Get Complete Project Material File(s) Now! »
Material requirements and availability for multi-terawatt deploy-ment of CZTS solar cells
The abundance of the elements that constitutes the CZTS material is one of its strongest selling points. The relatively rare elements of the absorber materials competing for market shares in the thin film market, In and Ga in CIGS and Te in CdTe make them unsuitable for multi-terawatt PV deployment. Further, the scarcity of them drive up the price of production, making the materials even less at-tractive for large scale PV deployment compared to CZTS solar cells (Schmalensee et al. 2015). Here, a brief discussion regarding the availability and price of the lim-iting element of the three mentioned materials and the implication for large scale PV deployment will be held.
Firstly, a premise for the discussion must be established. The report: « The Future of Solar Energy; An interdisciplinary MIT study » from 2015 will act as a founda-tion for discussion. In the study they estimate the installed peak capacity needed to cover 5 %, 50 % and 100 % of the electricity demand in 2050. Further, for their calculations they assumed cell eﬃciencies corresponding to the record cells pre-sented by the National Renewable Energy Laboratory (NREL) (NREL 2018) and used the International Energy Agency’s (IEA) projection from 2014, regarding the worldwide electricity demand in 2050 (IEA 2018). In their report, IEA predicts a demand of 33000 TWh in 2050 in their 2 C global warming scenario (ibid.). Fur-thermore, for Schmalensee et al. (2015) calculations they also assumed an annual-and global PV capacity factor of 15 %.
The limiting elements with regards to abundance in the earth crust and price on the market are Sn, In and Te for CZTS-, CIGS- and CdTe absorber layers respectively (ibid.). Out of the three, Sn is both the most abundant and the cheapest if comparing ton/$. According to their data, Sn is roughly one order of magnitude more abundant than In and roughly three orders of magnitude more abundant than Te (Schmalensee et al. 2015). With regards to price, Sn is roughly one order of magnitude cheaper than Te and roughly two orders of magnitude cheaper than In. Proceeding from the premise of PV covering 50 % of the global electricity demand in 2050, and assuming a constant rate of production, like the one today, for the limiting elements one can draw some conclusions regarding the diﬀerent technologies suitability for multi-terawatt deployment. According to Schmalensee et al. (ibid.) it would take roughly 700 years to accumulate the necessary amount of Te to meet the goal with the CdTe technology and about 250 years to accumulate enough In to meet the goal with the CIGS technology. Compared with roughly 1 year to accumulate enough Sn to meet the goal with the CZTS technology. Here it becomes apparent that with today’s extraction rates, only the CZTS technology is suitable for such a large deployment. With this said, several other factors play a part in the technologies suitability for large deployment. For example, markets competing for the resources have not been accounted for. It has been assumed that all extracted resources are available for PV applications, which does not hold true. Furthermore, to state that one PV technology, single handedly would cover 50 % or more of the global electricity demand is unreasonable. Rather a combination of several PV technologies is more suitable, with them having diﬀerent market shares as a result of the raw material availability. Also, other energy producing technologies need to be part of the global energy mix in order to achieve a global energy production based on renewables.
The CZTS material
The crystal structure of Cu2ZnSnS4, CZTS, in its ground-state can be seen in figure
1. It has a tetragonal kesterite structure where the S atom (anion) is surrounded by a Zn, a Sn and two Cu atoms (Dimitrievska 2015). The CZTS kesterite crystal is a quaternary compound with space group I4. Other possible configurations exist, like the stannite crystal structure where the cation order is diﬀerent, with alternating layers of ZnSn and Cu2 instead of CuSn and CuZn as in the kesterite phase and with space group I42m (Ito 2014). Stoichiometric CZTS has been shown to have a direct band gap of 1.5 eV (ibid.).
Figure 1: A picture representing the tetragonal kesterite crystal structure of CZTS in its ground state. In orange: Cu, in blue: Sn, in gray: Zn and in yellow: S. Permission is granted to copy, distribute and/or modify this document under the terms of the GNU Free Documentation License, Version 1.2 or any later version published by the Free Software Foundation; with no Invariant Sections, no Front-Cover Texts, and no Back-Cover Texts.
The to date most successful production route for CZTS thin films is a two stage process, including deposition of precursors and a recrystallization process (anneal). The formation of the CZTS phase during the anneal is described in reaction 1. In thin film solar cell applications, oﬀ-stoichiometric compositions have shown the highest performance. This introduces the possibility for diﬀerent structural defects in the Kesterite structure as well as formation of secondary phases (Dimitrievska 2015). The structural defects come in pairs or triplets with a neutral net charge, called defect complexes. This allows an oﬀ-stoichiometric composition without the crystal structure falling apart (Davydova et al. i.m.). Cu2ZnSnS4(s) )* Cu2S(s) + ZnS(s) + SnS2(s) (1)
One defect of particular interest in this study is the Cu-Zn cation disorder, CuZn + ZnCu which is present in high concentration in CZTS thin films (Ren 2017). This is due to the similar size of the two cations and the anti-site defect’s low formation energy, the ZuCu even having a negative formation energy (Ito 2014). The concentration can be reduced via low temperature thermal treatment (Scragg et al. 2014). Supplying a suﬃcient amount of thermal energy allows the cations to distribute randomly between their respective lattice positions (Davydova et al. i.m.). Alignment into their correct lattice positions starts when the temperature drops below the critical one, TC = 265 10 C, theoretically reaching a perfect order at 0 K (Scragg et al. 2014)(Rudisch et al. 2018). A temperature of 0 K is unachievable and thus some cation disorder remains after the thermal treatment. It is plausible that reducing the amount of disorder can increase the eﬀective band gap, Eg and improve the Voc deficit in finished CZTS solar devices (Scragg et al. 2014). This Cu-Zn disorder can be investigated using near-resonant (NR) Raman spectroscopy and be quantified with the use of the ordering parameters Q and Q’, as shown in Rudisch et al. (2016). Rudisch et al. (2018) concluded that the two mentioned ordering parameters convey the same information. Therefore, only the parameter Q will be used in this study for investigations regarding the cation disorder.
Secondary phase formation in CZTS
Secondary phases can arise from non-optimal film composition that lies outside the single phase region (SPR), as can be seen in figure 2. Based on the article by Siebentritt et al. (2012) a maximum deviation from stoichiometric composition of 1-2 at.-% allows the existence of a SPR. Secondary phases can also form during the annealing of CZTS precursors. The rather low stability of Sn (+IV oxidation state) makes the CZTS prone to decompose during the anneal due to reduction of Sn to oxidation state +II. The Sn-S bonds break and S atoms are extracted from the CZTS leading to the formation of secondary phases such as Cu2S, SnS and ZnS, as described in reaction 2 (Ren et al. 2017). The Sn-S phase decompose in steps from SnS2 to Sn2S3 to SnS via reaction 3. At the film surface the high vapor pressure of sulfur at typical annealing temperatures drives the evaporation of S2(g) and SnS(g), hence partial sulfur pressure in the process is an important parameter to control the decomposition of the Sn-S phase and therefore also the CZTS phase (Ito 2014). At the back of the film, S reacts with the Mo back contact and forms MoS2. The sulfur gets extracted from either the CTZS phase or the secondary phase SnS2 (ibid.). Cu2ZnSnS4(s) )* Cu2S(s) + ZnS(s) + SnS(g) + 1 S2(g) ; SnS(s) )* SnS(g) (2)
The formation of the ZnS phase is driven by diﬀusion that occurs when the con-centration of Zn reaches beyond the limit of solubility in CZTS. It has been shown to occur when the composition is Zn-rich, Zn/(Cu + Sn) > 13 suggesting that the formation of ZnS is mainly dependent on composition (Just et al. 2016). It forms at the back contact during the first minute of annealing before formation also starts at the surface as shown in Ren (2017). The formation of ZnS on the films surface is promoted by the surface decomposition of CZTS due to insuﬃcient partial sulfur pressure (Ren et al. 2017). The formation of the CuxS phase has also been linked to composition of the CZTS material with segregation starting at a Cu/Sn > 2 (Just et al. 2016).
Figure 2: Central part of the ternary phase diagram for CZTS. In the middle lies the SPR for CZTS and branching out are the compositional regions where secondary phases also are present.
The theory behind the chemical and optical characterization tools used in this study is presented in this section.
Energy dispersive x-ray spectroscopy
EDS relies on the interaction between charged particles, in this case electrons, and the samples atoms. Each unique atomic structure has a characteristic set of peaks in its electromagnetic emission spectrum. This fundamental principle is what allows identification of elements (Goldstein et al. 2017). In short, the incoming high energy particles have a possibility to excite electrons in the atom’s inner shells leaving an electron-hole behind. When that electron relaxes back to its original, unexcited state it releases a photon with energy depending on the diﬀerence in energy between the shells. This emitted photon yields a unique signal in the EDS spectrum depending on the atomic structure, much like a fingerprint (ibid.).
Photon-Crystal lattice interaction, Raman and photoluminescence
The Raman eﬀect
The Raman eﬀect, named after its discoverer C.V. Raman, originate from the interaction between incoming photons from a light source and phonons in the material (Loudon 1964). When light gets scattered by a material it mainly contains the wavelengths that were incident on the sample but some other wavelengths are also present at very low intensities: a few parts per million or less compared to the main signal. This represents an interaction with the material and is what is called the Raman eﬀect (Schroder 2006). In figure 3a basic scattering processes is shown and next to it in figure 3b the diﬀerence in energy shift for the mentioned scattering processes is shown. The Rayleigh scattering is an elastic process where the outgoing photon has the same wavelength as the incoming one. The Stokes scattering is an in-elastic process that imparts energy on the crystal lattice in the form of a phonon and in Anti-Stokes scattering a phonon from the lattice imparts its energy on an incoming photon (Rudisch 2016). It is possible for these in-elastic scattering processes to impart several phonons worth of energy on the crystal lattice or vice versa: with decreasing probability for each extra phonon. It is then called 1st order for one phonon and 2nd order for two and so on. These higher order scattered photons show up in the Raman spectrum as peaks at higher frequencies, exact position is dependent on the phonons imparted. The Anti-Stokes mode is much weaker than the Stokes mode, therefore it’s usually the Stokes scattering process that’s monitored in Raman spectroscopy by comparing the incoming light with the outgoing (Schroder 2006). The unique atomic structure of each compound results in characteristic shifts in energy of the scattered light. This can be seen in a Raman spectrum obtained via Raman spectroscopy as Raman peaks characteristic to the compound, each compound having its own unique « fingerprint » (Schroder 2006). Some compounds have overlapping Raman peaks due to similar atomic structure, making diﬀerentiation more diﬃcult.
(a) Basic scattering processes of light. The scattering can in short be described in three steps. 1: A photon excites an electron in the material. 2: The ex-cited electron interacts with the lattice structure and creates (Stokes) or ab-sorbs (Anti-Stokes) a phonon. 3: The electron recombines and releases a pho-ton.
(b) Intensity of energy shift of scattered light from the basic scattering processes seen in figure 3a. Starting from the left is the Stokes scattering, next is the Rayleigh scattering and last is the Anti-Stokes scattering.
Figure 3: Permission is granted by the copyright holder Katharina Rudisch to use and modify these pictures taken from her presentation on sub phase analysis of Cu2MnSnS4 with multi-wavelength Raman spectroscopy from 2016 (Rudisch 2016).
The use of Raman spectroscopy serves as a powerful tool to characterize crystal structures and identifying the presence of secondary phases. The standard exci-tation wavelength in Raman spectroscopy is 514 or 532 nm. In investigation of kesterite phase CZTS it yields a large number of Raman peaks with relativity low intensities making it hard to determine the diﬀerent peak positions and also identifying secondary phases. Instead, using an excitation wavelength of 785 nm results in a resonance eﬀect for certain kesterite phase CZTS Raman modes, mak-ing them several magnitudes more intense than the modes of secondary phases possible present in the material (Dimitrievska 2015). The resonance eﬀect occurs when the energy of the excitation source is near the required energy for electronic transition to the conduction band to take place (Strommen et al. 1977). A compound’s conduction band structure can be composed of several energy states to which an electronic transition can occur, leading to it having several possible points of resonance (Dimitrievska 2015). For CZTS these points of resonance occur at the band gap and at an energy level of 3.50 eV. Identification of secondary phases, using the resonance eﬀect, in kesterite phase CZTS requires the use of other ex-citation wavelengths, with the exception of the Cu3SnS4 phase which also shows a high intensity in Raman spectra when using 785 nm excitation wavelength due to the resonance eﬀect (ibid.). Using UV excitation (325 nm) for example yields a near resonance eﬀect for the compound ZnS which is a common secondary phase in oﬀ-stoichiometric CZTS materials. This makes identification of the phase rela-tively easy (ibid.).
The Raman spectra peak intensities and their corresponding position give valu-able information about the sample. The intensities give information about the material’s electronic properties, its band structure etc. and the peak position give information about the mechanical properties of the crystal like atomic masses, bond strengths, geometry etc. Analyzing the shift in observed peak position com-pared with the expected one can give information about strain and stress in the crystal structure. The width of the peak also contains information, giving a hint about the crystal quality of the material for example
Photoluminescence occurs when a photo-excited electron relaxes back to its orig-inal state and emits a photon with wavelength corresponding to the diﬀerence in the excited and unexcited energy state (Schroder 2006). This is what is called ra-diative recombination. Non-radiative recombination can also occur in a material. The excitation energy then dissipates via other forms of energy, such as heat or by inducing lattice defects in the crystal structure (Pelant et al. 2012). A material’s photoluminescent behavior can give an indication of how well it is suited for PV applications by measuring its radiative recombination. In impure materials several radiative recombination routes are possible, the most common ones for kesterite CZTS are shown in figure 4 (Schroder 2006).
At room temperature the band to band recombination, figure 4(a) dominates (ibid.). Recombination between the materials acceptor- and donor states, fig-ure 4(b) originating from the materials impurities or defects occur as well but are weakly pronounced at room temperature. For kesterite phase CZTS the abundant presence of structural defects and impurities seem to entail potential fluctuations in the band gap structure, altering its shape (Van Puyvelde et al. 2015). Therefore band to band recombination is seldom observed. Further, the high density of de-fects aggravate clear definition of donor- and acceptor energy states. In conclusion, what mode of recombination one observes in CZTS is hard to define. This topic is widely discussed in the research community, see for example the articles by Lang et al. (2017) and Van Puyvelde et al. (2015). Measuring photoluminescence is performed in the same way and with the same equipment as Raman spectroscopy, described in the section above. The obtained spectrum can be analyzed by iden-tifying the peak position and its intensity. Giving information about how much radiative recombination occurs in the material and comparing the peak position with the expected band gap for the investigated material can give information regarding the presence of defect states (Schroder 2006).
Table of contents :
1.1 Scope of the Project
2.1 The Kesterite CZTS
2.1.1 Material requirements and availability for multi-terawatt deployment of CZTS solar cells
2.1.2 The CZTS material
2.2 Secondary phase formation in CZTS
2.3 Analysis techniques
2.3.1 Energy dispersive x-ray spectroscopy
2.3.2 Photon-Crystal lattice interaction, Raman and photoluminescence
3.1 Preparation of samples
3.1.1 Precursor deposition
3.1.2 High temperature anneal
3.1.3 Post-anneal ordering treatment
3.2 Characterization of samples
3.2.1 Elemental analysis
3.2.2 Raman and photoluminescence spectroscopy setup
4 Results and discussion
4.1 Compositional analysis
4.2 Components of a CZTS Raman spectrum
4.3 Secondary phase identification
4.4 Distribution of secondary phases
4.5 Identification of the single phase region
4.5.1 Influence of annealing conditions/partial sulfur pressure on SPR and secondary phase formation
4.6 Investigation of the CZTS cation disorder
4.7 Photoluminescence of CZTS
4.7.1 The effect of composition and annealing conditions on PL intensity
4.7.2 The effect of composition and annealing conditions on PL peak position
4.7.3 The photoluminescence’s dependence on cation order