Journal of Colloid and Interface Science 415 (2014) 103–110
Contents lists available at ScienceDirect
Journal of Colloid and Interface Science www.elsevier.com/locate/jcis
Colloidally stable selenium@copper selenide core@shell nanoparticles as selenium source for manufacturing of copper–indium–selenide solar cells Hailong Dong a, Aina Quintilla b, Marco Cemernjak b, Radian Popescu c, Dagmar Gerthsen c, Erik Ahlswede b,⇑, Claus Feldmann a,⇑ a
Institut für Anorganische Chemie, Karlsruhe Institute of Technology (KIT), Engesserstraße 15, D-76131 Karlsruhe, Germany Zentrum für Sonnenenergie- und Wasserstoffforschung Baden-Württemberg (ZSW), Industriestraße 6, D-70565 Stuttgart, Germany c Laboratorium für Elektronenmikroskopie, Karlsruhe Institute of Technology (KIT), Engesserstraße 7, D-76131 Karlsruhe, Germany b
a r t i c l e
i n f o
Article history: Received 18 June 2013 Accepted 3 October 2013 Available online 17 October 2013 Keywords: Selenium Copper selenide Core–shell structure CIS Solar cell
a b s t r a c t Selenium nanoparticles with diameters of 100–400 nm are prepared via hydrazine-driven reduction of selenious acid. The as-prepared amorphous, red selenium (a-Se) particles were neither a stable phase nor were they colloidally stable. Due to phase transition to crystalline (trigonal), grey selenium (t-Se) at or even below room temperature, the particles merged rapidly and recrystallized as micronsized crystal needles. As a consequence, such Se particles were not suited for layer deposition and as a precursor to manufacture thin-film CIS (copper indium selenide/CuInSe2) solar cells. To overcome this restriction, Se@CuSe core@shell particles are presented here. For these Se@CuSe core@shell nanoparticles, the phase transition a-Se ? t-Se is shifted to temperatures higher than 100 °C. Moreover, a spherical shape of the particles is retained even after phase transition. Composition and structure of the Se@CuSe core@shell nanostructure are evidenced by electron microscopy (SEM/STEM), DLS, XRD, FT-IR and line-scan EDXS. As a conceptual study, the newly formed Se@CuSe core@shell nanostructures with CuSe acting as a protecting layer to increase the phase-transition temperature and to improve the colloidal stability were used as a selenium precursor for manufacturing of thin-film CIS solar cells and already lead to conversion efficiencies up to 3%. Ó 2013 Elsevier Inc. All rights reserved.
1. Introduction Thin-film solar cells represent an important alternative to crystalline silicon solar cells. Here, the most efficient thin-film technology is related to CuInSe2 (CIS: copper indium selenide) with record values above 20% if gallium is incorporated as well (CIGS: copper indium gallium selenide) [1].Conventionally, CIS/CIGS layers are deposited by applying high-vacuum techniques. Besides thermal co-evaporation of the relevant elements, including selenium, sequential processes are often used, where typically Se is deposited by thermal evaporation as a thin top coating on a layer stack of Cu-/In-/Ga-containing precursors [2]. In a second step, these layers are chemically converted in a thermal process at 550–600 °C – called ‘‘selenization’’ – into the desired chalcopyrite structure of the CIS/CIGS absorber. Ideally the Se cap acts as a sufficient Se source during the annealing step. Usually, an additional ⇑ Corresponding authors. Fax: +49 721 60844892 (C. Feldmann), fax: +49 711 7870230 (E. Ahlswede). E-mail addresses: [email protected] (E. Ahlswede), claus.feldmann@ kit.edu (C. Feldmann). 0021-9797/$ - see front matter Ó 2013 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.jcis.2013.10.001
treatment in harmful H2S atmosphere is applied during the heating step to increase the electronic quality of the absorber layer [3]. In recent years, vacuum-free deposition of suitable precursor pastes has attracted increasing attention in view of low-cost high-throughput production of CIS thin-film solar cells [4,5]. One possibility for paste formulation is to use nanoparticles, which have already been reported for elemental metals (Cu, In, CuIn/Cu11In9) [6–8], metal oxides [9], and metal selenides (CuInSe2, Cu2Se/ CuSe, In2Se3) [10–13]. For reaching high efficiencies, the precursor layers were always annealed in a selenium vapour atmosphere to compensate for a possible loss of selenium during the heating step or to reach denser polycrystalline layers. So far, only conventional sources of selenium were investigated, either as an evaporated capping layer or as additional selenium vapour in the gas phase. The use of inks containing Se nanoparticles is, however, a kind of a missing link. The availability of colloidally stable Se inks would allow avoiding vacuum-techniques and gas-phase deposition completely. Ideally, just a simple heat treatment of the as-deposited particulate metal and selenium precursor layers could generate dense and crystalline CIS layers. This option is however hampered due to the room-temperature phase transition of selenium (31 °C
104
H. Dong et al. / Journal of Colloid and Interface Science 415 (2014) 103–110
for bulk-Se) [14]. As a consequence, Se particles are of limited colloidal stability in suspension and show rapid agglomeration and merging of particles, followed by re-crystallization of large crystal needles. Such Se particles are not suitable for layer deposition and manufacturing of thin-film solar cells. In the following, we report on the synthesis of Se particles that are capped by CuSe to form Se@CuSe core@shell nanostructures. Based on this newly introduced CuSe capping, the phase-transition temperature of elemental selenium is shifted to >100 °C. As a result, colloidally stable suspensions of Se particles could be obtained and were manufactured - together with Cu11In9 nanoparticles - in a first conceptual study via a vacuum-free process to CIS thin-film solar cells. The strategy of repelling the Se phase transition and increasing the colloidal stability by metal selenide cappings, more generally, can also be interesting for application of Se particles in mechanical sensors, electrical rectifier, xerography, or tumour therapy [15,16].
2. Experimental section 2.1. Materials, synthesis and thin-film deposition General considerations: Selenious acid (H2SeO3, 98%), hydrazine monohydrate (N2H4H2O, 98%), copper(II) acetate (Cu(CH3COO)2, 98%) and polyvinylpyrrolidone (PVP, Mw = 40,000) were purchased from Sigma–Aldrich. Diethylene glycol (DEG, C4H10O4, 99%) was obtained from Alfa Aesar. All chemicals were of analytical grade and used as received. Synthesis of Se@CuSe core@shell nanoparticles: In a typical recipe, N2H4H2O (10–40 mmol, Table 1) and 200 mg PVP were added to 50 ml of DEG in a three-necked flask. This mixture was dispersed to form a homogeneous solution by constant stirring (solution I). Solution I was then cooled to 0 °C with an ice bath. Thereafter, 0.8 mmol of H2SeO3 in 3 ml of DEG (solution II) were injected into the precooled solution I. The reaction was allowed to proceed for different periods of time (t, Table 1). The gradual colour shift from a colourless solution to an orange suspension allows following the course of the reaction even by the naked eye. Finally, all residual N2H4H2O in the reaction solution was removed by vacuum distillation for 30 min at 0 °C, followed by 150 min at room temperature. Thereafter, 0.04 mmol of Cu(CH3COO)2 dissolved in 3 ml of DEG (solution III) were added dropwise. The addition of Cu(CH3COO)2 again resulted in a colour change, from orange via orangered to brick-red. Detailed information regarding the experimental parameters of the synthesis is listed in Table 1. Finally, the suspensions were diluted with demineralised water and the Se@CuSe core–shell nanoparticles were collected via centrifugation, washed three times with demineralised water, and then redispersed in demineralised water for further analysis and treatment. Preparation and processing of Se@CuSe inks: Suitable suspensions of Se@CuSe nanoparticles were obtained by redispersing the as-
prepared powders in ethanol with a solid load of typically 200 mg ml1. These precursor inks were prepared right before layer deposition in order to prevent particle agglomeration over time. Layer deposition was done via doctor blading, using an Erichsen film applicator and an adjustable blade by Zehntner, on Molybdenum covered sodalime-glass substrates of 1 mm thickness. Two different strategies were applied: Scenario 1 – Se@CuSe nanoparticles were co-dispersed with Cu11In9 nanoparticles in ethanol and deposited thereafter to form a mixed Se@CuSe/Cu11In9 nanoparticle film. Scenario 2 – Alternatively, a Cu11In9 layer was deposited first, followed by a stacked layer of the Se@CuSe nanoparticles on top of the Cu11In9 particle film. The synthesis of the Cu11In9 nanoparticles was performed as described in [8]. Selenization of thin-films and solar cell preparation: To convert the precursor films to chalcopyrite-type CIS absorber layers, selenization of the thin-films was performed in a tube furnace at 550 °C, eventually in nitrogen-diluted selenium vapour, as described in [8]. The CIS absorber layers were further treated by selective KCN etching to remove undesired binary CuSe phases, which would typically lead to shunted cell behaviour due to its good conductivity. The CIS absorber layers were finally completed by a wet-chemically processed CdS buffer layer, a sputtered ZnO buffer layer (i-ZnO) and a transparent ZnO:Al front-contact layer to obtain functional solar cells using standard procedures as described in [17]. 2.2. Materials characterization and analytical tools Dynamic light scattering (DLS) was performed with a Nanosizer ZS from Malvern Instruments (equipped with a He–Ne laser emitting at 633 nm), detection via non-invasive back-scattering at an angle of 173 °, 256 detector channels, polystyrene cuvettes). For analysis, the as-prepared Se@CuSe nanoparticles were redispersed in demineralised water by ultrasonic treatment for 15 min. Scanning electron microscopy (SEM) was performed on a Zeiss Supra 40 VP, using an acceleration voltage of 20 kV and a working distance of 2 mm to analyse the size distribution and shape of the Se@CuSe nanoparticles. All samples were prepared by evaporating a single drop of a dispersion of the as-prepared Se@CuSe nanoparticles in demineralised water at room temperature in air. The thinfilm morphology of precursor layers, selenized layers and completed solar cells were studied with a XL30 SFEG Sirion from FEI Company, using a 5 kV acceleration voltage and working distances of 5.9, 4.3 and 3.4 mm. Scanning transmission electron microscopy in the high-angle annular dark-field mode (HAADF STEM) was conducted with an aberration-corrected FEI Titan3 80–300 at 300 kV. Samples for STEM were prepared by evaporating water-based suspensions on an amorphous carbon (Lacey-) film copper grid. EDXS line profiles were recorded with an FEI Titan3 80–300 microscope by applying a drift-correction routine via cross correlation of several images, which yields a local precision better than
Table 1 Experimental conditions to adjust the mean size of Se@CuSe core@shell nanoparticles. Sample
N2H4H2O (mmol)
H2SeO3 (mmol)
cðN2 H4 H2 OÞ r cðH2 SeO3 Þ
Reaction time t (min)
Cu(Ac)2 (mmol)
Average diameter (nm)
A B C D E F
10 20 20 20 40 40
0.8 0.8 0.8 0.8 0.8 0.8
13 25 25 25 50 50
90 90 75 30 30 30
0.04 0.04 0.04 0.04 0.04 0.16
410(90) 310(63) 179(48) 162(48) 100(27) 105(29)
(r: Molar ratio of N2H4H2O and H2SeO3). (t: Reaction time of N2H4H2O and H2SeO3). (: Mean diameter according to DLS).
H. Dong et al. / Journal of Colloid and Interface Science 415 (2014) 103–110
1.0 nm. The drift-corrected EDXS line profiles were taken with a probe diameter of 0.5 nm and a distance of about 1.0 nm between two measuring points. With respect to a quantification of the EDX spectra from core–shell nanoparticles one has to keep in mind that only compositions are obtained that are averaged along the electron-beam direction. In concrete, the whole volume along the electron trajectory contributes to the detected X-ray signal. To evaluate the composition of different shells of a core–shell nanoparticle, a procedure was developed that is described in detail in [18]. X-ray powder diffraction (XRD) analysis was carried out with a Stoe Stadi-P diffractometer using Ge-monochromatized Cu Ka1 radiation. Differential scanning calorimetry (DSC) and thermogravimetry (TG) were performed with a STA 409C instrument from NETZSCH (Selb, Germany). Measurements were performed with the following conditions: 10 °C min–1 as heating rate; 30–500 °C as temperature range; 16.8 mg as sample weight; nitrogen atmosphere; alumina crucibles for sample and reference. Fourier-transform infrared spectra (FT-IR) were recorded on a Bruker Vertex 70 FT-IR spectrometer using KBr pellets. For direct comparison, 400 mg of dried KBr were carefully pestled with 0.3 mg of the sample and pressed to a thin pellet. In addition, spectra of the as-prepared Se@CuSe nanoparticles were subjected to a correction of scattering effects to allow for comparison to reference spectra. Current–voltage curves were measured using a Keithley 238 source-measuring unit under simulated AM 1.5 global solar irradiation with an ORIEL sun simulator at 100 mW cm–2 to extract the basic solar-cell characteristics from devices approximately 0.24 cm2 in size. External quantum efficiency measurements (EQE) were performed using a setup from Optosolar.
3. Results and discussion 3.1. Colloidal stability and phase transition of Se particles Colloidal particles of elemental selenium have been prepared by different types of synthesis. Most often, selenious acid (H2SeO3) is reduced by hydrazine (N2H4) [19,20]. Alternatively, reduction of selenious acid with ascorbic acid [21], dismutation of SeO2 [22], and melting of bulk-Se in high-boiling solvents [23] were applied. Based on these strategies mesoscaled Se particles with diameters
105
Fig. 1. Scheme illustrating the synthesis, agglomeration and crystallization of amorphous, red Se as well as the formation of Se@CuSe core@shell nanoparticles for stable inks and subsequent layer deposition.
of 200 nm to 2 lm were obtained most often [15,19,20,22]. Se particles with diameters of 100 nm or below are less often described [20]. Whatever approach was used to synthesize Se particles, the phase-transition temperature of Se (31 °C for bulk-Se [14]) is known as a general restriction. At this temperature the amorphous red phase of elemental selenium (a-Se) that is obtained - in accordance with Ostwald’s step rule – via low-temperature liquid-phase synthesis, shows a phase transition to the thermodynamically stable phase of crystalline, trigonal, grey selenium (t-Se). This phase transition, however, results in a rapid merging of the as-prepared a-Se particles, followed by instantaneous formation of micronsized crystal needles of t-Se (Figs. 1 and 2). For nano- and micronsized particles exhibiting huge specific surfaces, the a-Se ? t-Se phase transition is observed at even lower temperatures as compared to bulk-Se – thus, below room temperature [15,20–25]. In fact, even when kept at about 0 °C, the phase transition occurs after certain period in time (typically 1–3 days). The phase transition can be easily followed by the naked eye due to the colour change of suspensions/powders from orange to grey. This situation is well known and has been described frequently
Fig. 2. SEM images showing the merging (A) and crystallization (B) while amorphous, red selenium (a-Se) is passing through a phase transition to crystalline, grey selenium (t-Se).
106
H. Dong et al. / Journal of Colloid and Interface Science 415 (2014) 103–110
[15,20–25]. In sum, the phase transition severely restricts practical handling of Se particles in view of storing for certain period, gentle heating (e.g., due to centrifugation or redispersion) as well as regarding layer deposition or printing of thin-films. Thus, Se particles and suspensions thereof are not suited for manufacturing thinfilm solar cells as such. To obtain suspensions of Se particles with sufficient colloidal and phase stability that are suitable for thin-film deposition and manufacturing of CIS solar cells, we applied a hydrazine-driven reduction of selenious acid (H2SeO3) as suggested by Smith [19] and Xia [20] (Fig. 1). By variation of the N2H4:H2SeO3 ratio and the duration of the reaction as well as by addition of polyvinylpyrrolidone (PVP), the diameter of the as-prepared Se particles was controlled (Table 1). The synthesis results in uniform spherical a-Se particles that, however, show the well-known rapid phase transition with merging of particles and formation of large t-Se crystal needles (Figs. 1 and 2). To increase phase and colloidal stability, the as-prepared a-Se particles were capped with CuSe (Fig. 1). As CuSe typically is an impurity phase in view of many applications of selenium, the use of Se@CuSe core@shell nanostructures for manufacturing of CuInSe2-based solar cells is not a limitation. The strategy of synthesising Se@CuSe core@shell nanostructures is schematically depicted in Fig. 1. Subsequent to hydrazine-induced
synthesis of a-Se particles applying DEG as a solvent, the suspension was kept at 0 °C. This low temperature allows reducing the reaction rate, which supports particle growth and size uniformity. The formation of a-Se can be followed by the colour change from a colourless solution to an orange suspension. Prior to addition of Cu(CH3COO)2, residual N2H4H2O had to be removed via vacuum distillation, since Cu2+ would have been reduced to elemental Cu otherwise. The addition of Cu(CH3COO)2 again resulted in a colour change of the suspension from orange via orange-red to brick-red. Finally, the Se@CuSe nanoparticles were washed by repeated centrifugation and redispersion, and finally, obtained as a brick-red powder that can be easily redispersed (e.g. in ethanol) for manufacturing CIS solar cells. Powder samples as well as suspensions of Se@CuSe core@shell nanoparticles show sufficient stability in view of practical handling without any phase transition to t-Se.
3.2. Particle size and chemical composition of Se@CuSe core@shell nanoparticles Size and size distribution of the as-prepared Se@CuSe core@ shell nanoparticles were first verified via dynamic light scattering (DLS) based on powder samples that were redispersed in demineralised water (Fig. 3). To this concern, particles with relatively
Fig. 3. Particle size of Se@CuSe core@shell nanoparticles according to SEM and DLS analysis (samples A–E according to Table 1; DLS analysis in demineralised water).
H. Dong et al. / Journal of Colloid and Interface Science 415 (2014) 103–110
Fig. 4. Differential scanning calorimetry (DSC) and thermogravimetry (TG) of asprepared Se@CuSe core@shell nanoparticles (sample E, cf. Table 1) showing the onset of phase transition a-Se ? t-Se at 107 °C and the onset of the melting point at 217 °C (A); (B) crystallinity and spherical shape of Se@CuSe powders after sintering at 175 °C with XRD and SEM (reference: bulk-Se – ICDD No. 6-362; arrow indicating CuSe as the shell).
narrow size distribution and clearly different mean diameters of 410(90) nm (sample A), 310(63) nm (sample B), 179(48) nm (sample C), 162(48) nm (sample D) and 100(27) nm (sample E) were obtained. Size control is predominantly guaranteed by variation of the molar ratio N2H4:H2SeO3 (r) and the reaction time (t) (Table 1). At constant r (cf. samples B–D), the size of the Se@CuSe nanoparticles can be reduced by decreasing t (310 nm ? 162 nm). With t kept constant (cf. samples A/B, D/E), the average particle size can be decreased by increasing r (410 nm ? 310 nm, 162 nm ? 100 nm). Scanning electron microscopy (SEM) confirms particle size and size distribution of the as-prepared Se@CuSe samples (Fig. 3, cf. Fig. 7). Due to the relatively uniform size distribution, partial formation of dense-packed particle layers is
Fig. 5. FT-IR spectra of PVP-capped Se@CuSe core@shell nanoparticles (sample E, cf. Table 1; pure PVP as a reference).
107
observed. While the samples A to F were prepared to study the relevant experimental conditions as well as to perform materials characterization, sample E with the smallest diameter was used for solar-cell manufacturing. Composition and structure of the Se@CuSe nanoparticles were investigated by X-ray powder diffraction (XRD), differential scanning calorimetry (DSC) and Fourier-transform infrared spectroscopy (FT-IR). XRD of as-prepared brick-red powder samples, first of all, do not show any Bragg peaks, indicating the non-crystalline nature of a-Se (cf. Fig. 6). Moreover, no Bragg peaks of any CuxSey phase are visible likewise. DSC shows the phase transition a-Se ? t-Se for Se@CuSe with its onset at 107 °C (Fig. 4a). Thus, the phase-transition temperature of the nanomaterial is significantly increased, even when comparing to bulk-Se (31 °C) [14]. Thermogravimetry (TG), furthermore, indicates that no weight loss (e.g. due to evaporation of Se) occurs in a temperature range up to the melting point with its onset at 217 °C (melting point of bulkSe: 221 °C) (Fig. 4a). Even more surprisingly, the spherical shape of the Se@CuSe core@shell nanoparticles is still retained after sintering of powders at 175 °C, although the samples clearly show the crystalline structure of t-Se thereafter (Fig. 4b). In addition, the main Bragg peak of CuSe/Klockmannite becomes visible with low intensity at a two theta value of 28°. Altogether, the significantly increased phase-transition temperature and the spherical shape retaining after sintering impressively confirm the stability of the Se@CuSe core@shell nanostructure – and thus the effectiveness of the concept. The effect of the increased phase-transition temperature and the improved colloidal stability of the Se@CuSe core@shell nanoparticles, in sum, can be ascribed to CuSe serving as a protecting layer. On the one hand, CuSe obviously stabilizes the red a-Se modification. Traces of Cu2+ that might diffuse into the inner Se particle core could further support this effect. Moreover, the CuSe shell separates the nanoparticles from each other, so that a direct contact of the selenium cores to each other is excluded. Although Ag2Se-cappings – with relatively high thickness of about 50 nm – on Se particles were already used for Se-based photonic crystals with thermally switchable stop bands [26,27], the possibility to increase the phase stability and colloidal stability of Se particles is here shown for the first time. FT-IR spectra indicate the presence of PVP as a surface capping of the Se@CuSe nanostructure (Fig. 5). Thus, weak vibrational bands at 3050–2800 cm1 (v(CAH)), 1800–1550 cm1 (v(C@O)), and the fingerprint area (1500–600 cm1) are in good agreement with PVP reference spectra. The additional strong band at 3700– 3250 cm1 (v(OAH)) can be attributed to OH-containing solvents (water, ethanol) adhered on the particle surface. In order to prove the presence of a CuSe shell encapsulating the Se core, a copper-rich sample with a higher amount of CuSe (sample F:Se:Cu = 5:1) was prepared in addition to samples A–E exhibiting a lower Cu concentration (samples A–E: Se:Cu = 20:1, Table 1). According to DLS and SEM (Fig. 3), this Cu-rich sample F is very comparable to sample E in terms of particle size and size distribution (Table 1). XRD of the as-prepared Cu-rich sample F indeed evidences the presence of crystalline CuxSey phases (Fig. 6). Thus, CuSe/Klockmannite is visible as the predominant phase; Cu5Se4/Athabascaite is observed as a side-phase. When annealing the Cu-rich Se@CuSe sample F to 175 °C, crystalline t-Se is obtained as expected, in addition to CuSe/Klockmannite (Fig. 6). Finally, the core@shell nanostructure of Se@CuSe was validated based on HAADF STEM and EDXS line-profile analyses. Fig. 7c shows a HAADF-STEM image of a single particle of sample B with the dashed line indicating the position of an EDXS line-profile analysis. Quantification of the EDXS spectra yields the composition profile shown in Fig. 7d which demonstrates the core@shell structure. Here, a significant rising of the copper concentration is observed for the surface-region of the Se@CuSe core@shell
108
H. Dong et al. / Journal of Colloid and Interface Science 415 (2014) 103–110
Fig. 6. XRD patterns of as-prepared and sintered Se@CuSe core@shell nanoparticles (sample B and Cu-rich sample F; cf. Table 1) with reference diffractograms: greySe/t-Se – ICDD No. 6-362/black lines; CuSe/Klockmannite – ICDD No. 34-171/dark grey lines; Cu5Se4/Athabascaite – ICDD No. 21-1016/light grey lines.
Fig. 7. SEM overview (A and B) images, HAADF STEM image (C), concentration profile obtained from EDXS line-profile analysis (D, along dashed line as indicated in C), scheme with core and shell composition (E) of the Se@CuSe core@shell particle (sample B, cf. Table 1) after evaluation of the composition profile (D) according to the procedure described in [18].
nanostructure. The inner section of the particle, in contrast, shows a high Se concentration. Cu is nevertheless as well visible for the inner particle core since the electron beam still passes the CuSe shell (Fig. 7d) [18]. Using the evaluation procedure described in Ref. [18], the composition of the shell and core can be reconstructed as shown schematically in Fig. 7e. It is noted that the relatively large error for the composition given in Fig. 7e resulted
from the evaluation procedure and is not caused by the statistics of the EDX spectra. In sum, the as-prepared Se@CuSe core@shell nanoparticles exhibit an inner core of almost pure elemental selenium with a diameter of about 340 nm and a capping significantly enriched with Cu and a thickness of about 20 nm. Together with the XRD results, these findings validate the presence of a particle core consisting of almost pure Se and a shell of CuSe.
H. Dong et al. / Journal of Colloid and Interface Science 415 (2014) 103–110
109
3.3. Use of Se@CuSe core@shell nanoparticles as precursor for manufacturing of CIS thin-film solar cells As the diffusion of Cu2+/In3+/Se2- in the solid-state is slow, application of metal oxides (Cu2O/CuO, In2O3, CuInO2) or metal selenides (Cu2Se/CuSe, In2Se3, CuInSe2) as nanoscale CIS precursors can be accomplished by slow intermixing of phases, phase inhomogeneities, imperfect healing of pores between the initial nanoparticles, and a limited electrical contact between the grain boundaries. All these restrictions reduce the solar-cell performance and efficiency. In principle, this could be avoided by applying nanoscaled elemental metals (Cu, In, CuxIny) and elemental selenium particles as reactive precursors. Conversion to CuInSe2 would be instantaneously possible – ideally without any vacuum-deposition process or gas-phase selenization. Moreover, the interparticular pores should be filled due to the huge volume expansion when reacting metals to metal selenides. The potential of the here prepared Se@CuSe core@shell nanoparticles for manufacturing CIS solar cells is investigated as a conceptual study in combination with Cu11In9 nanoparticles, which have been already successfully developed and tested in CIS solar cells [8]. While sample E (Table 1) exhibits the smallest diameter, it was used for solar-cell manufacturing in the following throughout. The preparation of dense, homogenous films is not yet as convenient as for the Cu11In9 system but well feasible, as can be seen in Fig. 8. Although the Se@CuSe precursor together with Cu11In9 could make an additional Se atmosphere obsolete during the chemical conversion step, at the same time too much Se may cause problems with excessive MoSe2 formation at the solar cell’s back contact. Thus, precise optimization of the relevant conditions will be necessary as a next step. By now, different routes were evaluated, including mixed thin-films of Se@CuSe core@shell nanoparticles and Cu11In9 nanoparticles (scenario 1) and a capping-layer of Se@CuSe core@shell nanoparticles deposited on top of a Cu11In9 thin-film (scenario 2). To the latter concern, Fig. 8 exemplarily shows a thick layer of Se@CuSe on top of a thin Cu11In9 precursor layer. Notably, the diameters of the Cu11In9 particles (20 nm) and of the layer thickness (250 nm) are much smaller as compared to the Se@CuSe precursor (particle diameter: 100 nm, layer thickness: 600 nm). In addition to the above scenarios 1 and 2, different selenization environments were evaluated: The sample stacks were either heated in pure nitrogen or in selenium rich atmosphere for different times. Subsequent to sintering and selenization of the precursor layers at 550 °C [8,17], densification and crystal growth are clearly visible (Fig. 9). Crystal morphology and layer structure are significantly different from the precursor layer, indicating the reaction between Se@CuSe and Cu11In9. Taking into account that these first lab-cells were manufactured under non-optimized conditions (e.g. in terms of chemical composition, Cu:In:Se ratio, particle diameter, layer thickness and morphology, duration/temperature of sintering), the resulting CIS solar cells show promising performance for both scenarios of adding the Se@CuSe precursor. Thus, efficiencies of 1.8% for the Se@CuSe-Cu11In9 mixed film (scenario 1) and of 3.0% for the Se@CuSe capping layer (scenario 2) were obtained. Current–voltage characteristics of the best lab-cells are shown in Fig. 10. Regarding the different conditions of selenization that were tested, currently, the best results could still be obtained via the classical reference procedure (550 °C, 1 h) with an additional Se atmosphere during the tube furnace process [8,17]. In view of scenario 1 (mixed film of Se@CuSe and Cu11In9), there are hints that the formation of an incomplete CIS crust on top of a bottom-layer of much smaller grains could be avoided in favor of a more homogenous morphology (Fig. 9). Such double-layered crust structures – also visible for scenario 2 (Fig. 9) – are quite often observed and generally assumed as disadvantageous in terms of cell
Fig. 8. Doctor-bladed thin-film of Se@CuSe core@shell nanoparticles as capping on top of pre-deposited Cu11In9 nanoparticle thin-film (scenario 2).
Fig. 9. CIS absorber layers after selenization (550 °C, 1 h) – scenario 1: Se@CuSe core@shell nanoparticles and Cu11In9 nanoparticles mixed in suspension prior to thin-film deposition; scenario 2: Se@CuSe core@shell nanoparticles as a capping on top of a pre-deposited Cu11In9 nanoparticle layer.
Fig. 10. Current–voltage characteristics of the best CIS solar cell manufactured with Se@CuSe core@shell nanoparticles and Cu11In9 nanoparticles as precursors.
performance and efficiency [4–13]. The realization of a uniform and dense CIS crystal layer with crystals of similar size could thus be another beneficial aspect of introducing Se@CuSe core@shell nanoparticles as selenium precursor. Currently, scenario 1 is unfortunately accompanied by certain oxidation of the Mo back contact under formation of MoSe2. As a conceptual study with the function of lab-scale CIS solar cells being proven, the new Se@CuSe precursor already leads to promising results. Much more efforts and optimization of the experimental conditions (in terms of chemical composition, Cu:In:Se ratio, particle diameter, layer thickness
110
H. Dong et al. / Journal of Colloid and Interface Science 415 (2014) 103–110
and morphology, duration/temperature of sintering, etc.) are, however, necessary to improve the morphology of the CIS absorber layer and to increase the overall solar-cell performance.
Nanostructures (CFN) of the Deutsche Forschungsgemeinschaft (DFG) at the Karlsruhe Institute of Technology (KIT) for financial support.
4. Conclusions
References
Se@CuSe core@shell nanoparticles, 100–400 nm in total diameter, show a significantly higher phase stability as well as a higher colloidal stability as compared to non-capped selenium particles. Thus, the well-known phase transition of amorphous, red selenium (a-Se) to crystalline (trigonal), grey selenium (t-Se) occurs at 31 °C for bulk-Se and occurs even below room temperature for nano-/ micronsized Se. The as-prepared Se@CuSe core@shell nanoparticles with an inner core of elemental selenium (340 nm in diameter) and a CuSe capping (20 nm in diameter) show the onset of the phase transition at a temperature as high as 107 °C. These core– shell particles do neither show any merging of particles nor any formation of crystal needles of t-Se. Based on the increased phase and colloid stability, the Se@CuSe core@shell nanoparticles can be dispersed as a stable precursor suspension for thin-film deposition and manufacturing of thin-film CIS (copper indium selenide/CuInSe2) solar cells. First lab-scale CIS solar cells in a conceptual study show promising performances with conversion efficiencies of currently up to 3%. In addition to CIS solar cells, the concept of stabilizing Se nano-/mesoscaled particles by metal selenide cappings could be useful for other areas of application, such as the use of Se particles in mechanical sensors, electrical rectifiers, xerography, or tumour therapy. Acknowledgments The authors are grateful to the German Ministerium für Bildung und Forschung (BMBF) for funding within the project ‘‘Nanopartikuläre Dünnschicht-Solarzellen – Grundlagen und Prozesstechnologie (NanoPV)’’. Here, we explicitly thank Dr. Karen Köhler (Bayer Technology Services GmbH, Leverkusen) for fruitful collaboration. Finally, we acknowledge the Center of Functional
[1] P. Jackson, D. Hariskos, E. Lotter, S. Paetel, R. Wuerz, R. Menner, W. Wischmann, M. Powalla, Prog. Photovolt: Res. Appl. 19 (2011) 894. [2] V. Probst, W. Stetter, W. Riedl, H. Vogt, M. Wendl, H. Calwer, S. Zweigart, K.D. Ufert, B. Freienstein, H. Cerva, F.H. Karg, Thin Solid Films 387 (2001) 262. [3] S. Niki, M. Contreras, I. Repins, M. Powalla, K. Kushiya, S. Ishizuka, K. Matsubara, Prog. Photovolt: Res. Appl. 18 (2010) 453 (Review). [4] J. Hibberd, E. Chassaing, W. Liu, D.B. Mitzi, D. Lincot, A.N. Tiwari, Prog. Photovolt: Res. Appl. 18 (2010) 434 (Review). [5] T. Todorov, D.B. Mitzi, Eur. J. Inorg. Chem. (2010) 17 (Review). [6] F.S. Chen, J.S. Ma, C.H. Lu, Ceram. Int. 38 (2012) 5319. [7] J. Ungelenk, V. Haug, A. Quintilla, E. Ahlswede, Phys. Status Sol. 4 (2010) 58. [8] C. Kind, C. Feldmann, A. Quintilla, E. Ahlswede, Chem. Mater. 23 (2011) 5269. [9] V.K. Kapur, A. Bansal, P. Le, O.I. Asensio, Thin Solid Films 431–432 (2003) 53. [10] K. Kim, Y.-J. Eo, A. Cho, J. Gwak, J.H. Yun, K. Shin, S.K. Ahn, S.H. Park, K. Yoon, S.J. Ahn, J. Mater. Chem. 22 (2012) 8444. [11] A. Cho, S. Ahn, J.H. Yun, J. Gwak, S.K. Ahn, K. Shin, H. Song, K.H. Yoon, Solar Cells 109 (2013) 17. [12] Q. Guo, S.J. Kim, M. Kar, W.N. Shafarman, R.W. Birkmire, E.A. Stach, R. Agrawal, H.W. Hillhouse, Nano Letters 8 (2008) 2982. [13] V.A. Akhavan, B.W. Goodfellow, M.G. Panthani, D.K. Reid, D.J. Hellebusch, T. Adachi, B.A. Korgel, Energy Environ. Sci. 3 (2010) 1600. [14] R.A. Zingaro, W.C. Cooper (Eds.), Selenium, Van Nostrand Reinhold, New York, 1974. [15] X. Cao, Y. Xie, L. Li, Adv. Mater. 15 (2003) 1914. [16] L. Cai, N.C.Y. You, H. Lu, L.N. Mu, O.Y. Lu, S.Z. Yu, A.D. Le, J. Marshall, D. Herber, Z.F. Zhang, Cancer 106 (2006) 2345. [17] M. Powalla, B. Dimmler, Thin Solid Films 387 (2001) 251. [18] C. Kind, R. Popescu, R. Schneider, E. Müller, D. Gerthsen, C. Feldmann, RSC Advances 2 (2012) 9473. [19] T.W. Smith, R.A. Cheatham, Macromolecules 13 (1980) 1203. [20] U. Jeong, Y. Xia, Adv. Mater. 17 (2005) 102. [21] J.-L. Li, X.-Y. Liu, J. Nanosci. Nanotechnol. 8 (2008) 2488. [22] U. Lee, J. Choi, N. Myung, I.H. Kim, C.R.N. Chenthamarkshan, N.R. de Tacconi, K. Rajeshwar, Bull. Korean Chem. Soc. 29 (2008) 689. [23] Z. Quan, P. Yang, C. Li, X. Zhang, J. Yang, J. Lin, Cryst. Growth Des. 8 (2008) 3834. [24] B. Zhang, W. Dai, X. Ye, F. Zuo, Y. Xie, Angew. Chem. Int. Ed. 45 (2006) 2571. [25] L. Liu, Q. Peng, Y. Li, Nano Res. 1 (2008) 403. [26] U. Jeong, Y. Xia, Angew. Chem. Int. Ed. 44 (2005) 3099. [27] P.H.C. Camargo, Z.-Y. Li, Y. Xia, Soft Matter. (2007) 1215.
Contents lists available at ScienceDirect
Journal of Colloid and Interface Science www.elsevier.com/locate/jcis
Colloidally stable selenium@copper selenide core@shell nanoparticles as selenium source for manufacturing of copper–indium–selenide solar cells Hailong Dong a, Aina Quintilla b, Marco Cemernjak b, Radian Popescu c, Dagmar Gerthsen c, Erik Ahlswede b,⇑, Claus Feldmann a,⇑ a
Institut für Anorganische Chemie, Karlsruhe Institute of Technology (KIT), Engesserstraße 15, D-76131 Karlsruhe, Germany Zentrum für Sonnenenergie- und Wasserstoffforschung Baden-Württemberg (ZSW), Industriestraße 6, D-70565 Stuttgart, Germany c Laboratorium für Elektronenmikroskopie, Karlsruhe Institute of Technology (KIT), Engesserstraße 7, D-76131 Karlsruhe, Germany b
a r t i c l e
i n f o
Article history: Received 18 June 2013 Accepted 3 October 2013 Available online 17 October 2013 Keywords: Selenium Copper selenide Core–shell structure CIS Solar cell
a b s t r a c t Selenium nanoparticles with diameters of 100–400 nm are prepared via hydrazine-driven reduction of selenious acid. The as-prepared amorphous, red selenium (a-Se) particles were neither a stable phase nor were they colloidally stable. Due to phase transition to crystalline (trigonal), grey selenium (t-Se) at or even below room temperature, the particles merged rapidly and recrystallized as micronsized crystal needles. As a consequence, such Se particles were not suited for layer deposition and as a precursor to manufacture thin-film CIS (copper indium selenide/CuInSe2) solar cells. To overcome this restriction, Se@CuSe core@shell particles are presented here. For these Se@CuSe core@shell nanoparticles, the phase transition a-Se ? t-Se is shifted to temperatures higher than 100 °C. Moreover, a spherical shape of the particles is retained even after phase transition. Composition and structure of the Se@CuSe core@shell nanostructure are evidenced by electron microscopy (SEM/STEM), DLS, XRD, FT-IR and line-scan EDXS. As a conceptual study, the newly formed Se@CuSe core@shell nanostructures with CuSe acting as a protecting layer to increase the phase-transition temperature and to improve the colloidal stability were used as a selenium precursor for manufacturing of thin-film CIS solar cells and already lead to conversion efficiencies up to 3%. Ó 2013 Elsevier Inc. All rights reserved.
1. Introduction Thin-film solar cells represent an important alternative to crystalline silicon solar cells. Here, the most efficient thin-film technology is related to CuInSe2 (CIS: copper indium selenide) with record values above 20% if gallium is incorporated as well (CIGS: copper indium gallium selenide) [1].Conventionally, CIS/CIGS layers are deposited by applying high-vacuum techniques. Besides thermal co-evaporation of the relevant elements, including selenium, sequential processes are often used, where typically Se is deposited by thermal evaporation as a thin top coating on a layer stack of Cu-/In-/Ga-containing precursors [2]. In a second step, these layers are chemically converted in a thermal process at 550–600 °C – called ‘‘selenization’’ – into the desired chalcopyrite structure of the CIS/CIGS absorber. Ideally the Se cap acts as a sufficient Se source during the annealing step. Usually, an additional ⇑ Corresponding authors. Fax: +49 721 60844892 (C. Feldmann), fax: +49 711 7870230 (E. Ahlswede). E-mail addresses: [email protected] (E. Ahlswede), claus.feldmann@ kit.edu (C. Feldmann). 0021-9797/$ - see front matter Ó 2013 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.jcis.2013.10.001
treatment in harmful H2S atmosphere is applied during the heating step to increase the electronic quality of the absorber layer [3]. In recent years, vacuum-free deposition of suitable precursor pastes has attracted increasing attention in view of low-cost high-throughput production of CIS thin-film solar cells [4,5]. One possibility for paste formulation is to use nanoparticles, which have already been reported for elemental metals (Cu, In, CuIn/Cu11In9) [6–8], metal oxides [9], and metal selenides (CuInSe2, Cu2Se/ CuSe, In2Se3) [10–13]. For reaching high efficiencies, the precursor layers were always annealed in a selenium vapour atmosphere to compensate for a possible loss of selenium during the heating step or to reach denser polycrystalline layers. So far, only conventional sources of selenium were investigated, either as an evaporated capping layer or as additional selenium vapour in the gas phase. The use of inks containing Se nanoparticles is, however, a kind of a missing link. The availability of colloidally stable Se inks would allow avoiding vacuum-techniques and gas-phase deposition completely. Ideally, just a simple heat treatment of the as-deposited particulate metal and selenium precursor layers could generate dense and crystalline CIS layers. This option is however hampered due to the room-temperature phase transition of selenium (31 °C
104
H. Dong et al. / Journal of Colloid and Interface Science 415 (2014) 103–110
for bulk-Se) [14]. As a consequence, Se particles are of limited colloidal stability in suspension and show rapid agglomeration and merging of particles, followed by re-crystallization of large crystal needles. Such Se particles are not suitable for layer deposition and manufacturing of thin-film solar cells. In the following, we report on the synthesis of Se particles that are capped by CuSe to form Se@CuSe core@shell nanostructures. Based on this newly introduced CuSe capping, the phase-transition temperature of elemental selenium is shifted to >100 °C. As a result, colloidally stable suspensions of Se particles could be obtained and were manufactured - together with Cu11In9 nanoparticles - in a first conceptual study via a vacuum-free process to CIS thin-film solar cells. The strategy of repelling the Se phase transition and increasing the colloidal stability by metal selenide cappings, more generally, can also be interesting for application of Se particles in mechanical sensors, electrical rectifier, xerography, or tumour therapy [15,16].
2. Experimental section 2.1. Materials, synthesis and thin-film deposition General considerations: Selenious acid (H2SeO3, 98%), hydrazine monohydrate (N2H4H2O, 98%), copper(II) acetate (Cu(CH3COO)2, 98%) and polyvinylpyrrolidone (PVP, Mw = 40,000) were purchased from Sigma–Aldrich. Diethylene glycol (DEG, C4H10O4, 99%) was obtained from Alfa Aesar. All chemicals were of analytical grade and used as received. Synthesis of Se@CuSe core@shell nanoparticles: In a typical recipe, N2H4H2O (10–40 mmol, Table 1) and 200 mg PVP were added to 50 ml of DEG in a three-necked flask. This mixture was dispersed to form a homogeneous solution by constant stirring (solution I). Solution I was then cooled to 0 °C with an ice bath. Thereafter, 0.8 mmol of H2SeO3 in 3 ml of DEG (solution II) were injected into the precooled solution I. The reaction was allowed to proceed for different periods of time (t, Table 1). The gradual colour shift from a colourless solution to an orange suspension allows following the course of the reaction even by the naked eye. Finally, all residual N2H4H2O in the reaction solution was removed by vacuum distillation for 30 min at 0 °C, followed by 150 min at room temperature. Thereafter, 0.04 mmol of Cu(CH3COO)2 dissolved in 3 ml of DEG (solution III) were added dropwise. The addition of Cu(CH3COO)2 again resulted in a colour change, from orange via orangered to brick-red. Detailed information regarding the experimental parameters of the synthesis is listed in Table 1. Finally, the suspensions were diluted with demineralised water and the Se@CuSe core–shell nanoparticles were collected via centrifugation, washed three times with demineralised water, and then redispersed in demineralised water for further analysis and treatment. Preparation and processing of Se@CuSe inks: Suitable suspensions of Se@CuSe nanoparticles were obtained by redispersing the as-
prepared powders in ethanol with a solid load of typically 200 mg ml1. These precursor inks were prepared right before layer deposition in order to prevent particle agglomeration over time. Layer deposition was done via doctor blading, using an Erichsen film applicator and an adjustable blade by Zehntner, on Molybdenum covered sodalime-glass substrates of 1 mm thickness. Two different strategies were applied: Scenario 1 – Se@CuSe nanoparticles were co-dispersed with Cu11In9 nanoparticles in ethanol and deposited thereafter to form a mixed Se@CuSe/Cu11In9 nanoparticle film. Scenario 2 – Alternatively, a Cu11In9 layer was deposited first, followed by a stacked layer of the Se@CuSe nanoparticles on top of the Cu11In9 particle film. The synthesis of the Cu11In9 nanoparticles was performed as described in [8]. Selenization of thin-films and solar cell preparation: To convert the precursor films to chalcopyrite-type CIS absorber layers, selenization of the thin-films was performed in a tube furnace at 550 °C, eventually in nitrogen-diluted selenium vapour, as described in [8]. The CIS absorber layers were further treated by selective KCN etching to remove undesired binary CuSe phases, which would typically lead to shunted cell behaviour due to its good conductivity. The CIS absorber layers were finally completed by a wet-chemically processed CdS buffer layer, a sputtered ZnO buffer layer (i-ZnO) and a transparent ZnO:Al front-contact layer to obtain functional solar cells using standard procedures as described in [17]. 2.2. Materials characterization and analytical tools Dynamic light scattering (DLS) was performed with a Nanosizer ZS from Malvern Instruments (equipped with a He–Ne laser emitting at 633 nm), detection via non-invasive back-scattering at an angle of 173 °, 256 detector channels, polystyrene cuvettes). For analysis, the as-prepared Se@CuSe nanoparticles were redispersed in demineralised water by ultrasonic treatment for 15 min. Scanning electron microscopy (SEM) was performed on a Zeiss Supra 40 VP, using an acceleration voltage of 20 kV and a working distance of 2 mm to analyse the size distribution and shape of the Se@CuSe nanoparticles. All samples were prepared by evaporating a single drop of a dispersion of the as-prepared Se@CuSe nanoparticles in demineralised water at room temperature in air. The thinfilm morphology of precursor layers, selenized layers and completed solar cells were studied with a XL30 SFEG Sirion from FEI Company, using a 5 kV acceleration voltage and working distances of 5.9, 4.3 and 3.4 mm. Scanning transmission electron microscopy in the high-angle annular dark-field mode (HAADF STEM) was conducted with an aberration-corrected FEI Titan3 80–300 at 300 kV. Samples for STEM were prepared by evaporating water-based suspensions on an amorphous carbon (Lacey-) film copper grid. EDXS line profiles were recorded with an FEI Titan3 80–300 microscope by applying a drift-correction routine via cross correlation of several images, which yields a local precision better than
Table 1 Experimental conditions to adjust the mean size of Se@CuSe core@shell nanoparticles. Sample
N2H4H2O (mmol)
H2SeO3 (mmol)
cðN2 H4 H2 OÞ r cðH2 SeO3 Þ
Reaction time t (min)
Cu(Ac)2 (mmol)
Average diameter (nm)
A B C D E F
10 20 20 20 40 40
0.8 0.8 0.8 0.8 0.8 0.8
13 25 25 25 50 50
90 90 75 30 30 30
0.04 0.04 0.04 0.04 0.04 0.16
410(90) 310(63) 179(48) 162(48) 100(27) 105(29)
(r: Molar ratio of N2H4H2O and H2SeO3). (t: Reaction time of N2H4H2O and H2SeO3). (: Mean diameter according to DLS).
H. Dong et al. / Journal of Colloid and Interface Science 415 (2014) 103–110
1.0 nm. The drift-corrected EDXS line profiles were taken with a probe diameter of 0.5 nm and a distance of about 1.0 nm between two measuring points. With respect to a quantification of the EDX spectra from core–shell nanoparticles one has to keep in mind that only compositions are obtained that are averaged along the electron-beam direction. In concrete, the whole volume along the electron trajectory contributes to the detected X-ray signal. To evaluate the composition of different shells of a core–shell nanoparticle, a procedure was developed that is described in detail in [18]. X-ray powder diffraction (XRD) analysis was carried out with a Stoe Stadi-P diffractometer using Ge-monochromatized Cu Ka1 radiation. Differential scanning calorimetry (DSC) and thermogravimetry (TG) were performed with a STA 409C instrument from NETZSCH (Selb, Germany). Measurements were performed with the following conditions: 10 °C min–1 as heating rate; 30–500 °C as temperature range; 16.8 mg as sample weight; nitrogen atmosphere; alumina crucibles for sample and reference. Fourier-transform infrared spectra (FT-IR) were recorded on a Bruker Vertex 70 FT-IR spectrometer using KBr pellets. For direct comparison, 400 mg of dried KBr were carefully pestled with 0.3 mg of the sample and pressed to a thin pellet. In addition, spectra of the as-prepared Se@CuSe nanoparticles were subjected to a correction of scattering effects to allow for comparison to reference spectra. Current–voltage curves were measured using a Keithley 238 source-measuring unit under simulated AM 1.5 global solar irradiation with an ORIEL sun simulator at 100 mW cm–2 to extract the basic solar-cell characteristics from devices approximately 0.24 cm2 in size. External quantum efficiency measurements (EQE) were performed using a setup from Optosolar.
3. Results and discussion 3.1. Colloidal stability and phase transition of Se particles Colloidal particles of elemental selenium have been prepared by different types of synthesis. Most often, selenious acid (H2SeO3) is reduced by hydrazine (N2H4) [19,20]. Alternatively, reduction of selenious acid with ascorbic acid [21], dismutation of SeO2 [22], and melting of bulk-Se in high-boiling solvents [23] were applied. Based on these strategies mesoscaled Se particles with diameters
105
Fig. 1. Scheme illustrating the synthesis, agglomeration and crystallization of amorphous, red Se as well as the formation of Se@CuSe core@shell nanoparticles for stable inks and subsequent layer deposition.
of 200 nm to 2 lm were obtained most often [15,19,20,22]. Se particles with diameters of 100 nm or below are less often described [20]. Whatever approach was used to synthesize Se particles, the phase-transition temperature of Se (31 °C for bulk-Se [14]) is known as a general restriction. At this temperature the amorphous red phase of elemental selenium (a-Se) that is obtained - in accordance with Ostwald’s step rule – via low-temperature liquid-phase synthesis, shows a phase transition to the thermodynamically stable phase of crystalline, trigonal, grey selenium (t-Se). This phase transition, however, results in a rapid merging of the as-prepared a-Se particles, followed by instantaneous formation of micronsized crystal needles of t-Se (Figs. 1 and 2). For nano- and micronsized particles exhibiting huge specific surfaces, the a-Se ? t-Se phase transition is observed at even lower temperatures as compared to bulk-Se – thus, below room temperature [15,20–25]. In fact, even when kept at about 0 °C, the phase transition occurs after certain period in time (typically 1–3 days). The phase transition can be easily followed by the naked eye due to the colour change of suspensions/powders from orange to grey. This situation is well known and has been described frequently
Fig. 2. SEM images showing the merging (A) and crystallization (B) while amorphous, red selenium (a-Se) is passing through a phase transition to crystalline, grey selenium (t-Se).
106
H. Dong et al. / Journal of Colloid and Interface Science 415 (2014) 103–110
[15,20–25]. In sum, the phase transition severely restricts practical handling of Se particles in view of storing for certain period, gentle heating (e.g., due to centrifugation or redispersion) as well as regarding layer deposition or printing of thin-films. Thus, Se particles and suspensions thereof are not suited for manufacturing thinfilm solar cells as such. To obtain suspensions of Se particles with sufficient colloidal and phase stability that are suitable for thin-film deposition and manufacturing of CIS solar cells, we applied a hydrazine-driven reduction of selenious acid (H2SeO3) as suggested by Smith [19] and Xia [20] (Fig. 1). By variation of the N2H4:H2SeO3 ratio and the duration of the reaction as well as by addition of polyvinylpyrrolidone (PVP), the diameter of the as-prepared Se particles was controlled (Table 1). The synthesis results in uniform spherical a-Se particles that, however, show the well-known rapid phase transition with merging of particles and formation of large t-Se crystal needles (Figs. 1 and 2). To increase phase and colloidal stability, the as-prepared a-Se particles were capped with CuSe (Fig. 1). As CuSe typically is an impurity phase in view of many applications of selenium, the use of Se@CuSe core@shell nanostructures for manufacturing of CuInSe2-based solar cells is not a limitation. The strategy of synthesising Se@CuSe core@shell nanostructures is schematically depicted in Fig. 1. Subsequent to hydrazine-induced
synthesis of a-Se particles applying DEG as a solvent, the suspension was kept at 0 °C. This low temperature allows reducing the reaction rate, which supports particle growth and size uniformity. The formation of a-Se can be followed by the colour change from a colourless solution to an orange suspension. Prior to addition of Cu(CH3COO)2, residual N2H4H2O had to be removed via vacuum distillation, since Cu2+ would have been reduced to elemental Cu otherwise. The addition of Cu(CH3COO)2 again resulted in a colour change of the suspension from orange via orange-red to brick-red. Finally, the Se@CuSe nanoparticles were washed by repeated centrifugation and redispersion, and finally, obtained as a brick-red powder that can be easily redispersed (e.g. in ethanol) for manufacturing CIS solar cells. Powder samples as well as suspensions of Se@CuSe core@shell nanoparticles show sufficient stability in view of practical handling without any phase transition to t-Se.
3.2. Particle size and chemical composition of Se@CuSe core@shell nanoparticles Size and size distribution of the as-prepared Se@CuSe core@ shell nanoparticles were first verified via dynamic light scattering (DLS) based on powder samples that were redispersed in demineralised water (Fig. 3). To this concern, particles with relatively
Fig. 3. Particle size of Se@CuSe core@shell nanoparticles according to SEM and DLS analysis (samples A–E according to Table 1; DLS analysis in demineralised water).
H. Dong et al. / Journal of Colloid and Interface Science 415 (2014) 103–110
Fig. 4. Differential scanning calorimetry (DSC) and thermogravimetry (TG) of asprepared Se@CuSe core@shell nanoparticles (sample E, cf. Table 1) showing the onset of phase transition a-Se ? t-Se at 107 °C and the onset of the melting point at 217 °C (A); (B) crystallinity and spherical shape of Se@CuSe powders after sintering at 175 °C with XRD and SEM (reference: bulk-Se – ICDD No. 6-362; arrow indicating CuSe as the shell).
narrow size distribution and clearly different mean diameters of 410(90) nm (sample A), 310(63) nm (sample B), 179(48) nm (sample C), 162(48) nm (sample D) and 100(27) nm (sample E) were obtained. Size control is predominantly guaranteed by variation of the molar ratio N2H4:H2SeO3 (r) and the reaction time (t) (Table 1). At constant r (cf. samples B–D), the size of the Se@CuSe nanoparticles can be reduced by decreasing t (310 nm ? 162 nm). With t kept constant (cf. samples A/B, D/E), the average particle size can be decreased by increasing r (410 nm ? 310 nm, 162 nm ? 100 nm). Scanning electron microscopy (SEM) confirms particle size and size distribution of the as-prepared Se@CuSe samples (Fig. 3, cf. Fig. 7). Due to the relatively uniform size distribution, partial formation of dense-packed particle layers is
Fig. 5. FT-IR spectra of PVP-capped Se@CuSe core@shell nanoparticles (sample E, cf. Table 1; pure PVP as a reference).
107
observed. While the samples A to F were prepared to study the relevant experimental conditions as well as to perform materials characterization, sample E with the smallest diameter was used for solar-cell manufacturing. Composition and structure of the Se@CuSe nanoparticles were investigated by X-ray powder diffraction (XRD), differential scanning calorimetry (DSC) and Fourier-transform infrared spectroscopy (FT-IR). XRD of as-prepared brick-red powder samples, first of all, do not show any Bragg peaks, indicating the non-crystalline nature of a-Se (cf. Fig. 6). Moreover, no Bragg peaks of any CuxSey phase are visible likewise. DSC shows the phase transition a-Se ? t-Se for Se@CuSe with its onset at 107 °C (Fig. 4a). Thus, the phase-transition temperature of the nanomaterial is significantly increased, even when comparing to bulk-Se (31 °C) [14]. Thermogravimetry (TG), furthermore, indicates that no weight loss (e.g. due to evaporation of Se) occurs in a temperature range up to the melting point with its onset at 217 °C (melting point of bulkSe: 221 °C) (Fig. 4a). Even more surprisingly, the spherical shape of the Se@CuSe core@shell nanoparticles is still retained after sintering of powders at 175 °C, although the samples clearly show the crystalline structure of t-Se thereafter (Fig. 4b). In addition, the main Bragg peak of CuSe/Klockmannite becomes visible with low intensity at a two theta value of 28°. Altogether, the significantly increased phase-transition temperature and the spherical shape retaining after sintering impressively confirm the stability of the Se@CuSe core@shell nanostructure – and thus the effectiveness of the concept. The effect of the increased phase-transition temperature and the improved colloidal stability of the Se@CuSe core@shell nanoparticles, in sum, can be ascribed to CuSe serving as a protecting layer. On the one hand, CuSe obviously stabilizes the red a-Se modification. Traces of Cu2+ that might diffuse into the inner Se particle core could further support this effect. Moreover, the CuSe shell separates the nanoparticles from each other, so that a direct contact of the selenium cores to each other is excluded. Although Ag2Se-cappings – with relatively high thickness of about 50 nm – on Se particles were already used for Se-based photonic crystals with thermally switchable stop bands [26,27], the possibility to increase the phase stability and colloidal stability of Se particles is here shown for the first time. FT-IR spectra indicate the presence of PVP as a surface capping of the Se@CuSe nanostructure (Fig. 5). Thus, weak vibrational bands at 3050–2800 cm1 (v(CAH)), 1800–1550 cm1 (v(C@O)), and the fingerprint area (1500–600 cm1) are in good agreement with PVP reference spectra. The additional strong band at 3700– 3250 cm1 (v(OAH)) can be attributed to OH-containing solvents (water, ethanol) adhered on the particle surface. In order to prove the presence of a CuSe shell encapsulating the Se core, a copper-rich sample with a higher amount of CuSe (sample F:Se:Cu = 5:1) was prepared in addition to samples A–E exhibiting a lower Cu concentration (samples A–E: Se:Cu = 20:1, Table 1). According to DLS and SEM (Fig. 3), this Cu-rich sample F is very comparable to sample E in terms of particle size and size distribution (Table 1). XRD of the as-prepared Cu-rich sample F indeed evidences the presence of crystalline CuxSey phases (Fig. 6). Thus, CuSe/Klockmannite is visible as the predominant phase; Cu5Se4/Athabascaite is observed as a side-phase. When annealing the Cu-rich Se@CuSe sample F to 175 °C, crystalline t-Se is obtained as expected, in addition to CuSe/Klockmannite (Fig. 6). Finally, the core@shell nanostructure of Se@CuSe was validated based on HAADF STEM and EDXS line-profile analyses. Fig. 7c shows a HAADF-STEM image of a single particle of sample B with the dashed line indicating the position of an EDXS line-profile analysis. Quantification of the EDXS spectra yields the composition profile shown in Fig. 7d which demonstrates the core@shell structure. Here, a significant rising of the copper concentration is observed for the surface-region of the Se@CuSe core@shell
108
H. Dong et al. / Journal of Colloid and Interface Science 415 (2014) 103–110
Fig. 6. XRD patterns of as-prepared and sintered Se@CuSe core@shell nanoparticles (sample B and Cu-rich sample F; cf. Table 1) with reference diffractograms: greySe/t-Se – ICDD No. 6-362/black lines; CuSe/Klockmannite – ICDD No. 34-171/dark grey lines; Cu5Se4/Athabascaite – ICDD No. 21-1016/light grey lines.
Fig. 7. SEM overview (A and B) images, HAADF STEM image (C), concentration profile obtained from EDXS line-profile analysis (D, along dashed line as indicated in C), scheme with core and shell composition (E) of the Se@CuSe core@shell particle (sample B, cf. Table 1) after evaluation of the composition profile (D) according to the procedure described in [18].
nanostructure. The inner section of the particle, in contrast, shows a high Se concentration. Cu is nevertheless as well visible for the inner particle core since the electron beam still passes the CuSe shell (Fig. 7d) [18]. Using the evaluation procedure described in Ref. [18], the composition of the shell and core can be reconstructed as shown schematically in Fig. 7e. It is noted that the relatively large error for the composition given in Fig. 7e resulted
from the evaluation procedure and is not caused by the statistics of the EDX spectra. In sum, the as-prepared Se@CuSe core@shell nanoparticles exhibit an inner core of almost pure elemental selenium with a diameter of about 340 nm and a capping significantly enriched with Cu and a thickness of about 20 nm. Together with the XRD results, these findings validate the presence of a particle core consisting of almost pure Se and a shell of CuSe.
H. Dong et al. / Journal of Colloid and Interface Science 415 (2014) 103–110
109
3.3. Use of Se@CuSe core@shell nanoparticles as precursor for manufacturing of CIS thin-film solar cells As the diffusion of Cu2+/In3+/Se2- in the solid-state is slow, application of metal oxides (Cu2O/CuO, In2O3, CuInO2) or metal selenides (Cu2Se/CuSe, In2Se3, CuInSe2) as nanoscale CIS precursors can be accomplished by slow intermixing of phases, phase inhomogeneities, imperfect healing of pores between the initial nanoparticles, and a limited electrical contact between the grain boundaries. All these restrictions reduce the solar-cell performance and efficiency. In principle, this could be avoided by applying nanoscaled elemental metals (Cu, In, CuxIny) and elemental selenium particles as reactive precursors. Conversion to CuInSe2 would be instantaneously possible – ideally without any vacuum-deposition process or gas-phase selenization. Moreover, the interparticular pores should be filled due to the huge volume expansion when reacting metals to metal selenides. The potential of the here prepared Se@CuSe core@shell nanoparticles for manufacturing CIS solar cells is investigated as a conceptual study in combination with Cu11In9 nanoparticles, which have been already successfully developed and tested in CIS solar cells [8]. While sample E (Table 1) exhibits the smallest diameter, it was used for solar-cell manufacturing in the following throughout. The preparation of dense, homogenous films is not yet as convenient as for the Cu11In9 system but well feasible, as can be seen in Fig. 8. Although the Se@CuSe precursor together with Cu11In9 could make an additional Se atmosphere obsolete during the chemical conversion step, at the same time too much Se may cause problems with excessive MoSe2 formation at the solar cell’s back contact. Thus, precise optimization of the relevant conditions will be necessary as a next step. By now, different routes were evaluated, including mixed thin-films of Se@CuSe core@shell nanoparticles and Cu11In9 nanoparticles (scenario 1) and a capping-layer of Se@CuSe core@shell nanoparticles deposited on top of a Cu11In9 thin-film (scenario 2). To the latter concern, Fig. 8 exemplarily shows a thick layer of Se@CuSe on top of a thin Cu11In9 precursor layer. Notably, the diameters of the Cu11In9 particles (20 nm) and of the layer thickness (250 nm) are much smaller as compared to the Se@CuSe precursor (particle diameter: 100 nm, layer thickness: 600 nm). In addition to the above scenarios 1 and 2, different selenization environments were evaluated: The sample stacks were either heated in pure nitrogen or in selenium rich atmosphere for different times. Subsequent to sintering and selenization of the precursor layers at 550 °C [8,17], densification and crystal growth are clearly visible (Fig. 9). Crystal morphology and layer structure are significantly different from the precursor layer, indicating the reaction between Se@CuSe and Cu11In9. Taking into account that these first lab-cells were manufactured under non-optimized conditions (e.g. in terms of chemical composition, Cu:In:Se ratio, particle diameter, layer thickness and morphology, duration/temperature of sintering), the resulting CIS solar cells show promising performance for both scenarios of adding the Se@CuSe precursor. Thus, efficiencies of 1.8% for the Se@CuSe-Cu11In9 mixed film (scenario 1) and of 3.0% for the Se@CuSe capping layer (scenario 2) were obtained. Current–voltage characteristics of the best lab-cells are shown in Fig. 10. Regarding the different conditions of selenization that were tested, currently, the best results could still be obtained via the classical reference procedure (550 °C, 1 h) with an additional Se atmosphere during the tube furnace process [8,17]. In view of scenario 1 (mixed film of Se@CuSe and Cu11In9), there are hints that the formation of an incomplete CIS crust on top of a bottom-layer of much smaller grains could be avoided in favor of a more homogenous morphology (Fig. 9). Such double-layered crust structures – also visible for scenario 2 (Fig. 9) – are quite often observed and generally assumed as disadvantageous in terms of cell
Fig. 8. Doctor-bladed thin-film of Se@CuSe core@shell nanoparticles as capping on top of pre-deposited Cu11In9 nanoparticle thin-film (scenario 2).
Fig. 9. CIS absorber layers after selenization (550 °C, 1 h) – scenario 1: Se@CuSe core@shell nanoparticles and Cu11In9 nanoparticles mixed in suspension prior to thin-film deposition; scenario 2: Se@CuSe core@shell nanoparticles as a capping on top of a pre-deposited Cu11In9 nanoparticle layer.
Fig. 10. Current–voltage characteristics of the best CIS solar cell manufactured with Se@CuSe core@shell nanoparticles and Cu11In9 nanoparticles as precursors.
performance and efficiency [4–13]. The realization of a uniform and dense CIS crystal layer with crystals of similar size could thus be another beneficial aspect of introducing Se@CuSe core@shell nanoparticles as selenium precursor. Currently, scenario 1 is unfortunately accompanied by certain oxidation of the Mo back contact under formation of MoSe2. As a conceptual study with the function of lab-scale CIS solar cells being proven, the new Se@CuSe precursor already leads to promising results. Much more efforts and optimization of the experimental conditions (in terms of chemical composition, Cu:In:Se ratio, particle diameter, layer thickness
110
H. Dong et al. / Journal of Colloid and Interface Science 415 (2014) 103–110
and morphology, duration/temperature of sintering, etc.) are, however, necessary to improve the morphology of the CIS absorber layer and to increase the overall solar-cell performance.
Nanostructures (CFN) of the Deutsche Forschungsgemeinschaft (DFG) at the Karlsruhe Institute of Technology (KIT) for financial support.
4. Conclusions
References
Se@CuSe core@shell nanoparticles, 100–400 nm in total diameter, show a significantly higher phase stability as well as a higher colloidal stability as compared to non-capped selenium particles. Thus, the well-known phase transition of amorphous, red selenium (a-Se) to crystalline (trigonal), grey selenium (t-Se) occurs at 31 °C for bulk-Se and occurs even below room temperature for nano-/ micronsized Se. The as-prepared Se@CuSe core@shell nanoparticles with an inner core of elemental selenium (340 nm in diameter) and a CuSe capping (20 nm in diameter) show the onset of the phase transition at a temperature as high as 107 °C. These core– shell particles do neither show any merging of particles nor any formation of crystal needles of t-Se. Based on the increased phase and colloid stability, the Se@CuSe core@shell nanoparticles can be dispersed as a stable precursor suspension for thin-film deposition and manufacturing of thin-film CIS (copper indium selenide/CuInSe2) solar cells. First lab-scale CIS solar cells in a conceptual study show promising performances with conversion efficiencies of currently up to 3%. In addition to CIS solar cells, the concept of stabilizing Se nano-/mesoscaled particles by metal selenide cappings could be useful for other areas of application, such as the use of Se particles in mechanical sensors, electrical rectifiers, xerography, or tumour therapy. Acknowledgments The authors are grateful to the German Ministerium für Bildung und Forschung (BMBF) for funding within the project ‘‘Nanopartikuläre Dünnschicht-Solarzellen – Grundlagen und Prozesstechnologie (NanoPV)’’. Here, we explicitly thank Dr. Karen Köhler (Bayer Technology Services GmbH, Leverkusen) for fruitful collaboration. Finally, we acknowledge the Center of Functional
[1] P. Jackson, D. Hariskos, E. Lotter, S. Paetel, R. Wuerz, R. Menner, W. Wischmann, M. Powalla, Prog. Photovolt: Res. Appl. 19 (2011) 894. [2] V. Probst, W. Stetter, W. Riedl, H. Vogt, M. Wendl, H. Calwer, S. Zweigart, K.D. Ufert, B. Freienstein, H. Cerva, F.H. Karg, Thin Solid Films 387 (2001) 262. [3] S. Niki, M. Contreras, I. Repins, M. Powalla, K. Kushiya, S. Ishizuka, K. Matsubara, Prog. Photovolt: Res. Appl. 18 (2010) 453 (Review). [4] J. Hibberd, E. Chassaing, W. Liu, D.B. Mitzi, D. Lincot, A.N. Tiwari, Prog. Photovolt: Res. Appl. 18 (2010) 434 (Review). [5] T. Todorov, D.B. Mitzi, Eur. J. Inorg. Chem. (2010) 17 (Review). [6] F.S. Chen, J.S. Ma, C.H. Lu, Ceram. Int. 38 (2012) 5319. [7] J. Ungelenk, V. Haug, A. Quintilla, E. Ahlswede, Phys. Status Sol. 4 (2010) 58. [8] C. Kind, C. Feldmann, A. Quintilla, E. Ahlswede, Chem. Mater. 23 (2011) 5269. [9] V.K. Kapur, A. Bansal, P. Le, O.I. Asensio, Thin Solid Films 431–432 (2003) 53. [10] K. Kim, Y.-J. Eo, A. Cho, J. Gwak, J.H. Yun, K. Shin, S.K. Ahn, S.H. Park, K. Yoon, S.J. Ahn, J. Mater. Chem. 22 (2012) 8444. [11] A. Cho, S. Ahn, J.H. Yun, J. Gwak, S.K. Ahn, K. Shin, H. Song, K.H. Yoon, Solar Cells 109 (2013) 17. [12] Q. Guo, S.J. Kim, M. Kar, W.N. Shafarman, R.W. Birkmire, E.A. Stach, R. Agrawal, H.W. Hillhouse, Nano Letters 8 (2008) 2982. [13] V.A. Akhavan, B.W. Goodfellow, M.G. Panthani, D.K. Reid, D.J. Hellebusch, T. Adachi, B.A. Korgel, Energy Environ. Sci. 3 (2010) 1600. [14] R.A. Zingaro, W.C. Cooper (Eds.), Selenium, Van Nostrand Reinhold, New York, 1974. [15] X. Cao, Y. Xie, L. Li, Adv. Mater. 15 (2003) 1914. [16] L. Cai, N.C.Y. You, H. Lu, L.N. Mu, O.Y. Lu, S.Z. Yu, A.D. Le, J. Marshall, D. Herber, Z.F. Zhang, Cancer 106 (2006) 2345. [17] M. Powalla, B. Dimmler, Thin Solid Films 387 (2001) 251. [18] C. Kind, R. Popescu, R. Schneider, E. Müller, D. Gerthsen, C. Feldmann, RSC Advances 2 (2012) 9473. [19] T.W. Smith, R.A. Cheatham, Macromolecules 13 (1980) 1203. [20] U. Jeong, Y. Xia, Adv. Mater. 17 (2005) 102. [21] J.-L. Li, X.-Y. Liu, J. Nanosci. Nanotechnol. 8 (2008) 2488. [22] U. Lee, J. Choi, N. Myung, I.H. Kim, C.R.N. Chenthamarkshan, N.R. de Tacconi, K. Rajeshwar, Bull. Korean Chem. Soc. 29 (2008) 689. [23] Z. Quan, P. Yang, C. Li, X. Zhang, J. Yang, J. Lin, Cryst. Growth Des. 8 (2008) 3834. [24] B. Zhang, W. Dai, X. Ye, F. Zuo, Y. Xie, Angew. Chem. Int. Ed. 45 (2006) 2571. [25] L. Liu, Q. Peng, Y. Li, Nano Res. 1 (2008) 403. [26] U. Jeong, Y. Xia, Angew. Chem. Int. Ed. 44 (2005) 3099. [27] P.H.C. Camargo, Z.-Y. Li, Y. Xia, Soft Matter. (2007) 1215.
