the results of the two {112}-type facets also suggests that surface steps are an important feature for promoting step-flow growth and planarization over {112} facets. Relevance of this work to a variety of imperfect substrates for cost reduction of III-V devices will be discussed.

The record efficiencies in photovoltaic devices and hydrogen production via photoelectrochemical water splitting have been achieved in both cases by employing complex architectures of III-V alloys [1, 2], which has been enabled by the outstanding charge carrier transport properties, band gap and band edge engineering flexibility, as well as highly mismatch growth that can be achieved in III-V alloys by controlling their chemical composition.

It has been extensively demonstrated that Molecular Beam Epitaxy (MBE) and Metal-Organic Physical Vapor Deposition (MOPVD) techniques are capable of producing multinary III-V alloys with compositions beyond miscibility gaps due to their “far from equilibrium” nature. And although great progress has been made towards improving the growth rate in these techniques, III-V device synthesis cost is still significantly higher (one order of magnitude) than the incumbent Si solar cells manufacturing technologies [3]. In this context, it is necessary to develop novel synthesis methods with comparable growth quality capabilities, higher throughput, and lower operational costs than the afore mentioned ones. Hydride/Halide Vapor Phase Epitaxy HVPE-grown GaAs top cell incorporated in a GaAs/Si tandem cell is an example of an optimized photovoltaic device with 31.4% efficiency, showcasing the uttermost potential of this growth method [3].

More recently, the HVPE growth of GaSb_yP_1-y alloys has been studied with experimental and computational approaches, allowing the obtention of single crystal free-standing films with up to 6.7% antimony incorporation and growth rates around 500 µm/h. GaSb_yP_1-y alloys show lower onset potentials and higher fill factors compared to single crystal GaP when linear sweep voltammetries using 1M H₂SO₄ are carrier out. These results have been attributed to lower direct band gap transitions between 2.2 eV and 2.5 eV, and lower charge transfer resistance at the semiconductor/electrolyte interface of the GaSb_yP_1-y alloys versus pure GaP [4].

On the other hand, Vapor Liquid Phase Epitaxy (VLPE) is known for the low defect density materials it can produce and it has successfully been used for the growth of single crystal GaN at a growth rate of 1 µm/h with the use of ionic nitrogen gaseous species (plasma-assisted) and the careful control of their concentration in molten gallium by means of pulsing the applied radio frequency energy, preventing undesirable nucleation at the gallium/gas interface [5]. Furthermore, Ga-Nitride alloys have been synthesized with the same technique but including other V-group species ions, i.e., Sb+ and Bi+, that can substitute nitrogen forming GaSb_xN_1-x and GaBi_yN_1-y ternary 0-D or 1-D structures with up to 7% anion incorporation, leading to visible light absorption and band edge straddling of the Hydrogen Reduction and Oxygen Evolution electrochemical potentials.

References:
[1] W. H. Cheng et al., "Monolithic Photoelectrochemical Device for Direct Water Splitting with 19% Efficiency," Acs Energy Letters, vol. 3, no. 8, pp. 1795-1800, Aug 2018, doi: 10.1021/acsenergylett.8b00920.
[2] S. Essig et al., "Raising the one-sun conversion efficiency of III-V/Si solar cells to 32.8% for two junctions and 35.9% for three junctions," Nature Energy, vol. 2, no. 9, Sep 2017, Art no. 17144, doi: 10.1038/nenergy.2017.144.
[3] K. T. VanSant et al., "Toward Low-Cost 4-Terminal GaAs//Si Tandem Solar Cells," Acs Applied Energy Materials, vol. 2, no. 4, pp. 2375-2380, Apr 2019, doi: 10.1021/acsaem.9b00018.
[4] S. J. Calero-Barney, W. Paxton, P. Ortiz, and M. K. Sunkara, "Gallium antimonide phosphide growth using Halide Vapor Phase Epitaxy," (in English), Solar Energy Materials and Solar Cells, Article vol. 209, p. 9, Jun 2020, Art no. 110440, doi: 10.1016/j.solmat.2020.110440.
[5] D. F. Jaramillo-Cabanzo, J. B. Jasinski, and M. K. Sunkara, "Liquid Phase Epitaxy of Gallium Nitride," Crystal Growth & Design, vol. 19, no. 11, pp. 6577-6585, Nov 2019, doi: 10.1021/acs.cgd.9b01011.

Precision synthesis and processing of semiconductor compound alloys with controlled composition in multilayered device structures is critical to many energy conversion and information technologies including solid state light, power electronics, solar cells, satellite communications as well as light weight optics. As one example, a high indium content is required for the In_xGa_1-xN/GaN solid state light devices to fill the “green” light gap so that they can efficiently emit white light suitable for human eyes. Despite extensive studies, however, an abrupt reduction in the green light emission efficiency of In_xGa_1-xN films occurs when the indium content is increased to x > 0.2. As another sample, the CdS/CdTe-based solar cells can produce energy with a relatively low cost and therefore they can profoundly change the energy suppliers if their energy conversion efficiency achieved today can be further improved. The key to improve either the In_xGa_1-xN/GaN or the CdS/CdTe devices is to reduce or control the various defects formed during synthesis including dislocations and grain boundaries. To achieve this, the formation mechanisms of defects must be understood in atomistic details. These mechanisms are often difficult to explore using experimental approaches alone. Molecular dynamics simulation tools provide a powerful alternative approach to study atomistic assembly in greater details. In this presentation, we provide an overview of our past molecular dynamics simulation work on structure evolution during synthesis of semiconductor materials including multilayered compound alloys and some advanced knowledge gathered on processing of semiconductors. First, we describe molecular dynamics simulation methods of thin film growth. Next, we introduce an example semiconductor interatomic potential database. We then use the simulations methods to study a variety of problems relevant to the synthesis of semiconductor compound alloys including graphene growth, In_xGa_1-xN/GaN growth, CdTe/CdS growth, CdTe/GaAs growth, synthesis of Si crystals and SiC, atomistically-informed misfit dislocation theory, and dislocation migration in CdTe. MD has paved strong footings to a newer area of research to study “high pressure phase transformation” which seems to be most common mechanism of ductility in semiconductors beside dislocation nucleation. The impact of molecular dynamics simulations in semiconductor materials research is discussed.

Sandia National Laboratories is a multi-mission laboratory managed and operated by National Technology and Engineering Solutions of Sandia, LLC., a wholly owned subsidiary of Honeywell International, Inc., for the U.S. Department of Energy’s National Nuclear Security Administration under contract DE-NA-0003525. The views expressed in the article do not necessarily represent the views of the U.S. Department of Energy or the United States Government.

III-V semiconductor nanowires are promising building blocks in nanoscale energy conversion. Most prominently in photovoltaics and solid state lighting. In order to fabricate such pn-junction based devices we need to be able to p- and n-dope the nanowires but also understand the doping mechanisms. Here, thermodynamic knowledge of the nanowire-catalyst particle materials system is key.
We have previously demonstrated that we can grow p-type doped GaAs nanowires using aerotaxy [1]. This is a method where the nanowires are formed in the gas phase without substrate [2]. It is a continuous process, where a large number of nanowires per time unit can be produced. Thus, it is very promising for upscaling and for industrial production of nanowires. Moreover, the aerotaxy process also admits doping, which is necessary for electronic device fabrication. In this regard we have experimentally controlled the hole concentration by varying the Zn/Ga ratio during aerotaxy growth.
In this investigation we demonstrate that we can calculate the hole concentration in Zn-doped, gold alloy catalyzed GaAs nanowires grown by the VLS (vapor-liquid-solid) mechanism. We base the calculation on the defect formation energy proposed by Zhang and Northrup [3]. Using density functional theory (DFT), we calculate the energy of the defect, a Zn atom on a Ga site, using a supercell approach. The chemical potentials of Zn and Ga in the liquid catalyst particle are calculated from a thermodynamically assessed CALPHAD (CALculation of PHAse Diagrams) database including Au, Zn, Ga, and As. Using these quantities together with the chemical potential of the carriers, we calculate the hole concentration self-consistently.
Our calculations compare well with the experimental hole concentrations in the Zn-doped GaAs nanowires grown by aerotaxy. Specifically, it is interesting to note that the calculations confirm our experimental observation that it is difficult to achieve low and intermediate doping levels. Except for very low Zn concentrations in the catalyst particle and low growth temperatures, the hole concentration in the nanowire tends to be high (around 10¹⁹ cm^-3 and higher).
To conclude, we demonstrate accurate calculations of the doping concentration of VLS grown nanowires using a combination of DFT and CALPHAD. Such calculations are of great advantage when controlling and predicting the doping concentrations for a wide variety of experimental conditions.
References:
[1] F. Yang, et al., J. Cryst. Growth 414, 181–186 (2015)
[2] M. Heurlin, et al., Nature, 492, 90–94 (2012)
[3] S. B. Zhang and J. E. Northrup, Phys. Rev. Lett. 67, 2339-2342 (1991)

III-nitride materials have not reached their potential for energy conversion applications largely due to the difficulty in growing thick, high-quality, single-crystal ternary alloy films. One of the most prominent examples of this is the tendency for indium gallium nitride (InGaN) to separate into In-rich and In-poor regions during growth – a phenomenon typically referred to as phase separation. This has precluded the effective growth of this material for many of the applications which have motivated InGaN research since it was first synthesized (e.g., full spectrum or tandem-with-Si solar cells and RGB LEDs). Despite a large volume of research dedicated to this topic over the past three decades, the mechanisms that drive InGaN composition variability are not fully defined. In this work, we suggest that bulk diffusion and thus spinodal decomposition in the traditional sense, is of lesser concern than surface driven processes. Consequently, we propose that InGaN “phase separation” comes primarily from a combination of at least three distinct phenomena – thermal decomposition, vertical cation segregation (VCS), and lateral cation separation (LCS). Furthermore, thermal desorption complicates the control of composition in the grown film. A new 1D model describing the vertical segregation mechanism is proposed and evaluated using parameters extracted from state-of-the-art III-nitride epitaxy. Thermal decomposition and desorption are also included in the model to better replicate realistic growth conditions.
Via RHEED analysis, Moseley et al demonstrated that a critical dose of excess metal exists for epitaxy of InGaN beyond which a diffusion of indium away from the growth surface coupled with an equal gallium diffusion toward the growth surface (VCS) occurs [1]. This phenomenon can be expected to occur for any metal-rich growth condition. Our current understanding of vertical segregation suggests that it is not unique to InGaN and can be applied to any ternary III-nitride, and indeed, we have observed self-assembled super lattices in aluminum gallium nitride (AlGaN) grown by metal modulated epitaxy (MME), which does not naturally phase separate.
From the observation of VCS, we have built a model for the accumulation and consumption of metal adatoms during epitaxy of III-nitrides based on our current understanding of the surface kinetics. Using a system of coupled differential equations, we can describe the group-III (cation) adatoms as they adsorb onto and desorb from the surface, grow into and decompose from the crystal, and transfer between the pseudomorphic, laterally-contracted, and droplet adlayers. MME is particularly suited to evaluate this model as MME growth of III-nitrides can be conducted at substrate temperatures where decomposition and desorption are inhibited (even for InGaN), and lateral diffusion is enhanced leading to uniform adatom mixing. However, the use of high growth rates can prevent any significant LCS. This allows us to investigate growth kinetics at experimental conditions which isolate or emphasize the kinetic mechanisms in question. Further, the cyclic nature of MME makes the VCS effects easier to observe, as they occur repeatedly (every shutter cycle) during growth.
We found this model to be self-consistent and in line with our expectations for a realistic growth. We have also achieved further self-assembled super lattice structures in AlGaN by conducting growths at conditions conducive to vertical cation segregation. The simplicity of this model should make it easy to expand upon by adding other mechanisms like LCS and generalizing to other epitaxial techniques beyond MME and MBE.
[1] M. Moseley, B. Gunning, J. Greenlee, J. Lowder, G. Namkoong, and W. A. Doolittle, Journal of Applied Physics 112, 014909 (2012).

Thermophotovoltaic (TPV) cells are designed to convert photons emitted from hot surfaces into electrical power. In contrast to conventional PV cells that primarily absorb solar radiation in the visible and near infrared, TPV cells absorb the lower energy near-infrared to mid-infrared photons emitted from a high temperature source. Current TPV cells used in the range from 1500 - 1700K are based on narrow energy gap III-V semiconductors such as InGaAs which are limited by their high cost and challenges for scaling. Due to high throughput and maturity of fabrication technologies, Si PV cells cost orders of magnitude less than their III-V counterparts and benefit from excellent scalability and manufacturability, making Si an attractive and cost-effective alternative for TPVs. One disadvantage of Si for TPV cells is its larger band gap compared to InGaAs lattice-matched to InP (1.1 eV vs. 0.7 eV), thus requiring a higher source temperature (>2000K). To increase the power conversion efficiency (PCE) of Si TPV cells at lower temperatures, long wavelength out-of-band (OOB) photons must be reflected back to the thermal source to eventually be re-emitted as higher energy photons that are able to be absorbed by the cell. TPV cells using air-bridge back surface reflectors (BSRs) have been reported to increase OOB reflectance from 95% to 99% compared to conventional metal BSRs [1]. This corresponds to a 25% relative increase in PCE. In this work, we demonstrate a lateral PN junction silicon TPV cell with a 500 nm thick air-bridge BSR where the thin film semiconductor is suspended between Au metal grid lines. The devices exhibit low series resistances of 45-100 mΩ-cm² required to conduct the very high currents (> 1A/cm2) without significant loss. With the use of an anti-reflective coating, normalized external quantum efficiencies > 90% are achieved, along with an estimated OOB reflectance as high as 99.5%. This performance shows promise for high efficiency Si TPV cells that are both scalable and affordable, using environmentally friendly, earth-abundant materials.

[1] Fan, D., Burger, T., McSherry, S. et al. “Near-perfect photon utilization in an air-bridge thermophotovoltaic cell.” Nature 586, 237–241 (2020).

The addition of nanoscale defects to thermoelectric materials can significantly increase the figure of merit, ZT, by reducing the thermal conductivity. Unfortunately, these defects are also detrimental to the thermoelectric power factor in the numerator of the figure of merit ZT. We have derived strategies to recoup and even enhance the electrical performance of nanostructures Si containing additive SiC nanoparticles/ porosity by fine tuning the carrier concentration and through judicious design of the defects size and shape so as to provide energy selective electron filtering. A semiclassical Boltzmann transport equation is used to model the electrical transport properties of the Seebeck coefficient and electrical conductivity. The model is validated against a set of experiments on Si/SiC nanocomposites. The nanoinclusions lead to a significant improvement of the Seebeck coefficient. The theoretical analysis confirms that the enhancements in the thermoelectric properties are consistent with the energy selective scattering of electrons induced by the offset between the silicon Fermi level and the carbide conduction band edge. We also considered Si membranes containing discrete pores that are either spheres, cylinders, cubes, or triangular prisms. This model reveals three key results: The largest enhancement in the Seebeck coefficient occurs with cubic pores. The fractional improvement is about 15% at low carrier concentration up to 60% at high carrier population with characteristic length around 1 nm. To obtain the best energy filtering effect at room temperature, nanoporous Si needs to be doped to a higher carrier concentration than is optimal for bulk Si. Finally, in n-type Si thermoelectrics the electron filtering effect that can be generated with nanoscale porosity is significantly lower than the ideal filtering effect; nevertheless, the enhancement in the Seebeck coefficient that can be obtained is large enough to offset the reduction in electrical conductivity caused by porosity. This study proves that careful engineering of the energy-dependent electron scattering rate can provide a route towards relaxing long-standing constraints in the design of thermoelectric materials.

Wireless Sensor Networks (WSNs) devices are used in defense, healthcare, residential, medical, wearable, and industrial applications. WSNs currently use bulky primary batteries or rechargeable batteries, which must be replaced/charged after a certain time resulting in significant costs and down time. Therefore, we propose to use a lightweight, scalable, and flexible thermoelectirc generator (TEGs) that may potentially be used for charging the WSN batteries and served as long-lasting and self-sufficient wireless power supplies. These self-sufficient and long-lasting batteries can facilitate a widespread adoption of WSNs by assuring their safe, reliable, and maintenance-free operation. To address the global demand for low-cost, flexible thermoelectric generators, this work describes a novel, energy-efficient method of controlling the composite microstructure and resulting thermoelectric (TE) properties of p-type composite films. Combining (1) a small amount of naturally occurring binder that is sufficient to hold TE particles together without significantly decreasing electrical conductivity and (2) a wide distribution of heterogeneous (micro and nano) particles tightly packed on (3) application of mechanical pressure at (4) low-temperature curing for a short duration, this method yields big grains larger than the mean free path for charge carriers (for high electrical conductivity) and nano features smaller than the mean free path for phonons (for low thermal conductivity). The best properties were achieved with0.05 wt% of chitosan binder, heterogeneous-sized TE particles (100-mesh Bi_0.5Sb_1.5Te₃),200 MPa applied uniaxial pressure, and curing at 150 °C for 30 minutes to densify the as-deposited films. The 100-mesh chitosan-BST composite film achieved a ZT of 0.7, comparable to the best reported ZT of printed p-type TE films, but without using high-temperature and long-duration curing. The power output of our best-reported scalable 3-leg-BST TEG was 58 μW. The corresponding power density was 5.72 mW/cm² at a temperature difference of 38 K, significantly higher than the best reported values for single-leg TEGs. Furthermore, the flexibility of this prototype was performed by successfully bending at a 3-cm radius for 1000 cycles, thus demonstrating a high potential to be used as self-sufficient batteries for next-generation wearable devices and wireless sensor network applications.

The process of conversion of energy from primary carriers to different forms of end use always results in losses. The major form of energy loss is ‘heat’ and it has been found that almost 70 % of primary energy amounting to ~ 340 PJ is wasted in the form of heat. Among the several methods of recovering and converting this waste heat, thermoelectric conversion has been prominent. The recent development of several new materials with high figure-of-merit requires that their potential from different perspectives needs to be evaluated. The objective of this work therefore has been to estimate the materials requirement, their environmental impact and also assess the materials supply chain risk. Since the waste heat ranges in intensity from < 100 C to > 300 C, the thermoelectric materials required will be different for different ranges. Hence the waste heat has been classified into 3 separate categories – low < 373 K, medium 373 K – 573 K and high > 573 K. The candidate materials which exhibit the highest figure-of-merit in these temperature ranges have been considered for evaluation. It is found that elements such as Bi, Te, Sn, Co will be required in kilo tons capacity which pose supply chain risk. Also, processing the ores of these elements can have significant environmental impact in terms of embodied energy and CO₂ emissions. All these aspects will be discussed in detail in this paper.

Thermoelectric materials directly convert thermal energy to electrical energy and vice versa. To improve the energy conversion efficiency, accurate characterization of the atomic structure is critical, as this defines the charge- and phonon-transport. Scanning transmission electron microscopy (STEM) is widely used for defect characterization in thermoelectric materials, as structural defects and lattice strain often play a critical role in phonon scattering. However, the recently emerging high entropy thermoelectric materials are challenging objects for this characterization due to their very small strain and atomic-scale elemental mixture.
Here we show that these challenges can be solved by careful sample preparation, microscope fine-tuning, and the application of new experimental methods. Three STEM-based techniques are applied: atomic resolution energy-dispersive X-ray spectroscopy (EDS), geometric phase analysis (GPA), and nanobeam diffraction analysis (NBED) to understand the elemental positions and subtle lattice distortion, which are important for understanding the structure-property relationship.
Our high entropy material has a composition of Pb_0.89Sb_0.012Sn_0.1Se_0.5Te_0.25S_0.25 and was compared to the low entropy PbSe control material. The atomic positions of Pb, Sn, Te, Se were found by atomic resolution EDS mapping. Pb and Sn occupy the ‘Na’ position, while Te and Se occupy the ‘Cl’ position of the NaCl crystal structure, demonstrating that the elements were mixed homogenously at the atomic level.[1]
GPA can quantify the local strain distribution. However, weak strain fields require careful attention since the inevitable sample drift and scan noise will induce artifacts. To optimize this, we first wait a long time to minimize the sample drift, then we adopt fast scanning and take a stack of 30 frames, and finally, we use the ‘Smart Align’ software plugin to align the stacks together. This can simultaneously reduce the artifacts from sample drift and from scan noise, to give a strain analysis with 0.3% accuracy. With this, we find that the strain distribution in our high entropy alloy is much wider than that of low entropy alloy. The histogram shows that the statistical distribution in the high entropy alloy is much wider along with the normal <002> and shear <220> directions than in the low entropy alloy. This is important information, as the normal and shear strains are directly linked to the lattice thermal conductivity.
As GPA requires atomic resolution images, the observational area cannot be very large and NBED is used to overcome this shortage. We use a 4 nm parallel probe to scan a much larger area of 400 nm by 400 nm to obtain a series of electron diffraction patterns. With careful analysis, we are able to map the strain with an accuracy of 0.06%. The results of NBED indeed confirm that the strain distribution of the high entropy alloy is much wider than the low entropy alloy. However, the absolute values in the strain field by NBED are lower than what was obtained with GPA. This shows the strength of GPA at a very small scale: in this high entropy alloy, the local strain varies in areas smaller than 2 nm and the strain measured with NBED only gives an average of the probed 4 nm area. In the case of our high entropy alloy, a full picture of the strain fields will therefore only be obtained by a combination of the GPA and NBED results.

The authors acknowledge support from the Singapore Ministry of Education (MOE), AcRF Tier 1 (R-284-000-179-133) grant.

Reference:
[1] B. Jiang#, Y. Yu#, J. Cui#, X. Liu, L. Xie, J. Liao, Q. Zhang, Y. Huang, S. Ning, B. Jia, B. Zhu, S. Bai, L. Chen, S. J. Pennycook, J. He, High-entropy-stabilized chalcogenides with high thermoelectric performance. Science . 371, 830–834 (2021).

In recent years has been focused on GaAs based optical devices for IR detector, OPtical-fiber, ans solar cell at longer wavelengths (1.3 um - 1.5 um). Other groups have tried for reducing the strain In_xGa_1-xAs quantum well (QW) layer by embedding InAs quantum dots (QDs) to emissions at wavelengths above 1.3 um. Here, we inestigate the dot-in-a-well (DWELL) solar cell (SC) with InAs QDs and In_0.15Ga_0.85As QW through photoluminescence (PL) and photoreflectance (PR). QDs are grown through Stranski-Krastanov (S-K) and sub-monolayer (SML) methods. The results of PL spectra show that SML QDs (1.25 - 1.3 eV) PL emission energy is 0.2 higher than S-K QDs (1.03 - 1.1 eV) The difference in the emission energy is due to the quantum dot size effect. The activation energy of each sample is as follows: 90 meV for S-K DWELL SC and 29 meV for SML DWELL SC. The internal electric fields are investigated through PR spectroscopy. PR spectra show Franz-Keldysh oscillation (FKOs) due to the internal electric field. Regarding the relationship between the temperature and the internal electric fields of all sample. The internal electric fields increase on the temperature increases. Interestingly, the internal electric fields of the SML DWELL SC increases until 240 K; but it decreases above 240 K. The results are possibly due to the escape carriers from thermal energy contribute to the photovoltaic effect.

GaAs is the material of choice for the highest efficiency in solar energy conversion in a single junction device. The path to increase the efficiency is reached by integrating several III-V compounds in a multiple junction solar cell. Group III elements are scarce on the earth crust and expensive to extract. This renders III-V solar cells challenging to scale to extremely large production rates. Several paths have been explored to circumvent this. One of them consists in producing the solar cells in nanowire form on a silicon substrate. It has been shown that this results in about 1000 times in materials savings, with prospects to increase the efficiency even further than in thin film devices [1,2].
After reviewing the advantages of the III-V nanowire configuration, we will move to the path towards alternative compound semiconductors. We will show how Zn3P2, with a bandgap of 1.5eV, can be obtained both in nanowire and thin film in a defect-free manner. We will show how the substrate preparation prior to growth is essential to obtain high quality films [3-5]. Finally we will demonstrate how this material exhibits an absorption coefficient about 5 times higher than GaAs and high mobility at room temperature, rendering it an excellent alternative as a solar cell absorber [6]. We will also provide insights on device configurations that one could use with this material and provide perspectives for similar compounds made of earth abundant elements.
[1] P. Krogstrup et al, Nature Photon 7, 306 (2013)
[2] Y. Cui et al Nano Lett. 16, 6467 (2016)
[3] S. Escobar Steinvall et al. Nanoscale Horizons, 5, 274-282 (2020)
[4] M. Zamani et al J. Phys. Energy 3, 034011 (2021)
[5] S. Escobar Steinvall et al. Nanoscale Advances 3, 326 (2021)
[6] M.Y. Swinkels et al Phys. Rev. Appl. 14, 024045 (2020)

One of the key challenges in III-V growth for energy applications remains the high cost of manufacturing. This is driven by a variety of factors, but one of the most significant costs remains the epitaxial substrate. While polycrystalline materials grown through traditional vapor phase techniques offer the opportunity to eliminate the high-cost epitaxial substrates, the resulting materials quality is significantly lower than that obtained through III-V homoepitaxy. Thus, a major focus for the development of III-V energy harvesting devices has been (i) the use of epitaxial transfer approaches--to amortize the cost of the substrate over many devices--or (ii) growth of reduced dimensional structures on heteroepitaxial or non-epitaxial substrates--to reduce the defectivity of the resulting III-V materials. Here we will talk about a third approach enabling growth of high quality thin-film III-V's on non epitaxial substrates, Templated Liquid Phase (TLP) Growth.
We will first discuss the basics of this growth approach, which involves deposition of the group III element (e.g. indium) and a capping layer (e.g. SiO₂) on a non-epitaxial substrate. This is then heated to the desired growth temperature under vacuum, and then a group V precursor such as AsH₃, PH₃, or NH₃, may be flowed and reacted with the group III element. Uniquely, at the growth temperatures used, the deposited group III element is in the liquid phase, but through control over the surface energies as well as structure, it is possible for the liquid to maintain the geometry in which it was deposited. Eventually, this group III liquid becomes supersaturated and a III-V nucleus forms. By controlling the nucleation rate and density, it is possible to define large single crystal domains on the substrate, effectively minimizing the effect of grain boundaries, yielding significantly better material properties than vapor phase approaches. Critically, this enables the use of lower cost substrates such as metal foils, silicon wafers, and polymers.
Next, we will discuss the quality achievable with this approach for III-V materials on different substrates, and highlight the causes of varying quality on disparate substrates, despite the large single crystal domains being similar.in size. Using a combination of structural and electronic approaches, such as transmission electron microscopy, electron backscatter diffraction, photoluminescence, and Hall measurements, we will show that the current quality achieved by TLP approaches the quality exhibited by single crystalline counterparts, but without the epitaxial substrate. As a follow-up, we will show that the TLP grown III-V's can act as a 'remote substrate' for follow up metal organic chemical vapor deposition (MOCVD) growth of lattice matched hetrostructures, quantum wells, etc. Finally, we will discuss a variety of energy devices realized using the TLP grown materials.

The direct growth of III-V semiconductors on Si substrates provides an avenue for utilizing the tunable, direct bandgaps of III-Vs as well as existing economies of scale for Si. Direct, heteroepitaxial growth is more scalable than bonding. Our development of templated liquid-phase (TLP) heteroepitaxial growth of III-Vs on Si offers additional cost savings over metallorganic vapor-phase epitaxy (MOVPE) and, unlike vapor-liquid-solid (VLS) techniques, templated structures of III-V can be easily grown in orientations other than (111).

The TLP process uses evaporation—on a photolithographically-patterned surface—to supply a layer of group-III metal, then the sample is heated above the metal melting point and exposed to a group-V vapor-phase precursor, forming a liquid solution. Once saturated, solid III-V grows out of the solution. In this way, TLP growth differs from VLS nanowire growth, in which both group-V and group-III elements are supplied by gas phase and continually crystallized in solid wires. In TLP growth, a thin layer of solution is maintained during growth using dielectric capping layers. Without the dielectric caps, wire-like vertical growths or dewetting of the liquid metal can occur.

Our previous work has shown that TLP can result in relaxed, oriented growth of InP with large grain sizes in pre-defined, μm-scale structures [1]. Here, we investigate the presence of defects, which are prevalent due to the large (nearly 8%) lattice mismatch between InP and Si. We utilize a lateral overgrowth scheme in which the epi-layer takes the orientation of the substrate from a seed area where they are in contact, then growth propagates with minimal defects over a non-crystalline layer.

In our TLP growth process, substrates are patterned with photolithography to define template shapes. Indium metal is typically deposited via e-beam evaporation, though it can also be electroplated, and then followed by a thin capping layer of SiO_x. We find that spin-coating with solgel SiO_x is ideal for conformally coating and confining patterned indium. To convert to InP, the Si/In/SiO_x stacks are heated to 750 °C in an atmosphere of PH₃ diluted in H₂. Short anneal times are useful to identify nucleation sites and density, while longer anneals are used to show growth evolution.

We find that each template typically converts from a single nucleus of InP. X-ray diffraction shows that most InP is oriented to the (001) substrate, but some (111)-oriented InP is present. Electron channeling contrast imaging (ECCI) identifies crystalline defects and confirms that converted areas are typically single grains. Planar defects such as twins, which are very common in the epitaxy of III-Vs on Si, are observed spanning entire grains. Transmission electron microscopy (TEM) of the Si-InP interface shows that planar defects often propagate from roughened areas of the Si surface.

The growth and nucleation mechanisms in TLP growth differ from those of vapor-phase epitaxy techniques, but its use of vapor-phase flux still offers precise control of the process. With this improved understanding of the defect formation processes, TLP growth now offers an avenue toward lower-cost and faster III-V growth, as material quality is improved to compete with established vapor-phase growth methods.


References
[1] Schneble et al. JVST A. 2021

We demonstrate AlInP-passivated GaAs solar cells grown by dynamic-hydride vapor phase epitaxy (D-HVPE) with AM1.5G efficiencies of ~26%. D-HVPE is an epitaxial growth technique with potential for high-throughput, low-cost depositions of a variety of III-V optoelectronic devices¹. The incorporation of Al in III-V alloys enables high-quality, transparent heterobarriers for use in photovoltaics, transistors, LEDs, and other optoelectronics. However, difficulties with the growth of Al-containing materials in the HVPE environment, namely the incompatibility of AlCl with quartz reactor hardware and gettering of impurities like O and Si within Al-containing films, limit the full potential of D-HVPE-grown devices. We addressed this by developing an Al-source that generates AlCl₃ in-situ. AlCl₃ resists degradation at temperatures used in typical HVPE processes and results in Al-containing material deposition. High levels of uncracked group V hydrides enable the growth of AlInP and AlGaInP materials, despite the fact that thermodynamic predictions deemed it impossible to grow such materials². This work enables AlGaAs-passivated³ and AlGaInP-passivated⁴ solar cells. However, both of these materials absorb short wavelength photons more strongly than AlInP and thus they hinder the ultimate efficiency of a solar cell.
Here we focus on the growth and integration of AlInP as window layers for solar cells. Growth experiments varying input gases (HCl and H₂) to the Al-source, group V hydride and associated carrier flow rates, and substrate temperature indicate that there is a wide growth window for AlInP that results in smooth morphologies and compositional control. These studies offer insight into the AlCl₃ conversion efficiency in the Al source, Al incorporation efficiency, effect of free HCl on film growth, and potential control of impurities. Next, we demonstrate AlInP-passivated GaAs rear heterojunction solar cells with a > 1 mA/cm² increase in short circuit current density relative to GaInP-passivated cells. All of the improvement in photocurrent occurs at photon wavelengths < 660 nm due to the improved optical transparency of AlInP. These AlInP-passivated GaAs solar cells reach an efficiency of 25.7%, greater than the 25.5% efficiency of our best GaInP-passivated cell, at parity fill factor and slightly reduced open circuit voltage. We attribute the difference in open circuit voltage to a difference in architecture near the junction and not to any degradation stemming from the AlInP window. Implementation of this improved junction architecture with an AlInP window has the potential to yield a > 26.5% efficient GaAs solar cell. This result demonstrates that AlInP can be successfully integrated in HVPE-grown optoelectronic device structures, giving HVPE-grown solar cells the potential to reach efficiencies of state-of-the-art solar cells and making the full palette of III-V materials available for HVPE growth of any optoelectronic device structure.
¹ J. Simon, K. Schulte, K. Horowitz, T. Remo, D. Young, and A. Ptak, Crystals 9, 3 (2018).
² K.L. Schulte, W. Metaferia, J. Simon, D. Guiling, K. Udwary, G. Dodson, J.H. Leach, and A.J. Ptak, ACS Appl. Energy Mater. 2, 8405 (2019).
³ W. Metaferia, K.L. Schulte, J. Simon, D. Guiling, and A.J. Ptak, in 2020 47th IEEE Photovoltaic Specialists Conference (PVSC) (IEEE, Calgary, AB, Canada, 2020), pp. 0672–0674.
⁴ Y. Shoji, R. Oshima, K. Makita, A. Ubukata, and T. Sugaya, Prog Photovolt Res Appl pip.3454 (2021).

Direct growth of single crystal III-V thin film on top of metal has been desired to design a dynamic tuning photonics device such as metal-insulator-metal (MIM) resonator, where the top metal layer is defined into a sub-wavelength resonator element and bottom metal layer as a reflector. In previous research, heavily doped layer was used to mimic the metal layer by using either metal organic chemical vapor deposition (MOCVD) or molecular beam epitaxy (MBE), which is not only cost inefficient but having poor reflectivity. Here in this work, we have demonstrated InAs directly grown on metal, specifically on top of Mo and W to build a platform for the device application. The growth was carried at growth temperature of 300°C by low temperature templated liquid phase (LT-TLP) method. Unlike our previous grown templates on top of amorphous substrates, it is noteworthy that the size of the grown InAs area is over > 80um domain size with various InAs thickness (from ~100 nm up to μm scale) by tuning the initial indium template thickness. This approach enables us to control the size of the grown crystalline, potentially can be used to grow single crystalline size up to mmscale by optimizing the growth conditions and processes, opening the new door to breakthrough the large area III-V thin film growth for device application.

For III-V compound semiconductor solar cells to achieve scalability for very large-scale space and terrestrial photovoltaics, we need to radically reimagine how they are fabricated. We outline an approach for scalable synthesis of GaAs and InP thin film solar cells that bypasses epitaxial growth and eliminates or minimizes vacuum processing. Starting with thin film absorbers that are formed by mechanical exfoliation from GaAs and InP ingots bulk ingots, we demonstrate that high efficiency cells are possible using methods adopted from silicon and perovskite thin film photovoltaics, such as use of diffused junctions formed at ambient pressures, and solution processing of ohmic contacts and passivation layers using earth-abundant materials. We demonstrate diffused junction GaAs solar cells with 1-Sun AM1.5G efficiencies as high as 21.5%. We also performed a technoeconomic analysis indicating a pathway for III-V photovoltaics to achieve LCOE parity with future silicon photovoltaics, with a lower capital cost/Watt for manufacturing.

This article outlines a ~10 year (~2010 – 2020) effort within the Air Force Research Laboratory in the development and demonstration of advanced space solar cells in a physically flexible format with record high power output (~35% Air Mass Zero (AMO) sunlight conversion efficiency) and radiation tolerance (~28% after simulated 15 years in GEO). Essentially all spacecraft are powered by solar arrays, and the output of the arrays often dictate the ultimate capabilities of the payload.
At the initiation of this ATD, state-of-practice for space solar cells was ~30% efficient, three junction devices. This ATD produced the capability of large-area (>20 cm²), space-qualified, 32% inverted metamorphic multi-junction (IMM) cells with efficiencies of ~35%. Metamorphic refers to a crystalline growth of semiconductor layers that are slightly mismatched in terms of their atomic spacing. Inverted refer to the cell being grown up-side-down interference to the sunny side being deposited first. The IMM structure is the key for this ATD achieving higher efficiency and improved radiation resistance.

III–V/Si epitaxial tandems with a 1.7-eV GaAsP top cell promise stable power conversion efficiencies above the fundamental limit of Si single-junction cells. However, such cells historically suffered from limited minority carrier diffusion length in the top cell, leading to reduced short-circuit current densities (J_SC) and efficiencies. While conventional wisdom dictates that reduction of threading dislocation density in III–V/Si tandems is the key to higher efficiency, we have found that heterointerface design and growth sequence also play critical roles in reducing recombination losses. Our improved GaAsP cells make use of a wide-band gap AlGaAsP electron-blocking layer that forms a pristine interface with GaAsP, resulting in a 10%–20% (absolute) boost in quantum efficiency over earlier work in the critical red wavelength range (600–725 nm), despite similar threading dislocation density. Combining the improved top cell carrier collection with Si backside texturing, we have obtained 25.0% efficient GaAsP/Si tandem cells with a closely matched J_SC of 18.8 mA/cm². In this talk, I will discuss a roadmap to efficiencies of 27-28% and beyond for epitaxial GaAsP/Si tandem solar cells, including recent results describing the unexpectedly strong effects of doping on dislocations in GaP. I will also attempt to draw connections to recent developments in III-V on Si integrated photonics that may help to lower the barrier to entry for III-V on Si tandem solar cells.

The absorption edge of a GaAs solar cell can be tuned to longer wavelengths with strain-balanced quantum wells, as an alternative to metamorphic epitaxy. For one-sun photovoltaic applications, an optimum bandgap for both single and multijunction solar cells is near the broad water absorption peak at 930 nm and can be accessed by a Ga_0.9In_0.1As alloy. This alloy is strained with respect to the GaAs substrate, and so only very thin ~10 nm layers can be grown without relaxation via dislocation formation. However, solar cells require significant thickness for complete absorption of the incoming solar spectrum, requiring many such layers. We have designed quantum well superlattices with up to 300 sets of carefully tuned strain-balanced GaInAs/GaAsP pairs by limiting composition modulation and material degradation. We have incorporated quantum well solar cells into multijunction devices to achieve record 32.9% two-junction and 39.5% three-junction solar cells under standard one-sun illumination. We will describe the physics of the quantum well devices and the growth conditions that lead to sharp interfaces and good material quality, as well as the characterization of the efficiencies.

Group III-V ternary alloys indium gallium arsenide (InGaAs) and indium aluminum arsenide (InAlAs) have wide range of applications including quantum cascade lasers, photodetectors, high electron mobility transistors, and solar cells. Multi-junction solar cells based on epitaxially grown III-V compound semiconductors already exhibit very high efficiencies and continuing research are ongoing to increase the efficiency further. The performance of these epitaxial thin film solar cells depends on the optoelectronic properties of each component layer. The optical properties in the form of the complex dielectric function spectra of the metal-organic vapor phase epitaxy (MOVPE) InGaAs and InAlAs lattice matched to InP in the multilayer stacks have been measured by spectroscopic ellipsometry in the terahertz (THz), infrared (IR), and near infrared to visible (NIR-Vis) range and analyzed using physically realistic parametric models. The direct band gap of InGaAs and InAlAs are determined to be 0.730 and 1.44 eV respectively. Above gap critical point are observed at 1.21, 2.57, and 2.80 eV for InGaAs and at 1.74 and 2.95 eV InAlAs. Phonon modes are identified at 252 cm^-1 for InGaAs and 351 cm^-1 for InAlAs. Free carrier absorption in the IR and THz range is described using the Drude model from which resistivity, scattering time, carrier mobility, and carrier concentration are found to be 0.0038 Ωcm, 225 fs, 9444 cm²/Vs, and 1.7 × 10¹⁷ cm^-3 for InGaAs and 0.04 Ωcm, 244 fs, 5713 cm²/Vs, and 2.5 × 10¹⁶ cm^-3 for InAlAs, both of which are measured within multilayer stacks. Electrical Hall effect measurement of InGaAs yields similar values of mobility at 8088 cm²/Vs and carrier concentration at 1.4 × 10¹⁷ cm^-3. The measurement and data analysis approach provides a non-contacting method of determining electronic transport properties for epitaxial III-V films within devices.

The integration of diamond on gallium nitride (GaN) has significant potential for improving the thermal management of high-power high electron mobility transistors (HEMTs). However, there are several challenges associated with the whole process of direct diamond growth onto GaN through microwave plasma chemical vapour (MP-CVD). A prevalent issue is that GaN does not adhere very well to diamond and therefore requires an interlayer that can form a stable carbide such as an aluminium or silicon containing layer (eg. AlN, AlGaN, SiN or SiC)¹. Ultimately, one of the most significant challenges is that the diamond needs to replace the substrate (for example GaN-on-Si becomes GaN-on-Diamond), necessitating a lengthy wafer boding process to flip and expose the nitrogen polar side of a GaN/AlGaN stack for CVD diamond growth. This temporary carrier bonding and flipping process limits sample fabrication for principal studies to simply examine a CVD grown diamond on GaN/AlGaN interface.
We have presented several published works of a method that requires no wafer bonding to produce CVD diamond on GaN/AlGaN using membranes^2–4. In essence, GaN is grown on a Si substrate using metal organic chemical vapour deposition (MOCVD) and the Si substrate is selectively etched from the underside using a combination of photolithography and an inductively coupled plasma (ICP)⁴. This exposes very thin GaN/AlGaN membranes (<10 μm), suspended by a much thicker Si border which can be used to handle the sample. These fragile membranes can then be subjected to MPCVD for heteroepitaxial growth of thick polycrystalline diamond films (>20 μm). Using this approach, we demonstrate successfully adhered CVD diamond to the underside of the GaN/AlGaN.
In this work, we will present the fabrication process of CVD diamond on GaN membranes and the significant findings associated with this method⁴, not least the demonstration that these fragile structures survive the high induced thermal stresses during the CVD process² and the significant improvement in the measured thermal boundary resistance between a Diamond/AlGaN interface and a Diamond/SiC/AlGaN interface³. Additionally, we will present the challenges and limitations of this methodology, including membrane bow from thermal stress, the etching artefacts from using a Si border and the limitations on membrane size and shape.
References
¹ S. Mandal, E.L.H. Thomas, C. Middleton, L. Gines, J.T. Griffiths, M.J. Kappers, R.A. Oliver, D.J. Wallis, L.E. Goff, S.A. Lynch, M. Kuball, and O.A. Williams, ACS Omega 2, 7275 (2017).
² J.A. Cuenca, M.D. Smith, D.E. Field, F. C-P. Massabuau, S. Mandal, J. Pomeroy, D.J. Wallis, R.A. Oliver, I. Thayne, M. Kuball, and O.A. Williams, Carbon 174, 647 (2021).
³ D.E. Field, J.A. Cuenca, M.D. Smith, S.M. Fairclough, F.C.-P. Massabuau, J.W. Pomeroy, O.A. Williams, R.A. Oliver, I. Thayne, and M. Kuball, ACS Appl. Mater. Interfaces 1 (2020).
⁴ M.D. Smith, J.A. Cuenca, D.E. Field, Y. Fu, C. Yuan, F. Massabuau, S. Mandal, J.W. Pomeroy, R.A. Oliver, M.J. Uren, K. Elgaid, O.A. Williams, I. Thayne, and M. Kuball, AIP Adv. 10, 035306 (2020).

Efficient heat extraction in high power and high frequency devices is key towards full utilisation of materials like GaN, AlN and Ga₂O₃. The present technology of using SiC has allowed good device performances, but there is still room for considerable improvements. Replacing SiC(k_SiC ~ 360 – 490 W/m K) with diamond(k_Dia ~ 2100 W/m K) is one of the potential options. However, for any potential benefits to be had, the diamond layer should have thickness more than 50 microns. Here we will discuss the growth of diamond layers on wide band semiconductors. We have demonstrated the growth of diamond layer on GaN surfaces¹. While it is difficult to grow thick adherent layer on the GaN surfaces due to absence of any covalent bond between GaN and diamond, the problem can be overcome by growing the diamond layer on AlN surface². AlN acts as seed layer for GaN growth on silicon and a wide band gap semiconductor in itself as well. Finally, we have also demonstrated the growth of diamond on Ga₂O₃ single crystals³. In the case of Ga₂O₃, direct growth of diamond is not possible, and a thin interlayer is essential for the growth.
Advances in growth technologies makes it possible to grow diamond on variety of substrates, however, to grow diamond on non-diamond substrate, a seeding or nucleation step is essential⁴. In the results presented here we have used a seeding technique called electrostatic seeding. To accomplish this, it is essential to know the surface charge of the substrates. We have determined the zeta potentials (which is an indicator of the surface charge) of the wide band gap materials used in these works, which determined the type of diamond seeds (positive or negative) needed for efficient seeding. AFM was used to determine the seed density on the surface of the substrates. The seeded wafers were introduced in microwave plasma chemical vapour deposition system for growth of diamond. For thick diamond layers on AlN we have determined the thermal boundary resistance between diamond and AlN². The quality of diamond layers was determined by Raman spectroscopy, SEM etc.
References
¹ S. Mandal, E.L.H. Thomas, C. Middleton, et al., ACS Omega 2, 7275 (2017).
² S. Mandal, C. Yuan, F. Massabuau, et al., ACS Appl. Mater. Interfaces 11, 40826 (2019).
³ S. Mandal, K. Arts, H.C.M. Knoops, et al., Carbon N. Y. 181, 79 (2021).
⁴ S. Mandal, RSC Adv. 11, 10159 (2021).

Binary group-III-nitrides and their alloys have revolutionized optoelectronics, power electronics, and high frequency electronics over the previous two decades. This success is due to a wide band gap (which through alloying is theoretically tunable from IR, across the visible spectrum, and well into the UV – from 0.7 eV to 6.2 eV), high electron mobility, spontaneous polarization fields, bipolar dopability, and tolerance to high levels of extended defects such as threading dislocations. However, for optoelectronic applications in the spectral region beyond blue (green and beyond, ≥ 500 nm) such as green and amber light-emitting diodes (LEDs), the lattice mismatch between the GaN barriers and the InGaN emitters leads to large strain and therefore piezoelectric polarization fields in the quantum well (QW) active layers. These lead to a strong quantum-confined Stark effect significantly reducing electron-hole wavefunction overlap and therefore internal quantum efficiency. Significant research effort has surrounded improving the efficiency of these high In-composition alloys, including the use of semi-polar and non-polar substrates, although lattice mismatch and substrate cost remain an issue.
Alternatively, there has been increased research activity surrounding a family of wurtzite derived ternary nitrides, the II-IV-N₂ materials (II = Zn, Mg and IV = Si, Ge, Sn). This family of materials exists in a lattice-constant and band gap space which overlaps the III-N materials, potentially expanding the design space for nitride optoelectronics. This work focuses on ZnGeN₂, which is structurally and electronically similar to GaN, possessing a lattice mismatch of ~0.8% with respect to GaN, a theoretical band gap of 3.4 eV – 3.6 eV, and theorized use in longer-wavelength optoelectronic devices including green and amber LEDs. ZnGeN₂ epitaxial synthesis is still a new field, with only a few literature reports of ZnGeN₂ grown on GaN by MOCVD and MBE. ZnGeN₂-GaN alloys have also been grown epitaxially as a 50% alloy by MOCVD and are currently explored by bulk synthesis at low GaN compositions.
We will present the epitaxial growth of high-quality GaN/ZnGeN₂ and GaN/(ZnGe)_1-xGa_2xN₂ heterostructures, including single and multiple QW structures, using nitrogen plasma-assisted molecular beam epitaxy (MBE). Detailed structural and elemental analysis will be given, including X-ray diffraction (XRD), scanning transmission electron microscopy (STEM), energy dispersive X-ray spectroscopy (STEM-EDS), and atom probe tomography (APT). These methods demonstrate high-quality and abrupt interfaces in the heterostructures, even after multiple repeating heterointerfaces. APT and STEM-EDS confirm the presence of bulk Ga incorporation in the ZnGeN₂ QWs, thus demonstrating it is likely a GaN-ZnGeN₂ alloy. APT and STEM-EDS also confirm the presence of a confined 2-dimensional layer of Ge-rich material within the GaN barrier. Through changes in growth methodology, we have demonstrated methods to remove these unintentional impurities including associated improvements in structural quality.
Finally, the optical properties of these single and multiple QW structures will be explored in detail, including defect-assisted and band-to-band recombination mechanisms and signatures, through temperature-dependent and power-dependent photoluminescence (PL). Importantly, we find optical signatures of Zn impurities in GaN which are not observed in secondary ion mass spectrometry (SIMS), indicating unintentional Zn-doping below the SIMS detection limit of approximately 10¹⁶ cm^-2. We also observe optical signatures of degenerate n-doped GaN in samples which contain Ge-rich GaN barriers. Together, this data demonstrates both the promise of heteroepitaxially integrated hybrid ternary/binary nitride systems along with the challenges associated with growing such systems, including an outlook on methods to improve the materials and eventual devices.

In past decades traditional III-V materials and devices spurred major advances in optoelectronic technologies, including light-emitting diodes (LEDs), high-efficiency photovoltaics (PV), improved lasers, sensors and displays. However, despite these advances, lattice mismatch between GaN and InGaN in conventional III-V (In_xGa_1-xN) LEDs limits emission efficiency for wavelengths greater than ~500 nm. The strain caused by lattice mismatch leads to piezoelectric polarization creating reduced electron/hole wavefunction overlap and efficiency due to the quantum confined stark effect. The wurtzite II-IV-N₂ material system is currently being investigated as alternative for applications in optoelectronics, especially LEDs. The ternary structure allows for independent tuning of the band gap and lattice constants by cation disorder and alloying, allowing for wide material applications. In this work we explore an alternative material for green LEDs, ZnGeN₂.
ZnGeN₂ is proposed as a green-to-amber emitter, which, due to its similar structural properties to GaN, can be easily integrated into existing III-V heterostructures for faster industry deployment. The ZnGeN₂/GaN material system has smaller lattice mismatch and polarization than the traditional III-V system, allowing for increased wavefunction overlap internal quantum overlap. The theoretically predicted type II band alignment between GaN and ZnGeN₂ would improve hole confinement. ZnGeN₂ has been grown by sputtering, HVPE, and MOCVD. Our group has demonstrated high quality MBE ZnGeN₂ growth on GaN and AlN. This talk will focus on ZnGeN₂/GaN heterostructure device modelling and the experimental results that helped elucidate important input parameters for the simulations.
The modelling program, SiLENSe, is a 1D drift-diffusion model simulation tool used to determine the wavefunction and spectra of an LED from various inputs, including polarization, bias, and device design. We incorporated preliminary band offset measurements of ZnGeN₂ on GaN found in our experimental work, and these preliminary results will also be presented. These band offsets were found by an X-ray photoelectron spectroscopy (XPS) study of ZnGeN₂ band edge positions. We found that the valence band edge of ZnGeN₂ is at least 0.65 eV higher in energy than the valence band edge of GaN, enabling the expected type-II heterojunction. We are performing more experiments to improve the accuracy of these measurements by accounting for band bending and polarization. These band offsets are incorporated in device modelling that helps inform device architecture. Also included as input parameters for the modelling experiments were wurtzite lattice constants found during growth on AlN. GaN is the optimal substrate for ZnGeN₂ in LEDs, but due to their structural and optoelectronic similarity, growths on AlN are required to investigate structural, morphological, and optical properties. Growth on AlN also allowed for high quality lattice constant measurements of relaxed ZnGeN₂, which is useful for more precise device modelling. We will present how varying heterolayers thicknesses and positions affects wavefunction overlap and emission spectra region. The characterization and modelling presented here represent an important advancement for theorized GaN/ZnGeN₂ heterostructures in future nitride LEDs.

Highly durable piezoelectric nanogenerators (PENGs) with excellent output performance are receiving great attention. GaN nanowires (NWs) are promising candidates due to their flexibility, stability in bending, and non-toxicity. Current technology can only grow GaN NW on rigid substrates, and the NWs have been transferred to an external elastic substrate such as PET or PDMS to take full advantage of their flexibility. However, the transferred NW PENG exhibited poor electrical conductivity to the substrate and its mechanical stability was reduced due to the bonding with epoxy.
In this presentation, a foldable, scalable, durable, cost-effective, and high-performance PENG developed by the direct integration of GaN NWs on a Cu-foil is reported. Single-crystal GaN NW ensemble was achieved by metal-organic chemical vapor deposition with a graphene buffer layer where the direct growth of GaN on the metallic substrate played an important role in improving the stability of PENG. The PENG provides a durable and highly sensitive output compared to the previously reported GaN NW-based PENGs fabricated by transferring the NWs onto a foreign substrate. The reported PENG can harvest energy from a variety of ambient actuation sources such as bending, vibrations, air-flow, finger pressing, foot striking, fluid-flow, and normal force by weights, with the maximum piezoelectric output voltage and current-density recorded as 19.7 V and 1.9 mA.cm^-2, respectively. Due to its high conversion efficiency, the PENG can power several LEDs and thus can be used to power electronic devices. More importantly, the PENG retains its performance after more than 4 million actuation cycles, demonstrating the potential of our design for practical applications using biomechanical and ambient actuation sources for self-powered systems.

(In)GaN-based nanowires and nanofins are promising candidates for solar-to-fuel technologies due to their high surface-to-volume ratio, excellent crystal quality and tunable band gap with favorable band edge positions for water splitting and CO₂ reduction reactions. Regarding this application, a thorough understanding of the interaction between the exposed crystal facets and the environment is crucial, as ambient molecules influence the electronic states at the surface and the charge carrier recombination. Investigations of the photoluminescence response of (In)GaN nanostructures to various atmospheres is a powerful tool to investigate their environmental interaction and stability.^[1,2]
In this study, we investigate the role of the main air constituents nitrogen, oxygen and water on the efficiency of radiative recombination in GaN nanostructures grown by molecular beam epitaxy as a function of different surface treatments and at temperatures up to 200°C. Oxygen and water exposures result in a complex behavior as they can both act quenching and enhancing on the photoluminescence intensity, dependent on the temperature. For oxygen, we observe these characteristics already for low concentrations of below 0.5% in dry nitrogen. While the photoluminescence intensity changes induced by oxygen occur independently of illumination, the influence of water is light-induced: it evolves within tens of seconds under ultraviolet light exposure and is heavily influenced by the nanostructure pre-treatment. Combined measurements of the electrical current through GaN nanofins and their photoluminescence intensity reveal the environmental influence on the interaction of non-radiative recombination processes and changes in the surface band bending of the nanostructures.^[3]
This knowledge about the impact of ambient molecules on the charge carrier dynamics and the electronic band bending is necessary for the development of efficient and simultaneous stable photocatalytic devices.

[1] M. Hetzl, et al., Journal of Applied Physics 124, 35704 (2018)
[2] K. Maier, et al., ACS Sensors 3, 2254 (2018)
[3] M. Kraut, et al., Nanotechnology 32, 495703 (2021)

Dilute anion alloys of III-nitrides are interesting as their band gap and band edges can be engineered suitably for photoelectrochemical water splitting [1]. In addition, they are direct band gap semiconductors with high absorption coefficients, good mobility, and charge-transport characteristics, i.e., lower Auger recombination [2]. For instance, it has been conclusively shown through theory and experiment that the band gap of GaN can be reduced from 3.45 eV, down to 1.7 eV upon 2-8 % of antimony incorporation (nitrogen substitution) [1]. The described anion alloying promotes valence band [3] as opposed to conduction band offsets, the latter being undesirable as they would typically compromise the straddling of the hydrogen reduction potential.
Significant challenges exist in the production of devices based on III-V semiconductors due to limited or non-availability of suitable techniques for growing best quality films with III-Nitrides alloyed with anions. Towards that objective, a new synthesis technique has been developed for the growth of GaSb_xN_1-x and GaBi_yN_1-y 0-D and 1-D structures, the Plasma-Assisted Vapor Liquid Phase Epitaxy (PA-VLPE) method [4] enables the control of anion charged species concentration, i. e., N+, Sb+, Bi+, that diffuse through a molten gallium layer and crystallize at the gallium/substrate interface. Single crystal films with antimony and bismuth incorporation of up to 0.9% and 1.1%, respectively, have been obtained without conducing to delocalized broadening of the valence band Density of States of these alloys. Similarly, copper catalyzed nanowire growth with anion incorporation levels comparable to the ones obtained through MOCVD have successfully produced visible light absorbing Ga (Sb or Bi)N alloys, via PA-VLPE.
GaSb_yP_1-y alloys have been deposited on silicon substrates through Halide Vapor Phase Epitaxy with growth rates of up to 500 um/h and antimony incorporation between 3.7 and 6.7%. From experimental results, the effect of antimony on the optoelectronic properties of these alloys suggests the band gap reduction versus pure gallium phosphide does not occur for the indirect transition at 2.26 eV but rather for the direct one at 2.68 eV. Alloys with 2.2-2.5 eV direct bandgaps outperform single crystal 2.26 eV indirect gallium phosphide in onset potential and fill factor when tested in 1 M H₂SO₄and posses lower charge transfer resistance at the alloy/electrolyte interface even without the use of catalysts [5].
Acknowledgements: Some of the work was supported earlier with financial support from US Department of Energy (DE-FG02-07ER46375) and NSF (DMS1125909).
References
[1] S. Sunkara et al., "New Visible Light Absorbing Materials for Solar Fuels, Ga(Sb-x)N1-x," Advanced Materials, vol. 26, no. 18, pp. 2878-2882, May 2014, doi: 10.1002/adma.201305083.
[2] C.-K. Tan, "Dilute Anion III-Nitride semiconductor materials and nanostructures," Doctor of Philosophy in Electrical Engineering, Electrical Engineering, Lehigh University, 2016.
[3] N. A. Andriotis, R. M. Sheetz, R. Ernst, and M. Madhu, "Band alignment and optical absorption in Ga(Sb)N alloys," Journal of Physics: Condensed Matter, vol. 26, no. 5, p. 055013, 2014.
[4] D. Jaramillo, J. Jasinski and M. Sunkara, “Liquid Phase Epitaxial Growth of Gallium Nitride”, Crystal Growth and Design, 19, 11, 6577-6585(2019).
[5] S. J. Calero-Barney, W. Paxton, P. Ortiz, and M. K. Sunkara, "Gallium antimonide phosphide growth using Halide Vapor Phase Epitaxy," (in English), Solar Energy Materials and Solar Cells, Article vol. 209, p. 9, Jun 2020, Art no. 110440, doi: 10.1016/j.solmat.2020.110440.

Photoelectrochemical (PEC) water splitting is a potentially promising route to direct solar-driven hydrogen production. III-V multijunction, or tandem, semiconductor absorbers have demonstrated the highest solar-to-hydrogen (STH) efficiencies to date. To achieve the U.S. DOE’s hydrogen production target of $2/kg, in addition to high STH efficiency, the semiconductor absorber synthesis cost must come down 1-2 orders of magnitude from the current metal organic vapor phase epitaxial route. Hydride vapor phase epitaxy (HVPE) is a technique that has the potential to lower tandem III-V synthesis costs. Here we report the PEC performance of HVPE-grown GaInP/GaAs tandem photoelectrodes with 1.85/1.38 eV bandgaps. We sputtered a PtRu nanoparticulate hydrogen evolution cocatalyst on the III-V photocathode surface and used an IrOx anode for water oxidation. Upon illumination, the HVPE-grown structures were able to drive water electrolysis spontaneously (e.g., without an external bias) at STH efficiencies up to 10%. The durability of III-V photoelectrodes remains and unresolved challenge for cost effective hydrogen production via photoelectrolysis.

Photoelectrochemical (PEC) water-splitting couples light-absorbing semiconductor materials with water-splitting electrocatalysts to directly produce hydrogen from solar energy. Systems with tandem III-V semiconductor absorbers (e.g. GaInP₂/GaAs) have demonstrated the highest solar-to-hydrogen (STH) conversion efficiencies to date.^1,2,3 A GaInP₂ top junction is common to many of the highest performing systems, and the incorporation of a thin AlInP₂ window layer to decrease surface recombination and a semiconductor capping layer to protect the window layer from corrosion has been shown to improve top cell performance.^1,4 However, the stability of III-V-based PEC devices remains limited, and durability is a major challenge in the field of PEC.

In this work, we investigate the impacts of a window layer and capping layer in conjunction with an MoS₂ catalytic layer on durability and performance in single-junction GaInP₂ photocathodes. GaInP₂ photocathodes with a window layer only and both a window layer and a capping layer were fabricated and compared to a baseline MoS₂/GaInP₂ photocathode. PEC characteristics were probed via current-voltage, chronoamperometry, and integrated photon-to-current efficiency measurements under solar simulator illumination. Photocathode degradation was monitored by in situ optical microscopy and post-test XPS measurements, illustrating macroscopic degradation modes.

Overall, the photocathode with a capping layer and window layer gave both the best performance and longevity. This photocathode surface architecture can be translated to the top junction of other tandem absorber III-V systems to improve device stability and efficiency.

References:
(1) Young, J. L.; Steiner, M. A.; Döscher, H.; France, R. M.; Turner, J. A.; Deutsch, T. G. Direct Solar-to-Hydrogen Conversion via Inverted Metamorphic Multi-Junction Semiconductor Architectures. Nat. Energy 2017, 2, 17028.
(2) Khaselev, O.; Turner, J. A. A Monolithic Photovoltaic-Photoelectrochemical Device for Hydrogen Production via Water Splitting. Science 1998, 280, 425–427.
(3) Cheng, W.-H.; Richter, M. H.; May, M. M.; Ohlmann, J.; Lackner, D.; Dimroth, F.; Hannappel, T.; Atwater, H. A.; Lewerenz, H.-J. Monolithic Photoelectrochemical Device for Direct Water Splitting with 19% Efficiency. ACS Energy Lett. 2018, 3, 1795–1800.
(4) Steiner, M. A.; Barraugh, C. D.; Aldridge, C. W.; Alvarez, I. B.; Friedman, D. J.; Ekins-Daukes, N. J.; Deutsch, T. G.; Young, J. L. Photoelectrochemical Water Splitting Using Strain-Balanced Multiple Quantum Well Photovoltaic Cells. Sustain. Energy Fuels 2019, 3, 2837–2844, DOI: 10.1039/c9se00276f.

For photoelectrode used for solar fuel generation, one key characteristic is photoelectrode stability. They are efficient and have tunable bandgaps that could in principle allow in tandem structures for unassisted, high-efficiency water splitting and other applications. However, III-V based photoelectrodes are also reported to suffer continuous photo corrosion in aqueous solution. Even with the coating protection, catastrophic failure will still occur due to pitting corrosion at defect sites in protection layers, which limited the device stability to be less than 100 hours. In our recent work (doi:10.1021/acsenergylett.0c02521), we compare and discuss about the corrosion mechanism for continuous and discrete electrode, and then propose a new stabilization strategy by confining the nanoscale pinhole from propagating. Firstly, we recognize that the long-term failure mode of planar III-V materials involves pitting corrosion due to extrinsic defects in amorphous TiO2 protective layers. Then, we show that extended stability can be obtained in structures that isolated defects electrically and thus prevented deleterious corrosion from spreading laterally, undercutting the protection layers and active layers, and leading to catastrophic failure of the photoanode. When grown on insulating or self-passivating substrates, these nanowire arrays indeed exhibited extended stability and operated stably for over 600 hours of photoanodic water oxidation, as compared to the control experiments that yielded much less stability due to a lack of defect-tolerance in the device structure. This work employs comprehensive characterizations including multiscale SEM/EDS, TEM, and electrochemical methods to the electrodes before and after the stability test. We believe this discovery and study should help advance the understanding of corrosion mechanism for coated photoelectrode and the mitigation strategy.

The energetics alignment between the TiO₂-based protective coating and the photoanode is significant for efficient water splitting and other fuel conversion devices. Since different semiconductors have different valence band maximum energy supporting various oxidative reactions with different potentials, appropriate intermediate band (IB) positions are required to accomplish better charge transfer for efficient devices. This work found transition-metal cations doping can be used to fabricate TiMO_x (M standing for manganese, chromium, etc.), tuning the IBs by the atomic layer deposition method. Valence band X-ray photoelectron spectroscopy was conducted to measure the IB position and width, which illustrates that the IB is dependent on the doping element, concentration, and cluster size. To illustrate the photoelectrochemistry (PEC) performance, leaky TiO₂, TiO₂:Mn, and TiO₂:Cr coated gallium phosphide electrodes are tested in Fe(CN)₆^3-/4- redox solution under light illumination, which shows a large electron transfer rate compared to traditional leaky TiO₂ coating. With iridium oxide nanoparticles as cocatalyst deposited on the protective coating, the electrode conducts water oxidation reaction in pH=9 phosphate buffer solution and reached the light-limited current at 1-sun illumination, which shows high efficiency and stability. This success of achieving tunable IB protective coating expands the generality of the application of TiO₂-based protective coating for semiconductors and catalysts with different energy levels, thus providing a new pathway for designing stable and efficient PEC devices.

Atomic layer deposited (ALD) protection layers have been crucial for enabling the integration of chemically sensitive III-V semiconductors into photoelectrochemical energy conversion devices. Although high photovoltage systems have been demonstrated, there remain significant open questions regarding the roles of semiconductor/protection layer interfaces and defects within the ALD films on carrier injection, electronic transport, and recombination processes. Here, we aim to elucidate how the optical and electronic properties of TiO_x adlayers influence the PEC performance of III-V photocathodes. To unravel these factors, we report on the effect of ALD TiO_x defect energies and concentrations on interfacial charge transport and photoelectrochemical performance using p-InP/TiO_x/Pt structures as model systems. In particular, we demonstrate highly tunable stoichiometry and defect concentrations within ultrathin, conformal ALD TiO_x using different metalorganic precursors and oxidants. Photothermal deflection spectroscopy reveals defect-induced sub-bandgap optical absorption within TiO_x that can be tuned over three orders of magnitude. These results are corroborated by X-ray photoelectron spectroscopy, which confirms the presence of Ti³⁺ in sub-stoichiometric films and the emergence of electronic states within the bandgap. Consistent with this analysis, in-plane electronic transport measurements reveal conductivities that vary over five orders of magnitude for these films. In addition, in situ spectroscopic ellipsometry during deposition is used to monitor not just the growing TiO_x films but also the oxidation of the underlying InP, thereby allowing us to tailor ALD processes that suppress InPO_x interlayer formation and reduce the out-of-plane charge transport resistance. Photoelectrochemical testing of p-InP/TiO_x/Pt photocathodes under hydrogen evolution reaction conditions confirms the importance of avoiding interface oxidation. In addition, we find once interface oxidation is suppressed, a higher defect concentration within the TiO_x layer leads to the reduction of the HER onset potential. This can be explained by an increased recombination rate in those films, which is evident from surface photovoltage measurements. While high defect concentrations within protection layers significantly reduce photoelectrochemical performance, the interface characteristics rather than film properties play a dominant role for moderate and low defect content TiO_x films. These findings highlight the crucial role of controlling interface properties and defects of protection layers to not only preserve but even enhance the photoelectrochemical energy conversion efficiency of III-V semiconductor electrodes.

Photoelectrochemical (PEC) water splitting cells offer an environmentally friendly way to store solar energy in chemical form. However, increasing the efficiency of the current solutions requires detailed understanding of the physical processes on both sides of the solid-liquid interface. Accurate computational modeling offers information on the charge carrier dynamics of an operational cell, giving key insight required in developing new solutions.
While the performance of simple PEC cells can be adequately explained with semi-classical drift-diffusion theory, the increasing interest towards thin film cells, multi-layered devices and nanostructures in general requires theoretical treatment that can capture the quantum phenomena in these nanoscale features. Cells using III-V materials often make use of thin film protective layers, highly doped contact regions and multi-junction designs where tunnelling and other quantum effects have a significant impact on the charge transport inside the electrode. The non-equilibrium Green’s function (NEGF) approach offers an excellent compromise between the limiting assumptions included in classical transport equations and computationally heavy ab initio quantum chemical simulations.
We have implemented a numerical model of the semiconductor photoelectrode of a PEC cell based on the NEGF formalism, where The NEGF equations are solved self–consistently together with Poisson’s equation and electrochemical boundary conditions. The NEGF model allows us to compute an energy resolved solution of the charge carrier densities and corresponding currents inside the semiconductor electrode at nanoscale. The presented simulation offers important insight into the operation of PEC cells at nanoscale where quantum effects have a significant contribution to the charge transport, providing a pathway for designing more efficient and economically viable devices.
Acknowledgement: This research was funded by Academy of Finland

Cascade photoelectrocatalysis (PEC) is a possible method to improve the selectivity of solar-driven CO₂ reduction (CO₂R). This concept can be realized by coupling different CO₂R catalysts to different subcells in a multijunction photovoltaic (PV) stack. Efficient implementation will require careful tuning of the photocurrents and design of the photovoltages provided by the subcells to the CO₂R catalysts in such a way as to facilitate the target reaction. Here, we outline the design principles of the tandem PEC approach using two-step conversion of CO₂ to ethylene in aqueous, via a CO intermediate as a model system and use the open-source circuit simulator, SPICE, to simulate operating conditions. Cascade PEC can be realized in a three-terminal tandem (3TT) configuration using III-V-semiconductor based subcells coupled to Au (produces CO intermediate) and Cu (converts CO to ethylene). We identify conditions under which a 3TT configuration can have a higher solar to chemical conversion efficiency compared to a two-terminal two-junction tandem (2T 2J) with the same absorbers and a Cu catalyst only and demonstrate that 3TT PEC devices can be less sensitive to variations in catalyst activity compared to 2T devices. Three terminal tandem devices also allow for light and potential activated independent control of each subcell. To demonstrate this control, we made a GaInP/GaAs 3TT solar cell with different catalysts coupled to each site, each driving different reactions who’s respective E⁰ values differ (one produces the intermediate). Using different colored LED light sources and varying the potential applied while tracking the product current with online mass spectroscopy, we show independent and cascade control of the reduction sites on the 3TT PEC device.

Para-Xylene is an industrially significant aromatic feedstock, with an annually demand of 50 million tons. One of its oxidized forms, terephthalic acid is a primary component of polyesters, produced at ~7 million tons annually. The industrial synthesis is a high-temperature aerobic oxidation in acetic acid, which occurs in a three stage catalytic process.
Photocatalysis offers a solution via the use of efficient semiconductors for organic synthesis. Through adapting homogeneous reactions that are routinely performed in organic or photoredox synthesis, we can perform these reactions on the surface of photocatalysts. Through synergistic surface C-H bond activation and generation of reactive oxidative species (ROS) the catalytic cycle is compete. III-V Semiconductors coated with ternary TiO₂ based protective layers, coupled with selective co-catalysts can be used to photocatalytically oxidize para-xylene to terephthalic acid (TPA), promising a mild and distributed synthesis method at variable scale.
Oxidation of para-xylene is a model reaction to explore the translation of the principles of homogeneous catalysis to photocatalysis. TiO₂ surfaces have been shown to activate methanol to methoxy species, and photocatalytic partial oxidation of benzyl alcohol to benzaldehyde has been demonstrated, providing support for individual aspects of the work.
UV irradiation is problematic as it can lead to undesired side reactions, reducing selectivity. The addition of photo-inactive protective coatings can stabilize visible light absorbing semiconductors would be otherwise corrode from even trace water. It is necessary to understand the charge transfer process between surfaces and surface bound molecules and the energetics of the respective processes. The energetics of the holes and electrons indirectly control the surface reactivity, production rate, and desorption.
The surface chemistry should affect the energetics at the reductive and oxidative sites in the form of a surface dipole. This has been shown on Si photoanodes where the methoxy that is formed via methanol oxidation shifts the SI band position towards more negative potentials, increasing barrier heights. During operation, electrons and holes would exchanges charges with a variety of surface molecules, rather than one. As a consequence, the valence band positions of high charge-density coatings (>10²⁰) should shift, while low charge density (~10¹⁵)coatings such as amorphous “leaky TiO₂”, would remain fixed.
In this work, we relate product distribution with surface chemistry, molecular binding strength, and hole transfer energetics. Through tuning the Semiconductor band edge positions by varying the Ga, In, and Al content, along with well designed coating composition, the reactivity can be selectively tuned.

Photocatalytic CO₂ reduction (CO₂R) promises scalable conversion of atmospheric CO₂ to chemicals or fuels. Currently, most photocatalytic CO₂R systems require concentrated and gaseous CO₂ as feedstock. This process requires extra energy and carbon footprint, especially when separating 410 ppm CO₂ directly from the air. We demonstrate a design that enables photocatalytic CO₂R and local CO₂ production simultaneously using photocatalytic panels consist of III-V semiconductors and reversible quinone redox mediators. Our design, which draws inspiration from photosynthesis, consists of enriching CO₂ in a (bi)carbonate solution, spatially separating surface reaction sites and driving local pH swing to release CO₂ locally, and reducing CO₂ into CO₂R products eventually. We show that in the presence of quinone redox couples in a bicarbonate solution, CO is produced in a 1-atm Ar environment where the only source of CO₂ is the (bi)carbonate anions. Finally, we construct a COMSOL Multiphysics model that accounts for all the reaction kinetics to simulate the photocatalytic reaction during operation

Epitaxially grown III-V semiconductor nanowire (NW) arrays have become an emerging solar cell architecture with great promise for high-efficiency and low-cost solar energy harvest devices [1]. In NW arrays with dimensions comparable to the wavelength of solar radiation, the light-NW interaction can be controlled by both the NW geometry and the pitch of the NW array in order to enhance the solar cell efficiency beyond the Shockley-Queisser theoretical limit from a single planar junction solar cell [2]. In addition, the nanoscale footprints of NWs can accommodate lattice mismatch efficiently, resulting in a superior quality of the interface for heteroepitaxial integration and allow for high-performance III-V semiconductors on a Si platform in order to e.g. form dual-junction tandem structures monolithically on a Si bottom solar cell. Theoretical simulations indicate that a dual-junction tandem cell, using a III-V top cell on a Si bottom solar cell, can achieve efficiencies higher than 40 % [3]. For such dual tandem cells to be cost-effective and used on a large scale in the future, it is crucial that the III-V top cell is monolithically integrated on the Si cell (i.e., no use of III-V substrate), and that a minimum cost is used for the growth and processing of the III-V top cell, as this is at present the major cost issue holding back this technology [4].
In this work, we demonstrate a very effective use of the III-V photoconversion material leading to an ultra-high power-per-weight ratio by utilizing an axial junction GaAs/AlGaAs NW array grown by molecular beam epitaxy on a Si substrate. By analyzing single NW multi-contact devices, we firstly show that an n-GaAs shell is self-formed radially outside the axial p- and i-core of the GaAs NW during n-core growth, which significantly deteriorate the rectification property of the NWs in the axial direction. When employing a selective-area ex-situ etching process for the n-GaAs shell, a clear rectification of the axial NW p-i-n junction with a high on/off ratio was revealed. Such controlled etching process of the self-formed n-GaAs shell was further introduced to fabricate axial p-i-n junction GaAs NW array solar cells. Employing this method, a GaAs NW array solar cell with only ~ 1.3 % areal coverage of the NWs shows a photoconversion efficiency ~ 7.7 % under 1 sun intensity (AM 1.5G), which is the highest achieved efficiency from any single junction GaAs NW solar cell grown on Si substrate so far [5]. This corresponds to a power-per-weight ratio of the active III-V photoconversion material as high as 560 W/g, showing great promise for future high-efficiency and low-cost III-V/Si tandem solar cells.

References:
1. J. Wallentin et al., InP nanowire array solar cells achieving 13.8% efficiency by exceeding the ray optics limit. Science 339, 1057-1060 (2013)
2. G. Otnes and M.T. Borgström, Towards high efficiency nanowire solar cells. Nano Today 12, 31-45 (2017)
3. I. Almansouri et al., Supercharging silicon solar cell performance by means of multijunction concept. IEEE J. Photovolt. 5, 968–976 (2015).
4. S. Essig et al., Raising the one-sun conversion efficiency of III–V/Si solar cells to 32.8% for two junctions and 35.9% for three junctions. Nature Energy 2, 1-9 (2017).
5. A. Mukherjee et al., GaAs/AlGaAs nanowire array solar cell grown on Si with ultra-high power-per-weight ratio. ACS Photonics 8, 2355-2366 (2021).

The introduction of bismuth into GaAs results in a large band-gap bowing and a dramatic reduction in band-gap energy. This enables GaAs alloys containing dilute fractions of Bi to offer technologically useful absorption thresholds in multi-junction solar cells. Here we investigate the opportunity that GaAsBi holds as a component in an upright metamorphic solar cell architecture which, in principle, could offer a better match to the AM0 spectrum over the conventional InGaAs based UMM cell. We show that the strong and inherent alloy disorder associated with the dilute Bi fraction results in a degradation in the GaAsBi sub-cell by as much as 202 mV with 5.5% Bi incorporation. Despite this degradation in voltage we find that a potential GaAsBi based design that accounts for finite diffusion length and alloy disorder outperforms the conventional InGaAs UMM architecture at AM0, requiring only 2.8% Bi incorporation and 0.25% compressive strain.

III-nitride nanostructures, e.g., Ga(In)N nanowires, have emerged as one of the most promising material platform for artificial photosynthesis, including the chemical transformation of sunlight, water, and carbon dioxide into clean chemicals and fuels such as hydrogen and hydrocarbons. Recent studies suggested that the energy bandgap of GaN can be tuned across nearly the entire solar spectrum with the incorporation of indium as well as dilute anions (e.g., antimony). Moreover, the conduction and valence band edges of dilute anion III-nitride nanostructures can be tuned to straddle a broad range of redox potentials, including water splitting, carbon dioxide reduction and methane oxidation, which is essentially required for high efficiency artificial photosynthesis. Our recent studies further showed that the surfaces of GaN-based nanostructures can be terminated with nitrogen, not only for their top c-plane but also the lateral non-polar surfaces. Significantly, the lateral nonpolar surfaces can be further transformed to oxynitride, which can lead to improved performance, instead of degradation, during harsh photocatalysis reaction. This unique surface transformation and property has enabled the demonstration of photoelectrochemical solar water splitting with stability ~3,000 hrs without any performance degradation.

Moreover, epitaxial III-nitride nanostructures have shown great promise to overcome the sluggish kinetics related to various surface chemical reactions. As an example, the surfaces of epitaxial III-nitrides were found to enable unique interaction with carbon dioxide and other molecules/chemical species to significantly reduce the energy barrier in some otherwise difficult chemical reactions. With the integration of suitable co-catalysts, the selective production of high value chemicals and fuels can be potentially achieved, which includes the one-step light-driven conversion of carbon dioxide to methane, formic acid, syngas, and methanol with high Faradaic efficiency. Our studies also revealed the synergistic effect of light and thermal energy in significantly enhancing the efficiency, selectivity, and production rate.

III-nitride photocatalyst nanostructures have also enabled a series of achievements of light-driven organic transformations and nitrogen fixation, which are also critical for carbon neutrality and environmental sustainability. For example, by coordinating GaN with Ru nanoparticles, a metal/semiconductor interfacial Schottky junction was formed to effectively extract photoinduced electrons from GaN light absorber, rendering Ru nanoparticles with high electron density for facilitating the extremely inert N≡N bond cleavage with the use of hydrogen, resulting in mild ammonia synthesis powered by sunlight, in stark contrast with the harsh Haber-Bosch process.

By marrying III–V material systems with a quasi-one dimensional geometry, III–V nanowires comprise an ideal structure for solar energy harvesting. Resonant light trapping and waveguiding effects in individual nanowires of subwavelength diameter permit enhanced optical absorption of solar radiation. Multiple scattering events and in-plane waveguiding within arrays of vertically-standing nanowires can further enhance light absorption. The nanowire geometry also presents the opportunity for large interfacial areas allowing efficient charge carrier separation, for example at core–shell junctions in GaAs nanowire solar cells, and at photoelectrode–electrolyte interfaces in (In)GaN nanowire photoelectrochemical water splitting devices. These efficiency improvements can be achieved despite the microscopic quantity of III–V material present in the nanowire arrays, using up to 10⁴ times less material compared to planar structures [1].
To achieve the full potential of nanowires for solar cells and photoelectrochemical water splitting, a number of materials challenges must be addressed. Scalable approaches to growth are needed, for example III–nitride nanowire growth by metalorganic vapour phase epitaxy (MOVPE). We have investigated the self-assembly of GaN nanowires by MOVPE and found that nitridation of the growth substrate, and the formation of a thin AlN layer, leads to N-polar nanowire growth on sapphire substrates. A further challenge is the passivation of nanowire surfaces, which is essential for reducing non-radiative losses and reaching high open circuit voltages [2].
In addition, nanowire arrays can be embedded in a flexible transparent polymer and removed from the substrate [3]. This has the dual advantages of creating an ultra-thin flexible device, and permitting immediate re-use of templated substrates in subsequent growth runs. We have identified the mechanical constraints that dictate the choice of embedding polymer and limit the geometric properties of the nanowire array.
Together, these advances will enable free-standing membranes of nanowires optimised for solar photovoltaics and photoelectrochemical fuel generation.
[1] L Gao; Y Cui; J Wang; A Cavalli; A Standing; TT Vu; MA Verheijen; JE Haverkort; EP Bakkers; PH Notten, Nano Lett. 14:3715-9 (2014)
[2] N Jiang, HJ Joyce, P Parkinson, J Wong-Leung, HH Tan, C Jagadish, Front. Chem. 8: 1136 (2020)
[3] SA Baig, JL Boland, DA Damry, HH Tan, C Jagadish, HJ Joyce, MB Johnston, Nano Lett. 17:2603-2610 (2017)

We demonstrate Ga_1-xIn_xP compositionally graded buffers (CGBs) and metamorphic Ga_1-xIn_xAs solar cell devices grown by hydride vapor phase epitaxy (HVPE) as a route to high-throughput epitaxy of metamorphic devices. Metamorphic III-V epitaxy enables devices with tunable bandgaps, which have the potential to satisfy the demands of applications such as photovoltaics, thermophotovoltaics, solid state lighting, photodetectors and transistors. However, CGBs are usually the thickest regions in metamorphic devices and represent a significant cost due to traditionally low deposition rates of ternary materials by present growth technology. Faster growth rates increase throughput and reduce cost but are only useful if a low defect density can be maintained as rates increase. Steady-state models of plastic strain relaxation suggest that threading dislocation density (TDD) is proportional to growth rate in the limit in which the dislocation glide velocity controls the strain relaxation rate. In this work, we develop HVPE-grown Ga_1-xIn_xP CGBs with varying lattice constants between GaAs and InP and growth rates up to 1 µm/min. We study the effect of substrate offcut direction, growth rate, and strain grading rate on CGB defect structure, achieving TDD as low as 1.0 x 10⁶ cm^-2. Substrate offcut direction is the most important factor for obtaining CGBs with low defect density. (001) GaAs substrates offcut towards (111)B yield smoother CGBs with lower TDD compared to CGBs grown on substrates offcut towards (111)A. Transmission electron microscopy of CGBs grown on A and B-offcut substrates only finds evidence of phase separation in A-offcut CGBs, indicating that the B offcut limits phase separation, which in turn keeps TDD low. Reduced growth rates of ~0.3 µm/min lead to the appearance of CuPt-type atomic ordering, which affects the distribution of dislocations on the active glide planes but does not significantly affect TDD. Higher growth rates lead to smoother CGBs and do not appreciably increase TDD as otherwise predicted by steady-state models of plastic relaxation, indicating that glide velocity does not limit TDD in this case. We show TDD = 2.4 x 10⁶ cm^-2 in a grade from GaAs to InP grown at 1 µm/min, which suggests that HVPE can be used to develop cost-effective large-area InP virtual substrates on GaAs. We also develop metamorphic 1.0 eV Ga_0.7In_0.3As rear heterojunction solar cells grown on Ga_0.2In_0.8P CGBs as an example of an HVPE-grown metamorphic device. The CGB and device structure were grown in less than 10 minutes. We obtain open circuit voltages of up to 0.59 V in these cells, a result comparable to state-of-the-art metamorphic GaInAs solar cells grown by other methods. These results show HVPE’s promise for lattice-mismatched applications and low-cost InP virtual substrates on GaAs.

We demonstrate the growth of (110) GaAs solar cells by hydride vapor phase epitaxy (HVPE) on epi-ready and previously-spalled GaAs wafers as an advance towards a potentially low-cost (110)-based device platform. III-V solar cells exhibit higher conversion efficiencies than any other technology, but the high costs of epitaxial growth and growth substrates limit their application to high value applications such as space power. HVPE is a promising method to reduce the epitaxy costs associated with III-V solar cell growth, while selective-etchant-based epitaxial liftoff with substrate reuse is the industry-standard method to address substrate costs. However, epitaxial liftoff’s cost competitiveness for terrestrial applications is limited by low throughput and the periodic need for a chemical mechanical polishing (CMP) step to clear insoluble etch products. Current CMP cost estimates are $10-$25 per wafer, meaning that new substrate cost reduction methods are necessary to enable terrestrial use of III-Vs.
Controlled spalling, whereby a device is rapidly exfoliated from the substrate via crystalline cleavage, offers an alternative with the potential for lower cost. Ideally, spalling should not require a CMP step because a cleaved, contamination-free wafer surface is revealed from within the crystal after each cycle. We envision that devices can be grown on and spalled from a wafer numerous times, amortizing its cost. Previous work on the controlled spalling of (001) GaAs revealed challenges with surface faceting, because the spall propagates along {110} or {112} planes instead of the (001) plane. The facets, which can be 10 µm in size or more, waste material and require planarization before device regrowth. A promising solution to the faceting issue is to spall devices from (110) GaAs substrates, exploiting the highly-selective {110} GaAs cleavage planes by orienting them parallel to the substrate surface. Preliminary (110) wafer spalls do not exhibit the large features observed in (001) spalls, motivating the development of (110)-based devices. Epitaxy on (110) surfaces is not straightforward, however, as standard (001) growth conditions often produce faceted morphology and poor optoelectronic quality. Understanding of how to control the morphology of (110) growth surfaces is necessary to develop high-efficiency photovoltaic devices on that orientation. Furthermore, the fine structure of the spalled surface and its impacts on device regrowth must be studied.
In this work, we show progress towards the development of (110)-based high-efficiency solar cells and spalling. First, we show the effect of HVPE growth conditions on the morphology of (110) epitaxial growth. On (110) substrates offcut 3° towards (111)A, growth faceting develops under conditions in which the growth is Ga-limited, although Ga-limitation is not the only requirement for faceting. Low growth temperatures lead to a faceted morphology, while higher growth temperatures favor smooth growth and facet-free morphology. We show that the surface morphology is kinetically-controlled, and that the tendency to facet correlates with the growth rate. Overall, our results are consistent with models for step-bunching-induced surface faceting that invoke a negative Ehrlich–Schwoebel step-edge barrier. Using optimized growth conditions for GaAs and GaInP, we develop (110)-based GaAs solar cells on epi-ready substrates, demonstrating equal performance to devices grown on the standard (001) orientation. We also present spalling results from (110) substrates, revealing macroscopically planar surfaces with large terraces separated by steps with sub-µm height. Lastly, we show an initial GaAs solar cell device with nearly 16% efficiency grown on a previously spalled surface that was not subjected to any surface re-preparation. Together, these results provide preliminary evidence of a potentially low-cost path for terrestrial III-V photovoltaics via the (110) substrate orientation.

Improving integration III-Vs to Si is key to increase Photo-Voltaic Conversion Efficiency (PVCE). Improving PV technology also means increasing solar panels’ affordability and sustainability to accelerate its adoption without defeating its purpose of greening human activity and decreasing fossil fuel use.
Reaching higher PVCE while lowering manufacturing costs and greening fabrication can be achieved 1^st by lowering processing temperatures. These are presently ≥ 400°C in Metal-Organic Vapor Phase Epitaxy (MOVPE) or Molecular Beam Epitaxy (MBE). Another issue with GaAs/Si Tandem Solar Cells (TSC) is that their theoretical PVCE is 44% [[i]], but the highest achieved PVCE is 32.9% in the laboratory [[ii]], ¾ of the theoretical PVCE.
Three key issues limit achievable PVCE. First, bulk, surface, and interfacial defects are generated specifically during GaAs to Si integration, due to its present thermal budget. Second is residual oxygen on wafer surfaces. Third is thermal expansion during and after MOCVD and MBE at T ≥ 400°C which decreases carrier diffusion lengths, and PVCE. High temperatures not only raise costs in fabrication but limit PVCE.
Presently, GaAs/Si TSCs costs are 10 X higher than costs of Si single-junction solar cells [3], and only vastly cheaper GaAs-Si TSCs are viable for terrestrial applications. TSC costs can be reduced by replacing MOVPE or MBE with Nano-Bonding™ (NB) at ≤ 220°C with Surface Energy Engineering (SEE) [[iii]] which does not require costly, hard to maintain, power-consuming Ultra-High Vacuum (UHV) system, nor toxic gases handling hardware.
Nano-Bonding™ (NB) at ≤ 220°C with SEE at RT reduces the temperature for GaAs to Si bonding by 50%, and lowers the energy consumed by GaAs-Si integration by a factor 16.
The 1^st step in NB at ≤ 220°C is for SEE to planarize GaAs(100) and Si(100) wafer surfaces at the nano-, micro- and macro-scale via far-from-equilibrium (FFE), complementary 2D surface precursor phases (c2D-PP) [iii]. These are formed after GaAs and Si undergo complementary chemical etching processes, via chemisorption of Langmuir-Blodgett films. SEE reverses the initial hydro-affinity (HA) of GaAs from hydrophobic to hydrophilic, and vice-versa for Si to FFE states. The key property of c2D-PP is that they are stable at RT in air up to T ≤ 200°C but react into a uniform, 2D, defect-free, cross-bonding interphase between GaAs and Si when thermally activated with each other at T~220°C. They are characterized via their HA and Surface Energy (SE) on GaAs(100) and Si(100) by Three Liquid Contact Angle Analysis (3LCAA).
Concurrently with 3LCAA, High Resolution Ion Beam Analysis (HR-IBA) yields absolute Oxygen coverage, and X-Ray Photoelectron Spectroscopy (XPS) yields surface stoichiometry and chemistry. Combining 3LCAA with HR-IBA, and XPS yields an atomistic model for c2D-PP and helps optimize SEE and NB to lower bonding temperatures, remove oxygen and surface defects. SEE of GaAs use dilute NH₄OH:H₂O (1:10) , and modifies GaAs SE 30.4±1 mJ/m² to 60±2 mJ/m², while IBA detects a 50% decrease in surface Oxygen coverage from 7 ML to 3.5 ML after 30 min in air. Next, XPS shows that SEE decreases O-rich As₂O₅ on GaAs(100) by 13.5%, while As₂O₃ increases by at least 13.5%.
Nano-bonded GaAs-Si imaged via Surface Acoustic Wave Microscopy (SAWM) and Transmission Electron Microscopy (TEM). SAWM shows 98% of the GaAs wafer bonds to Si when under 60 kPa of mechanical compression, while 48% of the whole wafer area bonds without any compression. TEM reveals the reactivity of the 2D-PPs and contacted layers.
Future work involves measuring first the PVCE of the tandem solar cells formed from these structures and further improving time-to-characterization of c2D-PP and potential surface residual oxides prior to NB after SEE.
[ [i] ] De Vos A et al, (1981) APL.
[ [ii] ] NREL (2021)
[ iii ] Herbots et al, US Pat. 2005-17

The III-Nitride family of semiconductors (GaN, AlN, InN, and their ternary variants) provides a basis for designing photoelectrodes in contact with electrochemical interfaces. Three important aspects of the work in this material are (1) epitaxial growth of III-Nitrides on silicon and free-standing GaN, (2) design of the energetics of the III-Nitride stack on the substrate, and (3) integration of an electrocatalyst on the surface of the semiconductor to drive a reaction such as CO₂ reduction.
In this talk, we will present an overview of research at ARL exploring these three aspects of using III-Nitride materials for electrochemical junctions. First, both binary GaN and ternary InGaN are grown with epitaxial control on silicon and free-standing GaN to obtain single-crystals that in the case of silicon, would enable thin-film and micro/nanostructured heterojunctions for photoelectrodes, and in the case of free-standing GaN, would enable pseudomorphic growth to control the c-plane polarization fields. Second, one-dimensional drift-diffusion-Poisson modeling is used to design III-Nitride photoanode and photocathode junctions with controllable polarization from strain control in epitaxial films; these designs inform the importance of polarization field within a photoanode and photocathode, and opens up new ideas on using the field to generate kinetic energy in charge carriers for reducing an overpotential for oxidation and reduction reactions. Finally, an elemental copper source is integrated within a III-Nitride MBE system to enable Cu-based CO₂ reduction catalyst synthesis with both thermodynamic and kinetic control; this helps achieve the creation of a combined light absorber/catalyst photoelectrode in a UHV environment, and achieves a space for combining Cu with Ga and In to explore bimetallic and trimetallic catalysts in this environment.

Thermophotovoltaics (TPV) offer a direct method to translate the heat generated as a byproduct of other standard energy generation techniques into usable electricity. Yet, the current material options implemented for the optical emitters substantially limit the efficiency of this conversion process. We screen the optical behavior of >2,800 material combinations with melting point >2,000 oC, including refractory metals, silicides, borides, carbides, nitrides, and oxides, and find several options for GaSb-based TPV with theoretical efficiency >60%. Further, we down select feasible options by considering materials’ thermal expansion and thermochemical stability, often overlooked. Our emitter design is based on a scalable configuration, a dual- or triple-layer thin-film stack. We expand this approach and also identify optimal material combinations for InGaAsSb, InGaAs, Ge, GaSb, and Si. For each system, we also identify the optical emitters that provide the highest power output, which will be discussed in details.

PV-powered vehicle applications are very attractive for reducing CO₂ emission and creation of new market [1]. Development of high-efficiency (> 30%) and low-cost solar cell modules is very important. The III-V multi-junction (MJ) soar cells have shown high-efficiency of 37.9% and 39.2% with 3-junction and 6-junction solar cells [2]. Although Toyota Prius and Nissan eNV 200 demonstration cars installed with Sharp’s high-efficiency III-V 3-junction solar cell modules with an efficiency of more than 30% have demonstrated longer driving range of 26km/day average compared to 16km/day average for Sono Motors Sion installed with Si back contact solar cell modules with an efficiency of 21.5%, cost reduction of MJ solar cells is necessary for PV-powered vehicle applications. Development of Si tandem solar cells and high speed deposition such as hydride vapor phase epitaxy (H-VPE) is very promising for cost reduction. However, efficiencies of 35.9% with III-V/Si 3-junction tandem cells and 28.2% with H-VPE grown III-V 2-junction cells are lower compared to 37.9% with III-V 3-junction and 32.9% with MOVPE (metalorganic vapor phase epitaxy)-grown III-V 2-junction tandem solar cells. Therefore, it is necessary to clarify and reduce several losses of MJ solar cells. This paper presents high efficiency potential of III-V/Si 3-junction solar cells and H-VPE grown III-V 2-junction solar cells analyzed by using our analytical procedure [3] and discusses about non-radiative recombination, optical and resistance losses in those MJ tandem cells.
One of problems to attain the higher efficiency MJ solar cells is to reduce non-radiative recombination loss. The open-circuit voltage V_oc drop compared to bandgap energy (E_g/q - V_oc) is dependent upon non-radiative voltage loss (V_{oc, nrad}) that is expressed by external radiative efficiency (ERE). Open-circuit voltage is expressed by
Voc = Voc,rad + (kT/q)LN(ERE), (1)
where the second term shows non-radiative voltage loss, is radiative open-circuit voltage and 0.28V for III-V compound and 0.26V for Si were used as △Voc,rad (= E_g/q - Voc,rad ) in this study.
Correlation between V_oc values for III-V/Si 3-juction and H-VPE grown III-V 2-junction solar cells reported in the references and ERE values estimated by eq. (1) suggests that those MJ solar cells have larger non-radiative loss and further improvements in efficiency are thought to be possible by improving ERE. The ERE values estimated in this study are 0.51% for 35.9% III-V/Si 3-junction tandem solar cells and 0.17% for 28.2% H-VPE grown III-V 2-junction solar cells that are lower compared to 2.2% for 37.9% III-V 3-junction and 5.6% for 32.9% MOVPE-grown III-V 2-junction tandem solar cells. In addition, optical and resistance losses of H-VPE grown III-V 2-junction solar cells are higher compared to those of MOVPE-grown III-V 2-junction solar cells. According to our analysis [3], the 2-junction and 3-junction tandem solar cells have potential efficiencies of more than 36% and 42%, respectively.
Several loses such as non-radiative recombination, optical and resistance in MJ solar cells are discussed in this paper.
References
[1] M. Yamaguchi et al., Prog. Photovolt. 29, 684 (2021).
[2] M. Yamaguchi et al., J. Appl. Phys. 129, 240901 (2021).
[3] M. Yamaguchi et al., J. Mater. Res. 32, 3445 (2017).

As the power requirements for electronic circuits have steadily dropped, applications for stand-alone systems with a small form factor, long lifetime and high energy density have arisen. Stand-alone systems such as cardiac pacemakers, medical implants, environmental monitoring, anti-tampering circuits and deep space probes require autonomous power in the range between nanowatts and milliwatts. The energy and power requirements of these new systems are pushing the limits of battery technology. Nuclear power (in the form of radioisotopes) has energy density 3 to 4 orders of magnitude higher than current batteries (which have reached their theoretical limits) and half-life of 10 to 400 years. Radioisotopes of interest emit beta or alpha particles (high-energy electrons or helium 4 ions) and with proper shielding are considered safe. Surprisingly (for many) smoke detectors in the home are based on a radioisotope (Americium 241). Other products based on radioisotopes are currently in production. In this talk we will explain how radioisotopes can be used to directly produce electrical energy. We will compare the performance of nuclear batteries with current battery technology in some of the most demanding emerging applications. We will discuss the prospects for increasing the continuous power out put of these devices. We will also discuss the practical issues for implement of this technology cost, packaging and safety

Solar-driven approaches for the processing of solar fuels and materials are interesting, provided they can be efficiently, stably, scalably, and sustainably implemented. Especially photon-driven water splitting or CO₂ reduction has gained some attention. One way to increase the economic and sustainable competitiveness is the utilization of concentrated irradiation [1]. This pathway can also justify the utilization of rarer materials, given their significantly reduced area and material need.
Here, I will discuss room-temperature operation of concentrated photoelectrochemical devices for water splitting [1] and CO₂ reduction, highlighting the requirements on the photoabsorber performance and their best integration into an operational electrochemical device. I will then extend the discussion to photon-driven devices that operate at significantly higher temperatures (> 700 K) making use of a larger part of the solar spectrum and profiting from significantly enhanced kinetics [2]. Such devices require novel semiconductor-junctions that allow for effective charge separation. Specifically, we identify assemblies of GaAs/GaP and GaAs/GaInP-GaAs/GaP as potential candidates for such high-temperature solar cells. I will discuss how they can be best integrated into an operational high-temperature photoelectrochemical device, and compare their performance and potential with low-temperature approaches.
References
[1] S. Tembhurne, F. Nandjou, S. Haussener, Nature Energy, 10.1038/s41560-019-0373-7, 2019.
[2] R. Gutierrez, S. Haussener, Sustainable Energy & Fuels, 10.1039/D0SE01749C, 2021.

III-V nanocrystals offer great promise within the context of energy conversion technologies – in addition to the properties of bulk materials, nanocrystals also exhibit large surface-to-volume ratio for increased available sites for energy conversion, tunable bandgaps based on quantum size effects, and more versatile device manifestations such as on flexible substrates and as conformal coatings, to name a few. The challenge has been in devising robust, tunable, repeatable synthesis methods to create these nanocrystals, as traditional colloidal techniques for growing semiconductor nanocrystals, for instance with II-VI materials, are difficult to translate to the III-V set of materials. Recently there has been growing interest in using gas-phase methods for nanocrystal synthesis: gas-phase techniques can avoid the sometimes long processing times and multiple steps of colloidal methods. Low-temperature plasmas (LTPs) in particular have demonstrated tunable synthesis of III-V nanocrystals using feedstocks such as metalorganics and elemental sources. Here, we present on tunable synthesis of GaN and InN nanocrystals using a flow-through LTP reactor. We also demonstrate size control over the nanocrystals using flow parameters as well as a using pulsed plasma power.
GaN and InN nanocrystals were synthesized using trimethyl(gallium, indium) and ammonia as precursors. These feedstocks, along with argon as a background gas, were flown into a pyrex reactor tube encircled by three ring electrodes- a central powered electrode and two grounds. Radiofrequency (RF) power at 13.56 MHz and 50-150W was delivered to the powered electrode, to excite a plasma. The pressure in the reactor was kept below 20 Torr using a rotary vane vacuum pump and pressure-controlling orifice. The nanocrystals were collected at the exhaust to the reactor via inertial impaction upon substrates. We used X-ray Diffraction (XRD), Fourier-transform infrared spectroscopy (FTIR), PL and UV-Vis spectroscopy, and transmission electron microscopy (TEM), and electron paramagnetic spin resonance (EPR) to analyze the nanocrystal properties. We were able to demonstrate control over the average size of the nanocrystals between ~2-5nm by adjusting the gas flowrates and pressures in the reactor, aiming to modify the overall residence time of the nanocrystals in the plasma. However, achieving larger nanocrystal sizes was difficult using this approach. We additionally explored the use of pulsed plasma power as a method to use agglomeration/coagulation of primary nanoparticles followed by plasma annealing and further reactions for growth of nanocrystals with average size that exceeds that possible using typical flow-based controls. These results point to the promise of LTPs for III-V nanocrystal fabrication, including the tailoring of specific nanocrystal properties (surface, size, and composition) that will enable energy conversion technologies such as photocatalysis, luminescence, and others.

The accurate quantitative estimation of thermal conductivity still poses a big challenge especially in the case of thin films and complex structures that are inseparable part of present-day electronics and devices. In particular, the thermal conductivity plays a major role directly affecting the performance of thermoelectric devices^[1]. This report is focussed on understanding and quantifying the role of multiple interfaces on the thermal transport in Sb₂Te₃/MoS₂ multilayer heterostructure with nanoscale layer thickness. To study both the intrinsic thermal conductivity of the heterostructure and interfacial thermal resistance between the substrate and a multilayer sample we employ a novel approach of cross-sectional scanning thermal microscope (xSThM)^[2]. In xSThM a thermal transport in a nanoscale thin small angle wedge created via Ar-Ion nanopolishing is measured via sharp temperature sensitive SThM probe. By effectively probing a range of sample thickness, the xSThM facilitates the investigation of individual contribution of the in-plane and out-of-plane thermal conductivities in the case of multi-layer samples, as well as sample – substrate thermal resistance. By exploring variation of thickness and number of layers of MoS₂ we were able to optimize the Sb₂Te₃/MoS₂ heterostructure to achieve extremely low values of in-plane and cross-plane thermal conductivity 0.5767 and 0.2401 (W/mK), respectively, along with higher values of Seebeck coefficient (619 μV/K at 347 K). A major enhancement in the value of TE performance was observed due to the effective majority carrier filtering and phonon scattering at the potential barrier present due to multiple interfaces. This study provides new insight for understanding the thermal transport in multi-layered samples, as well as other hybrid or buried nanostructures. With xSThM allowing to independently quantify in-plane, cross-plane and interfacial thermal resistance in nanoscale structures, it provides a rare approach for exploring thermal transport in thin films of complex anisotropic materials and associated devices.
[1] X.-Y. Wang, H.-J. Wang, B. Xiang, L.-W. Fu, H. Zhu, D. Chai, B. Zhu, Y. Yu, N. Gao, Z.-Y. Huang, F.-Q. Zu, ACS Applied Materials & Interfaces 2018, 10, 23277.
[2] J. Spièce, C. Evangeli, A. J. Robson, A. El Sachat, L. Haenel, M. I. Alonso, M. Garriga, B. J. Robinson, M. Oehme, J. Schulze, F. Alzina, C. Sotomayor Torres, O. V. Kolosov, Nanoscale 2021, 13, 10829.