Atomic Layer Deposition of Cubic and ... - ACS Publications

Mar 20, 2017 - five ALD cycles were completed for each condition. Curves of the form a − b exp(−t/τ) were fitted, with saturation scale parameter...
0 downloads 0 Views 2MB Size
Article pubs.acs.org/cm

Atomic Layer Deposition of Cubic and Orthorhombic Phase Tin Monosulfide Oleksandr V. Bilousov, Yi Ren, Tobias Törndahl, Olivier Donzel-Gargand, Tove Ericson, Charlotte Platzer-Björkman, Marika Edoff, and Carl Hag̈ glund* Ångström Solar Center, Solid State Electronics, Uppsala University, Box 534, SE-751 21 Uppsala, Sweden S Supporting Information *

ABSTRACT: Tin monosulfide (SnS) is a promising lightabsorbing material with weak environmental constraints for application in thin film solar cells. In this paper, we present low-temperature atomic layer deposition (ALD) of high-purity SnS of both cubic and orthorhombic phases. Using tin(II) 2,4pentanedionate [Sn(acac)2] and hydrogen sulfide (H2S) as precursors, controlled growth of the two polymorphs is achieved. Quartz crystal microbalance measurements are used to establish saturated conditions and show that the SnS ALD is self-limiting over temperatures from at least 80 to 160 °C. In this temperature window, a stable mass gain of 19 ng cm−2 cycle−1 is observed. The SnS thin film crystal structure and morphology undergo significant changes depending on the conditions. High-resolution transmission electron microscopy and X-ray diffraction demonstrate that fully saturated growth requires a large H2S dose and results in the cubic phase. Smaller H2S doses and higher temperatures favor the orthorhombic phase. The optical properties of the two polymorphs differ significantly, as demonstrated by spectroscopic ellipsometry. The orthorhombic phase displays a wide (0.3−0.4 eV) Urbach tail in the near-infrared region, ascribed to its nanoscale structural disorder and/or to sulfur vacancy-induced gap states. In contrast, the cubic phase is smooth and void-free and shows a welldefined, direct forbidden-type bandgap of 1.64 eV.



uniform films with precisely controlled thicknesses can be accomplished over large areas. This is also applicable to nonplanar surfaces and allows in particular for the creation of very thin, conformal, and pinhole-free films on complex structures, with high reproducibility.28,29 The first report of SnS growth by ALD was by Kim and George,25 and they used the reaction of commercially available tin 2,4-pentanedionate [Sn(acac)2] and hydrogen sulfide (H2S). However, the growth rate was relatively low (0.24 Å cycle−1), and it was suggested that the high observed bandgap was due to incorporation of oxygen in the film. The optical properties and crystal structure were not studied in detail. Sinsermsuksakul et al.26 demonstrated growth of orthorhombic SnS (o-SnS) by ALD using bis(N,N′-diisopropylacetamidinato) tin(II) and H2S with higher growth rates of 0.86−0.90 Å cycle−1. However, as the precursor is more complex, a higher processing cost could be expected. SnS usually crystallizes in an orthorhombic phase,30 with a distorted rock-salt (NaCl) structure. Different studies2,10,26,31−33 have suggested that this material has an indirect optical bandgap of 1.1 eV and a direct bandgap of 1.3 eV. SnS may also take a cubic crystalline structure, which has a bandgap

INTRODUCTION Thin film solar cells have optimism-inducing prospects as a future resource for sustainable energy. Among materials investigated as absorbers for thin film solar cells, tin(II) monosulfide (SnS) is considered a promising semiconductor because of its suitable bandgap and its content of only nontoxic and earth-abundant elements.1−7 The simple binary composition of SnS is a general advantage for its production. SnS is, because of its high optical damping, also an excellent semiconductor candidate for the realization of ultrathin plasmonic solar cells.8 It has a high absorption coefficient and excellent carrier mobility,9−13 also for few-layer SnS, which promotes its use as a two-dimensional semiconductor in nanoelectronics.13 Moreover, nanostructured SnS has shown promise as a lithium battery anode material.14 SnS thin films may be prepared by a variety of methods, including chemical bath deposition,15,16 chemical vapor deposition,17 electrochemical deposition,18,19 spray pyrolysis,20,21 thermal evaporation,22,23 vapor transport deposition,24 and atomic layer deposition (ALD).25,26 The electrical and optical properties of SnS can be modified by changing the stoichiometry of the films27 or their doping, without changing the crystal structure.9 ALD has several advantages over other techniques as it is based on repeating a cycle of self-limiting gas−surface reactions. Through saturation of these reactions, conformal growth of © 2017 American Chemical Society

Received: December 15, 2016 Revised: March 9, 2017 Published: March 20, 2017 2969

DOI: 10.1021/acs.chemmater.6b05323 Chem. Mater. 2017, 29, 2969−2978

Article

Chemistry of Materials

Figure 1. ALD SnS mass gain per ALD cycle as a function of (a) Sn(acac)2 pulse time, (b) Sn(acac)2 purge time, and (c) H2S pulse time. Twentyfive ALD cycles were completed for each condition. Curves of the form a − b exp(−t/τ) were fitted, with saturation scale parameter τ indicated by vertical lines. Electron Microscopy. The surface morphology of the films was examined by field-emission scanning electron microscopy using a Zeiss 1550 instrument, using an acceleration voltage of 10 kV. The cross sections of thin as-grown SnS samples were characterized by highresolution transmission electron microscopy (HRTEM) from a Tecnai F30 ST instrument, with sample preparation by the in situ lift-out method.41 Compositional Analysis. To determine the content of Sn and S, Rutherford backscattering spectrometry (RBS) was performed using α particles accelerated by a tandem accelerator at Uppsala Tandem Laboratory to a kinetic energy of 2 MeV. The incidence angle of the α particles was perpendicular. The sample was slightly moved around its equilibrium position to avoid channeling. The detector was placed at a scattering angle of 170°. The SIMNRA software42 was used to fit the measured RBS energy spectra. RBS data were used to calibrate X-ray fluorescence (XRF) measurements, performed using a PANalytical Epsilon5 instrument. Grazing Incidence X-ray Diffraction. Grazing incidence X-ray diffraction (GIXRD) was performed with a Siemens D5000 instrument equipped with an X-ray mirror and parallel plate collimator. The powder diffraction files used for reference were mineral herzenbergite for o-SnS43 and π-SnS according to the recently resolved structure for the large cubic tin monosulfide polymorph.40 X-ray Reflectivity. X-ray reflectivity (XRR) was measured in a Philips X’Pert MRD system using an X-ray mirror and divergencematched parallel plate collimator. The reflectivity curves were fitted by using a specular model in the X’Pert reflectivity software. Raman Spectroscopy. Raman scattering measurements were performed in a Renishaw inVia confocal Raman microscope. A laser wavelength of 532 nm was used with a power of 0.2 mW for excitation, and a microscope lens with a numerical aperture of 0.35. Optical Characterization. A Woollam variable-angle spectroscopic ellipsometer was used to analyze the optical properties of the SnS layers. Standard ellipsometric measurements were performed at wavelengths from 260 to 1700 nm and angles of incidence of 65°, 70°, and 75°. The measurement data were analyzed using the WVASE32 software package. The diffuse and specular near normal incidence reflectance and transmittance were measured in an integrating sphere using a PerkinElmer Lambda 900 spectrophotometer. Safety. Note that, at the concentrations used in this work, hydrogen sulfide gas is both highly toxic and highly flammable.

around 1.7 eV.6,34−36 This polymorph was originally assigned to the zinc blende (ZB) crystalline structure.34,36,37 However, more recent studies have shown that the XRD pattern fits better with a cubic SnS (π-SnS) phase consisting of 64 atoms in the unit cell and a lattice parameter close to twice that of the zinc blende parameter.6,38−40 Despite these advances, there is still a need to explore the correlation of the SnS thin film growth conditions and properties. In particular, ALD of the πSnS phase has not been previously reported. In this work, we investigate the growth and properties of tin monosulfide films produced using ALD with Sn(acac)2 and H2S as precursors. As the growth mechanism is already established,25 we here focus on the film properties and show that they vary over a rather wide range depending on substrate temperature and other reaction conditions, and that this is connected to the growth of mixed π-SnS and o-SnS phases. Variation of the growth conditions allows for control over the polymorphism, morphology, and optical properties of the SnS thin films. In particular, it is shown that variation of substrate temperature and H2S dose can be used to produce ultrathin films ranging from more or less purely π-SnS to o-SnS. In addition to the crystal structure, morphology, and optical properties of the films, the elemental composition is characterized in detail.



EXPERIMENTAL SECTION

Thin Film Growth by ALD. A viscous flow ALD reactor (Microchemistry F-120) was used to grow SnS thin films on glass and Si(100) wafer substrates. Nitrogen gas, N2 (99.9999%), was used as the carrier gas for the ALD process and was maintained at a continuous flow of 500 sccm. Sn(acac)2 (Sigma-Aldrich, 99.99%) and hydrogen sulfide, H2S (Air Liquide, 99.5%), were used as precursors. The precursor dosing was controlled by means of needle valves and pulse times. Sn(acac)2 was placed in a bubbler and heated to 80 °C. N2 carrier gas was passed through the liquid Sn(acac)2 precursor and injected into the reaction chamber via heated tubing. The SnS ALD films were grown by a pulse sequence of Sn(acac)2, N2 purge, H2S, and N2 purge. Quartz crystal microbalance (QCM) experiments of SnS ALD were performed by mounting the QCM sensor in a housing placed a few millimeters from the substrate position in the ALD reactor. Unless otherwise stated, the substrate temperature was 120 °C. 2970

DOI: 10.1021/acs.chemmater.6b05323 Chem. Mater. 2017, 29, 2969−2978

Article

Chemistry of Materials

Figure 2. SEM images of SnS films prepared by 1000 ALD cycles at 120 °C using different H2S doses: (a) 0.4, (b) 1.6, (c) 2, (d) 3, (e) 5, and (f) 5 s with an ∼40% higher flow rate.



RESULTS AND DISCUSSION Growth Behavior. The influence of the ALD precursor pulse and purge times on the SnS growth rates was studied one at a time, with other process parameters kept fixed. We define the time sequence by t1−t2−t3−t4, where t1 is the Sn(acac)2 pulse time, t2 is the Sn(acac)2 purge time, t3 is the H2S pulse time, and t4 is the H2S purge time. As the hallmark of ALD is saturated growth, a study of the precursor doses required to reach this condition was performed at 80, 120, 160, and 200 °C. A 1.2 s−3.0 s−5.0 s−3.0 s sequence led to nearly saturated growth at all temperatures and was chosen as the baseline, from which each parameter was separately varied. QCM measurements over 25 cycles were used to obtain the mass gain per cycle for each variable of SnS ALD on the QCM sensor at each temperature, as shown in Figure 1. Because the QCM crystal was coated with SnS before the first run, the SnS growth monitored here is on the SnS itself, which allows steady state conditions (linear mass gain) to be quickly established. Because of the influence of temperature on the QCM sensor response, the saturation curves were calibrated against XRF measurements of SnS films grown at each temperature under the baseline conditions. As the H2S purge time required for saturated behavior was always shorter than the shortest times investigated (0.4 s), saturation curves for this parameter are not presented. The mass gain of the SnS films remains roughly constant over the range of Sn(acac)2 dose times at temperatures from 80 to 160 °C (Figure 1a), indicating saturated behavior for this part of the ALD cycle. This means that the surface reactions are completed and self-limited already for the smallest doses of the Sn precursor investigated here. At 200 °C, the growth per cycle increases from a relatively low level with an increasing Sn(acac)2 dose. This is presumably related to suppressed formation of SnS2 at this temperature as further discussed below. When the Sn purge time is varied, an initially decreasing mass gain is observed (Figure 1b). This saturates above approximately 3 s, indicating a nearly complete removal of the excess of the Sn(acac)2 precursor. An excess on, for instance, the sample surface in the form of loosely bound precursor could result in slow desorption and a less wellcontrolled component of the mass gain. With regard to the H2S dose (Figure 1c), saturated growth is observed at temperatures of up to 160 °C for H2S pulse lengths of ≥2 s.

From the QCM results, it is also clear that a large H2S dose is required to saturate the SnS growth rate. In fact, it takes doses that are approximately 3 orders of magnitude larger to saturate the reaction with H2S (∼50 Torr·L per pulse) compared to the reaction with Sn(acac)2 (∼0.025 Torr·L per pulse). This indicates a relatively low reactivity of H2S toward the SnS− Sn(acac) surface species formed by the first half-reaction of the ALD cycle.25 The saturation scale parameters (τ) fitted to the experiments in Figure 1 show that larger doses are generally required at higher temperatures. However, the saturated mass gain per cycle of ALD SnS films remains roughly constant from 80 to 160 °C. Lower or unsaturated values are observed at 200 °C. The hampered growth at 200 °C appears to be related to the formation of a SnS2 phase detected by XRD and Raman analysis, shown in the later part of the article. Separate QCM experiments on SnS2-coated substrates prepared by ALD with tetrakis(dimethylamino)tin(IV) and H2S as precursors confirm that Sn(acac)2 ALD is suppressed on the SnS2 surface by a factor of at least 4. Therefore, the low growth per cycle and precursor dose dependence at 200 °C may be understood from competition between the formation of SnS and SnS2 and the inhibited growth caused by the latter. Poor nucleation of Sn(acac)2 has previously been observed on other substrates.25 In the temperature interval from 80 to 160 °C, the saturated ALD mass gain is stable and around 19 ng cm−2 cycle−1, which is 60% higher than the value previously reported.25 Differences in the experimental conditions could, possibly, mean that a more complete saturation of the ALD half-reactions is reached in this case. Surface Morphology and Crystal Structure. SnS films prepared on Si and glass with varying H2S doses generally appeared shiny, more or less uniform, and pinhole-free over the entire substrate. However, analysis by scanning electron microscopy (SEM) reveals that the H2S pulse length has a clear influence on the surface morphology of these SnS films. A comparison of SnS films deposited using 1000 cycles and 0.4, 1.6, 2, 3, and 5 s pulses of H2S at 120 °C is provided in Figure 2. The SnS films are composed of small grains with two distinct morphologies. At a small H2S dose, we observe a rough top layer with elongated, bladelike grains of random orientation. As the H2S dose increases, the density and size of the blades 2971

DOI: 10.1021/acs.chemmater.6b05323 Chem. Mater. 2017, 29, 2969−2978

Article

Chemistry of Materials

these cases, the dominant peak is due to the (111) plane, just as it is in the present case (Figure 3a, for a 0.4 s H2S pulse length). The ratio of the intensities of peaks from π-SnS to those of oSnS planes is indicative of the relative abundance of each phase in the film. Of the two peaks at (i) 30.8° and (ii) 31.7°, π-SnS contributes more to (i) via the (400) planes and o-SnS contributes more to (ii) via the (111) planes. The peak fraction i/(i + ii) increases with the H2S dose, indicative of a higher content of the π-SnS phase (Figure 3b). A transition from oSnS to π-SnS with a larger H2S dose is seen also by the evolution of other peaks along with the SEM data. We therefore conclude that H2S exposure is vital not only for morphology but also for the crystal structure of ALD SnS. To further investigate the spatial distribution of the SnS phases, a sample of 2000 cycles was prepared with an intermediate H2S dose of 1.6 s. This resulted in a film surface with a notably bladelike SnS morphology, as shown in the SEM plan view in Figure 4b. GI-XRD patterns recorded at θ = 0.2°, 0.3°, 0.4°, and 0.5° are given in Figure 4a on an expanded scale for the region from 25° to 35° of 2θ. In the pattern with the smallest angle of incidence of 0.2°, which is smaller than that of the critical angle, the peak of o-SnS (111) is most pronounced. The peak of π-SnS (222) planes is absent, and the (400) peak has a low intensity, if any. However, as the sampling depth increases along with an angle of incidence above the critical angle, here up to 0.5°, the relative intensities of the π-SnS (222) and (400) peaks increase. This suggests a gradient with more πSnS phase near the bottom of the film and more o-SnS phase toward the SnS surface. To confirm this, high-resolution transmission electron microscopy was performed. The SnS layer is observed to be polycrystalline with randomly oriented grains, exhibiting high crystallinity and continuous lattice fringes over large areas (Figure 4c−f). The bottom of the layer is compact and void-free. HRTEM images on a single grain in this region of the SnS film, such as in Figure 4f, typically show lattice fringes corresponding to (222) planes of π-SnS with an interplanar d spacing of 0.34 nm. The top layer, on the other hand, exhibits a high roughness with protruding bladelike grains. Some of these elongated grains extend through the whole film and may have nucleated at an early stage of the ALD process. HRTEM on such an elongated grain shows an interplanar d spacing of 0.28 nm, in agreement with the (111) spacing for o-SnS. These results additionally support our interpretation of the GI-XRD data, namely, that SnS nucleates and initially grows in a predominantly π-SnS phase, which gradually undergoes a transition into the o-SnS phase with further growth. To explore the thickness dependence in a predominantly πSnS layer, films were prepared with a long, 5 s H2S pulse and varying numbers of ALD cycles on both Si and glass. Films deposited at a substrate temperature of 120 °C are found to be compact with few and relatively small grains protruding from the surface (Figure 5). The number of grains and their sizes increase with the number of ALD cycles, with grain sizes reaching 15−20 nm after 1000 cycles (Figure 5f). GI-XRD confirms that the films contain mainly the cubic phase (Figure 6a,b), regardless of the substrate. As the thickness of the film increases, no evolution of crystal structure or change in preferential orientation is observed. The HRTEM cross-sectional analysis shown in panels c and d of Figure 6 on a 1000-cycle film on Si shows that the film is much more uniform compared to that of a sample with a smaller H2S dose of 1.6 s

decrease and a smoother layer consisting of closely packed small grains appears. With the largest H2S dose, only few grains protrude from this more compact bottom layer. We thus observe a strong dependence of the SnS surface morphology on the H2S dose. To investigate the crystal structure, grazing incidence X-ray diffraction (GI-XRD) was performed for SnS thin films deposited with different H2S doses. By collecting the diffractograms at a small angle of incidence (here 0.5°), GIXRD allows for relatively high scattering also by films of nanoscale thicknesses. The data, presented in Figure 3a, reveal a

Figure 3. (a) GI-XRD pattern of SnS thin films prepared by 1000 ALD cycles at 120 °C using different H2S doses (pulse lengths of 0.4, 1.6, 2, 3, and 5 s). The principal diffraction peaks match those of PDF 39-0354 for o-SnS and π-SnS in the bottom panel. (b) Fraction of the mainly π-SnS-related peak intensity calculated as the ratio of the peak intensity at 30.8° to the sum of the peak intensities at 30.8° and 31.7°.

high degree of crystallinity, with a crystal structure that undergoes a notable evolution as the H2S dose varies. Films produced with a large H2S dose display a pattern of XRD peak positions and intensities in good agreement with those of the recently reported cubic SnS phase.38,40,44 The intensities of the XRD peaks for (222) and (400) planes are nearly the same, and the pattern shows good agreement with a large SnS cubic unit cell with a cell parameter of 11.59 Å, nearly twice that of a zinc blende structure.38 In contrast, for films produced using small H2S doses, a predominantly orthorhombic SnS structure is deduced by comparison with the standard pattern of the mineral herzenbergite (PDF 039-0354), with the following lattice parameters: a = 4.334 Å, b = 11.200 Å, and c = 3.987 Å. The oSnS structure has previously been observed for SnS thin films deposited by a wide range of methods.10,23,26,31−33 In most of 2972

DOI: 10.1021/acs.chemmater.6b05323 Chem. Mater. 2017, 29, 2969−2978

Article

Chemistry of Materials

Figure 4. (a) GI-XRD pattern of the SnS thin film prepared by 2000 ALD cycles at 120 °C taken at different incident angles of GI-XRD. (b) SEM image of the surface morphology. (c and d) Cross-sectional HRTEM images. (e) HRTEM image of the top layer. (f) HRTEM image of the compact bottom layer.

Figure 5. SEM images of π-SnS films on Si prepared by ALD at 120 °C for different number of cycles: (a) 25, (b) 50, (c) 100, (d) 200, (e) 500, and (f) 1000 cycles.

samples (Figure 7f), but from the evolution of the morphology and the XRD peaks, we conclude that an increasing substrate temperature favors formation of the orthorhombic phase over the cubic. This is consistent with how chemical bath-deposited SnS depends on temperature in the range from 20 to 40 °C.35 Film Composition. To investigate the film composition in further detail, the series of samples grown with a varying number of cycles and of mainly cubic phase (shown in Figures 5 and 6) were measured by Rutherford backscattering spectroscopy (RBS). The quantitative evaluation of the data indicates a Sn/S atomic ratio close to the stoichiometric value of tin monosulfide, although slightly sulfur rich with a mean ratio of 0.92 ± 0.03 and no discernible trend (see Table 1). This sulfur rich composition does not support a high oxygen content of the SnS films, and no oxygen impurities could be detected in the thickest films here. This is in contrast to previous results for ALD of SnS using Sn(acac)2.25 RBS is not ideal for detecting small amounts of oxygen on a Si substrate,

(Figure 4), but still highly crystalline with clearly observed lattice fringes in the grains. The grains are columnar and extend over the entire film thickness (42−43 nm) with more or less vertical grain boundaries, forming a compact layer. These results suggest that with a sufficiently high exposure to H2S during ALD, purely cubic SnS films of high quality can be achieved. Although only a weak temperature dependence of the growth per cycle is seen from 80 to 160 °C here, the substrate temperature does influence the SnS phase and morphology. For example, for an intermediate H2S pulse length of 1.2 s, grain sizes of SnS increase with temperature (Figure 7a−d). Small round-shaped grains prevail when the growth temperature is 80 °C, while at 100 and 120 °C, the grains grow larger and form a compact bottom layer with some protruding grains on top. At 175 °C, the surface is dominated by bladelike grains with a resulting high roughness. GI-XRD recorded at a 0.5° incident angle reveals a high degree of crystallinity for all of these 2973

DOI: 10.1021/acs.chemmater.6b05323 Chem. Mater. 2017, 29, 2969−2978

Article

Chemistry of Materials

Figure 6. GI-XRD pattern of π-SnS films prepared at 120 °C for different numbers of ALD cycles on (a) Si and (b) glass substrates. (c and d) Crosssectional HRTEM images of π-SnS prepared by 1000 cycles on the Si substrate.

Figure 7. (a−d) SEM plan views of SnS films prepared by 1000 cycles at different temperatures. An intermediate H2S dose of 1.2 s and otherwise saturated conditions were used. (f) GI-XRD pattern of SnS thin films prepared at different temperatures.

lower than the ideal crystal densities of o-SnS (5.197 g/cm3) and π-SnS (5.132 g/cm3). With a mass gain of 19 ng cm−2 cycle−1, the measured density value gives a (solid) thickness gain of 0.42 Å cycle−1. Additional measurements of composition were performed by X-ray fluorescence (XRF) spectroscopy. Using the RBS results as a calibration for the XRF data, with attenuation effects taken into account, the samples prepared at 120 °C and varying H2S doses were analyzed (Table 2). In this case, the composition varied from a slightly tin rich value (Sn/S ratio of 1.03) at small H2S doses to a sulfur rich composition (Sn/S ratio of 0.93) at large H2S doses. This is interesting, as it could indicate the possibility of influencing the point defect density and sulfur incorporation in the lattice, and thereby the SnS doping level. Phase Purity. The slight variations observed in film composition with varying H2S doses is not connected to the presence of additional phases, such as Sn2S3 or SnS2. This was concluded from Raman spectroscopy, which has a sensitivity to this type of phase impurity that is higher than that of XRD.31

Table 1. Sample Composition by RBS versus the Number of Cycles (thickness) for the Samples Shown in Figure 5, Prepared at 120 °C with 5 s H2S Pulse Lengths no. of ALD cycles

Sn/S ratio

25 50 100 200 500 1000

0.95 0.87 0.94 0.93 0.91 0.92

and possible traces of oxygen were seen in the 25- and 200cycle samples, but this could also be due to the native oxide of the substrate (seen in Figure 6c,d) or postdeposition surface oxidation of the thin SnS films. The density of the SnS film produced by 1000 ALD cycles, based on the atomic surface density obtained from RBS and the TEM measured film thickness, is ∼4.5 g/cm3. This is slightly 2974

DOI: 10.1021/acs.chemmater.6b05323 Chem. Mater. 2017, 29, 2969−2978

Article

Chemistry of Materials

XRD reflection at 15° in 2θ and the Raman peak at 310 cm−1. For these samples, the relative intensity of the π-SnS (222) diffraction peak also drops somewhat in comparison to those of films grown at 80 °C with a large H2S dose. This change is likely related to a certain degree of π-SnS texturing, for instance, due to the presence of the SnS2 phase. The appearance of SnS2 effectively puts an upper limit on the growth temperature of pure π-SnS and, as we have demonstrated, strongly reduces the rate of growth with Sn(acac)2 overall. Optical Characterization. Spectroscopic ellipsometry (SE) was used to characterize the optical properties of the deposited SnS films. The data analysis was complicated by the roughness, the mixed crystalline phases, and the gradients of these phases. To take this situation into account while limiting the number of fitting parameters, we employed a two-layer model for approximating the SnS film. The mainly π-SnS phase films deposited with a H2S dose of 5 s at 120 °C using a varying number of cycles were subjected to a multisample analysis. The optical properties of the solid content of the layers were constrained to be equal for all samples. The compact π-SnS in the bottom layer was fitted using three Gaussian oscillators and one polynomial spline function, such that the Kramers−Kronig relation was fulfilled. A thinner layer was used to model the mixture of air and (mainly) o-SnS grains at the surface (see Figure 10a) by means of Bruggeman effective medium theory.48 The solid SnS part was fitted using the same type of oscillator functions as the bottom layer but with other parameters. While this function was common for all samples, the filling factor of the effective medium was allowed to vary between samples. The thicknesses of the top and bottom layers were also fitted individually for each sample. The results for the refractive index and extinction coefficient of the compact bottom layer of the mainly π-SnS phase are presented in Figure 10c (see the Supporting Information for tabulated data). To validate the ellipsometry analysis, the fitted ellipsometry model was used to calculate the normal incidence absorptance by the transfer matrix method. Good agreement is found with the absorptance obtained from direct measurements of the reflectance and transmittance using an integrating sphere (Figure 10d). The absorption coefficient obtained from the SE analysis fits very well with the dependence of forbidden optical transitions over a direct bandgap (Figure 10e),49 in line with previous observations.15,36

Table 2. Sample Composition by XRF versus H2S Dose for the Samples Shown in Figure 2, Prepared at 120 °C Using 1000 ALD Cycles H2S pulse length (s)

Sn/S ratio

0.4 1.6 2 3 5

1.03 1.01 0.94 0.95 0.93

Raman characterization was performed on the series of samples grown at different temperatures as well as on the series for different H2S doses. Examination of the spectra for SnS prepared at a small H2S dose and a high temperature reveals characteristic peaks at 173, 187, and 224 cm−1 (see Figure 8). The peaks at 187 and 224 cm−1 match well with the Ag modes of o-SnS, while the peak at 173 cm−1 agrees well with the B3g mode of the same phase.26,45,46 As the H2S dose increases and the temperature decreases, such that the formation of π-SnS is favored, new Raman peaks that are not present in the spectrum of o-SnS appear at 110, 123, and 202 cm−1. These three peaks are unique to the π-SnS phase and were previously described for π-SnS nanoparticles.44 Also, the peak found at 187 cm−1 in the mainly o-SnS-containing samples is gradually overtaken by the cubic phase peak at 192 cm−1.44 The Raman spectra also show that the SnS phases are not accompanied by SnS2 and Sn2S3, which have their strongest peaks at ∼312 and ∼307 cm−1, respectively.47 Phase Dependence: Temperature versus H2S Dose. To probe the phase evolution under more extreme combinations of high and low temperatures and large and small H2S doses, four additional samples were prepared. At 80 °C and a large (5 s) H2S dose, a nearly pure π-SnS phase is formed, while a nearly pure orthorhombic phase is formed for a small (0.4 s) dose, as evidenced by GI-XRD at a 0.5° angle of incidence and Raman measurements (Figure 9). XRR measurements on the sample performed at a large H2S dose revealed a density of 4.5 g/cm3 and a thickness of 40 nm, confirming the mass gain of ∼19 ng cm−2 cycle−1 under saturated conditions. Similar results for the H2S dose dependence are obtained at 200 °C. This clearly indicates that the H2S dose dependence is strong enough to control the crystal structure and morphology of the SnS films, even at high temperatures. However, SnS films grown at 200 °C with a large H2S dose also contain an additional phase that can be attributed to SnS2 based on the GI-

Figure 8. Raman spectra of ALD SnS films prepared at (a) different temperatures and (b) different H2S doses. 2975

DOI: 10.1021/acs.chemmater.6b05323 Chem. Mater. 2017, 29, 2969−2978

Article

Chemistry of Materials

Figure 9. (a) GI-XRD pattern of SnS thin films and (b) Raman spectra of ALD SnS films prepared at 80 °C (0.4 and 5 s H2S doses) and 200 °C (0.4 and 5 s H2S doses).

Figure 10. (a and b) Sample models used for SE data analysis. (c) Refractive index (n) and extinction coefficient (k) of π-SnS and o-SnS. (d) Integrated sphere measurements () and ellipsometry results (---) for the absorptance of the π-SnS phase on glass. (e) Extrapolation of the direct forbidden-type bandgap for the π-SnS phase. (f) Exponential Urbach tails for the o-SnS phase shown on a logarithmic scale. The multipliers given in brackets were used for improved visibility.

reproduces this dependence, with tail widths (EU) of 0.29 and 0.31 eV, respectively. The Urbach dependence is fundamentally related to localization of the electronic wave function and is typically observed in disordered or amorphous semiconductors.50 Disorder may be static or dynamic positional or may be substitutional.51 The local variations in the electronic environment blur the band edges and create tails into the bandgap. In the case presented here, the width of the Urbach tail for o-SnS greatly exceeds typical values (∼50 meV) found for amorphous semiconductors52 and also the similarly strong thermal contributions to dynamic disorder in the lattice at room temperature. In line with this, a narrow width of 56 meV is indeed observed when the corresponding optical analysis is performed for π-SnS. It is hard to pinpoint the exact origin of the pronounced Urbach dependence observed in o-SnS. On one hand, XRD and HRTEM show that both phases of SnS are polycrystalline, but an obvious difference lies in their nanoscale morphology where o-SnS is more disordered with its bladed structure.

The extrapolated bandgap of 1.64 eV is also in accordance with previous reports for π-SnS.6,34−36 The optical properties of o-SnS vary more between samples. Focusing on samples prepared with the smallest H2S dose (0.4 s) that consist mainly of o-SnS, we find the bottom layer is again well fitted using three Gaussians and one polynomial spline function. The top layer is constructed using the SnS properties of the bottom layer in a mixed SnS/air Bruggeman effective medium (see Figure 10b). The resulting optical properties of o-SnS are presented in Figure 10c. A closer look at the absorption coefficient (α) for the sample produced at 120 °C reveals that it closely follows an exponential, Urbachtype dependence over two decades in α, thus fitting well to the equation α = α0 exp(hν/EU), where hν is the photon energy and EU (=0.37 eV) is the fitted tail width (Figure 10f). A bandgap is not clearly identified here as the tail extends above a photon energy of 2 eV, far beyond the expected bandgap for bulk o-SnS around 1.1 eV.2,10,26,31−33 The corresponding analysis of the o-SnS samples produced at 80 and 200 °C 2976

DOI: 10.1021/acs.chemmater.6b05323 Chem. Mater. 2017, 29, 2969−2978

Chemistry of Materials



Effects of wave function localization (quantum confinement) have been theoretically shown for few-layer SnS,13 and considering in addition the anisotropic properties of the orthorhombic phase,53 this could feasibly result in the Urbachtype absorption onset. The sensitivity of the optical properties to the morphology and thereby the growth conditions would then explain their cross-sample variability. On the other hand, according to the XRF measurement (Table 2), o-SnS is slightly sulfur deficient, and gap states associated with sulfur vacancies could thus also play a role in the band tailing. In support of this, π-SnS deviates more from stoichiometry, but on the sulfur rich side, and shows a much narrower tail similar to that of highquality semiconductors. In addition, in contrast to nanoscale disorder, S-vacancy-induced band tailing could possibly explain a similarly non-zero absorption tail observed for singlecrystalline o-SnS.53 Indeed, sulfur volatility was previously identified as an issue for the production of high-quality o-SnS.7 As the sulfur content is strongly connected to the evolution of SnS morphology in the SnS−ALD process, the considered contributions to band tailing are indistinguishable here. Further studies are thus required to shed light on whether nanoscale disorder or sulfur deficiency is the main reason for the observed Urbach dependence in o-SnS.

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. ORCID

Olivier Donzel-Gargand: 0000-0002-2101-3746 Carl Hägglund: 0000-0001-6589-3514 Funding

Supported by the Swedish Research Council (621-2014-5599). Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS The authors are grateful to Daniel Primetzhofer for assistance with the RBS measurements and to Fredrik Gustavsson for help with the TEM imaging.



REFERENCES

(1) Sinsermsuksakul, P.; Sun, L.; Lee, S. W.; Park, H. H.; Kim, S. B.; Yang, C.; Gordon, R. G. Overcoming Efficiency Limitations of SnSBased Solar Cells. Adv. Energy Mater. 2014, 4, 1400496. (2) Ramakrishna Reddy, K. T.; Koteswara Reddy, N.; Miles, R. W. Photovoltaic Properties of SnS Based Solar Cells. Sol. Energy Mater. Sol. Cells 2006, 90, 3041−3046. (3) Ikuno, T.; Suzuki, R.; Kitazumi, K.; Takahashi, N.; Kato, N.; Higuchi, K. SnS Thin Film Solar Cells with Zn1-xMgxO Buffer Layers. Appl. Phys. Lett. 2013, 102, 193901. (4) Wangperawong, A.; Hsu, P.-C.; Yee, Y.; Herron, S. M.; Clemens, B. M.; Cui, Y.; Bent, S. F. Bifacial Solar Cell with SnS Absorber by Vapor Transport Deposition. Appl. Phys. Lett. 2014, 105, 173904. (5) Steinmann, V.; Jaramillo, R.; Hartman, K.; Chakraborty, R.; Brandt, R. E.; Poindexter, J. R.; Lee, Y. S.; Sun, L.; Polizzotti, A.; Park, H. H.; Gordon, R. G.; Buonassisi, T. 3.88% Efficient Tin Sulfide Solar Cells Using Congruent Thermal Evaporation. Adv. Mater. (Weinheim, Ger.) 2014, 26, 7488−7492. (6) Garcia-Angelmo, A. R.; Romano-Trujillo, R.; Campos-Á lvarez, J.; Gomez-Daza, O.; Nair, M. T. S.; Nair, P. K. Thin Film Solar Cell of SnS Absorber with Cubic Crystalline Structure. Phys. Status Solidi A 2015, 212, 2332−2340. (7) Banai, R. E.; Horn, M. W.; Brownson, J. R. S. A Review of Tin (Ii) Monosulfide and Its Potential as a Photovoltaic Absorber. Sol. Energy Mater. Sol. Cells 2016, 150, 112−129. (8) Hägglund, C.; Zeltzer, G.; Ruiz, R.; Wangperawong, A.; Roelofs, K. E.; Bent, S. F. Strong Coupling of Plasmon and Nanocavity Modes for Dual-Band, Near-Perfect Absorbers and Ultrathin Photovoltaics. ACS Photonics 2016, 3, 456−463. (9) Parenteau, M.; Carlone, C. Influence of Temperature and Pressure on the Electronic Transitions in SnS and SnSe Semiconductors. Phys. Rev. B: Condens. Matter Mater. Phys. 1990, 41, 5227− 5234. (10) Hegde, S. S.; Kunjomana, A. G.; Chandrasekharan, K. A.; Ramesh, K.; Prashantha, M. Optical and Electrical Properties of SnS Semiconductor Crystals Grown by Physical Vapor Deposition Technique. Phys. B (Amsterdam, Neth.) 2011, 406, 1143−1148. (11) Koteeswara Reddy, N.; Ramakrishna Reddy, K. T. Preparation and Characterisation of Sprayed Tin Sulphide Films Grown at Different Precursor Concentrations. Mater. Chem. Phys. 2007, 102, 13−18. (12) Wang, W.; Leung, K. K.; Fong, W. K.; Wang, S. F.; Hui, Y. Y.; Lau, S. P.; Chen, Z.; Shi, L. J.; Cao, C. B.; Surya, C. Molecular Beam Epitaxy Growth of High Quality P-Doped SnS Van Der Waals Epitaxy on a Graphene Buffer Layer. J. Appl. Phys. (Melville, NY, U. S.) 2012, 111, 093520. (13) Xin, C.; Zheng, J.; Su, Y.; Li, S.; Zhang, B.; Feng, Y.; Pan, F. Few-Layer Tin Sulfide: A New Black-Phosphorus-Analogue 2D Material with a Sizeable Band Gap, Odd−Even Quantum Confine-



CONCLUSIONS We have demonstrated that ALD using Sn(acac)2 and H2S produces different phases of tin monosulfide depending on the process conditions. The recently discovered cubic SnS phase is favored in the early stages of growth but tends to evolve into the orthorhombic SnS phase for thicker films. The cubic phase is smooth and compact, while the orthorhombic phase has a bladelike morphology that results in a high surface roughness. The cubic phase dominates for fully saturating H2S doses, while the orthorhombic phase prevails with a subsaturated sulfur supply. The orthorhombic phase is also favored by increased substrate temperature. Saturated growth is confirmed as being stable at 19 ng cm−2 cycle−1, corresponding to 0.42 Å cycle−1, in the temperature interval of 80−160 °C. At 200 °C, small amounts of tin disulfide are observed in samples grown with large H2S doses. The presence of SnS2 reduces the growth rate and puts an upper limit on the ALD temperature window for purely cubic SnS growth. The cubic and orthorhombic SnS phases differ significantly in their optical properties. The cubic phase has a distinct, direct forbidden-type bandgap of 1.64 eV, while the orthorhombic phase displays a broad Urbach tail that extends up to ∼2 eV and has a width 0.3−0.4 eV. The origin of the tail is ascribed to the nanoscale structural disorder of the bladelike morphology, to sulfur vacancies of o-SnS, or to a combination thereof. Phase-controlled ALD of SnS demonstrated here may prove to be highly useful for applications in nanoelectronics, batteries, and thin film solar cells, including in ultrathin plasmonic solar cells and as the top-most absorber layer of tandem devices.



Article

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.chemmater.6b05323. Tabulated optical constants for cubic phase tin monosulfide (π-SnS) (PDF) 2977

DOI: 10.1021/acs.chemmater.6b05323 Chem. Mater. 2017, 29, 2969−2978

Article

Chemistry of Materials ment Effect, and High Carrier Mobility. J. Phys. Chem. C 2016, 120, 22663−22669. (14) Tripathi, A. M.; Mitra, S. Tin Sulfide (SnS) Nanorods: Structural, Optical and Lithium Storage Property Study. RSC Adv. 2014, 4, 10358−10366. (15) Avellaneda, D.; Delgado, G.; Nair, M. T. S.; Nair, P. K. Structural and Chemical Transformations in SnS Thin Films Used in Chemically Deposited Photovoltaic Cells. Thin Solid Films 2007, 515, 5771−5776. (16) He, H. Y.; Fei, J.; Lu, J. Optical and Electrical Properties of Pure and Sn4+-Doped N-SnS Films Deposited by Chemical Bath Deposition. Mater. Sci. Semicond. Process. 2014, 24, 90−95. (17) Sinsermsuksakul, P.; Hartman, K.; Bok Kim, S.; Heo, J.; Sun, L.; Hejin Park, H.; Chakraborty, R.; Buonassisi, T.; Gordon, R. G. Enhancing the Efficiency of SnS Solar Cells Via Band-Offset Engineering with a Zinc Oxysulfide Buffer Layer. Appl. Phys. Lett. 2013, 102, 053901. (18) Ichimura, M.; Takeuchi, K.; Ono, Y.; Arai, E. Electrochemical Deposition of SnS Thin Films. Thin Solid Films 2000, 361−362, 98− 101. (19) Cheng, S.; Chen, G.; Chen, Y.; Huang, C. Effect of Deposition Potential and Bath Temperature on the Electrodeposition of SnS Film. Opt. Mater. (Amsterdam, Neth.) 2006, 29, 439−444. (20) Dhanapandian, S.; Gowri Manohari, A.; Manoharan, C.; Santhosh Kumar, K.; Mahalingam, T. Optimization of Spray Deposition Parameters for the Formation of Single-Phase Tin Sulfide Thin Films. Mater. Sci. Semicond. Process. 2014, 18, 65−70. (21) Warrier, A. R.; Sajeesh, T. H.; Kartha, C. S.; Vijayakumar, K. P. Determination of Thermal and Electronic Carrier Transport Properties of SnS Thinfilms Using Photothermal Beam Deflection Technique. Mater. Res. Bull. 2012, 47, 3758−3763. (22) Devika, M.; Ramakrishna Reddy, K. T.; Koteeswara Reddy, N.; Ramesh, K.; Ganesan, R.; Gopal, E. S. R.; Gunasekhar, K. R. Microstructure Dependent Physical Properties of Evaporated Tin Sulfide Films. J. Appl. Phys. (Melville, NY, U. S.) 2006, 100, 023518. (23) Devika, M.; Reddy, N. K.; Ramesh, K.; Ganesan, R.; Gunasekhar, K. R.; Gopal, E. S. R.; Reddy, K. T. R. Thickness Effect on the Physical Properties of Evaporated SnS Films. J. Electrochem. Soc. 2007, 154, H67−H73. (24) Wangperawong, A.; Herron, S. M.; Runser, R. R.; Hägglund, C.; Tanskanen, J. T.; Lee, H.-B.-R.; Clemens, B. M.; Bent, S. F. Vapor Transport Deposition and Epitaxy of Orthorhombic SnS on Glass and NaCl Substrates. Appl. Phys. Lett. 2013, 103, 052105. (25) Kim, J. Y.; George, S. M. Tin Monosulfide Thin Films Grown by Atomic Layer Deposition Using Tin 2,4-Pentanedionate and Hydrogen Sulfide. J. Phys. Chem. C 2010, 114, 17597−17603. (26) Sinsermsuksakul, P.; Heo, J.; Noh, W.; Hock, A. S.; Gordon, R. G. Atomic Layer Deposition of Tin Monosulfide Thin Films. Adv. Energy Mater. 2011, 1, 1116−1125. (27) Ran, F.-Y.; Xiao, Z.; Hiramatsu, H.; Hosono, H.; Kamiya, T. Growth of High-Quality SnS Epitaxial Films by H2s Flow Pulsed Laser Deposition. Appl. Phys. Lett. 2014, 104, 072106. (28) George, S. M. Atomic Layer Deposition: An Overview. Chem. Rev. (Washington, DC, U. S.) 2010, 110, 111−131. (29) Leskelä, M.; Ritala, M. Atomic Layer Deposition (ALD): From Precursors to Thin Film Structures. Thin Solid Films 2002, 409, 138− 146. (30) Wiedemeier, H.; von Schnering, H. G. Refinement of the Structures of Ges, Gese, SnS and SnSe. Z. Kristallogr. - New Cryst. Struct. 1978, 148, 295−303. (31) Mathews, N. R.; Anaya, H. B. M.; Cortes-Jacome, M. A.; Angeles-Chavez, C.; Toledo-Antonio, J. A. Tin Sulfide Thin Films by Pulse Electrodeposition: Structural, Morphological, and Optical Properties. J. Electrochem. Soc. 2010, 157, H337−H341. (32) Vidal, J.; Lany, S.; d’Avezac, M.; Zunger, A.; Zakutayev, A.; Francis, J.; Tate, J. Band-Structure, Optical Properties, and Defect Physics of the Photovoltaic Semiconductor SnS. Appl. Phys. Lett. 2012, 100, 032104.

(33) Hartman, K.; Johnson, J. L.; Bertoni, M. I.; Recht, D.; Aziz, M. J.; Scarpulla, M. A.; Buonassisi, T. SnS Thin-Films by RF Sputtering at Room Temperature. Thin Solid Films 2011, 519, 7421−7424. (34) Ahmet, I. Y.; Hill, M. S.; Johnson, A. L.; Peter, L. M. Polymorph-Selective Deposition of High Purity SnS Thin Films from a Single Source Precursor. Chem. Mater. 2015, 27, 7680−7688. (35) Garcia-Angelmo, A. R.; Nair, M. T. S.; Nair, P. K. Evolution of Crystalline Structure in SnS Thin Films Prepared by Chemical Deposition. Solid State Sci. 2014, 30, 26−35. (36) Avellaneda, D.; Nair, M. T. S.; Nair, P. K. Polymorphic Tin Sulfide Thin Films of Zinc Blende and Orthorhombic Structures by Chemical Deposition. J. Electrochem. Soc. 2008, 155, D517−D525. (37) Greyson, E. C.; Barton, J. E.; Odom, T. W. Tetrahedral Zinc Blende Tin Sulfide Nano-and Microcrystals. Small 2006, 2, 368−371. (38) Rabkin, A.; Samuha, S.; Abutbul, R. E.; Ezersky, V.; Meshi, L.; Golan, Y. New Nanocrystalline Materials: A Previously Unknown Simple Cubic Phase in the SnS Binary System. Nano Lett. 2015, 15, 2174−2179. (39) Nair, P. K.; Garcia-Angelmo, A. R.; Nair, M. T. S. Cubic and Orthorhombic SnS Thin-Film Absorbers for Tin Sulfide Solar Cells. Phys. Status Solidi A 2016, 213, 170−177. (40) Abutbul, R. E.; Garcia-Angelmo, A. R.; Burshtein, Z.; Nair, M. T. S.; Nair, P. K.; Golan, Y. Crystal Structure of a Large Cubic Tin Monosulfide Polymorph: An Unraveled Puzzle. CrystEngComm 2016, 18, 5188−5194. (41) Langford, R. M.; Clinton, C. In Situ Lift-out Using a Fib-Sem System. Micron 2004, 35 (7), 607−611. (42) Mayer, M. Simnra User’s Guide. In Report IPP; Max-PlanckInstitut für Plasmaphysik: Garching, Germany, 1997; Vol. 9/113. (43) Sugaki, A.; Kitakaze, A.; Kitazawa, H. Synthesized Tin and TinSilver Sulfide Minerals; Synthetic Sulfide Minerals (XII). Sci. Rep. Tohoku Univ., Ser. 3 1985, 16, 199−211. (44) Abutbul, R. E.; Segev, E.; Zeiri, L.; Ezersky, V.; Makov, G.; Golan, Y. Synthesis and Properties of Nanocrystalline π-SnS - a New Cubic Phase of Tin Sulphide. RSC Adv. 2016, 6, 5848−5855. (45) Sohila, S.; Rajalakshmi, M.; Ghosh, C.; Arora, A. K.; Muthamizhchelvan, C. Optical and Raman Scattering Studies on SnS Nanoparticles. J. Alloys Compd. 2011, 509, 5843−5847. (46) Raadik, T.; Grossberg, M.; Raudoja, J.; Traksmaa, R.; Krustok, J. Temperature-Dependent Photoreflectance of SnS Crystals. J. Phys. Chem. Solids 2013, 74, 1683−1685. (47) Price, L. S.; Parkin, I. P.; Hardy, A. M. E.; Clark, R. J. H.; Hibbert, T. G.; Molloy, K. C. Atmospheric Pressure Chemical Vapor Deposition of Tin Sulfides (SnS, Sn2S3, and SnS2) on Glass. Chem. Mater. 1999, 11, 1792−1799. (48) Niklasson, G. A.; Granqvist, C. G.; Hunderi, O. Effective Medium Models for the Optical Properties of Inhomogeneous Materials. Appl. Opt. 1981, 20, 26−30. (49) Stenzel, O. The Physics of Thin Film Optical Spectra: An Introduction, 2nd ed.; Springer International Publishing: Cham, Switzerland, 2016; Vol. 44, p 359. (50) John, S.; Soukoulis, C.; Cohen, M. H.; Economou, E. N. Theory of Electron Band Tails and the Urbach Optical-Absorption Edge. Phys. Rev. Lett. 1986, 57, 1777−1780. (51) Soukoulis, C. M.; Economou, E. N. Electronic Localization in Disordered Systems. Waves in Random Media 1999, 9, 255−269. (52) Sritrakool, W.; Sa-yakanit, V.; Glyde, H. R. Band Tails in Disordered Systems. Phys. Rev. B: Condens. Matter Mater. Phys. 1986, 33, 1199−1202. (53) Banai, R. E.; Burton, L. A.; Choi, S. G.; Hofherr, F.; Sorgenfrei, T.; Walsh, A.; To, B.; Cröll, A.; Brownson, J. R. S. Ellipsometric Characterization and Density-Functional Theory Analysis of Anisotropic Optical Properties of Single-Crystal A-SnS. J. Appl. Phys. 2014, 116, 013511.

2978

DOI: 10.1021/acs.chemmater.6b05323 Chem. Mater. 2017, 29, 2969−2978