Ki-Joong Kim,
Richard P. Oleksak,
Changqing Pan,
Michael W. Knapp,
Peter B. Kreider,
Gregory S. Herman and
Chih-Hung Chang*
School of Chemical, Biological & Environmental Engineering, Oregon State University, Corvallis, OR 97331, USA. E-mail: Chih-hung.chang@oregonstate.edu
First published on 25th March 2014
A continuous synthetic method in a micro-tubular reactor is introduced for synthesizing mono-disperse and solution-stable chalcopyrite colloidal copper indium diselenide nanocrystal (CuInSe2 NC) inks with potential scalability. It was found that the morphologies of the CuInSe2 NCs were dependent on the Cu/In/Se composition. The NC morphology changed from spherical to hexagonal to trigonal with increasing In or Se content, whereas trigonal morphologies synthesized at high temperature yielded chalcopyrite CuInSe2 NCs. A laboratory-scale photovoltaic device with 1.9% efficiency under AM1.5G illumination was also fabricated to verify the utility of these inks.
The ability to produce stable, high-quality NC inks in solution with high-throughput is a key step in the development of low-cost thin film and QD-based solar cells19 and light emitting diodes.17,20–22 The ability to control the uniformity of the size, shape, composition, crystal structure and surface properties of the NCs is also of technological interest.23–25 Several methods have been reported in the literature for the synthesis of CuInSe2 NCs, however synthesis of high quality NCs has typically been limited to small batch reactors (∼25 mL) leading to low production rates.12 Furthermore, vessel temperature and mass transfer characteristics are not well defined for these small batch reactors and significant variation in product results is commonly observed, especially if the batch size is increased. These problems make it difficult to scale up of CuInSe2 NC production based on synthetic processes developed for small batch reactors, and may limit the commercial scale fabrication of solar cells.
Several techniques have been reported for the synthesis of CuInSe2 NCs. Among these, the most common utilize organic solvents either in a high pressure environment, i.e. the solvothermal method26 or via small scale inert environment Schlenk line experiments where precursors are either mixed at room temperature followed by slow heating,27 or hot-injected28 to give burst nucleation. In particular, the hot-injection method has been widely used to produce mono-dispersed CuInSe2 NCs with good control over NC size. The most thoroughly studied hot-injection route utilizes oleylamine (OLA) as the sole reaction medium acting as both solvent and ligand.9,12,29 In a typical hot-injection method a room temperature Se precursor solution is injected into CuCl and InCl3 precursor solutions held at high temperature leading to the kinetically determined formation of CuInSe2 NCs of disordered sphalerite phase.9,29 These investigations have shown that reaction condition is a critical factor in the crystal phase determination.
Recently, we have reported a scalable, continuous flow microreactor for the synthesis CuInSe2 NCs.30 This method combines the high synthesis quality with the potential scalability and reduced batch-to-batch variation of a continuous flow process. Herein we introduce a continuous synthetic method in a micro-tubular for generating mono-disperse and solution-stable chalcopyrite CuInSe2 colloidal NC inks and demonstrate their use in PV application. Furthermore, we elucidate a relationship between the composition and morphology of the CuInSe2 NCs and propose a morphology based reaction pathway.
![]() | ||
Fig. 1 Schematic diagram of (A) continuous synthetic system and (B) detailed tee in an oil bath indicated by red-colored in (A). |
In a typical synthesis, each solution was continuously pumped using a peristaltic pump (REGLO Digital, Ismatec) into Tygon tubing (1.52 mm i.d., Upchurch Scientific) at a flow rate of 0.3 mL min−1 and 0.1 mL min−1 for solutions A and B, respectively. Both precursors were connected with Ar gas in a Tedlar bag while pumping in order to keep an inert environment. Solution A initially flowed into a stainless steel tee immersed in an oil bath maintained at the desired reaction temperature. Solution B was injected using a stainless steel gauge needle (o.d.: 0.91 mm, i.d.: 0.6 mm) which was inserted into a stainless steel tee with the needle tip located at the exit of the tee. As shown in Fig. 1, both precursor solutions were heated to reaction temperature in the oil bath prior to mixing. The reaction was carried out in a 4 m long PTFE tubing (o.d.: 3.18 mm, i.d.: 1.59 mm, wall: 0.79 mm) corresponding to a residence time of 20 min. The reaction product was collected in a 30 mL glass vial for 10 min, and then 10 mL of ethanol was added to precipitate the NCs, followed by centrifugation at 7000 rpm for 10 min. The supernatant was discarded and 10 mL of toluene and 5 mL of ethanol were added to the NCs followed by an additional 7000 rpm for 10 min. The supernatant was decanted, and the final product was re-dispersed in 5 mL of toluene or dried under vacuum for further characterization. Typically, CuInSe2 NC powder was obtained at a rate of ∼42 mg h−1.
The XRD results indicate that the lower temperature reaction products were comprised primarily of copper selenide intermediates (i.e. Cu3Se2 and Cu2−xSe). No indium selenide intermediates were observed for these reaction conditions, suggesting the NC formation follows a similar pathway to that which was reported previously for the synthesis of CuInSe2 using an OLA solvent.29 That is, the thermodynamically favorable chalcopyrite CuInSe2 phase is generated primarily through the solid–liquid reaction of copper selenides and dissolved InCl3 for extended reaction times. It has also been shown that synthetic routes using OLA leads to CuInSe2 NCs with disordered sphalerite structure in the case of hot-injection method, whereas the chalcopyrite phase is formed in the case of slow heating of all dissolved precursors in one pot.28 In our study, both Cu/In and Se precursor solutions are maintained above the nucleation temperature prior to mixing. This allows for the thermodynamically controlled formation of the stable chalcopyrite phase. In a prior study of continuous synthetic method of CuInSe2 NCs using a solvent system of oleic acid (OA) and tri-n-octylphosphine (TOP), it has shown that both copper selenide and indium selenide binary intermediates were formed and that the rate limiting step for synthesis of CuInSe2 was formation of indium selenide.30 In the current study, we observed no evidence of indium selenide formation in the final products. This is not surprising due to the significantly different solvent systems used which will undoubtedly play a role in NC formation. In particular, herein Se dissolved in OLA is immediately available for NC nucleation upon mixing of the precursors, whereas in the prior study30 decomposition of TOPSe molecules precedes nucleation. Furthermore, it is expected that the different bonding strengths of OLA to dissolved precursors and nucleated NCs in comparison to OA and TOP will affect the reaction pathway and thus the synthesis herein is more likely to follow that reported for a previous detailed spectroscopic study of CuInSe2 formation in OLA29 as discussed above.
As indicated by XRD, the Cu2−xSe phase is converted to the chalcopyrite CuInSe2 phase at higher temperatures which provide sufficient activation energy for phase conversion. These findings are consistent with those of previous reports in literature that indicate the Cu ions diffuse out of the Cu2−xSe phase into the solvent with increasing temperature, allowing for the inclusion of In ions into the structure of Cu2−xSe, which leads to the formation of chalcopyrite CuInSe2 phase.31
The absorption spectra of the CuInSe2 NCs synthesized at different reaction temperatures were measured using UV-Vis absorbance spectroscopy (Fig. 2B). A broad and intense absorbance peak for the NCs synthesized at 200 °C was observed at 1050 nm, which is attributed to transitions involving the indirect band gap of copper selenide.32,33 These absorbance peaks gradually decreases with an increase in reaction temperature due to the decrease in copper selenide intermediates, which is in agreement with the XRD data. The composition in the ternary I–III–VI2 compounds can significantly influence their band gap properties.34 For each reaction condition, the band gap was determined by extrapolating the linear region of a plot of the squared absorbance versus the photon energy (Fig. 2B inset). Band gap values are seen to vary in the range of 2.45–1.10 eV as the reaction temperature increases from 200 °C to 240 °C. Although the band gap of copper selenide is not well defined because of the wide variety of stoichiometric ratios, bulk copper selenide such as Cu3Se2 and Cu2−xSe are typically reported to possess a direct band gap of 2.1–2.4 eV and an indirect band gap of 1.2–1.7 eV.33 The obtained band gap values are consistent with a decrease in copper selenide with increasing reaction temperature. The band gap of the CuInSe2 NCs synthesized at 240 °C was found to be 1.10 eV, which is in a good agreement with that of bulk chalcopyrite CuInSe2 (1.04 eV).35
The morphologies of NCs synthesized at different reaction temperatures were observed by TEM and are found to change with increasing reaction temperature (Fig. 3). It can also be seen that the NCs are uniform in size with a mixture of spherical, hexagonal, and trigonal shapes. At 200 °C, a large fraction of spherical/hexagonal NCs were observed with an average size of 14.7 nm (Fig. 3A). Similar morphologies are observed for the 220 °C reaction temperature (Fig. 3B) with a slight increases average size of 15.6 nm. Very uniform trigonal NCs were observed at 240 °C reaction temperature with an average size of 16.2 nm and a 1.8 nm standard deviation indicating a fairly narrow size distribution (Fig. 3C). All these samples exhibit good monodispersity as indicated by the histograms of their size distributions as shown in Fig. 3E–G.
![]() | ||
Fig. 3 (A–C) TEM images and (D) an example of EDS spectrum of mono-disperse NCs synthesized at different reaction temperatures for 20 min by continuous synthetic method (see more details in the Table 1). The average particle sizes of the NCs synthesized at 200 °C, 220 °C and 240 °C were 14.7 ± 2.9 nm, 15.6 ± 3.1 nm and 16.2 ± 1.8 nm, respectively. (E–G) Histograms of the NCs synthesized at different reaction temperatures ((E): 200 °C, (F): 220 °C and (G): 240 °C). |
A prior study has reported the synthesis of CuInSe2 NCs with trigonal shape using a batch reactor with a selenourea (SeC(NH2)2) Se precursor and OLA solvent.36 The existence of trigonal shapes is thus likely related to the preferred growth of CuInSe2 NCs along certain crystal facets in the OLA solvent system. The presence of exclusively trigonal shapes for the 240 °C reaction therefore suggests the reaction has gone to completion and consists exclusively of CuInSe2, in agreement with XRD and EDS measurements. The Cu/In/Se compositions of the NCs synthesized at different reaction temperatures were analyzed by EDS (Fig. 3D), which showed varying ratios of Cu:
In
:
Se compositions at different reaction temperatures. The compositions of Cu/In/Se tend to approach the desired stoichiometry of 1
:
1
:
2 when reaction temperature increases to 240 °C, with a slight excess of Se (Table 1).
Reaction temperature (°C) | Compositions (atomic%) | Atomic ratios | ||
---|---|---|---|---|
Cu | In | Se | ||
200 | 41.4 ± 0.4 | 9.6 ± 0.4 | 49.0 ± 0.1 | (1.00![]() ![]() ![]() ![]() |
220 | 31.3 ± 1.0 | 18.6 ± 0.5 | 51.3 ± 0.6 | (1.00![]() ![]() ![]() ![]() |
240 | 23.6 ± 0.1 | 23.3 ± 0.1 | 53.1 ± 0.2 | (1.00![]() ![]() ![]() ![]() |
HAADF-STEM EDS analysis was conducted to ascertain the Cu/In/Se compositions for individual NCs of different shape (Fig. 4). The 220 °C reaction product was used since this contained all three types of the observed morphologies (spherical, hexagonal, and trigonal). EDS results reveal the absence of In ions for spherical NCs which showed a Cu/In/Se = 1.00 ± 0.02:
0.00
:
0.78 ± 0.01 which is attributed to the Cu3Se2 or Cu2−xSe phase; while the hexagonal NCs consists of Cu, In, and Se with an atomic ratio of 1.00 ± 0.03
:
0.22 ± 0.15
:
1.16 ± 0.04. Based on the XRD data, initially Cu3Se2 is formed, followed by the formation of Cu2−xSe, and finally the gradual incorporation of In ions into the structure of the Cu2−xSe NCs, where an increase in the reaction temperature leads to an increase in In ions content. This led to an insufficient In ions content for CuInSe2 NCs synthesized at the low temperature. The composition of trigonal shaped NCs showed a Cu/In/Se = 1.00 ± 0.07
:
1.03 ± 0.09
:
1.99 ± 0.01 suggesting they are CuInSe2. These results also suggest the shape of synthesized NCs may be correlated to their Cu/In/Se compositions.
To gain more insight into the crystal structure for individual NCs, HRTEM (Fig. 5A–C) and Fourier transform (FT) (Fig. 5D–F) analysis were used. The two fold symmetric spots in FT images indicate that all the individual NCs are single crystalline in structure. Fig. 5A shows the HRTEM image of the spherical NC. The resolved interplanar spacing is 0.322 nm, corresponding to the (200) lattice plane (Fig. 5D) of tetragonal phase Cu3Se2 (JCPDS no. 47-1745). The lattice spacing of hexagonal NC (Fig. 5B) with an insufficient In content are 0.333, 0.288, and 0.203 nm, respectively. The lattice plane of the hexagonal NC is also given by the FT pattern (Fig. 5E). Longer distances of d-spacing indicate that the inclusion of In ions causes distortion of the copper selenides structure.
The spacing of the lattice planes of the trigonal NC in the HRTEM images (Fig. 5C) shows predominant spots in the FT image (Fig. 5F). These spots could be indexed to the tetragonal phase of CuInSe2 with a = b = 0.5782 nm and c = 1.1619 nm lattice parameters using the CaRine Crystallography software. This is evidence that the morphology change to the trigonal NCs is accompanied by the phase conversion to chalcopyrite structure.
Several groups have reported the synthesis of mono-disperse and solution-stable CuInSe2 NCs using batch reactors,9,12–14,36 however the relationship between morphology and composition has not yet been reported. Here we propose the evolution of CuInSe2 NCs from spherical/hexagonal and finally trigonal shapes as the reaction temperature is increased and the reaction proceeds to completion. A schematic diagram of the process is presented in Fig. 5G. Morphologies of the CuInSe2 NCs are observed to be dependent on Cu/In/Se ratio. As reaction temperature is increased, the spherical NCs are transformed to hexagonal NCs which gradually incorporate In and/or Se ions, finally resulting in trigonal shaped CuInSe2 NCs.
Laboratory-scale PV device was also fabricated to verify the utility of these inks. Chalcopyrite CuInSe2 NCs synthesized herein were used to fabricate a PV device via drop casting on Mo-coated glass and was subjected to annealing at 500 °C in a selenium environment. The final stack consisted of glass/Mo/CuInSe2/CdS/i-ZnO/n-ZnO:Al. XRD patterns of the CuInSe2 NCs synthesized at 240 °C and CuInSe2 thin film after selenization on Mo/glass substrate shows only sharpening of existing chalcopyrite CuInSe2 peaks indicating the annealed CuInSe2 thin film exhibited significant grain growth while remaining phase pure (Fig. 6). The diffraction peaks from Mo and MoSe2 are detectable from the substrate and interfacial layers directly under the CuInSe2 thin film. Peaks arising from the Mo substrate and the MoSe2 interlayer are indicated.
![]() | ||
Fig. 6 XRD patterns of (A) CuInSe2 NCs synthesized at 240 °C by continuous flow hot-injection method and (B) CuInSe2 thin film on Mo/glass substrate after selenization at 500 °C for 30 minutes. |
Fig. 7 shows the current–voltage (I–V) characteristics of the PV device and an FE-SEM image of the cell cross section is included as Fig. 7 inset. As seen in the cross section the CuInSe2 NC film experienced significant crystallization with grains as large as 1–2 μm. The device exhibited a power conversion efficiency of 1.9% under AM1.5 illumination (open circuit voltage = 347 mV, short circuit current density = 19.8 mA cm−2, fill factor = 28%). This efficiency is comparable to previously reported CuInSe2 thin film solar cells using single phase CuInSe2 NCs as precursors ink.7,9,12,37–39 One of the primary efficiency limitations of this device is the low fill factor suggesting a high series resistance which is likely related to the large film thickness of 4.5 μm, which is 3–4 times the thickness typically required to absorb all incident photons. Higher efficiency cells have also been reported using multiphase NC phase ink.39 Further optimization of the PV device fabrication and modification of the precursor components should lead to significant increases in efficiency using these inks.
This journal is © The Royal Society of Chemistry 2014 |