SEARCH

SEARCH BY CITATION

Keywords:

  • copper;
  • p-type;
  • transparent conductor;
  • transparent diode;
  • zinc sulfide

Abstract

  1. Top of page
  2. Abstract
  3. 1 Introduction
  4. 2 Theory
  5. 3 Experimental
  6. 4 Results and discussion
  7. 5 Conclusions
  8. Acknowledgements
Thumbnail image of graphical abstract

Copper alloyed ZnS was investigated as a p-type, transparent conducting material composed of earth-abundant elements. Thin films of Cu-alloyed ZnS were synthesized using pulsed laser deposition with Cu contents in the range of x = 0.06–0.27 (Cu content x is reported as the fraction of cation present). Thermopower and Hall effect measurements show that the films are p-type. We find that transparency and conductivity are comparable to some of the best reported p-type materials with our best films exhibiting conductivities of 54 S cm−1 and optical transmission of 65% at 550. The hole conduction mechanism is discussed in terms of possible Cu acceptor doping of the ZnS and hole conductivity in a minority Cu2S phase. Transparent rectifying p-CuZnS/n-ZnO diodes were fabricated.

Photo demonstrating the high transparency of a rectifying heterojunction diode made with p-type Cu-alloyed ZnS and n-type ZnO.


1 Introduction

  1. Top of page
  2. Abstract
  3. 1 Introduction
  4. 2 Theory
  5. 3 Experimental
  6. 4 Results and discussion
  7. 5 Conclusions
  8. Acknowledgements

Transparent conducting materials (TCMs) are essential components of electronic devices such as thin film photovoltaics (PV) and light emitting diodes (LED). At present, the predominant material used for transparent top contacts in PV is tin-doped indium oxide (ITO). Scaling PV to the terawatt levels may not be possible with ITO owing to the scarcity and cost constraints imposed by the presence of In as a main component in such films. For this reason, alternatives to ITO are being developed such as Al-doped ZnO 1, metal nanowire meshes 2, and graphene 3. As with ITO, however, these are typically n-type; materials that possess both high optical transparency and high conductivity are almost universally n-type 4. Hole conductors (p-type) of similar quality could open a new design space for optoelectronic devices and transparent electronics and therefore have been of great scientific interest 5. In the area of photovoltaics, p-type TCMs have the potential to facilitate the cost effective synthesis of heterostructured thin film devices with intrinsically n-type materials. In another application, p-type TCMs could improve the performance of multi-junction solar cells by serving to collect photogenerated charges laterally from each absorber layer independently, improving the overall efficiency. In the area of consumer electronics, better p-type TCMs would improve the performance of transparent thin film transistors which are required for high performance transparent display technologies 6.

At present, there are only a few p-type TCMs with performance comparable to n-type materials 7, 8. Some of the best performing materials have the delafossite structure and include CuAlS2 with a conductivity of 63.5 S cm−1 and transmittance in the visible spectrum of 80% 9 and BaCu2S2 with a conductivity of 17 S cm−1 and transparency of 90% at a wavelength of 650 nm 10. The motivation for this investigation is to expand the range of available p-type TCMs, both in terms of performance and in crystal structure.

Although examined for its properties as a potential phosphor material 11, the electrical transport properties of Cu-alloyed ZnS thin films have largely been uninvestigated. ZnS has two stable crystalline phases at room temperature: wurtzite (α phase) and sphalerite. The α phase was chosen for this investigation as it can act as a template for the subsequent synthesis and investigation of chalcogenide absorber materials for PV and there currently exist no p-type wurtzite TCMs 12. In addition, α-ZnS has acceptable lattice matching with a number of n-type absorbers such as WSe2 which has a direct bandgap of 1.46 eV 13 and MoSe2 with a direct bandgap of 1.43 eV 14. The α phase of ZnS has a band gap of 3.7 eV making it completely transparent in the optical regime of interest for PV 15; however, intrinsic ZnS is a band insulator with a low carrier density and is thus unsuitable as a contact. Both n- and p-type extrinsic doping have been investigated with ZnS. For n-type doping, Al has been reported to yield a conductivity of 10−3 S cm−1 and transmittance of 75% in the visible spectrum for the sphalerite phase 16. A recent theory investigation has suggested dual-donor codoping with Sn and F could be favorable 17. P-type doping has been reported with Li and N leading conductivities of 17.2 and 38.4 S cm−1, respectively; this, however, is once again for the sphalerite phase of ZnS. Optical transparency could not be investigated in these studies because of the use of opaque GaAs substrates 18, 19. The prospects of Cu2S as a p-type TCM have also been investigated; hole conductivities of 1500 S cm−1 can be achieved 20. However, due to the strong absorption of Cu2S (Eg = 1.2 eV), obtaining >50% transmission in the visible requires films of 10 nm or less in thickness.

From simple valence considerations Cu1+ should be a suitable substitutional acceptor when replacing Zn2+ in ZnS. Indeed, Zn1−xCuxS solid solutions have been reported to generate H2 under illumination in an aqueous solution for x ∼ 4%, implying p-type behavior 21. In this work, we investigated the hole conductivity limits in Cu-alloyed ZnS. Our first principle calculations show that degenerate hole concentrations should be achievable under sulfur-rich synthesis conditions. The formation of Cu-alloyed ZnS is thermodynamically unfavorable but non-equilibrium phases have been well known to be synthesizable by a host of techniques such as sputtering, molecular beam epitaxy, and pulsed laser deposition (PLD) 22. Using PLD we synthesized thin films of Cu-alloyed ZnS with a Cu content (reported as a fraction of the total metal content in the film, x) varying between 0.06 and 0.27 and under these conditions found excellent hole conductivity (up to 54.4 S cm−1) and good transparency across the visible portion of the solar spectrum.

2 Theory

  1. Top of page
  2. Abstract
  3. 1 Introduction
  4. 2 Theory
  5. 3 Experimental
  6. 4 Results and discussion
  7. 5 Conclusions
  8. Acknowledgements

The possibility of forming p-type conductivity in ZnS through the introduction of Cu substitutional acceptors was examined theoretically in order to evaluate the validity of a subsequent experimental investigation.

2.1 Optical properties

In order to determine the optical properties of Cu-alloyed ZnS, the density of states for the ZnS system for 12.5% and 18.75% Cu incorporation was calculated (Fig. 1(a)). We see the Cu 3d orbitals form new states above the valence band maximum (VBM) leading to a slight narrowing of the band gap while remaining clear of any mid-gap states which could reduce the material's transparency in the visible spectrum.

thumbnail image

Figure 1. (online color at: www.pss-a.com) (a) Graph of the ab initio density of states calculations of CuxZn1−xS with x = 0, 0.125, and 0.1875. The primary contribution to the density of states from the Cu incorporation (magenta lines) is at the valence band edge, implying a role for these states in the observed hole conduction. (b and c) The calculated formation energy change of CuZn antisite, S vacancy VS and Zn interstitial Zni as the Fermi energy shifts from the valence band maximum (0 eV) to the conduction band minimum (3.68 eV), where only the line of the lowest energy charge state for each element is plotted. (b) Plot for S rich, Zn poor and Cu rich conditions and (c) plot for S poor, Zn rich and Cu rich conditions.

Download figure to PowerPoint

2.2 Electronic properties

The limit to the p-type conductivity of heavily Cu-alloyed ZnS will be the formation of compensating donor defects. Cu-doping in ZnS should form equation image antisites (Cu1+ substituting for Zn2+), which are acceptors and produce holes. These holes may be compensated by intrinsic donor defects such as Zn interstitials equation image and S vacancies equation image. In Kroger–Vink notation, the compensation reactions are:

  • $$2{\rm Cu'}_{{\rm Zn}} {\rm } + {\rm Zn}_{{\rm i}}^{ \bullet \bullet } = {\rm null}$$, ((1))
  • $$2{\rm Cu'}_{{\rm Zn}} + {\rm V}_{{\rm S}}^{{\rm \bullet \bullet }} = {\rm null}$$. ((2))

If (1) and (2) are energetically favorable, the holes released by equation image antisites will be “killed” by the hole killers equation image and equation image, thus no p-type conductivity can be obtained. This situation exists in ZnO, and it is thought to explain the associated difficulty with p-type doping. In ZnO, density functional theory calculations have shown that the formation energy of equation image or equation image in ZnO decreases to negative values and becomes lower than that of equation image as the Fermi energy approaches the VBM 23–25. This means the formation of the hole “killers” and the accompanying compensation reactions will occur spontaneously, limiting the Fermi level far above the VBM and making it difficult to obtain p-type ZnO. Here we show that a completely different situation exists for ZnS, and this type of compensation can be avoided under S rich conditions.

As seen in Fig. 1b, the two donor defects equation image and equation image are in their neutral state (q = 0) when the Fermi energy is high, near the conduction band minimum (CBM), and change to the q = +2 charged state which has a decreasing formation energy as the Fermi energy shifts down from the CBM to the VBM. Under the S poor and Zn rich growth condition (Fig. 1c, where the Zn chemical potential is taken from Zn metal), the formation energy of equation image drops to negative values when the Fermi energy is at the VBM, 2 eV lower than that of the equation image antisite. Therefore, the spontaneous formation of q = +2 charged Zni will compensate all free holes, making p-type doping unobtainable under this condition. However, under the S rich and Zn poor growth condition (where the S chemical potential is taken from sulfur-6 crystal), the formation energy of equation image and equation image increases significantly relative to that of equation image, and when the Fermi energy is at the VBM, equation image is about 1 eV lower than equation image. Therefore, it is expected that heavy Cu-doping under these conditions should lead to p-type behavior. Although the results shown in Fig. 1b are only for the impurity limit, they show qualitatively that it is much easier to form a p-type ZnS alloy than ZnO, probably because the VBM level of ZnS is much higher than that of ZnO.

2.3 Calculation methods

Density of states calculations were computed by Vienna Ab initio Simulation Package (VASP) using local density approximation (LDA). Material compositions of CuxZn1−xS where x = 0.125 and x = 0.1875 were calculated using 2 Cu atoms and 3 Cu atoms in a 2 × 2 × 2 ZnS wurtzite supercell, respectively.

The change in formation energy in our study was calculated using the supercell method 23–27 where one defect α in the charge state q is placed in a ZnS supercell, and its formation energy equation image is calculated according to

  • equation image((3))

where equation image and equation image is the calculated total energy of the supercell with and without a defect respectively, equation image is the change of the number of the atom i (i = Cu, Zn, or S) in the supercell caused by the formation of the defect α, equation image is the energy per atom of the element i in its bulk phase (Cu and Zn in the face centered cubic structure, and S in the sulfur-6 structure), equation image is the chemical potential of the element i relative to its bulk phase (equation image = 0 means the element i is as rich as in the pure bulk phase), and equation image is the Fermi energy relative to the valence band maximum equation image of ZnS. For equation image of charged defects, the first-order Makov–Payne correction is taken for the spurious electrostatic interaction caused by the limited size of the supercell, as discussed in Ref. 22.

For our formation energy calculations S rich and Zn poor conditions mean equation image = 0 eV, and correspondingly equation image = equation image = −2.0 eV in order to make ZnS thermodynamically stable. Zn rich and S poor conditions mean equation image = 0 eV and equation image = equation image. For the dopant element Cu, we assume it can be as rich as possible provided that the secondary phases CuS and Cu2S cannot be formed, which means equation image = −0.55 eV when S is rich (equation image= 0 eV), and equation image<0 eV when S is poor (equation image = equation image).

The calculated formation energy of defects in semiconductors differs significantly when different approximations to the exchange-correlation functional are used 22–24. To avoid the band gap errors and their influence on the formation energy as a result of using the generalized gradient approximation (GGA), we used the Heyd–Scuseria–Ernzerhof (HSE) hybrid functional for our formation energy calculations. In addition, GGA calculations were performed to compare the results from two different approximations. Both GGA and HSE calculations show that Cu-doping in ZnS gives rise to p-type conductivity under the S rich and Zn poor condition, while p-type doping is unobtainable under the S poor and Zn rich condition.

3 Experimental

  1. Top of page
  2. Abstract
  3. 1 Introduction
  4. 2 Theory
  5. 3 Experimental
  6. 4 Results and discussion
  7. 5 Conclusions
  8. Acknowledgements

3.1 Sample growth

Cu-alloyed ZnS (x = 0.06, 0.10, 0.16, 0.21, 0.26, see below) thin films (40–800 nm) were deposited by pulsed laser deposition. One inch diameter targets were made by cold isostatic pressing of Cu2S and sphalerite ZnS powders with a Cu/(Cu + Zn) atomic ratio of 20%. The powders were obtained from Sigma–Aldrich and used as received. The targets were placed 6 cm away from 5 mm × 5 mm Al2O3 (0001) double side polished substrates. A rectangular mask and UV transparent lens with a focal length of 16.5 cm was used to focus the beam to a 10 mm2 rectangular spot on the target surface. The laser source was a pulsed KrF excimer laser operating at 248 nm maintained at a 30 ns pulse width with repetition rate varying between 5 and 10 Hz. The energy per pulse was varied for the depositions from 75 to 250 mJ per pulse while the substrate temperature was held at 550 °C.

The Cu content in the deposited films was found to be reproducibly dependent on the laser fluence, with higher fluences producing greater Cu incorporation. The higher volatility of S species relative to Cu and Zn in our target led to the desired S-rich deposition conditions and subsequently S-rich films. Films were deposited with a substrate temperature of 550 °C. At lower temperatures, the crystallinity of the films was poor, while at higher temperatures the films were found to be substantially S deficient due to S outgassing. Intrinsic ZnS control films were deposited in the same manner from a ZnS target.

3.2 Characterization

Normal θ–2θ X-ray diffraction (Panalytical X'Pert Pro Diffractometer) was used for phase identification. Profilometry (Veeco Dektak IIA) and atomic force microscopy (Veeco-DI equipped with a Nanoscope IV Controller and silicon tip from MikroMasch) were used to determine film thicknesses and surface roughness, respectively. X-ray photoelectron spectroscopy (PHI 5400 XPS system) was used to determine the bonding environment and oxidation state of the introduced Cu species. Energy dispersive spectrometry (EDS) in a scanning electron microscope (JEOL JSM-6490LV equipped with Oxford INCAx-Sight) was utilized to determine film composition. Optical transmission characteristics were determined using a Hitachi U-3000 Spectrophotometer. Resistivity, mobility, and carrier concentration of the thin film samples were determined using the Van der Pauw contact configuration with the Ecopia HMS-3000 Hall Measurement System. Raman measurements were performed with a Horiba LabRam HR microprobe using 532 and 632 nm laser excitation.

4 Results and discussion

  1. Top of page
  2. Abstract
  3. 1 Introduction
  4. 2 Theory
  5. 3 Experimental
  6. 4 Results and discussion
  7. 5 Conclusions
  8. Acknowledgements

Film compositions were determined using EDS. The film composition is reported as CuxZn1−xSy where x represents the content of Cu as a fraction of the total available cation in the sample. The S content in the conducting films was found to be approximately 50 at. %. The presence of Cu2S as a minority phase in the films is discussed in the following sections.

The diffraction pattern obtained from a representative x = 0.21 Cu-alloyed ZnS thin film grown on sapphire (001) along with the reference pattern from a undoped ZnS film is shown in Fig. 2. The XRD pattern of the target is also included for reference. In both ZnS and CuZnS films the diffraction pattern shows the expected peak for the wurtzite phase. The existence of sphalerite ZnS in the films (the form of ZnS in the target) was precluded by performing azimuthal ϕ-scans. No pattern corresponding to any phase of Cu2S was observed in the films (compare the reference pattern from β-Cu2S and γ-Cu2S found in the target). In the normal θ–2θ scans, we observed no significant shift in the (002) peak between α-ZnS and Cu-alloyed ZnS. If Cu is substituting on the Zn site we would not expect a large lattice constant shift as the crystal radii of Cu in the +1 valence state and Zn in the +2 valence state with IV coordination are both 74 pm 28. The out-of-plane lattice parameter of both films (obtained from the 2θ positions) is 6.21 Å, indicating a relaxed film. Atomic force microscopy revealed that the films have an RMS roughness of ∼2.5 nm as seen in Fig. 3.

thumbnail image

Figure 2. (online color at: www.pss-a.com) θ–2θ patterns showing the presence of the wurtzite crystal phase for ZnS and ZnS with 21% Cu doping (a). Patterns from the pulsed laser deposition target (sphalerite ZnS and Cu2S) are shown in (b) for reference. The ZnS feature in the films appears at the expected angle for α-ZnS. No features corresponding to Cu2S phases are found in the thin films.

Download figure to PowerPoint

thumbnail image

Figure 3. (online color at: www.pss-a.com) Atomic force microscopy image of the surface topology of a 100 nm thick Cu0.21Zn0.79S thin film. The RMS roughness is 2.5 nm.

Download figure to PowerPoint

X-ray photoelectron spectroscopy (XPS) was used to investigate the Cu oxidation state (Fig. 4). The binding energies of the observed photoelectron peaks of Cu 2p1/2 and 2p3/2 were 932.5 and 952.4 eV, respectively and no satellite or “shake up” peaks were found in the higher binding energy direction which is consistent with the standard Cu2S reference XPS spectrum for Cu1+. We find a small negative shift in the binding energies for the Zn 2p spectra of CuZnS when compared to α-ZnS (Fig. 4b). The lower binding energy is expected for Cu in ZnS as the Cu1+–Zn2+ interaction is weaker than that of Zn–Zn found in ZnS. This result suggests that some Cu may be substituting for Zn in the wurtzite lattice.

thumbnail image

Figure 4. (online color at: www.pss-a.com) XPS plots of the Cu 2p (a) and Zn 2p (b) binding energies. The Cu ions in the film were found to be in the +1 oxidation state. A small negative shift in the Zn 2p binding energy in the Cu alloyed ZnS films suggests that some Cu ions may be substituting for the Zn ions in the wurtzite ZnS lattice.

Download figure to PowerPoint

Raman spectroscopy revealed the presence of Cu2S in the films for the entire composition range. We note that with the laser excitation used (2.4 eV), scattering from Cu2S (Eg = 1.2 eV) is strongly favored compared to ZnS (Eg = 3.7 eV) due to resonance enhancement. This observation, combined with the absence of a Cu2S pattern in XRD, suggests that Cu2S is present as a minority phase, possibly in a disordered phase at the ZnS grain boundaries.

The presence of both Cu and sufficient S were required to produce a conducting film; both ZnS control films and S deficient Cu-alloyed films grown at higher deposition temperatures were highly insulating. All conducting Cu-alloyed ZnS films were p-type as confirmed with room temperature Hall and Seebeck measurements (Seebeck data shown in Fig. 5). As shown in Fig. 6a, conductivity increases with Cu content, reaching a maximum of 54.4 S cm−1 for x = 0.21 at room temperature.

thumbnail image

Figure 5. Room temperature Seebeck coefficient measurement of Cu0.21Zn0.79S. The Seebeck coefficient was found to be +11.6 µV K−1 indicating p-type conductivity.

Download figure to PowerPoint

thumbnail image

Figure 6. (online color at: www.pss-a.com) (a) Plot of film conductivity versus Cu content. The highest conductivity was found in films with 21% Cu content. (b) Transmission spectrum of Cu-alloyed ZnS thin films of varying thickness with 10% Cu content. Transmission as high as 70% across the visible spectrum is demonstrated for film thicknesses of 100 nm. See Table 1 for a summary of sheet resistance Rs and transmission as a function of thickness for films with a Cu content of x = 0.10.

Download figure to PowerPoint

The effect of film thickness on the visible transparency was investigated by holding the composition constant at a Cu content of x = 0.10 and varying thickness between 40 and 800 nm. Figure 6b shows that, as expected from our theoretical considerations, we observed good transparency across the visible spectrum with transmission as high as 70% for film thicknesses of 100 nm. The smoothly varying dependence of the film conductivity on thickness indicates that our films are isotropic and conduction is not occurring as a result of an Cu2S interface layer (Table 1).

Table 1. Summary of electrical and optical properties as a function of thickness for a series with a Cu content x = 0.10
film thickness (nm)T at 550 nm (%)σ (S cm−1)Rs (Ω/□)
409911.022,750
808118.26875
1007623.34300
2006524.42015
4004027.0925
8002430.3412

Absorption onset measurements were taken for Cu-alloyed ZnS thin films with differing Cu content (Fig. 7). As expected from our density of states calculations, the Cu defect states at the VBM appear to narrow the band gap as we see a shift in the position of the absorption onset to lower energies with increasing Cu content. Absorption in the visible spectrum becomes most noticeable in the films with a Cu content of x = 0.21 and above. This behavior may be associated with the absorption edge of Cu2S.

thumbnail image

Figure 7. (online color at: www.pss-a.com) Absorption onset measurements of CuxZn1−xSy with different Cu contents. We observe a shift of the onset of absorption to lower energies with increasing Cu content which is indicative of the states introduced at the VBM through Cu incorporation. The higher absorbance seen in films with 21% Cu content and higher may be associated with the Cu2S absorption edge.

Download figure to PowerPoint

In films with a Cu content of x = 0.21, we observe a transmission of 65% at 550 nm and a sheet resistance of 1900 Ω/□. As with all TCMs, we see a trade-off between transparency and conductivity and, therefore, the selection of suitable film thickness will be application dependent.

A heterojunction diode between the p-type CuZnS with x = 0.21 Cu content and undoped n-type ZnO was fabricated. Polycrystalline ZnO thin films were deposited in a similar fashion by PLD at a substrate temperature of 550 °C. The lattice mismatch between ZnS and ZnO is large at 15% but research on heterojunction nanowires has demonstrated good interfaces 29. Figure 8a shows the IV curve for the device. Rectifying behavior is demonstrated while maintaining a high optical transparency. From the IV characteristic of the device, the diode blocks up to a reverse bias of 6 V. The on–off ratio was found to be 75 at ±3 V. The sluggish rise of the IV behavior in the forward bias direction can be attributed to the high series resistance of the device coming predominantly from the poor conductivity of undoped ZnO. Nevertheless, our unoptimized diode demonstrates comparable performance to some other examples of transparent diodes 30, 31.

thumbnail image

Figure 8. (online color at: www.pss-a.com) (a) IV curve of Cu-alloyed ZnS-based diode structure. (b) Schematic of the diode structure. A 100 nm layer of Cu0.21Zn0.79S was grown on ZnO on an Al2O3 substrate. Al/Au contacts were deposited on both the ZnO and Cu0.21Zn0.79S layers. (c) Photo demonstrating the diode's high transparency.

Download figure to PowerPoint

5 Conclusions

  1. Top of page
  2. Abstract
  3. 1 Introduction
  4. 2 Theory
  5. 3 Experimental
  6. 4 Results and discussion
  7. 5 Conclusions
  8. Acknowledgements

A p-type wurtzite TCM was synthesized from earth abundant elements. The potential application of this TCM was demonstrated through its use in a rectifying transparent diode. By utilizing a wurtzite transparent p-type conductor, photovoltaic cells could potentially be manufactured using new types of materials. In addition, the development and optimization of such a p-type TCM could open new frontiers in the development of transparent transistors, light emitting diodes, electro-chromic windows, and a wide range of additional optoelectronic devices.

Acknowledgements

  1. Top of page
  2. Abstract
  3. 1 Introduction
  4. 2 Theory
  5. 3 Experimental
  6. 4 Results and discussion
  7. 5 Conclusions
  8. Acknowledgements

The authors thank Morgan Trassin, Ian Sharp, Ajay Yadev, Jayakanth Ravichandran, Dennis Meier, and John Heron for insightful scientific discussion and characterization work. Deposition work was performed in the Helios Solar Energy Research Center, which is supported by the Director, Office of Science, Office of Basic Energy Sciences, Materials Sciences and Engineering Division of the U.S. Department of Energy under Contract No. DE-AC02-05CH11231. Characterization work was performed through a collaboration with the Joint Center for Artificial Photosynthesis, a U.S. Department of Energy Innovation Hub, supported through the Office of Science of the U.S. Department of Energy under Award Number DE-SC0004993.