1
MSc Chemistry: Science for
Energy and Sustainability
By: Roxanne Strijdhorst
12308242
Master Thesis (60 EC)
November 2019 – August 2020
Synthesis and Characterization of
the Chalcogenide Perovskite: BaZrS
3
Supervisor:
prof. dr. E.C. Garnett
Second examiner:
prof. dr. J.N.H. Reek
2
Abstract
Chalcogenide perovskites have emerged as a promising semiconductor material for photovoltaic applications, however, not many examples of thin-film and nanocrystal synthesis exist which hinders device application. Herein the synthesis and characterization of thin-film or nanocrystal chalcogenide perovskite BaZrS3 is reported. Four synthesis routes were considered: chemical vapor deposition (CVD), molecular sieve
encapsulation, hot-injection synthesis and intercalation. Of the four, BaZrS3 was successfully synthesized by
high-temperature sulfurization of SiO2 molecular sieves loaded with precursor salts. Crystallite sizes of 3.45
– 7.94 nm were achieved due to the confined pores of the molecular sieves that encapsulated the BaZrS3
nanocrystals. Through XRD measurements other Zr-phases were found. Bandgap calculated from UV-Vis absorption obtained 1.89-2.31 eV. CVD synthesis produced nanodiscs of ZrS2 and nanowires that showed
weak absorption which is unlike BaZrS3, however, similar photoluminescence. Solution-synthesis routes
(hot-injection and intercalation) could not achieve the high temperatures that are necessary to synthesize BaZrS3.
3
Table of Contents
1. Introduction ... 4 2. Background ... 5 2.1 Perovskite... 5 2.1.1 Structure ... 52.1.2 Perovskites for Photovoltaics... 5
2.2 Chalcogenide Perovskite ... 6
2.2.1 Structure and properties ... 7
2.2.2 Zr-chalcogenides ... 7
2.3 Synthesis Methods... 12
2.3.1 Chemical Vapor Deposition ... 12
2.3.2 Molecular Sieve Encapsulation ... 13
2.3.3 Hot-Injection Synthesis ... 14
2.3.4 Intercalation ... 15
3. Materials and Methods ... 16
3.1 General Procedures ... 16
3.2 CVD ... 16
3.3 Molecular Sieve Encapsulation ... 17
3.4 Hot-Injection Synthesis ... 17
3.5 Intercalation ... 17
4. Results and Discussion ... 18
4.1 CVD ... 18
4.2 Molecular Sieve Encapsulation ... 22
4.3 Hot-Injection Synthesis ... 26
4.4 Intercalation ... 28
5. Conclusion and Outlook ... 29
6. Acknowledgements ... 29
7. References ... 31
8. Appendices ... 35
A1. Tauc Plots ... 35
A2. Scherrer Equation ... 36
4
1. Introduction
By 2050, global energy consumption is expected to increase 50% in comparison to 2018, more than half of this coming from the industrial sector. A challenge in meeting this increased consumption is doing it in such a way that limits human impact on climate change, as energy-related CO2 comprises two-thirds of all
greenhouse gases1,2. Analysis of different energy pathways show that it is possible to achieve a global
energy transition while avoiding climate change, however, combinations of new technologies and policies are necessary1.
Driven by the demand for clean energy, the fastest growing energy form is renewable energy2. New
additions of renewable energy technologies installed have reached a record high due to the increased competitiveness and falling costs, especially solar photovoltaics (PV) and wind power1,3. PV is particularly
attractive for energy generation due to its quietness, scale flexibility, and simple maintenance and significantly less damages to the local environment of its energy generation and can be placed in urban areas, for instance, on roofs. Therefore, PV has an the potential to be a significant contributor to the global energy transition4.
Efficient materials for PV devices are good semiconductors with a suitable optical bandgap. Many materials have successfully been developed into solar cells including Si, GaAs, CdTe, and CuIn1-xGaxSe2. Si solar cells
are currently the most common type of solar cell, however, it is a poor light absorber and requires thick, high quality layers to absorb enough light to generate charge carriers5. Research on PV materials is focused
on decreasing manufacturing cost and increasing efficiency. One strategy to achieve this is finding materials that can serve as an addition or an alternative to Si devices6.
Since 2009, organic-inorganic halide perovskites have broken through as a promising material for PV devices due to their unprecedented increase in power conversion efficiency (PCE). Efficiencies of >20% for these devices have been attained, largely due to their high charge carrier mobilities and long diffusion length. Additionally, their bandgap is easily tunable by changing the organic or halide components, or through doping7. They consist of abundant elements and can be readily synthesized using mild conditions.
Both factors make these materials ideal for low-cost, large scale manufacturing8.
Though organic-inorganic perovskites have appealing optical and electrical properties, their commercialization is hindered by the use of lead, a toxic heavy metal, as the inorganic component and the instability of the material under typical humidity levels9,10. Chalcogenide perovskites containing chalcogen
atoms (S, Se and Te) are a lead-free and more stable alternative to lead halide perovskites. They exhibit optimal bandgaps that allow them to have a theoretical maximum power conversion efficiency of ~30%. However, their properties, such as the carrier type, absorption coefficient and defect properties, have remained largely unexplored. This is due to the lack of thin film and nanocrystal samples as the synthesis of chalcogenide perovskites generally obtains bulk powder samples. The lack of understanding of these properties becomes an obstacle for device application11. Hence, this research project aims to synthesize thin
film or nanocrystal chalcogenide perovskites. Techniques such as chemical vapor deposition, hot-injection synthesis, intercalation and molecular sieve encapsulation will be attempted.
5
2. Background
2.1 Perovskite
Traditionally, the term perovskite is used for calcium titanate ore (CaTiO3). It currently describes a class of
materials that exhibit the same structure as CaTiO3, namely, the ABX3 chemical formula11. The most common
type of perovskites are oxide perovskites. There is a huge variety of oxide perovskites, as their crystal structure is extremely versatile and can accommodate around 90% of metallic elements. Oxide perovskites are leading materials in condensed matter science due to their wide range of functional properties, for example, catalytic, superconducting, piezoelectric, pyroelectric, ferromagnetic and multiferroic properties. These properties have strong correlation with their crystal structure12.
2.1.1 Structure
In this ABX3 configuration, an octahedral coordination is formed between the B-site cation and the six
X-site anions to form [BX6]n- octahedra (Figure 1). A 3-D framework of corner-sharing octahedra is formed
with the X-site anion in the corner. This results in a network of B-X-B-X-B-X bonds in three directions. Hence, the optical and electrical properties of the perovskite is largely controlled by the corner-sharing octahedra. The A-site cation fill the voids created by this octahedra and form a cuboctahedron by coordinating with 12 X-site anions. Their purpose is to fine-tune the electrical an optical properties by ensuring structural stability and charge neutrality11,13.
The structure of perovskites is largely governed by the Goldschmidt tolerance factor, which can be defined by the following equation:
t = 𝑟𝐴+ 𝑟𝑋
√2(𝑟𝐵+ 𝑟𝑋)
wherein rA, rB, and rx are the radii of the A-, B- and X- site ions, respectively. The ionic radii value should be
taken from the list of Shannon’s effective ionic radii. An ideal perovskite structure, with a B-X-B-X bond angle of 180° (Figure 1a) would have a 0.9<t<1. The Goldschmidt tolerance factor of a distorted perovskite structure (Figure 1b) would be in the range of 0.71<t<0.911. The structural flexibility allows for different
elements to be incorporated into the A-, B- and X- sites, each with its own set of interesting properties. The corner sharing octahedra is less favorable for t<0.71 or t>1, at this point is when hexagonal or needle-like phases form. Similar to the perovskite structure, the A-site cation is 12-coordinate while the B-site cation is 6-coordinate in the hexagonal phase. In needle-like phase, the coordination of the A-site cation is reduced to 9 while that of the B-site cation remains the same11.
2.1.2
Perovskites for Photovoltaics
Halide perovskites have recently been established as strong contenders to be materials for conventional semiconductors. There are two classes of halide perovskites, the alkali-halide perovskite and the organo-metal halide perovskite. The main difference is the A-site cation. Alkali-halide perovskites have alkali organo-metals such as Li+, Na+, K+, and Cs+ as their cation. Meanwhile the organo-metal halide perovskites have organic
cations like aliphatic or aromatic ammonium ions. Both types, however, have divalent metals as their B-site cation (ex: Zn2+, Ge2+, Sn2+, Pb2+, Fe2+, Co2+) and halides as their X-site anion7.
The efficiency of solar cells made with halide perovskites have already reached 25.2% in 201914 from 3.8%
in 200915. This efficiency has already surpassed other types of commercial semiconductor materials such as
CdTe and CuIn(1-x)Ga(x)Se216. The success of halide perovskite solar cells is from a combination of factors.
6
synthesized quickly and at low-temperature conditions through a variety of solution and evaporation methods17.
Figure 1 Different phases for ABX3 compounds (a) Ideal perovskite phase (b) distorted perovskite phase (c) hexagonal phase (d) needle-like phase. Grey spheres represent the A-site cation and the brown spheres represent the X-site anion. The B-site cation is located in the middle of the green octahedra. Image from ref. [13]
Halide perovskites, compared to other conventional semiconductors, have properties like low carrier concentrations (~1013 cm-3) and long carrier lifetimes (~1 µs)16. Additionally, they benefit from relatively
high carrier mobilities, long charge carrier diffusion lengths (up to 1 µm), and absorption stronger than Si7.
These properties originate from the interesting structure of halide perovskites. Halide perovskites are ionic, meaning that the Coulomb attraction is maximized between the cations and anions as a result of the higher coordination. This ionicity is also responsible for the minimization of deep-level defect formation, which are responsible for non-radiative recombination16. The characteristics of the M-X bond largely determines the
electronic structure of halide perovskites, therefore, the material properties, such as the bandgap, are easily tunable by changing the components of the M/X-site ion. Because of this tunability, halide perovskites can be used as a device on its own or in tandem with another material7.
Most interest has gone towards perovskites with metal halides from the fourth main group (group 4A: Ge2+,
Sn2+, and Pb2+) due to their good optoelectronic properties and easy, low temperature fabrication. However,
halide perovskites are relatively unstable. The material degrades when subjected to moisture, oxygen, radiation, pressure, and elevated temperatures. This, combined with the use of lead, a toxic heavy metal, hinders its potential for commercial PV applications7.
2.2 Chalcogenide Perovskite
Chalcogenide perovskites are more similar to the classic oxide perovskites. They feature the chalcogens S and Se as the X component, and the A and B components represent metals with a +2 and +4 valence, respectively13. Though reports of their synthesis have existed since the 1950s18, chalcogenide perovskites
have received very little attention from the PV community. Publications on these materials are scarce, hence, there is an insufficient amount of knowledge available on their physical properties, limiting their application in PV devices19.
7
2.2.1 Structure and properties
Experimentally, chalcogenide perovskites are known to exist in three phases: the distorted phase, the hexagonal phase and the needle-like phase. Most chalcogenide perovskites crystallize with the space group
Pnma in the distorted perovskite phase (or GdFeO3 structure)20. There have been some reports, for example
BaTiS3, of chalcogenide perovskites existing in the hexagonal phase (BaNiO3 structure) with the space group
of P63/mmc21. Examples of needle-like phase (or NH4CdCl3 structure) materials have been reported for
BaZrSe3, SrZrS3 and SrZrSe322,23. The distorted perovskite structure is preferred for PV applications. The
needle-like and hexagonal phases tend to display low carrier mobility due to the heavy electron and hole masses as a result of the localized conduction. This localization arises from the lack of octahedra connections in certain directions9. In general, the distorted perovskite phase has the largest bandgap and the hexagonal
phase has the smallest13.
2.2.2 Zr-chalcogenides
Though different materials have been established as optimal by different studies, BaZrS3 (BZS), SrZrS3(SZS)
and CaZrS3 have been repeatedly identified as promising for PV applications by several computational
screening studies13,24,25. Recent literature on chalcogenide perovskites mainly focuses on BZS9,13,19,26.
However, SZS has also been extensively studied in both its existing phases: α-SZS and β-SZS19,22,26. Hence,
BZS and both phases of-SZS will be focused highlighted in the next section. 2.2.2.1 Structure and Synthesis
Synthesis of BZS has only resulted in the distorted perovskite phase. Meanwhile, SZS can be synthesized in the α- and β-phase, which are in the needle-like phase and distorted phase, respectively26.
Typical synthesis routes that have been employed for chalcogenide perovskites is through the high temperature sulfurization of the oxide counterparts using CS2 or H2S or by conventional solid state synthesis
by a binary sulfide mixture26. However, similar to using DFT calculations, the properties of the material vary
with different synthesis routes. For instance, Meng et al performed solid-state synthesis of BZS using binary mixtures (BaS and ZrS2) followed by repeated annealing and a bandgap of 1.85 eV was obtained9.
Meanwhile, Perera et al. obtained a bandgap of 1.7 eV through the sulfurization of BaZrO3 with CS219. Both
methods achieved pure phases as demonstrated by the EDX spectrum in Figure 2c. Figure 2a and b show SEM images of BZS and its starting oxide counterpart19. The powder changed color from white to black,
suggesting that a considerable part of the visible light spectrum is absorbed26. Perera et al. also synthesized
SZS in the same manner. This only resulted in the formation of α-SZS. Therefore, Perera concluded that SZS is not a promising candidate as the needle-like α phase demonstrates small bandgaps outside the range of their UV-Vis spectrometer19.
Figure 2 SEM images of (a) polycrystalline BaZrO3 powder (b) resulting polycrystalline BZS (c) EDX of BZS. Scale bars
8
Niu et al. developed a novel synthesis of BZS and SZS using iodine as a catalyst. This techniqued enabled both α-SZS and β-SZS to be synthesized separately. Iodine is added to a conventional solid-state synthesis reaction of the earth metal sulphide (SrS or BaS) and an elemental source (S and Zr). Iodine acted as a flux to enhance the reactivity of the precursors. Thus, the reaction was carried out over several hours instead of several days. The bulk chalcogenides were synthesized with high-quality as can be seen with the EDX and XRD in Figure 326. BaCl2 was also used successfully by Niu et al to create a flux in the solid-state synthesis
of BZS27.
2.2.2.2 Stability
A major concern regarding the use of perovskites in solar cells is their stability. As previously mentioned, organic-inorganic halide perovskites are not stable under certain conditions7. Furthermore, these types of
perovskites are based on lead which will have a negative impact if released into the environment10. Hence,
chalcogenide perovskites will be evaluated in the following section based on their stability in moisture, air, light and heat.
Niu et al. examined the α- and β-SZS and BZS under ambient conditions. The materials showed no color change or degradation that was measurable after one year28. To compare, the longest amount of time that
an organic-inorganic halide solar cell was stable is 1000 hours29. Furthermore, differential scanning
calorimetry (DSC) and thermogravimetric analysis (TGA) up to 1200 °C were performed on the chalcogenides by Niu et al to evaluate thermal stability. All materials remained stable at temperatures well above 500 °C. α-SZS was the first to be oxidized at 550 °C. The two distorted phase materials, β-SZS and BZS, were both oxidized at approximately 650 °C. The degradation products were studied with XRD and EDX. A mixture of ZrO2, BaSO4 and BaZrO3 were obtained as degradation products for BZS. SZS also
provided a mixture of SrZrO3, SrSO4, and ZrO228.
That α-SZS was subject to oxidation at lower temperatures proves that lower symmetry phases (needle-like phase) are less stable than that of higher symmetry phases (distorted perovskite phase)28. However, both
types of chalcogenides are more stable than hybrid perovskites. CH3NH3PbI3 films have been reported to
degrade at 85 °C29. Though it is useful to know that chalcogenide materials are stable on their own, it is also
important to test the thermal stability, stability in humid conditions, and stability under illumination of a device containing these materials for future implementation. No reports like this exist for chalcogenide perovskites.
2.2.2.3 Material Properties
The bandgap of a material is critical in defining the efficiency limit of a solar cell. The relationship between efficiency and the size of bandgap was calculated by Shockley and Queisser, they demonstrated that to reach a theoretical efficiency limit of >25% for single junction solar cells, the bandgap must be in the ideal range between 0.9 eV and 1.6 eV. This is known as the Shockley-Queisser limit30. For tandem solar cells, the
optimal bandgap for the top cell should have a bandgap of 1.7 eV if Si is the bottom cell to reach maximum efficiency31. Hence, BZS and SZS will be evaluated based on their bandgap.
9
Figure 3 Images of the synthesized (a) SZS powder (b)B-SZS (c) BZS. XRD patterns and Rietvield analysis of (d) α-SZS © β-SZS and (f) BZS. Simulated intensity profiles are represented by the black lines. Raman Spectra of (g) α-SZS (h) β-SZS and (i) BZS at room temperature. Image from ref. [26]
The bandgap of BZS was determined to be at a range of 1.7 – 1.85 eV, both experimentally and theoretically9,13,16,19,26. Figure 4a shows the photoluminescence (PL) spectra from which the bandgap was
identified by Niu et al. This range of bandgaps are the most suitable as a top cell for a tandem solar cell device32. An exception to this range was determined by Kuhar et al. in their computational screening study,
wherein they predicted a bandgap of 2.25 eV25. Meanwhile, α-SZS, in the needle-like phase, has an
experimentally determined bandgap of 1.53 eV26. Computational studies have determined the bandgap to
be 1.46 eV25 or ~1.35 eV13. β-SZS has a bandgap experimentally measured to be 2.13 eV26. Computationally
predicted bandgap is 2.50 eV25 and ~1.85 eV13.
Figure 4(a) Photoluminescence spectra for α-SZS, β-SZS and BZS (b) Band structure of α-SZS, β-SZS and BZS Image from ref. [26]
10
The type of bandgap is also an important factor in solar cell performance. Direct bandgaps have been known to perform better than indirect bandgaps. For this reason, Si, which is an indirect bandgap material, requires thick layers and CdTe, a direct bandgap material, can be made very thin. Figure 4b shows the band structures of α-SZS, β-SZS and BZS. The band structures show that all three perovskites are direct band bandgap materials as all their valence band maximums and conduction band minimums fall on the Γ point26. The
band structures of both distorted perovskite structure materials, β-SZS and BZS, are fairly similar. The large dispersion of their conduction band minimum suggests good electron transport19. On the contrary, α-SZS
and β-SZS have different band structures, despite being of the same chemical composition26. Niu et al.
suggested that the lack of a corner sharing octahedra in the needle-like phase is the cause of this discrepancy. Therefore, α-SZS is a weaker absorbing material as compared to β-SZS and BZS26.
The potential of chalcogenide perovskites to be solar absorber materials was further evaluated by Nishigaki et al. The chalcogenide perovskites in the study were synthesized via conventional solid-state reactions, they were all of distorted perovskite phase. The absorption coefficient (α) of the materials are exceptionally high exceeding 105 cm-1 in the band edge region as can be seen in Figure 5a33. The dashed line in the figure
represents the α values obtained from a study by Wei et al., a high α was also obtained, however, at ~1.97 eV, which is largely redshifted compared to that of Nishigaki16,33. This is a result of the different techniques
used to obtain the values. Nigishaki et al derived their Eg values from Tauc plots based from high-sensitivity
ellipsometry technique and Wei used UV-Vis absorption instead16,33.
Figure 5 (a) Absorption coefficient of chalcogenide perovskites. Circles display experimental spectra, the solid lines display DFT spectra. (b) Absorption coefficient of chalcogenide perovskites in comparison to other PV absorbers. This figure shows the the variation of α from Eg, a nominal position defined at α = 103 cm-1 (Eg’). Image from ref. [33]
The band-edge α of chalcogenide perovskites were compared to other solar absorber materials in Figure 5b. E’g is an energy that corresponds to α = 103 cm-1 and the change of α from ΔE = E – E’g is obtained. The
figure shows that chalcogenide perovskites have extraordinarily high α in comparison to GaAs, CuInSe2, and
halide perovskites CH3NH3PbI3. The high band-edge absorption allows for thinner films to be used, which
improves the effectiveness of charge carrier concentration without sacrificing absorption. Thinner films also benefit from a higher open circuit voltage (Voc). Additionaly, these perovskites show a sharp absorption
curve, which is important as tail-state absorption reduces the Voc. Nishigaki, using DFT calculations,
determined that tail-state absorption is negligible for these materials33.
To produce efficient solar cells, the formation of deep-level defects should be avoided as they inhibit charge transport by trapping electrons and holes34,35. This can be done by promoting growth conditions that
11
facilitate formation of the necessary shallow-level defects for suitable doping while avoiding deep-level defects. Meng et al. investigated BZS in terms of its defect levels using DFT calculations (Figure 6). 12 possible intrinsic defect points were considered, namely, three vacancies (VBa, VZr, VS), three interstitials (Bai,
Zri, Si), two cation substitutions (BaZr, ZrBa), and four antisite substitutions (BaS, ZrS, SBa, SZr). The transitions
energies that were calculated are displayed in Figure 6. The red lines mark acceptor-like defects and the blue lines mark donor-like defects. From the 12 points, 5 were found to produce deep-level defects: Si, SBa,
SZr, ZrS and Zri. The other points produce shallow-level defects9. From the calculations, Meng et al. proposed
that using near stoichiometric conditions suppresses the formation of deep-levels defects while maintaining charge carrier density and hence is the most desirable in the synthesis of BZS perovskites9. There is currently
no study on the defect formation of SZS, however, chalcogenide perovskites are, in general, less likely to form them due to their strong ionicity19.
Figure 6 Transition energy levels (eV) of intrinsic defects in BZS as calculated by Meng et al. Red bars represent acceptor levels and blue bars represent donor levels. Image from ref.[9]
2.2.2.3.1 Thin Film Properties
Though chalcogenide perovskites show strong potential for PV applications, synthesis methods previously discussed focus on obtaining powder or single crystal bulk samples16,26. One factor contributing to the rapid
success of organic-inorganic halide perovskites is the easy and high-quality synthesis method of solution processed thin film growth25. Lack of thin film samples limits knowledge on how the perovskite would
behave as a solar cell device16,26.
So far only two examples of chalcogenide thin films exist: LaYS3 and BZS. LaYS3 was the first thin film to be
synthesized. It was done by the co-sputtering of La and Y onto a fused silica substrate followed by sulfurization in a quartz tube furnace. The bandgap experimentally obtained was 2 eV. Though this is too high for PV applications, Kuhar et al. showed that it is a promising candidate for a tandem device for photochemical water splitting25.
BZS was first synthesized in thin film form in 2019. Wei et al used puled laser deposition to deposit BaZrO3
on sapphire substrates at 800 °C. Sulfurization using CS2 followed. A polycrystalline film with no preferential
orientation was obtained. Grain size and electron mobility increased with sulfurization temperature, an increase in grain size is advantageous in suppressing carrier scattering. A slight sulfur deficiency, likely
12
caused by sulfur vacancies, was observed at different locations on the film. A bandgap of 1.82 eV was obtained16.
Hall measurements showed that the film was n-type, suggesting that electrons are the dominant carriers. This is potentially due to the sulfur vacancies as each of them will donate 2 excess electrons. The measurements showed a carrier density in the range of 1019-1020, suggesting considerably high doping
levels. Hall mobility, which was dependent on sulfurization temperature, ranged from 2.1 – 13.7 cm2/Vs, this
is expected as the grain size and crystallinity increased with sulfurization temperature, suppressing carrier scattering. 13.7 cm2/Vs is comparable to that of MAPbI3. The limiting factor is the carrier concentration,
however, that could be improved with proper passivation and reduction of carrier density. As mentioned previously, the α obtained was greater than 105 cm-1. Temperature dependent conductivity measurements
indicated shallow donor levels. Their results suggested that the sulfur vacancies provide the shallow defect levels that are ionized readily at room temperature. Further studies are needed to elaborate on defects and carrier transport mechanism. Their measurements show that BZS is a promising candidate for optoelectronic applications16.
Even more recently, Comparotto et al. used sputtering of BaS, and Zr in an H2S atmosphere and subsequent
annealing to synthesize a thin film of BZS. Industrially, sputtering is widely used for large-area thin film production. This synthesis method produced high-quality films, with small traces of ZrO2. However, to
achieve reasonable crystalline quality, annealing at 800-900 °C was required. They saw this as evidence for a large energetic barrier for the nucleation and growth of the processes employed which could be linked to the high chemical stability of BZS in comparison to halide perovskites32.
2.3 Synthesis Methods
It remains unclear whether the high reaction temperatures are an inherent characteristic of BZS or if an alternative synthesis method may be employed at milder reaction conditions. This is the aim of the project and the synthesis routes used in the study will be elaborated on further in this section.
2.3.1 Chemical Vapor Deposition
Chemical vapor deposition (CVD) is a prevalent bottom-up approach to produce thin films due to its ability to produce large-scale coatings of uniform thickness36,37. The film is grown through a series of chemical
reactions between precursor molecules in the gaseous phase38. The volatile precursors then decompose on
a heated substrate surface to deposit the thin film36. Precursors are held at the ideal volatilization
temperature while a carrier gas acts as which is both a diluting gas, similarly to a solvent, and to transport the precursors to the substrate36,38. Carrier gases often comprise of hydrogen, nitrogen, argon or any
mixture of these gases38.
There are many sequential process steps in a CVD reaction (Figure 7). The first step, (a) involves transport of the gaseous precursors to the boundary layer which is a stagnant layer surrounding the substrate. The gaseous precursors are then transported to the substrate surface (b), the precursors then adsorb onto the substrate surface (c) wherein chemical reactions take place (d). These chemical reactions may be surface reactions between the adsorbed reactants, between the adsorbed reactants and the reactants still in the vapor phase or between reactants in the vapor phase. In e), the initial nucleation starts. Reaction by-products desorb in f) and are transported to the boundary layer where they are carried away37.
13
Figure 7 Steps of a CVD process. Image from ref. [37]
CVD is easily tunable, by varying reaction conditions such as the amount/type of carrier gas, substrate temperature, substrate material, and total pressure, a wide variety of materials with different physical and chemical properties can be grown37. This is because these conditions control the rate of diffusion of the
starting precursors in the vapor phase. Crystallinity and reaction rate are also affected by the temperature. In general, high temperatures favor crystalline films and lower temperatures favor amorphous films. CVD, however, has many advantages, such as: possibility for large-area film production, high deposition rates, good coverage of 3-D structures, flexibility, and the ability to work without high vacuums. Drawbacks include the high deposition temperatures, use of toxic precursors, the production of toxic by-products and the difficulty to precisely control film thickness36.
In spite of its drawbacks, CVD has already been used to synthesize lead halide perovskite thin films. Tavakoli et al, for example, used methylammonium iodide and lead iodide to synthesize a thin film of MAPbI3 on a
glass substrate coated with TiO217. Additionally, there are examples of CVD synthesis of transition metal
dichalcogenides (MX2, X being a chalcogen) such as MoS2 and WS2. Wang et al used low-melting,
evaporable chloride and sulfur to synthesize, in a large-scale, atomically thin ZrS2(Figure 8). S was placed
outside of the furnace, separate from ZrCl4 due to its lower melting point39. Hence, with this knowledge,
CVD will be used to try to synthesize BZS.
Figure 8 Schematic of the CVD synthesis of ZrS2. Image from ref. [39]
2.3.2 Molecular Sieve Encapsulation
To achieve good stability and maintain the excellent properties of halide perovskites, Zhang et al recently developed a facile synthesis for ceramic-like, stable and luminescent CsPbBr3. This was achieved first by
soaking PbBr2 and CsBr salts in water with the molecular sieves (MS) derived from silica. Upon heating (600
14
demonstrated robust stability, even against harsh hydrochloric acid solutions for 50 days and maintained their photoluminescence quantum yield (PLQY)40.
Figure 9 Synthesis of CsPbBr3 nanocrystals into a molecular sieve (SiO2). Image from ref. [40]
2.3.3 Hot-Injection Synthesis
Hot-injection synthesis is a robust and easily scalable method for the production of high-quality nanocrystal inks. These inks can be deposited on substrates to achieve low-cost PV. Hot-injection was first used in 1993 for cadmium chalcogenide nanocrystals. Following this report, it was extended to many other kinds of nanocrystal materials such as metal oxides and semiconductor materials. It is a well-established technique due to its simplicity and fast reaction times. Hot-injection also provides control over crystal size, composition, morphology, and phase formation.
The hot-injection technique is based on a nucleation and growth mechanism. Many parameters influence the nucleation and growth such as the concentration of the precursors, the type of ligand, the molar ratio of the precursor and capping ligand, and the time and temperature of the reaction41. Figure 10 shows a
schematic representation of the hot-injection method. This method follows the Oswald Ripening Principle which states that the dissolution and re-deposition of smaller crystals onto the surface of larger crystals take place when the driving force for crystallization is low42.
It begins with the rapid injection of a cold solution containing precursors into a hot solution of surfactants. This causes nuclei to form instantaneously and causes a supersaturation of nanocrystals. Nucleation is quenched by the reaction mixture cooling and the sharp decrease in monomer concentration41,43. It is
inevitable, in this stage, to have a broad crystal size distribution as nucleation and crystal growth are happening simultaneously, promoted by supersaturation. If the duration of nucleation is much shorter than the growth process, it ensures that the nanocrystals reach the same size during growth. During the synthesis, the nanocrystals are capped with the surfactant molecules in solution to prevent agglomeration and provides them with colloidal stability42.
Growth of nanocrystals involves diffusion of the monomers (an anion-cation pair) to the surface of the nuclei which subsequently react. It has been done before with many types of sulfide materials, for example, Cu(In1-yGay)(S1-xSex)2 (CIGSSe), Cu2ZnSn(S1-xSex)4 (CZTSSe), and Cu2Zn(Sn1-yGay)(S1-xSex)4 (CZTGSSe). The
hot-15
injection synthesis of these materials entail injection of a cold sulfur-oleylamine solution into a hot solution of metal precursors (typically chlorides, bromides, acetlyacetonates, or acetates), reaction temperatures typically range from 220 to 285 °C44.
Figure 10 Mechanism of synthesis of CZTS based nanocrystals using hot-injection synthesis. Image from ref. [41]
Oleylamine (OLA) is a primary amine with a long alkyl chain, it is the most commonly used coordinating solvent due to its high boiling point (364 °C) and high coordinating ability. Furthermore, it can act as an electron donor at high temperatures, therefore, it can act as the capping agent to stabilize the nanocrystal surface, and the reagent that activates the metal precursors. Though it has been widely used, some concerns arise due to the toxicity and price of OLA. Trioctlyphoshine oxide (TOPO) can also be used on its own, or in combination with OLA. It can help provide better conditions for growth and facilitate size and crystal structure. This is due to the steric hindrance it provides once the surface curvature of the nanocrystals increases41.
Octadecene (ODE) is a cheap, safe, and high boiling (~315 °C) solvent. It is a non-coordinating solvent that can be used to tune the reactivity of the monomers by making a balance between nucleation and growth. Additional capping agents are necessary when using ODE. Dodecanethiol (DDT), however, is a strong coordinating ligand. It assists in passivating the nanocrystals by coordinating with the metal cations on the surface. OLA is necessary to control the morphology of the nanocrystals. DDT, unlike the other solvents, can act both as a ligand and a sulfur source. Thiourea has also been used in lieu of DDT41.
Hot-injection synthesis will be performed with varying combinations of OLA, ODE, DDT and thiourea in attempt to synthesize BZS.
2.3.4 Intercalation
Bulk layered materials are materials of a unique structure which provides interesting chemical and physical properties that have been extensively studied in the last several decades. One example of this class of materials are transition metal dichalcogenides (TMD, MX2). The layers typically display strong covalent
bonding in-plane and weak Van-der-Waals interactions out-of-plane through the interlayer gap. The large interlayer gap that TMD exhibit allows easy intercalation45.
Intercalation is the chemical process by which an atom, ion or molecule may insert into the crystal gap of a bulk layered material (Figure 11). Several applications have been found for TMDs such as field effect transistors, optoelectronic devices, piezoelectric devices, superconductors and electrochemical electrodes. It is most famously used in Li-ion batteries wherein Li ions are intercalated through the van der Waals gaps upon charging and discharging. Intercalation has been achieved successfully through electrochemical methods, chemical vapor transport, ion exchange and oxidation-reduction reactions45.
16
Figure 11 Intercalation schematic for TMD layered materials. Image from ref. [45]
Energy conversion applications such as photovoltaics or thermoelectrics are also possible with intercalation of TMDs. For example, SnS has been intercalated into TiS2 to create a natural superlattice to form a
high-performance thermoelectric material46. Furthermore, Lerf et al demonstrated that ternary chalcogenides can
easily undergo solvation reactions with inorganic or organic solvents of decent polarity according to the following equation:
AxMnXm + y (solvent) ⇌ A+x (solvent)[MnXm]x-
Wherein A is a group 1A or 2A metal, M is a transition metal and X is a chalcogen. The solvated cations have high mobility in the interlayer space; therefore, they can act as electrolyte solutions to perform ion-exchange reactions47.
Though Lerf et al had previously synthesized the ternary chalcogenides in their report, the reaction is a reversible one. Therefore, in this report, intercalation will be attempted as a synthesis method. This will be done by intercalating various Ba sources (and additional S sources if necessary) into ZrS2 in different
solvents.
3. Materials and Methods
3.1 General Procedures
Chemicals were purchased from chemical suppliers and used as received. Sample characterization for all methods were performed using a scanning electron microscope (FEI Verios 460) with an X-ray detector (Oxford Instruments EDX detector), an X-ray diffraction (Bruker D2 Phaser, Cu Kα radiation, wavelength = 1.5418 Å), an UV/VIS/NIR Spectrophotometer (Perkin Elmer, L750), a confocal imaging microscope (WITec alpha300 SR equipped with a 532 nm laser), and a and a dark field optical microscope (Zeiss axio imager.A2m).
3.2 CVD
Sulfur (S, 99.5%) and Barium Chloride (BaCl2, 99.999%) were purchased from Sigma Aldrich. Zirconium (IV)
sulfide (ZrS2, 99%) was purchased from STREM Chemicals Inc. A single-zone furnace and a horizontal quartz
tube comprised the CVD system.
0.2 – 0.4 mmol of BaCl2 and 0.2 mmol of ZrS2 were placed on opposite ends of a 10 cm quartz boat. 15
mmol of S was placed in another quartz boat. The boat containing the metal precursors was moved into a quartz CVD tube and ZrS2 was moved to the center of the furnace with BaCl2 behind it. S was placed right
17
outside the furnace wherein the temperature was ~200 °C. 9 Si wafer substrates that were sputter coated with 2 nm of Au using a Leica EM ACE600 (current: 60 mA, sputter rate: ~0.35 nm/s) were lined from the center until the end of the furnace. Prior to sputter coating, the substrates were cleaned by ultrasonication in acetone then isopropanol and dried with N2 gas. The CVD tube was flushed with 25 sccm Ar and 25 sccm
H2 prior to reaction for 10 minutes. The furnace was set to heat at 950 °C at a rate of 20 °C/minute under
the same gas flow. The temperature remained at 950 °C for 20 minutes. Afterwards, the CVD was cooled to room temperature while maintaining the gas flow.
3.3 Molecular Sieve Encapsulation
Zirconium tetrachloride (ZrCl4, >99.5%) was purchased from Alfa Aesar. MCM-41 was purchased from ACS
Material, LLC. Anhydrous dimethyl sulfoxide (DMSO, (CH3)2SO, >99.9%) was purchased from Sigma Aldrich.
1.75 mmol of ZrCl4 and 1.75 mmol of BaCl2 were weighed in the glovebox and dissolved in 25 mL of DMSO.
The mixture was taken out of the glovebox and stirred for 30 minutes at 80 °C. MCM-41 was added to the mixture at a 2:1 MCM-41: BaCl2 + ZrCl4 weight ratio and stirring continued for 1 hour. Excess DMSO was
removed through decantation. The remaining powder was dried in the oven for 3 days at 80 °C. Once completely dried, 0.3 g of the MS/salt mixture was sulfurized with 30 sccm H2S at 700, 800, 900 and 1000
°C for 4 hours (named BZS-MS-700, BZS-MS-800, BZS-MS-900, BZS-MS-1000, respectively). After sulfurization, the resulting product was purified through washing and centrifuging three times with DMSO and dried in the oven overnight.
3.4 Hot-Injection Synthesis
Thiourea (SC(NH₂)₂, >99%), Oleylamine (OLA, C18H35NH2, 98%), and 1-octadecene (ODE, C18H36, 90%) were
purchased from Sigma Aldrich and OLA and ODE were degassed with the freeze-pump-thaw method and stored in the glovebox. Barium Acetate (BaAc2, 99%) was purchased from Alfa Aesar.
5 mmol of S/thiourea was dissolved in 2 mL of OLA and heated for 30 minutes at 80 °C on a Schlenk line under N2 and cooled to room temperature. 1 mmol of ZrCl4 and 1 mmol of BaCl2/BaAc2 were added to 6
mL of OLA or a 1:1 mixture of OLA:ODE in the glovebox and transferred to a Schlenk line. The metal precursor mixture was heated to 300 °C. The cold S+ OLA mixture was injected into the flask. The reaction was kept at 300 °C for 1 hour. The resulting mixture was purified through solvent exchange with hexane, ethyl acetate, and ethanol and centrifuged.
3.5 Intercalation
Anhydrous acetonitrile (ACN, CH3CN, >99.9%) was purchased from Sigma Aldrich and anhydrous
tetrahydrofuran (THF, (CH2)4O, >99.9%) was purchased from VWR International BV.
Two batches of 0.06 mmol of ZrS2 and 0.06 mmol of BaS were weighed and placed in a vial with 4 mL of
either ACN, THF, or DMSO. One vial from each solvent was placed in an ultrasonicator and another was heated at 80 °C for one week.
18
4. Results and Discussion
4.1 CVD
CVD was tried with changing combinations of different precursors for each metal, for the barium-source: BaS, Ba(OAc)2, and BaCl2 and for the zirconium source: ZrS2 and ZrCl4. Excess S was added to create a
sulfur-rich environment to facilitate the formation of the perovskite. Furthermore, S vaporizes at lower temperatures (444.6 °C) than any of the precursors so an excess was required to ensure there was enough S to last the duration of the reaction.
In all the reactions, Si wafers were used as a substrate for deposition. Initially, the substrates were used as received and only cleaned, however, the precursors were not depositing on the substrate. BaS and ZrS2
both completely vaporize at temperatures above 1000 °C, therefore, the substrates were sputter-coated with 2 nm of Au. At elevated temperatures, the Au form nanoislands (Figure 12) that can act as a preferential collection site for the limited amount of gaseous Zr and Ba species. The CVD reaction was performed with the substrates annealed at 300 °C for one hour prior to the reaction in inert atmosphere and with without annealing. The results of the reaction were similar so it was decided not to anneal in order to reduce the steps necessary.
Reactions were initially done with ZrCl4 and either BaS or BaCl2 in the center of the furnace (the “hot zone”)
and S positioned at approximately 200 °C. Material that deposited on the substrates, independent of the Ba-source used, were white in color and suspected to be ZrCl4. This could not be confirmed as XRD spectra
showed no identifying peaks. However, Wang et al. experienced a similar issue in the CVD synthesis of ZrS2
with ZrCl439. They resolved this by flowing a burst of Ar (300 sccm) at 500 °C, after the melting point of ZrCl4
(437 °C). Replicating this deposited no material. This could be because the ZrCl4 vaporizes at much lower
temperatures that BaS and BaCl2 and is blown away before the vapor species are able to react. Mixing the
precursors on the boat instead of keeping them separate offered the same result. Switching BaS/BaCl2 with
Ba(OAc)2 which has a melting point closer to that of ZrCl4 (450 °C) also did not afford any deposition.
The reaction was performed in the same conditions, however, ZrCl4 was substituted for ZrS2. The resulting
nanowires from a CVD reaction using BaS and ZrS2 with additional S at 950 °C are displayed in Figure 13b
and c. The XRD of the resulting nanowires, the substrate with Au nanoislands and the desired BZS product can be seen in Figure 13a. The spectra of BZS in the figure shows BZS mixed with another Zr phase, however, this does not affect the positions of the peak, only the width and intensity of the peaks, this will be elaborated on in the next section. Therefore, the BZS XRD spectra will only be used to compare peak position in this analysis.
The XRD peaks at 70 ° on the nanowire and the Au-coated substrate result from the Si substrate. Au has a distinct peak at 38 °. Additionally, the substrate XRD displays an amorphous hump at 20 – 30 °. The XRD of the nanowires has two distinct peaks that do not belong to Au at 27 ° and at 46 °. Neither of these peaks match the XRD spectrum of BZS, further EDX analysis shows that no Ba is present in the composition of the wires, therefore it cannot be the perovskite or any Ba sulfide or oxide. BaS has a much higher melting point Figure 12 SEM image (5 kV, 100 pA) Au nanoislands after
19
.Figure 13 a) XRD spectrum of the resulting nanowires on the Si Substrate (black line), the substrate coated in Au (red line), BZS (blue) and ZrS2 (green) b) SEM image of the product scale bar represents 10 µm (29 kV, 1.6 nA) c) SEM
image of a single wire scale bar represents 5 µm (29 kV, 1.6 nA) d) XRD of ZrS3 taken from ref [48] the experimental
spectra (red) was taken with Cu x-rays and the reference (black) was taken with Mo x-rays b) a) c) d)
10
20
30
40
50
60
70
Si2Theta (degree)
Nanowires on substrate Si (111)Int
ensity (
a.u.
)
Substrate coated with Au
(333) (242) (123) (202) (022) (121) BZS (004) (201) (103) (111) (110) (102) (003) (002) (101) (101) (100) (001) ZrS2
20
than that of ZrS2 so it is possible that not enough material has vaporized and the reaction could not be
finalized. The two peaks present on the XRD, however, are difficult to identify. The peaks do not match the XRD of ZrS2, with the exception of the peak at 26 °. That peak on the XRD of ZrS2 is a rather weak peak so
it cannot be said with certainty that what was made was ZrS2.
Because of the presence of excess S, it is possible that ZrS3 was formed. The experimental ZrS3 measured
with Cu x-rays is displayed in Figure 13b (red line). The reference XRD of ZrS3 was measured with Mo x-rays
and so will not be used as a comparison (black line). ZrS3 does also have a peak at 26 ° and at 46 °. Although
those two peaks are not the strongest peaks of the XRD. The strongest peaks of ZrS3 is positioned at 3448.
This material does have a peak at that position; however, it is a very weak one. Therefore, more studies are necessary to identify the material more thoroughly.48. This material does have a peak at that position;
however, it is a very weak one. Therefore, more studies are necessary to identify the material more thoroughly.
The same conditions were applied using BaCl2 to combat the issue of the lack of vaporization from BaS.
Initial reactions provided no deposition on the substrate. However, when a new, clean CVD tube was used, nanowire growth was achieved, highlighting the importance of utilizing an individual CVD tube for each set of precursors. Figure 14a shows the SEM image of the deposition on the substrate. As can be seen, there are different phases present: there are nanowires, but also discs. Figure 14b shows an isolated nanowire. The XRD spectrum (Figure 14c) looks similar to that of BZS, however, upon closer inspection, all the peaks are shifted around 3-5 ° higher than the XRD of BZS. The peak at 70 ° represents the Si substrate. The XRD also does not match that of ZrS2 or ZrS3. ZrO2 has a similar XRD to BZS and would have peaks matching BZS
at 30 ° and 50 °, therefore it is also does not match ZrO249.
The UV-Vis absorption (Figure 14d) also does not conform with that of BZS, BZS would have a strong absorption onset at ~680 nm. The material synthesized has a weak absorption and with an absorption onset from 800 nm. The PL spectrum of this material, however, is exactly what is expected from BZS. ZrS2 has a
bandgap of 1.4 eV, therefore, the PL would be expected to be at higher wavelengths that are not in the range of the spectra. It may be that there was not enough material on the quartz substrate to measure a decent absorption spectrum as the material had to be scratched from a thin film on a Si substrate.
21
Figure 14 a) SEM image of wires and discs on Si substrate (29 kV, 1.6 nA) scale bar represents 25 µm b) SEM image of isolated wire on Si substrate (29 kV, 1.6 nA) scale bar represents 10 µm c) XRD spectrum of CVD reaction with BaCl2
and ZrS2 (black) and BZS (blue) d) XRD of tetragonal ZrO2 (t-ZrO2) and monoclinic ZrO2 taken from ref. [50] e) UV-Vis
Absorbance f) PL intensity (532 nm excitation) of CVD reaction with BaCl2 and ZrS2
a) b)
e)
c) d)
22
Upon conducting EDX analysis on the discs and nanowires separately (Table 1), it was realized that the discs were composed of 32.59% Zr and 63.40% S, therefore, it is likely that the discs are ZrS2 that had deposited
on an Au nanoisland. The nanowires, however, had a different composition. No Au was present as an isolated wire was measured from the center of the wire rather than the tip, where the Au would be. It shows an atomic percentage similar to that of the disc phase, only including 1.63% Ba. The Zr and S atomic% appears to also have formed ZrS2. This is unexpected as the XRD more
closely matches that of BZS than ZrS2. Additionally, the PL spectra is
suitable to show BZS formation. It may be that the peaks on the XRD have shifted due to strain in the material and that the nanowires were, indeed the perovskites, however, this cannot be said with certainty. Though it is unclear what materials were made from the CVD experiments, CVD appears to be a promising technique for nanowire or nanocrystal growth. Further research should try to find a more suitable Ba precursor so that it vaporizes at the desired temperature and reacts with ZrS2. Additionally, better characterization should be
done to properly identify the materials made in CVD synthesis to elucidate the reaction precursors and conditions needed to incorporate more Ba and synthesize the material.
4.2 Molecular Sieve Encapsulation
As described previously, MS was soaked in a solution of precursor salts and then dried. The resulting powder was heated at high temperatures to collapse the pores of the MS and simultaneously sulfurize the salts. At this temperature, BZS nanocrystals were synthesized inside the pores of the MS which gradually collapsed and encapsulated the perovskite nanocrystals in SiO2 forming a dense BZS-MS solid.
Figure 15b shows the colors of the BZS-MS when sulfurized at different temperatures. The color gradually changed from white at 700 °C to dark orange at 1000 °C. The XRD patterns (Figure 15a) confirm the successful formation of BZS from 900 °C and higher. The broad shoulder observed at around 23 ° arises from the amorphous phase of SiO2. Using the Scherrer Equation (Appendix 2), the average crystallite size
for BZS-MS-900 was calculated to be 3.45 nm and 7.94 nm for BZS-MS-1000. The broad XRD peaks are the result of small crystallite sizes, as the crystals were confined by the pore size oft he MS.The increase in crystallite size with increasing temperature can be attributed to the softening and collapsing of the MS pores at higher reaction conditions. This allowed for growth of larger nanocrystals due to a weaker template confinement effect40.
The XRD shows crystal orientations in accordance with the distorted perovskite structure. BZS is isostructural with GdFeO3, therefore, it is orthorhombic with the space group Pnma. However, the intensity of the peak
at 30 ° and the splitting of the peak at 50 ° indicate that another structure is present in the mixture along with BZS. These peaks can more clearly be seen in the XRD of BZS-MS-1000 (Figure 15c). Both these peaks are in accordance with that of ZrCl4, therefore, not all the precursor salts have reacted. The peak at 25 ° can
already be seen in the XRD spectra of BZS-MS-800, from this, it can be said that the perovskite is present in the reaction mixture.
Discs Wire Element Atomic % S 63.40 64.86 Zr 32.59 33.51 Ba 0.00 1.63 Au 4.00 0.00 Total: 100.00 100.00 Table 1 Atomic % of the circular and nanowire phases of the product of BaCl2 and ZrS2 CVD reaction as determined through EDX analysis
23
This was confirmed with EDX mapping (Table 2) which showed that the atomic ratio of Zr was higher than that of Ba and S. It also shows the presence of 1.20 at% Cl. It is possible that the pores collapsed before all the Zr had been reduced by H2S. To determine if that was indeed happening, the reaction was redone with
a ramp rate of 5 °C/min instead of 20 at 900 °C (BZS-MS-900-slow). Additionally, 60 sccm of Ar + H2 (1:1)
was included. The EDX map of the BZS-MS-900-slow showed that almost all the Cl was removed indicating that both Zr and Ba were reduced by H2S.
Furthermore, the amount of S present in the mixture had increased. The amount of Zr, however, was still too high to have BZS alone in the product.
Figure 16 compares the XRD spectra of the slow and fast ramping rate. The XRD of both have peaks present in the same position, although the peaks of the product from slow ramping appear to be more defined. This could mean that the product is more crystalline. The crystallite size of BZS-MS-slow was calculated to be 3.77 nm which is larger than the 3.45 nm of BZS-MS-900. BZS-MS-900 BZS-MS-900-slow Element Atomic % S 32.74 50.03 Cl 1.20 0.27 Zr 55.70 40.05 Ba 10.36 9.65 Total: 100.00 100.00
Figure 15 a) XRD spectra and b) photograph of resulting BZS-MS synthesized at 700 °C, 800 °C, 900 °C, and 1000 °C c) XRD spectra of BZS-MS-1000 and its corresponding crystal orientations in black. Crystal orientations displayed in red are that of ZrCl4
20
30
40
50
60
70
80
2Theta (degree)
BZS-MS-700 BZS-MS-800Int
ensity (
a.u.
)
BZS-MS-900 BZS-MS-1000 a) b) c)Table 2 Composition of BZS-MS-900 and BZS-MS-900-slow as determined by EDX 20 30 40 50 60 70 80 (200) (125) (333) (242) (123) (242) (123) (202)
Int
ensity (
a.u.
)
2Theta (degree)
(121) (022) (121)24
The peak splitting at 50 ° and the intense peak at 30 ° still remain in BZS-MS-900-slow despite the reduction in Cl. This could be indication of another Zr crystal structure. ZrCl4 may have been reduced to Zr metal in
inert atmosphere and the encapsulation has protected the metal from oxidation, or the Zr was oxidized into ZrO2. These peaks may be attributed to either Zr metal
or ZrO2. The XRD of the hexagonal phase of Zr is
displayed in Figure 16. It would exhibit a peak at 44 ° from its 101 orientation and a peak at 57 ° from the 102 orientation51. Though the peaks overlap with BZS at
those positions, it does not explain the sharp peak at 30 ° and the peak splitting at 50 °. Alternatively, ZrO2
exhibits a sharp peak at 30 ° from its 111 orientation and a peak at 50 ° from its 220 orientation from the tetragonal phase of ZrO2 (Figure 14d). ZrCl4 reacts easily
with water to form Zr oxides, therefore, traces of water in the DMSO or in the flow gas may have formed the oxide. This was difficult to confirm through EDX mapping as the presence of oxygen in the substrate used in the measurement and in the SiO2 from the MS
would interfere with the measurement of oxygen in the pores.
Absorption measurements were performed on the powders before and after washing three times with DMSO then dried. This was done to determine if the nanocrystals were really incorporated into the MS pores and to wash off excess external salts. Figure 17 shows that the UV-Vis absorption of the BZS-MS materials of all temperatures actually increased after washing and drying. This is a suggestion that the pores had encapsulated the salts and protected the nanocrystals and the external salts had been removed successfully. Furthermore, it is evidence that the material is resistant to DMSO.
The absorption of the BZS-MS increases with temperature which can be expected as visually it can already be seen that the product of heating at 1000 °C is significantly darker in color than that of 700 °C. The band edge absorption does not change following to washing, however, it does change with changing temperature. It increases from 550 nm at BZS-MS-700 to >650 nm at BZS-MS-1000. Tauc plots calculated (Appendix 1) a bandgap of 2.31 eV for BZS-MS-900 and 1.89 eV for BZS-MS-1000, both of which are higher than previously calculated bandgaps of BZS (~1.8 eV). It is expected that the reaction at higher temperature has a higher bandgap. The larger crystallite sizes of BZS-MS-1000 red-shift the bandgap closer to the bulk.
20 30 40 50 60 70 80
Intensity (a.u.)
2Theta (degree)
BZS-MS-900 BZS-MS-900-slow
Figure 16 XRD spectrum of a) BZS-MS-900-slow (red) and BZS-MS-900 (black) b) XRD of hexagonal Zr metal taken from ref. [51]
a)
25
Figure 17 UV-Vis absorption of a) BZS-MS-700 b) BZS-MS-800 c) BZS-MS-900 d) BZS-MS-1000 e) BZS-MS-900-slow. Lines in black are absorption before washing and lines in red are after washing and drying with DMSO. f) PL intensity of BZS-MS-900, BZS-MS-1000, BZS-MS-900-slow d) 600 650 700 750 600 800 1000 1200 1400 1600 1800 Int ensity ( a.u. ) Wavelength (nm) BZS-MS-900 BZS-MS-1000 BZS-MS-900-slow 300 400 500 600 700 800 0.0 0.1 0.2 0.3 0.4 0.5 0.6 Abs (a. u.) Wavelength (nm)
MS-900-Slow after wash. MS-900-Slow before wash.
a) b)
c)
26
The absorption was also taken for MS-900-slow (Figure 17e). The absorption is less than that of BZS-MS-900 and 1000. The bandgap for this material was calculated to be 2.06 eV. This is lower than the 2.31 eV calculated for its faster heated counterpart. This could again be potentially due to the difference in crystallite size observed in the XRD.
ZrO2 is white in color and an extremely good insulator, hence has a wide bandgap of ~5 eV. However, a
large fraction of oxygen vacancies can produce a reduced ZrO2 (ZrO2-x). This material is black and showed
absorption much stronger than its white counterpart. The bandgap of black zirconia is 1.5 eV, it was synthesized through the magnesiothermic reduction of ZrO2 in a tube furnace. Therefore, it cannot be ruled
out that H2S has reduced ZrO2 to form black ZrO249.
PL measurements are displayed in Figure 17f. The material with the strongest PL is BZS-MS-900 slow, followed by BZS-MS-1000. BZS-MS-900 has almost no PL. That BZS-MS-1000 has weaker luminescence than BZS-MS-900-slow is unexpected as the XRD of the resulting powder synthesized at 1000 °C appears to be more crystalline and more intense indicating better quality nanocrystals. These variations, however, may be due to spatial inhomogeneity in material quality, loading or light in/outcoupling. Therefore, nothing conclusive can be said about difference in PL of the materials. Unfortunately, wavelengths above 740 nm were out of the scope of the WITec, therefore, the symmetry of the peaks cannot be commented on. Additionally, the interaction between the MS and the BZS plays an important role in the optical properties of the materials, which could be the reason for BZS-MS-900-slow to have a stronger PL response. Hence, further research in needed to make a clear conclusion. This could entail measuring PL in different positions of the sample and exciting at higher laser power.
Overall, MS encapsulation is a successful method to synthesize BZS nanocrystals due to the confined space of the pores. However, more work needs to be done to remove the Zr impurities. This could be done by preparing the salt-filled sieves in the glovebox entirely and storing them there until sulfurized. This may reduce the oxidation of the ZrCl4 into ZrO2 prior to the sulfurization step, if this is indeed the cause for the
Zr impurities. Additionally, it appears that BaCl2 are not filling the pores as readily as ZrCl4 based on the
higher Zr-ratio compared to Ba. The Zr4+ cation of is smaller than the Ba2+ cation which enhances the
Coloumb interaction, therefore, it is also possible that the difference in charge of Ba and Zr is affecting the interaction and subsequently the loading of the salts into the sieves. Also, the bond energy of Zr-O is higher than that of Ba-O (776.1 kJ mol-1 and561.9 kJ mol-1 , respectively), this proves that Zr has a stronger chemical
interaction with the oxygen in the sieves than Ba. To resolve this, loading BaCl2 before ZrCl4 would be an
option, or increasing the ratio BaCl2 instead of 1:1 ratio of both salts. Further research should also be done
to determine the ideal mass ratio of ZrCl4 + BaCl2:MS as the mass ratios have profound effects on the optical
properties of the BZS-MS materials due to the availabilities of pores that may encapsulate the nanocrystals.
4.3 Hot-Injection Synthesis
Solution synthesis was considered as a route towards the formation of BZS. The synthesis was performed with Ba(OAc)2 or BaCl2 and ZrCl4 in OLA as one precursor solution and S or thiourea in OLA as another.
Variations of the synthesis included changing the order of injection, the temperature, the sulfur source, the solvent and the time of reaction.
Reactions wherein cold thiourea in OLA were injected into a hot solution of OLA and a combination of OLA and ODE (1:1) containing the Ba and Zr salts were unsuccessful. XRD spectra of the reactions did not show any crystalline peaks indicating that the perovskite had formed.
27
Using S instead was more successful, although, not with OLA and ODE combination. That provided the same outcome as using thiourea. However, injection of cold S in OLA into a hot solution of the precursor Ba and Zr salts had a different outcome. The resulting XRD can be seen in Figure 18. The figure shows that many different possible phases are present in the mixture. It is difficult to say with certainty if all those crystal structures are existing as S, BZS and BaS are difficult to dissolve and, therefore, difficult to separate.
Performing this reaction again and purifying with different solvents produced different results as displayed in Figure 19. The same set of spectra were obtained when 160 °C solution of metal precursors in OLA was injected into a 300 °C solution of S in OLA. Purifying the reaction product by washing three times with ethyl acetate produced sharp, crystalline peaks in the XRD spectrum. This was identified to be ammonium chloride. The same spectrum resulted when purified with hexane. The formation of ammonium chloride is an indication that there is some decomposition of the OLA during the reaction. This is unexpected as OLA is a high boiling solvent (boiling point: 364 °C), however, there is no other potential source of nitrogen in the reaction mixture, and this result appeared on multiple occasions. It was originally
thought to have been impurities in the solvent, however, the solvent was purified through the freeze-pump-thaw method and produced the same outcome. Therefore, it is possible that the nanoparticles synthesized in the hot-injection reaction have the potential to stimulate decomposition in OLA52.
Using ethanol afforded a different reaction product. When washing the reaction mixture with ethanol, the mixture became black suggesting that a reaction had taken place. This could be a result of the destabilization and subsequent decomposition of Ba/Zr metal-oleate complexes or the destabilization of the decomposition products of OLA. In spite of this, the XRD displayed broad peaks (Figure 19). Similarly to the MS encapsulation experiments, the peak at 30 ° is much more intense than the one for BZS, BZS would instead have an intense peak at 25 ° and another at approximately 41 ° (Figure 13b). The XRD is more closely fitting with the tetragonal phase of ZrO2 than
BZS (Figure 14d). Hence, what was repeatedly synthesized is potentially ZrO2.
Figure 18 XRD spectrum of S injection in metal precursor solution
Figure 19 XRD spectra of the most common reaction outcome in hot-injection synthesis when washed in ethanol, ethyl acetate and without purification
20 30 40 50 60 70
Int
ensity (
a.u.
)
2Theta (degree)
ZrO2 BaS BZS S 20 30 40 50 60 70 802Theta (degree)
No Purification (300) (210)Int
ensity (
a.u.
)
NH4Cl (ethyl acetate) (100) (110) (111) (211)(220) (131) (022) t-ZrO2 (ethanol) (111)28
Solution synthesis is difficult to react Zr and S. Previous reports of ZrS2 solution synthesis use CS2 as the
sulfur source. Though ZrS2 could be synthesized with S, the resulting product were unstable and of poor
crystallinity, Jeong et al attributed this to S forming a highly reactive radical at these high working temperatures, and, therefore, a stable, crystalline product could not be formed. They observed this with other group IV metal sulfides53. Additionally, Zr has little affinity towards S, this is because S is significantly
larger than oxygen, which fits much better in the interstitial site of Zr. Hence, Zr is very reactive towards oxygen. Zr-S compounds have only been known to form at temperatures higher than 500 °C54. Therefore,
it is likely that ZrO2 had formed.
4.4 Intercalation
Another type of solution synthesis was performed. Intercalation experiments were performed with ZrS2 and
BaS. The vials containing the precursors in ACN, THF and DMSO were heated at 80 °C or placed in an ultrasonicator for a week. Figure 20 displays the XRD spectrum of the product of the reaction heated at 80 °C in ACN kept for a week, however, the spectra of all reactions looked similar to this (Appendix 3). It shows that even after a week, the precursors are still clearly visible in the spectrum. Images from the optical microscope (Figure 20) shows that there was no interaction between the two precursors. The visible white particles are that of BaS and the more orange are ZrS2. Peaks at 20 ° and 22 ° are not present in BaS and
ZrS2, however, they are also not present in BZS.
The solubility of BaS is really low in all the solvents, which limits its intercalation into ZrS2. For this reason,
the reaction was kept for a week as it was assumed that partial solubility and long reaction time would facilitate intercalation. However, this was not the case. Therefore, the reaction was tried again with different precursors, namely, NaS and BaCl2 in DMSO for a week in the same conditions. DMSO was chosen as both
NaS and BaCl2 are soluble in it. The XRD of the resulting product was again a reflection of the precursors. It
is possible that more energy is necessary to form the perovskite as heating was only done at 80 °C or only with ultrasonication.
