Room-temperature synthesized copper iodide thin film as ... - PNAS

2 downloads 0 Views 1MB Size Report
Nov 1, 2016 - high conductivity and transparency boosts the figure of merit of a ... copper iodide thin film | p-type transparent conductor | figure of merit |.

Room-temperature synthesized copper iodide thin film as degenerate p-type transparent conductor with a boosted figure of merit Chang Yanga,1, Max Kneiβa, Michael Lorenza, and Marius Grundmanna a

Institut für Experimentelle Physik II, Universität Leipzig, 04103 Leipzig, Germany

A degenerate p-type conduction of cuprous iodide (CuI) thin films is achieved at the iodine-rich growth condition, allowing for the record high room-temperature conductivity of ∼156 S/cm for asdeposited CuI and ∼283 S/cm for I-doped CuI. At the same time, the films appear clear and exhibit a high transmission of 60–85% in the visible spectral range. The realization of such simultaneously high conductivity and transparency boosts the figure of merit of a p-type TC: its value jumps from ∼200 to ∼17,000 MΩ−1. Polycrystalline CuI thin films were deposited at room temperature by reactive sputtering. Their electrical and optical properties are examined relative to other p-type transparent conductors. The transport properties of CuI thin films were investigated by temperature-dependent conductivity measurements, which reveal a semiconductor–metal transition depending on the iodine/argon ratio in the sputtering gas.

|

copper iodide thin film p-type transparent conductor reactive sputtering room-temperature growth

|

T

| figure of merit |

ransparent conductors (TCs) are well known for their wide use in passive electronic applications, such as transparent electrodes for solar cells, flat-panel displays, and light-emitting diodes. For active applications, the combination of p- and n-type transparent semiconductors contributes to the science and technology field of transparent electronics (1, 2). Commonly used TCs are n-type wide-bandgap semiconductors including In2O3, SnO2, ZnO, GaN, and their highly doped versions. However, p-type doping of n-type TCs is extremely difficult, which is primarily related to self-compensation in wide-bandgap semiconductors (3, 4). For example, as-prepared p-type ZnO usually shows poor conductivity and easily transforms into n-type conduction after several days or weeks. The only successful case is Mg-doped GaN exhibiting a stable p-type conduction, yet limited to epitaxial materials and a poor conductivity of ∼5 S/cm (5, 6). Hence, the search for high-performance p-type TCs has been a major challenge in recent decades. Since the discovery of highly transparent p-type CuAlO 2 − x thin films in 1997, much attention has been focused on CuI-based oxides such as delafossite CuMO2 (M: Al+3, Ga+3, In+3, Sc+3, Y+3) (7–11), Cu2SrO2 (12), and layered oxychalcogenide LnOCuCh (Ln: La+3, Pr+3, Nd+3, Sm+3, Gd+3, Y+3; Ch: S−2, Se−2) (13–17). To date, the highest conductivity of a p-type TC is ∼220 S/cm for Mg-doped CuCrO2, however at the cost of low visible transmittance down to ∼30% (18). Recently, perovskite Sr-doped LaCrO3 has been reported as a novel p-type TC, exhibiting a moderate conductivity of ∼54 S/cm; however, the transmittance of 42.3% is still fairly low (19). Results on p-type TCs from the literature are summarized in Table S1. When retaining a high optical transparency (transmittance T > 70%) in the visible spectral range (band gap Eg > 3.0 eV), the conductivities of p-type TCs are usually several orders of magnitude lower than their n-type counterparts, primarily due to the much lower mobility of valence-band–derived carriers (holes) compared with that of conduction-band–derived carriers (electrons). Another problem for these aforementioned oxide TCs is the generally high synthesis temperature (>400 °C) that is neither cost-effective nor www.pnas.org/cgi/doi/10.1073/pnas.1613643113

compatible in many device applications. Consequently, the lack of high-performance p-type TCs has been the main obstacle for both passive and active electronic applications, and a high-conductivity, high optical transmittance material with corresponding low-temperature synthesis techniques is desired for practical use. The first TC was zincblende copper iodide (CuI), which was discovered by Bädeker as early as 1907 (20, 21). It has a simple binary composition consisting of two abundant elements (Cu and I). The wide direct bandgap of CuI (3.1 eV at room temperature) allows for a high transparency in the visible spectral range. Moreover, its exciton binding energy (62 meV) is similar to that of ZnO (22). CuI is a native p-type conductor due to copper vacancies with a high hole mobility (>40 cm2V−1·s−1 in bulk) owing to the fairly small effective mass of 0.30m0 for the light holes (22, 23). It is also possible to improve the hole conductivity further by appropriate iodine doping (24). In addition, n-type doping is possible in Zn-doped CuI (25, 26), which would greatly extend its applications. On the other hand, CuI does not require high synthesis temperatures like other p-type TCs. The stable cubic structure makes it easy to synthesize CuI thin films near room temperature using a number of chemical and physical methods, including vapor iodization (20), thermal evaporation (27), pulsed laser deposition (PLD) (28), and sputtering (29–31). Therefore, CuI is readily compatible in organic electronics, for example as (part of) the electrode in organic light-emitting diodes and organic solar cells (32–34). Recent work on the roomtemperature heteroepitaxy of CuI demonstrates its compatibility Significance The lack of high-performance p-type transparent conductors (TCs) has been a main obstacle for the development of transparent electronics. In this study, we overcome this challenge in achieving simultaneously high conductivity and transparency for p-type TCs by developing the degenerate wide-bandgap semiconductor cuprous iodide (CuI). We propose industrially applicable techniques including room-temperature physical deposition by reactive sputtering for thin-film growth and iodine doping of CuI. The obtained CuI polycrystalline thin films exhibit record high conductivity as well as a high transparency, resulting in a 100× improvement of the figure of merit compared with any other p-type transparent conducting material. These results indicate significant advances in the development of p-type TCs toward their practical use. Author contributions: C.Y. and M.G. designed research; C.Y. performed research; M.K. and M.L. contributed new reagents/analytic tools; C.Y. and M.K. analyzed data; and C.Y. wrote the paper. The authors declare no conflict of interest. This article is a PNAS Direct Submission. A.R. is a Guest Editor invited by the Editorial Board. 1

To whom correspondence should be addressed. Email: [email protected]

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. 1073/pnas.1613643113/-/DCSupplemental.

PNAS Early Edition | 1 of 5

APPLIED PHYSICAL SCIENCES

Edited by Angus Rockett, University of Illinois, Urbana, IL, and accepted by Editorial Board Member John A. Rogers October 6, 2016 (received for review August 19, 2016)

with most crystalline electronic materials (35). The realization of highly rectifying bipolar diodes of CuI/ZnO (current on/off-ratio above 1 × 109 at ±2 V) opens up the application of CuI in transparent electronics (35, 36). Such advantageous properties make CuI a promising p-type transparent electronic material. However, the performance of CuI as a highly conducting p-type TC has not been well explored, because it is commonly recognized as a nondegenerate semiconductor with a roomtemperature conductivity limited to values below 100 S/cm (27). In contrast, conventional n-type TCs are usually degenerate semiconductors, such as indium tin oxide, which act as metallic conductors with conductivity above 103 S/cm. Here, we report the degenerate p-type conductivity of CuI thin films grown by reactive sputtering at room temperature. The sputtering technique is extensively used for thin-film growth and is cost-effective with the possibility of large-area deposition. A mixture of iodine vapor and argon is supplied as the sputtering gas. It is noted that the ionization of molecular iodine during sputtering greatly enhances the chemical reaction between iodine and sputtered copper metal, making it feasible to synthesize CuI thin films at room temperature. In addition, such highly reactive iodine leads to an iodine-rich equilibrium growth condition, which is favorable for inducing holes in CuI by creating copper vacancies (31). Results and Discussion The obtained CuI thin films are textured with (111)-orientation in the pure γ-phase, which is consistent with our previous report (35). The crystal structure characterization is shown in Fig. S1. The dependence of the room-temperature p-type conductivity of (as-deposited) CuI thin films as a function of the total sputtering gas pressure (Psput) is shown in Fig. 1A; the conductivity increases with increasing iodine/(iodine + argon) ratio. This behavior is dominated by the increase in hole concentration at iodine-rich growth conditions, whereas the hole mobility is rather similar for all of the CuI samples. The CuI thin film obtained with the lowest pressure tested, Psput of 0.005 mbar, shows a high conductivity of ∼156 S/cm at room temperature. Such a high hole conductivity is already much greater than that of previously

reported CuI (5–100 S/cm) (27–29, 36, 37). This significant improvement is probably related to the highly reactive ionized iodine involved in reactive sputtering, suppressing the formation of iodine vacancies or interstitials that are believed to be harmful to the formation and transport of holes in CuI. In addition, a straightforward room-temperature iodine doping process using the Bädeker vapor route leads to a dramatically elevated conductivity of up to ∼283 S/cm for p-type TCs. We note that although many iodides (e.g., KI, SnI2, PbI2. . .) are known to be highly hygroscopic, CuI possesses a low water solubility of 0.00042 g/L which allows stable p-type conductivity of CuI thin films exposed to air for months. As shown in Fig. S2, the p-type conductivity of a CuI thin film stayed almost constant for a week and slightly reduced from 156 to 134 S/cm after 3 months. Simply storing in a dry box or protecting with a cap layer would ensure a better stability of CuI under ambient conditions. However, detailed investigations on aging are subject to further investigations. Generally, semiconducting behavior is observed for most p-type TCs such as delafossites (7, 8, 18), layered oxychalcogenides (12), and perovskite Sr-doped LaCrO3 (19), which show thermally activated conductivity near room temperature and variablerange hopping at low temperatures due to the strong localization of carriers. In contrast, the transport properties of the p-type CuI thin films reported here are quite unusual, as can be seen from the temperature (T) dependence of the conductivity (σ) in Fig. 1B. Degenerate p-type conductivity is observed near room temperature for all CuI samples. With the lowest pressure tested (Psput = 0.005 mbar), a high hole concentration of 1.2 × 1020 cm−3 leads to a monotonous metallic conduction of CuI down to low temperatures. For larger Psput, a transition from metallic to semiconducting conductivity occurs at temperatures below ∼180 K, suggesting that more than one transport mechanism is operative. Within the metallic region, the conductivity follows the power law, σ(T) = σ 0T−2, according to the ionized impurity scattering model for degenerate semiconductors (38). On the other hand, a perfect fit of the semiconducting σ–T curves is obtained using the fluctuation-induced tunneling conductivity model (FITC) with the dependence σ(T) = σ 1exp[−T1/(T + T0)]

Fig. 1. Electrical properties of as-deposited CuI thin films. (A) Total sputtering gas pressure (Psput) dependence of conductivity (σ), carrier density (p), and Hall mobility (μ) at room temperature. (B) Temperature (T) dependence σ for CuI thin films prepared with Psput of 0.005, 0.02, and 0.1 mbar, respectively.

2 of 5 | www.pnas.org/cgi/doi/10.1073/pnas.1613643113

Yang et al.

Quantitatively, the TC performance can be evaluated by the figure of merit (FOM), being the ratio of the electrical conductivity (σ) to the visible absorption coefficient (α), σ/α = −1/[Rs ln (Tvis + R)], where Rs is the sheet resistance (Ω/sq) given by 1/σd, with d being the film thickness (41). A larger value of FOM indicates better performance of the TC. Due to the absent R values in most studies, the FOM is usually simplified by the approximation R = 0 and using FOM ∼ −1/(Rs lnTvis) (19). As shown in Fig. 3, the FOM for p-type TCs has improved over the years from 1996 until now. Except for CuI, the FOM for most p-type TCs is lower than 200 MΩ−1 unless sacrificing their transmittance below 50%. In contrast, the as-deposited CuI in this study shows an FOM about 50× higher, up to 9,500 MΩ−1, while providing a high Tvis above 70%. This FOM value can be further doubled to 17,000 MΩ−1 without affecting the transparency by iodine doping. These results demonstrate the superior TC performance of our CuI thin films with respect to all measures. The CuI reported here exhibits significantly higher FOM (10×) compared with previously reported CuI and p-type ZnO, and even 100× greater compared with any other p-type TC. The FOM values for p-type TCs are summarized in Table S1. Considering the approximation with R = 0, the FOM values are actually underestimated here. The accurate calculation using the intrinsic absorption coefficient results in a much larger FOM of CuI thin film up to 1.3 × 106 MΩ−1. Finally, we note that degenerate conductivity is common for n-type TCs, but is rarely found for p-type TCs (42). This metallic hole conduction of CuI allows for the record high p-type conductivity of ∼156 S/cm for undoped CuI and ∼283 S/cm for I-doped CuI. Considering the fact that the obtained CuI films are in the polycrystalline state, further improvement in the hole

Fig. 2. (A) Optical properties. Transmittance (T), reflectance (R), and intrinsic absorption coefficient (α) for CuI thin film deposited at a sputtering pressure of 0.005 mbar. (B) Graphical representation of room-temperature electrical conductivity (σ) and averaged visible transmittances (Tvis) for CuI thin films (red rhombuses), other p-type (black circles), and n-type TCs (blue squares). P-type TCs: SrCrO (19), CuCrMgO (18), CaCoO (46), CuScO (10), LaSrCrO (19), CuYCaO (11), ZnRhO (47), AgCoO (45), MgGaN (6, 43), MgCrO (48), CuAlO (44), SrCuO (12), CuGaO (8), ZnON (49), iodi-CuI (prepared by iodization) (36), sput-CuI (prepared by sputtering) (29), evap-CuI (prepared by thermal evaporation) (27), and PLD-CuI (prepared by PLD) (28). The superscripts of p-type TCs indicate the FOM. For n-type TCs, we selected Al- and Ga-doped ZnO that are also prepared at room temperature for comparison (50, 51).

Yang et al.

PNAS Early Edition | 3 of 5

APPLIED PHYSICAL SCIENCES

rather than the commonly applied variable-range-hopping model (39). The fitting parameters σ 0, σ 1, T0, and T1 are listed in Table S2. The FITC model for disordered materials is reasonable here considering the polycrystalline state of the CuI samples. Fig. 2A shows the optical transmittance (T), reflectance (R), and intrinsic absorption coefficient (α) for the CuI thin film deposited with a sputtering pressure of 0.005 mbar. The transmittance within the wavelength region of 400–800 nm is ∼60–85%, which is slightly lower than that of typical n-type TCs (80–90%) due to the significant reflections at the surface of the films and the film/substrate interface. The interference-free absorption of CuI thin films can be determined by self-consistently solving T/(1 − R) (40). The intrinsic absorption coefficient is extremely small, with α < 200 cm−1 at wavelength over 450 nm, indicating the high transparency of the obtained CuI thin films in the visible and nearinfrared (NIR) regions. In Fig. 2B we summarize the electrical conductivity (σ) and visible transmittance (Tvis) (i.e., the transmittance averaged over the visible spectral range 400–800 nm) for CuI thin films as well as for other TCs. For the purpose of achieving simultaneously high σ and Tvis, there is obviously a huge gap between p- and n-type TCs. It is now filled by CuI. For a high Tvis >70%, our p-type CuI thin films exhibit a very high conductivity, two or three orders of magnitude higher than that of any other p-type TC of such transparency. We would like to point out that, although the data of p-type ZnO have been included here as a comparison, there are still many doubts about the correctness of the doping-type assignments and the electrical parameters of reported p-type ZnO.

Materials and Methods

Fig. 3. Graphical representation of FOM over time for p-type TCs with Tvis > 30%. References are noted in brackets.

conductivity can be expected from films with optimized structure such as larger grains and, in particular, by developing epitaxial techniques for highly doped CuI. Our results underscore the great potential of CuI applied as p-type transparent electrodes. CuI is also compatible with many other n-type TCs for active device structures, such as transparent CuI/ZnO bipolar diodes (35, 36). Besides diodes, CuI/n-type TC tunneling contacts also appear attractive and could be a further direction of research. The sputtering technique at room temperature used here can be upscaled for applications of CuI in transparent electronics. These results may motivate further systematic studies of CuI as one of the most promising p-type TCs. 1. Thomas G (1997) Invisible circuits. Nature 389:907–908. 2. Grundmann M, et al. (2016) Oxide bipolar electronics: Materials, devices and circuits. J Phys D Appl Phys 49:213001. 3. Tsur Y, Riess I (1999) Self-compensation in semiconductors. Phys Rev B 60:8138–8146. 4. Look DC, et al. (2011) Self-compensation in semiconductors: The Zn vacancy in Ga-doped ZnO. Phys Rev B 84:115202. 5. Amano H, Kito M, Hiramatsu K, Akasaki I (1989) P-type conduction in Mg-doped GaN treated with low-energy electron beam irradiation (LEEBI). Jpn J Appl Phys 28: L2112–L2114. 6. Gunning B, Lowder J, Moseley M, Doolittle WA (2012) Negligible carrier freeze-out facilitated by impurity band conduction in highly p-type GaN. Appl Phys Lett 101: 082106. 7. Kawazoe H, et al. (1997) P-type electrical conduction in transparent thin films of CuAlO2. Nature 389:939–942. 8. Ueda K, et al. (2001) Epitaxial growth of transparent p-type conducting CuGaO2 thin films on sapphire (001) substrates by pulsed laser deposition. J Appl Phys 89: 1790–1793. 9. Yanagi H, Hase T, Ibuki S, Ueda K, Hosono H (2001) Bipolarity in electrical conduction of transparent oxide semiconductor CuInO2 with delafossite structure. Appl Phys Lett 78:1583–1585. 10. Duan N, Sleight AW, Jayaraj MK, Tate J (2000) Transparent p-type conducting CuScO2+x films. Appl Phys Lett 77:1325–1326. 11. Jayaraj MK, Draeseke AD, Tate J, Sleight AW (2001) p-Type transparent thin films of CuY1−xCaxO2. Thin Solid Films 397:244–248. 12. Kudo A, Yanagi H, Hosono H, Kawazoe H (1998) SrCu2O2: A p-type conductive oxide with wide band gap. Appl Phys Lett 73:220–222. 13. Ueda K, Inoue S, Hirose S, Kawazoe H, Hosono H (2000) Transparent p-type semiconductor: LaCuOS layered oxysulfide. Appl Phys Lett 77:2701–2703. 14. Yanagi H, Tate J, Park S, Park C-H, Keszler DA (2003) p-Type conductivity in wideband-gap BaCuQF (Q=S,Se). Appl Phys Lett 82:2814–2816. 15. Ueda K, Hosono H (2002) Band gap engineering, band edge emission, and p-type conductivity in wide-gap LaCuOS1−xSex oxychalcogenides. J Appl Phys 91:4768–4770.

4 of 5 | www.pnas.org/cgi/doi/10.1073/pnas.1613643113

Both reactive and nonreactive forms of direct current (dc), radio frequency (rf), and magnetron sputtering techniques can be used for the deposition of CuI thin films (29, 31). Reactive sputtering has an advantage over conventional (rf or dc) sputtering from CuI targets because of the higher plasma density due to the high conductivity of the Cu metal target, leading to greater uniformity of the films. In addition, problems in sintering the CuI target pellets can be avoided by this method. Here, the CuI films are deposited on Corning 1737 glass substrates by reactive sputtering in a dynamically pumped chamber with a base pressure of ∼1 × 10−5 mbar. A high-purity (99.999%) copper disk is used as the dc sputtering target. For the synthesis of CuI, iodine vapor is introduced by a needle valve connected to a heated iodine source. This iodine source is a stainless steel tube filled with iodine particles which is kept at ∼180 °C to sustain a sufficient iodine vapor pressure. The bearings of the turbomolecular pump are protected by nitrogen gas. The throttle gate valve between the chamber and turbomolecular pump is partially closed to adjust the iodine partial pressure to ∼1 × 10−3 mbar. Argon is then introduced to adjust the total sputtering gas pressure Psput. Presputtering is conducted with a power of 30 W for 20 min with the shutter closed. The CuI samples are then sputtered at 30 W at room temperature. After the sputtering process, some of the as-deposited CuI thin films are selected for iodine doping. This is a simple room-temperature iodine doping process using the Bädeker vapor route, which has been designed for synthesis of CuI thin films via the chemical reaction of a copper thin film with saturated iodine vapor (20, 37). We use the Bädeker vapor route for iodine doping by simply placing the as-deposited CuI thin films together with volatile iodine particles in a covered Petri dish in the atmosphere at room temperature. This doping process is conducted for a short time (10 min), which is sufficient to improve the conductivity without affecting the optical transparency. The thickness of the CuI thin film is determined as ∼200 (± 10) nm with a Dektak profilometer. Either the room temperature or temperature-dependent Hall effect is measured in van der Pauw geometry with a magnetic field of 0.4 T. The optical transmission spectra are measured by a PerkinElmer Lambda 40 UV-VIS-NIR spectrometer between wavelengths of 200 and 2,000 nm. ACKNOWLEDGMENTS. We thank Holger Hochmuth for technical support with the CuI sputter system, and Ulrike Teschner for the optical spectra measurements. This work was funded by Deutsche Forschungsgemeinschaft Grant GR 1011/28-1. 16. Ueda K, Hosono H, Hamada N (2004) Energy band structure of LaCuOCh(Ch = S, Se and Te) calculated by the full-potential linearized augmented plane-wave method. J Phys Condens Matter 16:5179–5186. 17. Liu ML, Wu LB, Huang FQ, Chen LD, Ibers JA (2007) Syntheses, crystal and electronic structure, and some optical and transport properties of LnCuOTe (Ln=La, Ce, Nd). J Solid State Chem 180:62–69. 18. Nagarajan R, Draeseke AD, Sleight AW, Tate J (2001) p-Type conductivity in CuCr1−xMgxO2 films and powders. J Appl Phys 89:8022–8025. 19. Zhang KHL, et al. (2015) Perovskite Sr-doped LaCrO3 as a new p-type transparent conducting oxide. Adv Mater 27(35):5191–5195. 20. Bädeker K (1907) Über die elektrische Leitfähigkeit und die thermoelektrische Kraft einiger Schwermetallverbindungen. Ann Phys 327:749–766. 21. Grundmann M (2015) Karl Bädeker (1877-1914) and the discovery of transparent conductive materials. Phys Status Solidi A: Appl Mater Sci 212:1409–1426. 22. Chen D, et al. (2010) Growth strategy and physical properties of the high mobility p-type CuI crystal. Cryst Growth Des 10:2057–2060. 23. Wang J, Li JB, Li SS (2011) Native p-type transparent conductive CuI via intrinsic defects. J Appl Phys 110:054907. 24. Tennakone K, et al. (1998) Deposition of thin conducting films of CuI on glass. Sol Energy Mater Sol Cells 55:283–289. 25. Zhu JJ, Gu M, Pandey R (2013) Structural and electronic properties of CuI doped with Zn, Ga and Al. J Phys Chem Solids 74:1122–1126. 26. Xia M, et al. (2015) Electrical and luminescence properties of Zn2+ doped CuI thin films. J Mater Sci Mater Electron 26:2629–2633. 27. Grundmann M, et al. (2013) Cuprous iodide - a p-type transparent semiconductor: History and novel applications. Phys Status Solidi A: Appl Mater Sci 210:1671–1703. 28. Sirimanne PM, Rusop M, Shirata T, Soga T, Jimbo T (2002) Characterization of transparent conducting CuI thin films prepared by pulse laser deposition technique. Chem Phys Lett 366:485–489. 29. Tanaka T, Kawabata K, Hirose M (1996) Transparent, conductive CuI films prepared by rf-dc coupled magnetron sputtering. Thin Solid Films 281-282:179–181.

Yang et al.

41. Gordon RG (2000) Criteria for choosing transparent conductors. MRS Bull 25:52–57. 42. Hiramatsu H, et al. (2003) Degenerate p-type conductivity in wide-gap LaCuOS1-xSex (x=0–1) epitaxial films. Appl Phys Lett 82:1048–1050. 43. Sato H, Minami T, Yamada E, Ishii M, Takata S (1994) Transparent and conductive impurity-doped GaN thin films prepared by an electron cyclotron resonance plasma metalorganic chemical vapor deposition method. J Appl Phys 75:1405–1409. 44. Yanagi H, et al. (2000) Electronic structure and optoelectronic properties of transparent p-type conducting CuAlO2. J Appl Phys 88:4159–4163. 45. Tate J, et al. (2002) p-Type oxides for use in transparent diodes. Thin Solid Films 411: 119–124. 46. Aksit M, Kolli SK, Slauch IM, Robinson RD (2014) Misfit layered Ca3Co4O9 as a high figure of merit p-type transparent conducting oxide film through solution processing. Appl Phys Lett 104:161901. 47. Dekkers M, Rijnders G, Blank DHA (2007) ZnIr2O4, a p-type transparent oxide semiconductor in the class of spinel zinc-d6-transition metal oxide. Appl Phys Lett 90: 021903. 48. Farrell L, et al. (2015) Conducting mechanism in the epitaxial p-type transparent conducting oxide Cr2O3:Mg. Phys Rev B 91:125202. 49. Nian H, Hahn SH, Koo K-K, Shin EW, Kim EJ (2009) Sol–gel derived N-doped ZnO thin films. Mater Lett 63:2246–2248. 50. Fortunato E, et al. (2003) Growth of ZnO:Ga thin films at room temperature on polymeric substrates: Thickness dependence. Thin Solid Films 442:121–126. 51. Matsubara K, Fons P, Iwata K, Yamada A, Niki S (2002) Room-temperature deposition of Al-doped ZnO films by oxygen radical-assisted pulsed laser deposition. Thin Solid Films 422:176–179.

APPLIED PHYSICAL SCIENCES

30. Reichelt K, Mair G (1979) Preparation of CuI films on NaCl single crystals by reactive sputtering and characterization by electron microscopy and Rutherford backscattering. J Vac Sci Technol 16:896–898. 31. Maurer RJ (1945) Deviations from stoichiometric proportions in cuprous iodide. J Chem Phys 13:321–326. 32. Lee JH, Leem DS, Kim JJ (2009) High performance top-emitting organic light-emitting diodes with copper iodide-doped hole injection layer. Org Electron 9:805–808. 33. Zhou Y, et al. (2012) Glancing angle deposition of copper iodide nanocrystals for efficient organic photovoltaics. Nano Lett 12(8):4146–4152. 34. Rand BP, et al. (2012) The impact of molecular orientation on the photovoltaic properties of a phthalocyanine/fullerene heterojunction. Adv Funct Mater 22: 2987–2995. 35. Yang C, Kneiß M, Schein FL, Lorenz M, Grundmann M (2016) Room-temperature domain-epitaxy of copper iodide thin films for transparent CuI/ZnO heterojunctions with high rectification ratios larger than 109. Sci Rep 6:21937. 36. Schein FL, von Wenckstern H, Grundmann M (2013) Transparent p-CuI/n-ZnO heterojunction diodes. Appl Phys Lett 102:092109. 37. Bädeker K (1909) Über eine eigentümliche Form elektrischen Leitvermögens bei festen Körpern. Ann Phys 334:566. 38. Meeks T, Krieger JB (1969) Temperature dependence of the resistivity of degenerately doped semiconductors at low temperatures. Phys Rev 185:1068–1072. 39. Sheng P (1980) Fluctuation-induced tunneling conduction in disordered materials. Phys Rev B 21:2180–2195. 40. Hishikawa Y, et al. (1991) Interference-free determination of the optical absorption coefficient and the optical gap of amorphous silicon thin films. Jpn J Appl Phys 30: 1008–1014.

Yang et al.

PNAS Early Edition | 5 of 5

Supporting Information Yang et al. 10.1073/pnas.1613643113 Phase Purity of CuI Thin Films Fig. S1A shows X-ray diffraction (XRD) 2θ-ω-scan of the CuI thin film grown on glass. The diffraction peaks correspond to the (111), (222), (333) plane of CuI in zincblende structure, suggesting the growth of γ-phase CuI along the [111] direction. On the other hand, the X-ray scattering intensity of CuI (022) plane in a ϕ-scan is constant over the angle, as shown in Fig. S1B, indicating that no preferred azimuthal orientation of CuI grains is present. Hence, we conclude that the obtained CuI thin film is textured with (111) orientation, and no second phases have been detected. This result is consistent with our previous report (35). Effects of Substrate Temperature on Structural Properties of CuI Thin Films Room-temperature deposition techniques are of great interest for integration of CuI with organic materials. However, for systematic investigation on the growth nature of CuI thin films using reactive sputtering, various substrate temperatures are used during the deposition. In Fig. S3 we compare the (111) diffraction peaks according to the XRD patterns for CuI thin films grown at various temperatures. The full width at half maximum (FWHM) of these (111) peaks monotonously reduces with increasing substrate temper-

ature, suggesting the improved crystallinity of CuI thin film via substrate heating. For crystalline semiconductors, higher crystallinity usually leads to higher Hall mobility. On the other hand, interestingly, substrate heating also strongly affects the microstructure of CuI thin film. Fig. S4 shows the scanning electron microscopy (SEM) images of CuI thin films grown at various temperatures. It reveals that a compact thin film can be obtained at room temperature. When the substrate is heated, the obtained samples comprise loosely compacted micrometer-sized crystallites. In this case, these micro- and nanocrystallites cause significant light scattering, resulting in degraded transparency for CuI films. The grain size increases with rising substrate temperature, which is consistent with XRD results. This phenomenon can be explained by a three-dimensional (island) growth mode of CuI on glass substrate. Heating of the substrate may cause a high surface mobility of copper and iodine atoms. Hence, two-dimensional layer-by-layer growth techniques, e.g., atomic layer deposition and molecular beam epitaxy, are more suitable to obtain compact CuI thin films at elevated temperatures. Although substrate heating is not preferred for fabricating highly transparent thin films using sputtering in this work, it should be a facile way to produce micro- and nanostructures of CuI.

Fig. S1. XRD (A) 2θ-ω-scan and (B) ϕ-scan of as-deposited CuI thin film grown on glass. The XRD patterns were measured using a Philips X’Pert X-ray diffractometer equipped with a Bragg–Brentano powder goniometer using divergent/focusing slit optics and Cu Kα radiation.

Yang et al. www.pnas.org/cgi/content/short/1613643113

1 of 3

Fig. S2. The p-type conductivity over time for a typical CuI thin film obtained in this work directly exposed to air without any specific storage or protection treatment.

Fig. S3.

(A) The (111) diffraction peaks according to XRD 2θ-ω-scans and (B) the FWHM of (111) peaks for CuI thin films grown at different substrate temperatures.

Fig. S4. SEM images of sample surfaces for CuI grown at different substrate temperatures. The SEM measurement was performed using an FEI NOVA Nanolab 200.

Yang et al. www.pnas.org/cgi/content/short/1613643113

2 of 3

Table S1. Crystal structure, transmittance, conductivity, and FOM for our CuI thin films and other p-type TCs p-type TCs

Structure

Mg-doped GaN N-doped ZnO CuAlO2 CuGaO2 CuScO2 + x CuY1 − xCaxO3 CuxCr1 − xMgxO2 AgCoO2 SrCu2O2 Ca3Co4O9 ZnRh2O4 MgxCr2-xO3 La0.5Sr0.5CrO3 SrCrO3 CuI† CuI‡ CuI§ CuI{ CuI I-doped CuI

Wurtzite Delafossite

Tetragonal Misfit-layered monoclinic Spinel Corundum Perovskite Zincblende Zincblende

σ, S/cm

Tvis, %

D, nm

Rs, Ω/sq

FOM, MΩ-1

Reference

148 2,000 22 0.9 180 11 4,564 4.3 2.6 155

6, 43 49 44 8 10 11 18 45 12 46

70* 80 70 80 40 40 30 50 80 31.3

5.3 15 0.34 0.02 15 1 220 0.2 0.048 18

100 300 230 100 110 100 250 150 120 100

1.9 × 10 2.2 × 103 1.3 × 105 5.0 × 106 6.0 × 103 1.0 × 105 182 3.3 × 105 1.7 × 106 5.5 × 103

50 65 42.3 29 70 80 49 75 72 72

2 0.333 56 750 18.5 10 5 32 156 283

150 150 50 50 100 ∼500 500 240 200 200

3.3 × 104 2.0 × 105 3.5 × 103 267 5,400 2,000 4,000 1,300 320 177

4

44 12 314 3,050 518 2,240 350 2,670 9,500 17,000

47 48 19 19 29 28 36 27 This work This work

*Because of the absent Tvis value of p-type GaN in the literature, Tvis of n-type GaN is adopted here instead for a reference only (43). It is expected that the Tvis values for p-type and n-type GaN are in the same level. † Prepared by sputtering. ‡ Prepared by PLD. § Prepared by iodization. { Prepared by thermal evaporation.

Table S2. Fitting parameters for temperature-dependent conductivity of CuI thin films shown in Fig. 1B using a power law, as well as the FITC model with the fitting function σ(T)−1 = σ 0-1T2 + σ 1-1exp[T1/(T + T0)] Psputt, mbar

σ 0, S/cm

σ 1, S/cm

T0 , K

T1 , K

0.1 0.02 0.005

1.7 × 10 9.1 × 107 8.3 × 107

54.1 94.3 181.8

180.5 111.2 0

134.1 33.8 0

Yang et al. www.pnas.org/cgi/content/short/1613643113

7

3 of 3