Scalable method of fabricating high-quality transition metal chalcogenide materials

Abstract

Improved fabrication of chalcogenide structures is provided by reacting a chalcogen gas with a transition metal structure. Optionally, the transition metal structure could be patterned before the reaction, which would result in a similarly patterned chalcogenide structure. In one example, we demonstrated rapid fabrication of multi-atomic layer WSe2 with good material quality. Such chalcogenides have various applications for optoelectronic devices, such as solar cells and photoelectrochemical cells.

Description

[0001] Scalable method of fabricating high-quality transition metal chalcogenide materials by

[0002] Katie Neilson

[0003] Koosha Nassiri Nazi f

[0004] Alwin S . Daus

[0005] Eric Pop

[0006] Sarallah Hamtaei

[0007] FIELD OF THE INVENTION

[0008] This invention relates to improved fabrication of chalcogenide materials , especially semiconductor chalcogenides for optoelectronic devices .

[0009] BACKGROUND

[0010] Current fabrication methods of transition metal dichalcogenide ( TMD) solar cell are not scalable by any standard . These methods mostly rely on exfoliating small flakes from a bulk TMD crystal . There is a signi ficant body of work in the nanoelectronics industry on synthesis of high-quality monolayer ( ~ 0 . 7 nm thick) TMDs , but such methods cannot be used to produce high-quality multilayer ( 10-200 nm) TMDs needed for high-ef ficiency TMD solar cells .

[0011] SUMMARY

[0012] Our work ( synthesis method) addresses this issue by producing high-quality multilayer chalcogenides at scale . More specifically, we have developed a scalable massproduction-friendly synthesis method for high-quality, multilayer transition metal chalcogenides based on selenization / sulfurization of transition metals with JMSe or the like. These high-quality films could be used to produce high-efficiency solar cells.

[0013] Applications include high-specific-power ultrathin transition metal dichalcogenide (TMD) solar cells to be deployed in emerging technologies such as drones, low- earth-orbit satellites, electric vehicles, Internet of things (loT) and wearable electronics.

[0014] Significant advantages are provided. Higher material quality and scalability (ability to be mass-produced at low cost) , as well as ability to form multilayer structures (rather than monolayers, which is the main focus of the nanoelectronics industry) .

[0015] BRIEF DESCRIPTION OF THE DRAWINGS

[0016] FIG. 1 shows two exemplary embodiments of the invention .

[0017] FIGs. 2A-C show an example of selenization of W with H2Se to provide WSe2.

[0018] FIGs. 3A-C shows some materials characterization results for WSe2 made by selenization of W with H2Se.

[0019] FIG. 4A shows carrier lifetime results for these WSe2 layers at several different excitation wavelengths.

[0020] FIG. 4B show calculated dependence of power conversion efficiency on carrier lifetime for a solar cell having a WSe2 absorber that is 26 nm thick. DETAILED DESCRIPTION

[0021] Section A describes general principles relating to embodiments of the invention, and section B considers a detailed example .

[0022] A) General principles

[0023] FIG . 1 shows an exemplary embodiment of the invention . The top shows the result of depositing a transition metal structure 104 on a substrate 102 . In the simplest version of the process ( left path) , the next step is flowing a chalcogen gas over the transition metal structure 104 such that a chemical reaction between the chalcogen gas and the transition metal structure 104 occurs to form a transition metal chalcogenide structure 106 . Here the chalcogen gas includes one or more compounds of sul fur, selenium or tellurium .

[0024] In the version shown on the right path of FIG . 1 , transition metal layer 104 is patterned to provide a patterned transition metal structure 104a, which leads to a corresponding patterned transition metal chalcogenide structure 106a . In this version of the process , depositing a transition metal structure can be regarded as having two steps : depositing the layer and then patterning that layer .

[0025] Suitable chalcogenide gases include , but are not limited to : hydrogen sul fide , dimethyl sul fide , diethyl sul fide , C4H4S , hydrogen selenide , dimethyl selenide , diethyl selenide , C4H4Se , chalcogen amides , and mixtures thereof .

[0026] The chemical reaction can be performed at one or more temperatures in a range from 300C to 900C and at one or more pressures in a range from 5 mbar to 2000 mbar . The chemical reaction can be performed for a duration in a range from 2 minutes to 1 hour Notably, this is significantly faster than some reports in the literature where thick-layer chalcogenide synthesis can take many hours .

[0027] The chemical reaction can be catalyzed. Suitable catalysts include, but are not limited to: KOH, NaOH, NaCl, KI, NaF, RbF, CsF, alkali hydroxides and alkali halides. If used, the catalyst can be provided in various ways, such as: including the catalyst in the transition metal structure, including the catalyst in the chalcogen gas, and incorporating the catalyst into the transition metal chalcogenide structure in a post-deposition process.

[0028] Suitable transition metal chalcogenide materials for this work include, but are not limited to: WSe2, WS2, MoSe2, M0S2, InSe, InS, and PtSe2. Preferably, the transition metal chalcogenide structure includes a transition metal chalcogenide having a layered or van der Waals microstructure .

[0029] The transition metal chalcogenide structure can be part or all of a solar cell or a photoelectrochemical cell.

[0030] Bl) Introduction

[0031] Semiconducting transition metal dichalcogenides (TMDs) , e.g., M0S2 and WSe2, offer excellent electronic and optical properties for use in a variety of applications from nanoelectronics to photovoltaics . These include good charge carrier mobility in atomically thin (sub-l-nm) layers, ultrahigh optical absorption coefficients, nearideal band gaps for solar energy harvesting, surfaces without dangling bonds, and facile integration on rigid and flexible substrates. In particular, TMDs hold great promise in high-specific-power (i.e. high-power-per-weight ) applications, e.g., energy harvesting in high-altitude drones and wearable electronics, where light weight and high power output are desired. These emerging photovoltaic markets are growing at a rapid pace, potentially exceeding $100B by 2027.

[0032] Recent studies have shown that optimally-designed thin (<100 nm) TMD solar cells made of sulfides or selenides could achieve ~25% power conversion efficiency (PCE) in a single- j unction structure and ~32% PCE in a TMD-silicon tandem structure, on par with current perovskite-silicon tandem solar cells. As a result, single- j unction TMD solar cells can offer excellent specific power (power-per-weight ) greater than 40 W g-1, about 10x higher than established thin-film solar cell technologies cadmium telluride (CdTe) , copper indium gallium selenide (CIGS) , amorphous silicon (a-Si) , and III-Vs. In addition, while thin-film solar cell technologies face challenges such as high cost (III-V single and multi- j unction cells) , low durability (perovskite, organic, dye-sensitized photovoltaics ) , and the use of rare (CIGS) or toxic materials (CdTe, PbS quantum dot, lead-based perovskite) , TMD solar cells provide a potentially low-cost, durable, eco-friendly, and bio-compatible solution which could reach future large- scale adoption.

[0033] The majority of TMD synthesis research over the past decade has been focused on monolayer (< 1 nm) films, particularly for use in next-generation transistors. In contrast, high-quality TMD multilayers are most often obtained in a non-scalable fashion through the mechanical exfoliation of micron-scale flakes from bulk crystals grown by chemical vapor transport (CVT) . Selenization, sulfurization, or tellurization of metallic precursor (s) could be employed as a low-cost method to synthesize TMD films at industrial scale, similar to CIGS. In this approach, the substrate size is, in principle, limited only by the furnace dimensions and can be easily scaled to larger areas with the use of industrial tools.

[0034] In practice, while sulfurization has achieved more uniform films, selenization and tellurization have proven more difficult. This has been mainly attributed to a lower chemical reactivity of selenium and tellurium, which requires the presence of a reducing agent when growing selenide and telluride films. While synthesis by tellurization has made progress, the band gaps of semiconducting tellurium-based TMDs are smaller (~0.7 eV) and less optimal for solar applications than those of selenides. On the other hand, previous studies on selenization synthesis of WSe2 films have displayed suboptimal properties for photovoltaics , being either too thin, too rough, and / or suffering from morphological inhomogeneity. The selenization process itself could also exceed 20 hours, rendering it unfavourable from an industrial perspective. An ideal film growth would have controllable thickness that is atomically smooth and oriented in the (001) direction, with sufficiently large grain sizes and good charge carrier lifetimes.

[0035] Herein, we demonstrate the scalable synthesis of multi-atomic layer (15-30 nm) WSe2 on up to 150 mm wafers by selenization of sputtered tungsten films using (for the first time) low-thermal-budget H2Se precursors at 650 °C, in 1-2 hours, leading to optimal characteristics for use as an absorber layer in ultrathin high-specific-power solar cells. At such thicknesses, WSe2 has a band gap of 1.2- 1.3 eV, which is near-ideal for solar energy harvesting under both AMO and AMI .5G spectra. Additionally, our approach results in films with superior characteristics compared to previous selenization reports, including van der Waals (vdW) layered structure, smooth and uniform surfaces, grain sizes comparable to the film thickness, Hall mobilities up to 8 cm2V1s-1, and minority carrier lifetimes up to 100 ns, over 10x greater than previous reports on large-area TMD films prepared by liquid exfoliation. Here, we specifically focus on the absorber material due to the separate challenges associated with solar cell design; however, with improved contacts and doping, these WSe2 films can provide efficiency up to 21.7% in an optimized solar cell design. More importantly, upon integration on ultrathin flexible substrates, such solar cells are calculated to provide specific power of ~62 W g-1in a packaged cell, over 10x higher than established thin- film solar cell technologies. Fully packaged modules, which include considerations such as interconnect weight and encapsulation, show specific power of up to 3 W g-1, approximately 5x higher than the record of 0.7 W g-1, demonstrated in a multi- j unction III-V-based solar module.

[0036] B2) WSe2 Growth by Selenization of Tungsten

[0037] To prepare the WSe2 film, we sputter 5-10 nm of tungsten using a DC source (Kurt J. Lesker) at 50 W onto ~500 pm-thick silicon substrates at a pressure of 10 mTorr and an argon (Ar) flow rate of 50 seem (202 on FIG. 2A) to provide an optimal film porosity. These films can either be pre-patterned (via lithography and lift-off) or blanket- deposited. The tungsten films are then selenized (204 on FIG. 2A) into multilayer (15-30 nm) WSe2 using hydrogen selenide (JASe) at 650 °C, as shown in FIGs. 2B-C (and 206 on FIG. 2A) . The final film thickness is mainly controlled by the starting thickness of the sputtered W, with the caveat that very thin starting films (< a few nm) require significantly shorter growth times to remain as continuous films versus as islands. The growth rate is relatively consistent after a W thickness of a few nm, for a final film thickness of approximately 2.1 times the starting thickness. J Se provides a potentially more homogenous selenization and higher chemical reactivity, therefore lowering the synthesis temperature and thermal budget.

[0038] For the growth with JASe, a rapid thermal annealing tool is employed with N2 and JASe as process gases (FIG. 2B) . The sample faces up on a silicon carbide (SiC)- coated graphite susceptor. Other features of this tool include cold wall chamber 212, gas inlet 214 and halogen lamps 216. The growth is done by a two-stage anneal (FIG. 2C) ; the chamber is filled with 488 Torr of N2 base pressure, which is then heated up to 400 °C. The sample is then annealed for 20 minutes, with 75 Torr of diluted JhSe gradually fed into the chamber during the first 4 minutes, at 20 seem. The flow rate is chosen to provide sufficient carrier gas, while allowing for the films to remain planar on the substrate (versus island formation or delamination in situ) . The chamber is subsequently heated to 650 °C, where the sample is held for 10 minutes before cooling to room temperature. Immediately after the two annealing stages (at 400 °C and 650 °C) , the chamber pressure is 600 and 638 Torr, respectively. All heat-up and cool-down ramps are set to 0.5 °C / s.

[0039] Selenization of W leads to multilayer WSe2 films whose thickness can be tuned by the amount of sputtered W. The substrate size is only limited by the furnace dimensions, which in our study is 150 mm for gas-source selenization. Industrialized furnaces would enable growth on even larger substrates, i.e. 300 mm or greater. The WSe2 films are highly uniform, with van der Waals structure, and can be easily transferred to other rigid or flexible substrates, e.g. to build flexible, high-specific-power solar cells. For example, we have transferred a pre-patterned WSe2 film onto a flexible, lightweight polyimide (PI) substrate, and we have spectroscopically demonstrated growth uniformity on a 150 mm wafer using the (E12g) Raman peak of WSe2 normalized to the silicon substrate below.

[0040] B3) Material Characterization

[0041] Several characterization methods were used to assess the quality of the grown WSe2 films. Initial scanning electron microscopy (SEM) images (FIG. 3A) show smooth WSe2 grown with JRSe. Energy dispersive X-ray spectroscopy (EDS) is used to determine the Se to W ratio of the films either as-grown on silicon or after transfer to PI. EDS spectra are taken under 5 kV acceleration voltage, to ensure data acquisition at the surface. The measurement is first done at the edge of the chip, as the smoothness of the films makes it difficult to focus near the center, followed by two more measurements at randomly chosen and wide apart locations. The EDS data from WSe2 on PI substrates indicate high spatial homogeneity, with a coefficient of variation below 3% for Se and W atomic percentages.

[0042] A transmission electron microscopy (TEM) study showed that WSe2 films prepared with JRSe have average grain sizes of ~8 nm (FIG. 3B) . These JRSe films show in-plane XRD peaks, and it is expected that further optimization of growth parameters (such as growth promoters, carrier gas, or pressure optimization) could provide more ideal film orientation, and larger grains, as the partial pressure of Se during growth can determine the film orientation. We also display Raman spectra in FIG. 3C, normalized to the Si peak of the substrate. Here "H2Se" refer to selenization as described above with H2Se, and "SS-Se" refers to a selenization process carried out with an elemental Se source to provide samples for comparison. These provide information about the vibrational modes, which correlate with the relative quality of the films. Solid source grown films exhibit a distinct degenerate Aig / EX2g peak and the 2LA(M) signature. H2Se films show an increased left shoulder to the E22gpeak, which could be due to a higher presence of defects in these films or to degenerate peak splitting due to an increased presence of strain, as indicated by the shifted out of plane (00O peaks in the x-ray diffraction spectra.

[0043] Square WSe2 samples measured by the van der Pauw method are used to determine the majority carrier type, density, and their Hall mobility. The films show p-type doping, with hole density ~1017cm3and average Hall mobility of ~5 cm“2V1s“1at room temperature. Across various samples, we measured Hall mobilities and hole densities up to 8 cm2V_1s-1and 4.2 x lO17cm-3, respectively. These are among the best mobility results reported to date for selenized WSe2. SS-Se samples exhibit a tight distribution of Hall mobility and hole density whereas H2Se samples show wider variation. This could be explained by the mobility anisotropy in WSe2 and the more varied orientation of WSe2 layers observed in the H2Se growths causing sample-to-sample variation. In contrast, we note that SS-Se growths have better horizontal layering within the WSe2 films.

[0044] B4) Flash-photolysis time-resolved microwave conductivity

[0045] (TRMC) measurement The flash-photolysis time-resolved microwave conductivity (TRMC) method is adopted to determine the charge carrier lifetime in our selenized films, particularly as this dictates the efficiency limit in photovoltaic applications. TRMC is a contactless microwave spectroscopy technique where a thin film sample is excited by a laser inside a resonant cavity, probed by microwave radiation, and measured as a function of time. TRMC experiments provide the product of free-charge carrier yield and the sum of both electron and hole mobilities, known as the yield-mobility product <> + / Ai), where is the yield of charges per photon, whileeand h are the electron and hole mobility, respectively. Both electrons and holes are produced by photoexcitation, but the degree to which carriers experience trap sites (and the precise energies of those traps) is unclear to date, in TMDs . Thus, it is challenging to deconvolve to estimate the photogenerated carrier mobilities. Instead, we focus on the carrier lifetimes.

[0046] To characterize TMD photoconductivity, we first examine the optical absorption and steady-state microwave conductivity (SSMC) action spectra associated with films for which we could measure a sufficiently good response at low photon fluences. The optical absorption and SSMC action spectra provide the three excitation wavelengths (at 500, 580, and 780 nm) for the TRMC study, as follows: 1) bulk transitions above the WSe2 band gap (500 nm or 2.48 eV) , 2) transitions into the B exciton (580 nm or 2.14 eV) , and 3) transitions into the lowest A exciton (780 nm or 1.59 eV) . FIG. 4A displays carrier lifetimes in WSe2 films at a fluence of 109cur2, representing AM 1.5 G solar illumination. Selenization on textured substrates may allow for even larger grain sizes and subsequently larger lifetimes .

[0047] B5) Projected efficiency and specific power of WSe2 solar cells and modules

[0048] The ultimate performance limit of a single- unction solar cell is governed by the optoelectronic characteristics and synthesis quality of its absorber material. We perform a realistic modeling of singlejunction solar cells performance limits with WSe2 absorber layers by accounting for both intrinsic and extrinsic properties. The model incorporates measured absorption data as well as radiative, Auger, and defect-assisted Shockley- Read-Hall (SRH) recombination.

[0049] FIG. 4B shows the simulated current density vs. voltage and power conversion efficiency (PCE) limit of single- j unction solar cells with 26 nm-thick WSe2 absorber layers at various film quality levels, as represented by SRH recombination lifetimes (TSRH) . This thickness is chosen here because it is that of our selenized films, and is the thickness which maximizes PCE for lifetimes in the 10 ns to 1 ps range. Our model shows that TSRH of 10 ns, 100 ns, and 1 ps lead to PCE limits of 18.2%, 21.7%, and 25.1%, respectively. In the absence of SRH recombination (defect- free WSe2, TSRH =00) , these 26 nm thin WSe2 solar cells reach a PCE of 29.5% at the Tiedj e-Yablonovitch limit (or up to ~30.8% in films thicker than 100 nm8) . Carrier lifetimes associated with radiative and Auger recombination in intrinsic or lightly-doped multilayer WSe2 are on the order of tens of micro-seconds. Thus, measured carrier lifetimes below 1 ps correspond to SRH dominating recombination, which is the case for our selenized WSe2 films (FIG. 4A) . According to our realistic model, the selenized WSe2 films in this study, which exhibit TSRH up to 100 ns, can achieve a PCE of 21.7% upon design optimization, which is on par with incumbent solar technologies such as Si, CdTe, and CIGS .

[0050] More notably, after integration on ultrathin flexible substrates, selenized WSe2 solar cells are expected to provide specific power of ~62 W g-1in a packaged cell, over 10x higher than established thin-film solar cell technologies such as III-Vs, CdTe and CIGS. These packaged WSe2 solar cells could be adopted in size-constrained, low- power applications such as Internet of Things (loT) and wearable electronics. Higher-power applications such as satellites and electric vehicles require larger, fully- packaged modules, with higher areal weight densities due to additional interconnects and module encapsulation. Such fully-packaged WSe2 modules may have specific power of up to 3 W g-1, over 4x higher than the present record of 0.7 W g-1in multi- j unction III-V solar modules.

[0051] B6) Conclusion

[0052] We report scalable methods for large-area WSe2 film synthesis with gaseous precursors, for use as absorber layers in ultrathin high-specific-power solar cells. The films show van der Waals layered structure and grain sizes on the order of the film thickness (17 nm) . We report some of the highest Hall mobilities to date, up to 8 cm2V1s-1, in uniform, wafer-scale, multilayer films. These films also have charge carrier lifetimes up to 100 ns. Simulations show that such lifetimes can lead to solar cell efficiency of ~21.7%, and specific power of ~62 W g-1in a packaged cell or 3 W g-1in a fully-packaged module. These results pave the way for large-scale, commercial TMD solar cells with high specific power, which could address the growing need for lightweight, flexible energy harvesters for applications in loT, wearables, electric vehicles, and aerospace.

[0053] B7) Methods

[0054] Growth -- Gas source selenization was done in a coldwall rapid thermal processing furnace (Annealsys AS- One 150) , capable of housing 150 mm wafers. The heat is provided by infrared halogen lamps on the chamber' s lid, and the temperature is monitored using a thermocouple attached to the susceptor. Processing gases were nitrogen, and 10% diluted JASe in N2.

Claims

CLAIMS1. A method of making a transition metal chalcogenide structure, the method comprising: depositing a transition metal structure on a substrate ; flowing a chalcogen gas over the transition metal structure such that a chemical reaction between the chalcogen gas and the transition metal structure occurs to form the transition metal chalcogenide structure, wherein the chalcogen gas includes one or more compounds of sulfur, selenium or tellurium.

2. The method of claim 1, wherein the depositing a transition metal structure includes: depositing a transition metal layer; and patterning the transition metal layer to form the transition metal structure.

3. The method of claim 1, wherein the chalcogenide gas is selected from the group consisting of: hydrogen sulfide, dimethyl sulfide, diethyl sulfide, C4H4S, hydrogen selenide, dimethyl selenide, diethyl selenide, C4H4Se, chalcogen amides, and mixtures thereof.

4. The method of claim 1, wherein the chemical reaction is performed at one or more temperatures in a range from 300C to 900C.

5. The method of claim 1, wherein the chemical reaction is performed at one or more pressures in a range from 5 mbar to 2000 mbar.

6. The method of claim 1, wherein the chemical reaction is performed for a duration in a range from 2 minutes to 1 hour .

7. The method of claim 1, further comprising catalyzing the chemical reaction.

8. The method of claim 7, wherein the catalyst is selected from the group consisting of: KOH, NaOH, NaCl, KI, NaF, RbF, CsF, alkali hydroxides and alkali halides.

9. The method of claim 7, wherein the catalyst is provided to the chemical reaction by a method selected from the group consisting of: including the catalyst in the transition metal structure, including the catalyst in the chalcogen gas, and incorporating the catalyst into the transition metal chalcogenide structure in a postdeposition process.

10. The method of claim 1, wherein the transition metal chalcogenide structure includes one or more materials selected from the group consisting of: WSe2, WS2, MoSe2, M0S2, InSe, InS, and PtSe2.

11. The method of claim 1, wherein the transition metal chalcogenide structure includes a transition metalchalcogenide having a layered or van der Waals microstructure .12 . The method of claim 1 , wherein the transition metal chalcogenide structure is part or all of a solar cell or a photoelectrochemical cell .

Images