1
Study of the purification process and the surface-ligand interactions in
cesium lead halide perovskite nanocrystals via vibrational spectroscopy
Master thesis Physics and Astronomy:
Science for Energy and Sustainability
Tsoulfas Christos
Vrije Universiteit Amsterdam (VU) & University of Amsterdam (UVA)
Supervisor: Dr. Elizabeth von Hauff
Second reader: Dr. Simon Boehme
2
Abstract
Lead halide perovskite nanocrystals can be used for numerous applications such as light-emitting diodes (LEDs) for display applications or such as chemical detectors. Their manufacturing as well as their purification process plays a crucial role for the stability and the performance of the perovskite nanocrystals. In this study, firstly, the focus is placed on the washing procedure of the nanocrystals. The goals are to investigate the efficiency of the washing procedure, to clarify the role of the bonding between the Oleic Acid and Oleylamine to the perovskite’s surface and lastly to examine the binding motifs to perovskite nanocrystals with different halides. The results of the Raman measurements indicate that the first step of the washing procedure is successfully removing
(
70
4
)
% of the Octadecene. However, during the whole washing procedure the majority of the Oleylamine and the Oleic Acid are also washed out indicating a serious probability of destabilizing the nanocrystals. Moreover, the results which concern the role between the Oleic Acid and Oleylamine of the nanocrystals are indicating that as the perovskite lattice spacing is decreasing (from CsPbI3 to CsPbCl2Br) the major binding is between theperovskite cell and the carboxylate group of the Oleate. Finally, the results on the binding motif indicate that each halide is binding with the Oleate with the bridging bidentate coordination.
3
Table of Contents
1. Introduction
2. Lead-halide perovskite nanocrystals 2.1. Crystal structure and unit cell 2.2. Nanocrystal surface and ligands 2.3. Synthesis and byproducts 2.4. Purification procedure
3. Vibrational spectroscopy: Raman and FTIR 4. Research in this thesis
4.1. Research question
4.2. Methodology and sample preparation 5. Results and Discussion
5.1. Spectral markers 5.1.1 Octadecene
5.1.2 Oleylamine and oleic acid
5.2. Purification process of CsPbCl2Br nanocrystals (Raman)
5.3. Purification process of CsPbBr3 nanocrystals (FTIR)
5.4. Halide-dependent purification process and ligand binding motif of CsPbX3 nanocrystals
(FTIR)
6. Conclusions and recommendations 7. Acknowledgements
4
1 Introduction
The latest years colloidally synthesized nanocrystals (NCs) are in the centre of the academic attention as future optoelectronic materials. The reason is that they have a number of advantages. First of all, they have high photoluminescence quantum yield (PLQY) which can reach 90%[1] and tunable bandgaps which renders them the ideal
candidate for applications such as light-emitting diodes (LEDs), lasers, photo-detectors and photovoltaics(PVs)[1],[2]. Moreover, colloidally synthesized nanocrystals have the
advantages of low cost synthesis as well as easy film deposition techniques. This is the case because the NCs can be processed in solutions which allow their manipulation with simple techniques such as drop-casting, spay-coating and spin-coating [3]. The goals for the
majority of the studies are to clarify the nanocrystal’s surface-ligand bonding as well as the effect of the nanocrystals size. The aim is to maximize the lifetime and the efficiency of the nanocrystals as well as to find alternative materials which can be used either as solutions for processing for the nanocrystal core or as ligands. Until now, the results of the studies are very promising as efficient photovoltaic devices with certified power conversion efficiencies are approaching 20%[1]. Still a number of drawbacks render the perovskite
nanocrystals a material which is challenging to become commercially available especially for PVs applications. The most important reasons are the intrinsic instability of the perovskite nanocrystal to polar solvents such as water and oxygen which limits their lifetime. Encapsulation of the perovskite nanocrystal film can partially counter this issue. However, with encapsulation the price of the device is increasing leading to prohibiting high prices and hence cannot be a commercial success.
5
2 Lead-halide perovskite nanocrystals
2.1 Crystal structure and unit cell
In general, perovskite materials are described by the formula ABX3 where
A and B are cations and X is an anion. In this project, A is Cesium (Cs), B is Lead (Pb) and X is the halide which can be an atom of Iodide (I), Bromide (Br) or Chloride (Cl). The crystal structure of perovskites at room temperature is depicted in Figure 2.1.1. However, atoms bond in their charged forms as shown in the same figure. Lead Halide Perovskites (LHPs) nanocrystals have both intrinsic advantages and disadvantages. First of all, in order to create a stable perovskite nanocrystal you need to passivate the “free” surface bonds. Figure 2.1.2 illustrates the unit cell of the perovskite nanocrystal as well as the passivation of the “free” bonds. All the atoms, both cations and anions are depicted in their charged forms. The molecules which are used in order to passivate the surface bonds are Oleic Acid (OA) and Oleylamine (OLA), the next section includes more information about them. LHPs materials with formula ABX3 have significant
optoelectronic properties and can be applied for numerous applications as mentioned above. However, they suffer for a number of problems. The most important of them are the following: They have intrinsic instability against polar solvents such as moisture and oxygen which limit their lifetime[3],[6].
Figure 2.1.1: Schematic of the cubic and orthorhombic perovskite crystal structure. Image from Akkerman et al.[4]
Figure 2.1.2: Schematic of the perovskite unit cell and binding of Oleate (R-COO-) and Oleylammonium (R-NH3+)
ligands to Pb2+ and X- surface sites, respectively, with
6
Secondly, in our case, as Pb is a major constituent, Pb toxicity issues during device fabrication, deployment and disposal make the perovskite nanocrystals a material which needs specific processing methods[7]. Furthermore, the ligand system that stabilizes the
surface is highly dynamic, resulting in exposure of the nanoparticle surface, possibly resulting in disintegrations during the purification or during processing[8],[9]. Figure 2.1.3
illustrates the ligand dissociation and the nanocrystals’ dissolution which may happen during either at the production phase or at the purification process of the NCs[4].
Figure 2.1.3: Schematic of the ligand dissociation and NC dissolution. Image from Akkerman et al.[4]
Finally, the nanoparticles are thermally instable due to their low melting point[4].This means
that during the manufacturing process, superstructures can be created which will affect the whole performance of the perovskites[10],[11]. The size unit cell of the perovskite nanocrystal
which is dependent on the halide plays a crucial role as can affect the bonding of the nanocrystals surface with the ligands and change the dynamics of the system.
2.2 Nanocrystal surface and ligands
As mentioned in section 2.1, in order to create a colloidal stable nanocrystal the “free” surface bonds need to be passivated. This is a major issue as nanoparticles have a larger surface area compared to bulk materials and it is logical to have more surface defects. Capping agents or ligands are used in order to do passivate the surface states and hence make the nanoparticle more stable. Also, they are used in order to control the growth size of the nanocrystals during the production phase. This is a reliable method in order to keep them colloidally dispersed in a solvent as well as to achieve long-term stability. In this way the properties of the nanoparticles can be optimized[8]. In this study, OLA and OA were
7
The ligands have similar structure as they are both organic molecules with a long non-polar carbon tail. They are consisted of 18 carbon atoms and one double bond at the ninth carbon atom. The difference between them is on the factional anchor group. For OA it is a carboxylic acid group and for OLA it is an amine group (NH2). However, ions
can be produced which affect the binding motifs between the ligands and the perovskite nanocrystals. From OA, the ion which is called Oleate can be produced when the anchor carboxylic
acid group becomes deprotonated. As for OLA, the ion which is called Oleylammonium can be produced when the anchor amine group becomes protonated. Oleate binds with Pb atoms and Oleylammonium binds with Cs atoms in order to create the passivated nanocrystal. The next section includes information about the purification procedure and hence about the samples as well as about their properties.
2.3 Purification procedure and synthesis
Figure 2.3.1 illustrates the purification process (washing procedure) of the nanocrystals. The complete washing procedure is as follows: First of all, the synthesis mixture was cooled to room temperature. Then, 3mL of anhydrous toluene was added to the mixture and then the solution was centrifuged for 10 minutes at 3000rpm. Then, under inert atmosphere, the supernatant (named hereafter ‘supernatant 1’, and depicted in blue to the left of Figure 2.3.1) was removed and the solids (‘which is named: raw synthesis product’, depicted in light green to the left of Figure 2.3.1) were resuspended in 2mL of anhydrous hexane. The produced sample was thereafter stored in a nitrogen filled glove box. For the washing procedure, 1mL of methyl acetate was added per mL of nanocrystal solution into the sample. The suspension was thereafter centrifuged at 4500rpm for 5 minutes. The supernatant was decanted yielding the ‘supernatant washed 1’, and the solids were resuspended in the same volume of the nanocrystal solution yielding the ‘precipitate washed 1’. This washing step was repeated twice yielding ‘supernatant washed 2’, ‘precipitate washed 2’, ‘supernatant washed 3’, and ‘precipitate washed 3’ which is the final product of the whole purification process.
Figure 2.2: Schematic illustration of the chemical structure of OA and OLA.
8
Figure 2.3.1:Schematic of the purification process (washing procedure) of the nanocrystals.
The purification process may create more pure nanocrystals by removing all the unnecessary materials but also creates some negative issues regarding the PLQY. Figure 2.3.2 illustrates the decrease on the PLQY for the first and the second washing step for different halide series of perovskite nanocrystals. It needs to be noted that these series are not the ones which were examined for this thesis. The goal is to create colloidal stable nanocrystals with high PLQY so as to be able to become a commercial success for a number of applications as mentioned in the introduction. Of course, depending on the application the focus between achieving high PLQY or maximum purity can be optimized.
Moreover, figure 2.3.3 shows the perovskite nanocrystal samples (precipitates) washed 1 time for each halide series. As explained in the purification process, every series starts with the raw synthesis and the raw supernatant and leads to product washed 3 times and the supernatant washed 3 times. Hence, every series is constituted of 8 samples.
Figure 2.3.2: PLQY measurements for different series of lead halide perovskite nanocrystals. Figure from PLQY data courtesy of Jence Mulder (TU Delft).
9
Figure 2.3.3: Illustration of the samples of product washed 1 for all the different perovskite series.
From figure 2.3.3, it can be clearly seen that some samples such as the CsPbBrI2 and CsPbI3 are not in good condition. Practically, this means that either the measurements which concern these samples should be avoided or that the measurement is going to have a huge error bar. In any case, the best possible scenario is to remake these measurements with new samples.
3. Vibrational spectroscopy: Raman and FTIR
Vibrational spectroscopy is used in order to obtain a fundamental insight of the characteristic vibrational energies for example of a molecule. Two techniques are usually used, the Raman spectroscopy and the Fourier-transform Infrared Spectroscopy (FTIR). These techniques are complementary due to the different physical principles which are based on.
First of all, when light interacts with any material a number of events can occur. The light can be transmitted, reflected, absorbed or scattered. When light is absorbed or scattered, the most probable scenario which can happen is called Rayleigh scattering. In this case, the incident photon and the emitted photon have the same energy. However, none of the two techniques are based on this simple phenomenon.
The physical mechanism for the Raman spectroscopy is a two-photon inelastic light scattering event which is called stokes scattering[12]. The photon which comes from a
monochromatic source (laser) has a specific energy. This energy is larger than the than the vibrational quantum energy of the molecule. When the incident photon interacts with the molecule, it is losing some of its energy which is given to one of the molecules’ electrons. This electron goes from the ground state of the molecular vibrations to the virtual energy states. Then the same electron relaxes to one of the molecular vibration states. When it relaxes, it emits another photon and two cases can happen. Case 1, the emitted photon has
10
either a lower frequency and hence lower energy compared to the incident photon, or case 2, higher frequency and hence higher energy. In case 1, this phenomenon is called stokes scattering while in case two the phenomenon is called anti-stokes scattering. Figure 3, shows all the possible cases .However, anti-Stokes scattering is orders of magnitude less probable to happen than the Stokes scattering. Hence, the Raman spectroscopy which is used for the sake of this project is bases on the Stokes Raman scattering. The energy difference between the energy of the incident photon and the measured photon gives us the energy of the vibrational mode of the molecule[12]. Experimentally, the measured vibrational
spectra has for x axis the wavenumber and for y axis the intensity.
Figure 3.1: Schematic illustrations of all the possible types of scattering, where a) is the Stokes Scattering (or Raman scattering), b) is the Rayleigh scattering and finally c) is the anti-Stokes scattering. Image from Lupoi et at,2015[13]
The physical mechanism that leads to an IR response is the absorption of electromagnetic (EM) radiation by molecular vibrations[13],[14]. The first
difference with the Raman spectroscopy is that a polychromatic light source is used. For FTIR a Michelson Interferometer is used which is consisted from the light source, two mirrors, the detector and the samples’ case. Figure 3.2 shows a typical Michelson interferometer. In this set up, firstly, the photons undergo positive interference and then are going through the sample. The electric dipole moment of the vibrational mode of the molecule/sample is getting changed as it is interacting with the photons. Furthermore, some photons with specific energy match the molecular
Figure 3.2: Schematic illustration of the Michelson
Interferometer where M1 is the movable mirror and M2 is the tiltable mirror. Image from Jeremy Ong. 2010[16]
11
vibration modes of the molecule and hence these photons are becoming absorbed[15]. The
photodetector measures the intensity of the transmitted light for all the wavelengths while the mirror of the interferometer moves. For each mirror position, an interferogram is recorded. Then, these interferograms goes under Fourier transformation and give the spectra which is the absorbance (or transmittance) versus the frequency or the wavenumber. The intensity of the IR band that is measured is proportional to the square of the change in electric dipole moment[15].
The most important difference which renders these techniques complementary is that Raman spectroscopy can be used best for symmetric or in-phase vibrations and non-polar groups. This is because they do not yield a change in the dipole moment and hence they are not IR active. On the other hand, FTIR spectroscopy can be used for asymmetric or out-of-phase vibrations and hence can detect polar groups[15]. In addition, another difference is
that Raman signals are very weak compared to FTIR signals. This is because as mentioned above Raman spectroscopy is based on stokes scattering and stokes scattering gives an intensity which is an order of 10−6 less than the intensity given from the Rayleigh scattering[14]. However, Raman spectroscopy is easier to perform experimentally than IR
spectroscopy. This is because at IR spectroscopy the thickness of the sample plays the most crucial role. If you have a sample which is too thick then the majority of the photons cannot propagate the sample and hence cannot reach the detector. On the other hand, if the sample is too thin then you cannot detect any difference as the majority of the photons are not inferring with the sample. Also, in order to obtain the ideal FTIR signal, the proper substrate needs to be used as well as free water and carbon dioxide environment are required.
4. Research in this thesis
4.1 Research question
The objective of this research is firstly to prove that the specific purification process (washing procedure) is indeed the proper one for dealing with the perovskite nanocrystals and then to obtain insights on the ligand system of perovskite nanoparticles. The Raman spectra of the pure ligands as well as ligands on different molecular concentrations were studied for this reason. Furthermore, the halide dependence was also investigated with the aim of understanding the surface-ligand bonding depending on the halide.
12
4.2 Methodology and sample preparation
It is well know that many chemical species are active in the mid-infrared region. Both the Bruker Optics - IFS 66/S FTIR spectrometer which was used for the FTIR experiment and the Renishaw inVia confocal Raman microscope are set to probe to the mid infrared region. The measured spectra are in the range of (100-4000) cm-1 for the FTIR spectra and from
(150 to 3200) cm-1 for the Raman spectra. However, in order to maximize the quality of the
experimental outcome, the focus has been on the proper spectral region.
The Bruker Optics - IFS 66/S FTIR spectrometer has specific steps in order to create a reliable experimental measurement. Firstly, the deuterated triglycine sulfate (DTGS) detector needs to be cooled with liquid nitrogen at least 30 minutes before the first measurement, in order to prevent any influence of external infrared radiation. For each measurement, the first step is to refill the detector compartment. Then, the set up needs a minimum of ten minutes to stabilize before measuring. This is needed because the air in sample chamber needs to be free of substances like carbon dioxide and water vapor, which have a strong IR absorption. Nitrogen gas is flushing out the substances so as to minimize the impact of the IR signal of air. However, some parameters affect the needed time period. These parameters are the time period which is needed in order to refill the sample chamber and hence the time period which was without the nitrogen flushing. In addition, the substrate thickness plays an important role because some proportion of water vapor or carbon dioxide can be trapped inside the substrate. Moreover, the ideal scenario is to take a background spectrum before each measurement. This method will neglect the impact of the substrate as well as it will minimize the impact of the light source fluctuations. Finally, a paper made mask was made in order to prevent any negative impact because of the different positioning of the sample. The mask was placed between the light source and the sample. This practically means that the photons which reach the detector go through a specific spatial location of the CaF2 substrate.
On the other hand, the Raman set up does not have any of these needs. The only challenging part for the experimental measurement is that the focus needs to be the best it can be so as to obtain a spectrum which has as less noise as possible with well defined peaks.
For both techniques the method of preparing the samples is the drop-casting method. However, some samples have only been drop-casted one time with 20μL, while others more times. For the FTIR technique the thickness of the sample is crucial because if it is too thick (or too thin), the number of photons which reach the detector are almost zero (or 100%) and hence cannot provide enough information.
The analysis for both methods become with the program named Origin. Raman spectra were corrected for the influence of the substrate and the objective via equation 1.
13
Where ISample,real is the real spectrum that needs to be measured, ISample,raw is the actual
measurement of each sample, Isub,raw is the measurement of the substrate which was used,
Iobj,raw is the measurement of the objective which was used and a is a parameter which need
to be optimized by hand for each measurement.
Background correction for FTIR spectra was performed by (arbitrarily) subtracting a polynomial background, for example to remove obvious artifacts. For the measurements with a high error bar, the reason is that for the fitted plots at origin, the area for the measured peak and the error of the same measurement has almost the same value leading to high uncertainty. Appendix 1 provides more information about the fitting.
In addition, for both FTIR and Raman measurements, NC films on CaF2 substrates were
prepared via drop casting. With drop casting, the quantity (on μL) can be measured, however the thickness of the samples is not the same as a number or parameters are influencing the outcome. These parameters are: the concentration of each sample (before the drop casting) (see figure 2.3), the area of dispersion on the substrate and the rate of evaporation of each sample so as to become solid. Finally, is order to achieve the best vibrational spectra different number of drop-casting steps was performed for each measurement. Also, in order to minimize the impact of the signal due to the substrate, either for comparing different washing steps of the same perovskite nanocrystal or different halide series, the same substrate where used. CaF2 was the material of the substrate.
Apart from the fitting and the sample preparation information, it needs to be noted that the samples which were used for this thesis have different origins. First of all, the perovskite nanocrystal samples where synthesized and provided from the group of Liberato Manna at the Institute Italiano di Tecnologia in Genova in Italy by Jence Mulder and they are in liquid state. Secondly, Oleic acid and Oleylamine are received from Leyre Gómez Navascués from the group of Tom Gregorkiewicz at the University of Amsterdam. Furthermore, OA, OLA and Octadecene (ODE) are bought from the company named Sigma-Aldrich.
5 Results and discussion
5.1 Spectral markers
As explained above, different chemical species have different vibrational modes. ODE, OLA and OA are the main components of the solvent phase of the perovskite nanocrystals dispersions, (see the synthesis protocol in section 2.3.1). Also hexane and methyl acetate were present at the washing procedure of the perovskite nanocrystals. Their IR spectra and the assignment of their peaks are the first steps in order to obtain their spectra markers and hence to be able to state if these compounds exist in the samples throughout the washing procedure. Then, the second step is to compare different halide perovskite nanocrystals
14
with the aim of proving both the preference of the ligand system and then the different binding motifs depending on the halide.
5.1.1 Octadecene
First of all, ODE is almost the same molecule as OLA or OA with the only difference that it does not have any functional group attached to it, it is just a hydrocarbon chain. The Raman spectrum for ODE is obtained and can be seen at figure 5.1.1. As it is clear from the figure, ODE has many peaks at the region 600-2000cm-1. The peak at 1641 cm−1 is assigned to
the (C=C) stretching mode and will be used as a spectral marker of the ODE[17]. As ODE
and OLA, OA are similar molecules the assignment of the peaks is also similar. The peaks at 1300 and 1438cm-1 correspond to CH 2 bending mode. 600 800 1000 1200 1400 1600 1800 2000 0 10000 20000 30000 40000 50000 60000 70000 80000 90000 100000
Int
en
sity a.
u.
Wavenumber cm-1 ODE 1641Figure 5.1.1: Raman spectra of Octadecene at the range of 600-2000 cm-1.
15
5.1.2 Oleylamine and Oleic Acid
Commercially available OLA and OA samples have some impurities which give additional Raman peaks, these impurities can be attributed to two isomers. In the case of OA the two isomers are the oleic acid and the elaidic acid. For convenience, elaidic acid and oleic acid are called here ‘trans oleic acid’ and ‘cis oleic acid’, respectively. In the case of OLA the two isomers are the Oleylamine and Elaidylamine. The Figure 5.1.2 illustrates the difference between the two molecules for the case of carboxylic acid, the cis and trans isomers have the same chemical formula which is: CH3(CH2)7CH=CH(CH2)7COOH[18]. The
only difference is on the spatial arrangement of the molecules as the
trans-isomer is a straight line and the cis-isomer is bent.
Figure 5.1.3 illustrates the Raman spectra of the pure OLA and OA as well as mixes of different molecular ratios between them.
Figure 5.1.2: Schematic illustration of the trans and cis isomers.
16 1200 1250 1300 1350 1400 1450 1500 1550 1600 1650 1700 1750 0,0 0,2 0,4 0,6 0,8 1,0 1,2 1,4 1,6 1,8 2,0 2,2
Int
en
sity a.
u.
Wavenumber cm-1 100% OA 75%OA 25% OLA 50% OA 50% OLA 25% OA 75% OLA 100% OLA cis:1654 trans: 1671 1265 1300 1438Figure 5.1.3: Raman spectra of OA and OLA for different molecular ratios. The black line corresponds to 100% OLA the red to 25% OA and 75% OLA, the blue to 50% OA and 50% OLA, the green to 75% OA and 25% OLA and finally the purple to 100% OA.
First of all, for figure 5.1.3, the assignment of the peaks is as following: the peak at 1265cm−1 is assigned to the cis-CH symmetric rock mode, at 1300cm-1 to the CH
2 bending
mode, the peak at 1438cm-1 corresponds to CH2 bending mode, both peaks at 1654 cm-1
and at 1671 cm-1 corresponds to the C=C stretching mode, the first correspond to the cis
isomer stretching mode and the second corresponds to the trans isomer stretching mode[17],[18].
Moreover, it is clear thatthe cis and trans isomers are both present inside the sample of the OLA. However, for OA it seems that only the peak of the cis-isomer has significant presence. Also, from the same figure it is clear that as the molecular concentration goes from 100% OLA to 100% OA the relative intensity of the peak of the trans isomer is decreasing, indicating that the concentration of the trans isomer is getting lower. For 100% OA the peak which corresponds to the trans isomer stretching mode can been hardly seen. In addition, in order to analyze the presence of the trans and cis isomers a step further, a quantitave approach is needed. First step is to fit the peaks and obtain their properties such as the area of the peaks and their intensity. In order to so, the Gaussian fitting was chosen and the program named Origin was used. Then, by analyzing the ratio of the area of the cis and trans peaks, as
peak cis the of area peak rans thet of area peak trans the of area ratio . . . . . . . . . . . . + = , a quantitative
approach can be realized. Another approach is to compare the intensity of the peaks which should give the same results. Appendix 2 has the information about the intensity of the
17
peaks and the results are in accordance to each other. Figure 5.1.4 illustrates the differences on the ratio for different molecular concentration of OLA and OA.
From figure 5.1.4 , it can be seen that as the molecular concentration goes from 100% OA to 100% OLA the area of trans/(cis+trans) isomers is increasing and it should be a linear relationship. Given the error bars, this statement seems to be valid. Under the assumption that the cis and trans isomers are bonding with the nanocrystal cell on the same way, the above ratio can be used in order to track the presence of OA and OLA into the sample or at least define which of the two substances prevails at concentration in the sample.
5.2 Purification process of CsPbCl2Br nanocrystals (Raman)
The first perovskite nanocrystals series which is studied is the CsPbCl2Br. Figure 5.2.1,
illustrates the Raman spectra for all the washing steps of the CsPbCl2Br series.
100% OA 25% OLA 75% OA 50% OLA 50% OA 75% OLA 25% OA 100% OLA 0,00 0,02 0,04 0,06 0,08 0,10 0,12 0,14 Area of Trans / (Ci s + T ran s ) iso mers a.u.
Molecular concentrations of OLA and OA trans/(cis+trans)
Figure 5.1.4: Illustration of the area of trans/(cis+trans) isomers versus different molecular concentrations on OA and OLA.
18 1200 1250 1300 1350 1400 1450 1500 1550 1600 1650 1700 1750 0,0 0,2 0,4 0,6 0,8 1,0 1,2 1,4 1,6 1,8 2,0 2,2 2,4 2,6 2,8 3,0 3,2 3,4 3,6 3,8 4,0
Int
en
sity a.
u.
Wavenumber (cm-1) SW3 CsPbCl2Br SW2 CsPbCl2Br SW1 CsPbCl2Br S1 CsPbCl2Br PW3 CsPbCl2Br PW2 CsPbCl2Br PW1 CsPbCl2Br P1 CsPbCl2Br 1438 1300 ODE: 1641 cis: 1655 trans: 1671 1265Figure 5.2.1: Raman spectra of CsPbCl2Br synthesis for all the steps of the washing procedure normalized at 1438 cm-1. {Where P1 is the raw synthesis (brown line), PW1 is the precipitate washed 1 (light blue line), PW2 is the precipitate washed 2 times (yellow line), PW3 is the precipitate washed 3 times (final product- purple line), S1 is the supernatant (green line), SW1 is the supernatant washed 1 time (blue line), SW2 is supernatant washed 2 times (red line) and finally SW3 is supernatant washed 3 times (black line)}.
All the peaks which were present at the OLA and OA are still present. The only difference is the ODE peak which is present to the sample. A focused region, on the range of 1550-1750cm-1 can provide more details as far as the isomers are concerned. Figure 5.2.2,
19 1550 1575 1600 1625 1650 1675 1700 1725 1750 0,0 0,2 0,4 0,6 0,8 1,0 1,2 1,4 1,6 1,8 2,0 2,2 2,4 2,6 2,8 3,0 3,2 3,4 3,6 3,8 4,0
Int
en
sity a.
u.
Wavenumber (cm-1) SW3 CsPbCl2Br SW2 CsPbCl2Br SW1 CsPbCl2Br S1 CsPbCl2Br PW3 CsPbCl2Br PW2 CsPbCl2Br PW1 CsPbCl2Br P1 CsPbCl2Br ODE: 1641cis:
1655
trans: 1671Figure 5.2.2: Raman spectra of CsPbCl2Br synthesis for all the steps of the washing procedure focused on the region 1550-1750 cm-1 normalized at 1438cm-1. Where P1 is the raw synthesis (brown line), PW1 is the precipitate washed 1 (light blue line), PW2 is the precipitate washed 2 times (yellow line), PW3 is the precipitate washed 3 times (final product- purple line), S1 is the supernatant (green line), SW1 is the supernatant washed 1 time (blue line), SW2 is supernatant washed 2 times (red line) and finally SW3 is supernatant washed 3 times (black line).
From figure 5.2.2, it is clear that the relative intensity of the peak of the cis- isomer at 1655cm-1 keeps almost the same value through the washing procedure and the peak of the
trans-isomer at 1671cm-1 is clearly observed for precipitate washed 1 (light blue line) and
precipitate washed 2 (yellow line). Also, from the same figure we can conclude that the relative intensity of the peak at 1641 cm-1 is getting decreased both for the precipitates and
the supernatants through the washing procedure. This means that the ODE is getting washed out. The final product which is the precipitate washed 3 times and the supernatant which is washed 3 times have almost no the peak at 1641 cm-1. A more detailed analysis
including the quantitative approach mentioned above is following.
Figure 5.2.3a and b, shows the area of the ODE peaks both for the precipitates and the supernatants of the CsPbCl2Br series, respectively. From the same figure, it can be clearly
stated that the main quantity of ODE is getting washed out at the first step of the washing procedure which represents
(
70
4
)
% of the ODE to be washed out. Also, supernatant 1 and supernatant washed 1 have the largest area of the ODE peak which proves the above statement.20
Figure 5.2.3: Area of ODE peak through the washing procedure, P1 is the raw synthesis, PW1 is precipitate washed 1,PW2 is precipitate washed 2, PW3 is precipitate washed 3(final product), S1 is supernatant 1, SW1 is supernatant washed 1, SW2 is supernatant washed 2, SW3 is supernatant washed 3.
Moreover, figures 5.2.4.a and 5.2.4.b, shows the area of the area of Trans / (Cis + Trans) isomer peaks both for the precipitates and the supernatants. From the same figure, it can be seen that the area of Trans / (Cis + Trans) isomer peaks for the precipitates is getting decreased throughout the washing procedure. If the measurement for the raw synthesis is excluded as it has a huge error bar, then a clear trend can be seen. This trend seems to be the same as the trend with the trend of the figure 5.1.4 (different molecular ration of OLA and OA). Following the same assumption which is that the trans and the cis isomer bond to the surface of the nanocrystal at the same way, it is clear that throughout the washing procedure the concentration of OLA is getting decreased in comparison with OA. This indicates that higher amounts of OLA are getting washed out compared to OA. Furthermore, as far as the supernatants are concerned, the washed out OA and OLA should give the opposite trend. However, this cannot be clearly stated. Supernatant 1 and supernatant washed 1
P1 PW1 PW2 PW3 0 1 2 3 4 5 6 7 Area of ODE pe ak a .u. Washing Steps ODE a. S1 SW1 SW2 SW3 0 1 2 3 4 5 6 7 8 Area of th e O DE pea k a. u. Washing Steps ODE b.
Raw synthesis Step 1 Step 2 Step 3 0,0 0,1 0,2 0,3 0,4 0,5 0,6 tra ns / (cis+ tra ns) ar ea a. u. Washing Steps trans/(cis+trans) Washing steps for the precipititates of CsPbCl2Br
a
Raw synthesis Step 1 Step 2 Step 3
0,0 0,1 0,2 0,3 0,4 0,5 0,6 tra ns/ (cis+t ra ns) ar ea a. u. Washing Steps trans/(cis+trans) b Washing steps for the supernatants of CsPbCl2Br
Figure 5.2.4: a. Area of Trans / ( Cis + Trans) isomer peaks through the washing procedure for the precipitates of CsPbCl2Br synthesis, b. Area of Trans / ( Cis + Trans) isomer
peaks through the washing procedure for the supernatants of CsPbCl2Br synthesis.
21
have huge error bars which need to be taken into serious consideration. By excluding these measurements, then indeed the opposite trend can be seen. The physical reason behind these error bars was that the supernatant samples were not homogeneous. There was some precipitation in the samples. Stirring up was the method which was used in order to restore the sample into its original form. Probably, this is not enough. New samples and then new measurements would be the best approach for countering this issue.
5.3 Purification process of CsPbBr3 nanocrystals (FTIR)
The first series of perovskite nanocrystals which was examined with the FTIR set up is the CsPbBr3. Figure 5.3, shows all the steps of the washing procedure apart from supernatant
washed 3 times. Supernatant washed 3 times is missing due to lack of material.
1200 1250 1300 1350 1400 1450 1500 1550 1600 1650 1700 1750 1800 0,0 0,5 1,0 1,5 2,0 2,5 3,0 3,5
Absor
ba
nce
a.
u.
Wavenumber (cm
-1)
(SW2) (SW1) (S1) (PW3) (PW2) (PW1) (P1) 1465 1641 1575 1524 1550 1710 1402 1730Figure 5.3: FTIR spectra of CsPbBr3 synthesis normalized at 1465 cm-1 for all the steps of the washing
procedure. Where P1 is the raw synthesis (black line), PW1 is the precipitate washed 1 (red line), PW2 is the precipitate washed 2 times (orange line), PW3 is the precipitate washed 3 times (final product- yellow line), S1 is the supernatant (green line), SW1 is the supernatant washed 1 time (blue line), SW2 is supernatant washed 2 times (brown line).
22
The assignment of the peaks goes as following: The peaks at 1402 is attributed to the (COO-)symmetric stretching modes[18], the peak at 1465cm-1 on the CH
2 bending mode, the
peaks between (1524-1550) cm-1 can be attributed to the antisymmetric (COO-) stretching
more[19] ,[20] , the peak at 1575cm-1 can be attributed to the ammonium twist mode[21] and the
1710 and 1731 cm-1 can be attributed to the (C=O) stretching mode[20].
First of all, the remarks about the ODE peak (1641 cm-1) are the following: ODE is clearly
present at samples: Precipitate 1 and precipitate washed 1, (brown and red line) respectively. Supernatant 1 (green line), supernatant washed 1 (blue line) and supernatant washed 2 (purple line). However it is logical to assume that supernatant washed 3 times should not have a high absorbance peak at 1641 cm-1 as ODE is not present at significant
amounts at the precipitate washed 2 times. This means that the majority of the ODE is was getting washed out at the first two steps of the washing procedure. This seems strange because for the Raman measurements 70% of the ODE was washed out at the first step of the washing procedure. However, here the results are only qualitative which renders it hard to make one to one comparisons.
In addition, the most important differences are at the peaks which appear at 1524-1550, 1575 and 1710, 1731 cm-1. As the washing procedure is evolving from step 1 to step 3, the
absorbance ratio of the antisymmetric COO- stretching versus the ammonium twist mode is
getting lower. This indicates that OA is getting washed out leading to higher concentration of OLA into the sample. Hence, as the washing procedure is taking place, there is an indication that the binding which prevails is between the nanocrystals’ surface and the Oleylammonium. In addition, there is a small blue shift for the peak of the antisymmetric COO- stretching mode , the physical reason behind it is that some lead atoms might “share”
one carboxylate, resulting in a different coordination and hence slightly different binding energy[22] and thus, different vibrational energy mode. This gives an indication that as the
wavenumber is getting decreased, the binding motif goes from Bridging to Chelating binding motif[22]. However, the difference between (COO-) symmetric and antisymmetric
stretching mode is not that large in order to prove 100% the above indication. Moreover, the peaks at 1710 and 1731 cm-1 appears for the precipitates washed 2 and 3 times as
well as for the supernatants washed 2 and 3 times. For the precipitates, seems to be a very broad peak which may include more peaks inside that broad spectral region. The appearance of a peak at 1710 cm-1 indicates that OA is present in its protonated form, i.e.
as Oleic acid (instead of Oleate). This practically means that at least some amount OA and the nanocrystal are not bound.
23
5.4 Halide-dependent purification process and ligand binding motif of
CsPbX3 nanocrystals (FTIR)
Different halides are expected to have different ligand binding motifs. In order to determine the binding motif of each perovskite series a straight comparison between different halides is needed. Figure 5.4 provide information about different lead halide perovskite series. For all the series, Precipitate washed 1 is the step of the washing procedure which is chosen. This step is chosen for a number of reasons. The first one is that the major quantity of ODE is already washed out and it should not give a high signal. Secondly, as mentioned above during the washing procedure OA and OLA are getting washed out leading to destabilizing the perovskite nanocrystal. The final reason is that the samples of this particular washing step seemed to be in better condition that the others leading to more clear results. So, the sample which was produced after the first washing step is the ideal candidate.
1200 1300 1400 1500 1600 1700 1800 0,0 0,5 1,0 1,5 2,0 2,5 3,0 3,5
Absor
ba
nce
a.
u.
Wavenumber cm-1 Cl2Br Br3 Br2I BrI2 I3 1528 1580 Halide Dependance 1402 1465 1711symmetric CH2 stretching mode
antisymmetric CH2 stretchingmode
(CH2 ) bending NH 3
+
twist (C=O) streching
Figure 5.4: FTIR spectra of different halide synthesis of precipitates washed 1 time normalized at 1465 cm-1.
The black line corresponds to the CsPbI3, the red line to CsPbBr2I, the orange line to CsPbBr2I, the green line
to CsPbBr3, the blue to CsPbCl2Br and the brown to CsPbCl2Br.
First of all, the assignment of the peaks is the same as it was for the previous chapter. Furthermore, from figure 5.4, it can be stated that as the perovskite series goes from CsPbI3 to CsPbCl2Br which means from larger to smaller lattice spacing, (and therefore
24
smaller Pb-Pb distance), there is minimal ligand-ligand separation at the surface. Also, the ratio of the relative absorbance between the two peaks which are NH3+ twist at 1580cm-1
and the anti-symmetric COO- stretching mode at 1528cm-1 is changing. As the ratio of the
antisymmetric stretching mode versus the NH3+ twist is getting lower (from CsPbCl2Br to
CsPbI3), it is implied that there is an increasing ammonium to carboxylate ratio. This is an
indication that maybe there is a stronger binding between the perovskite cell and the amine group of Oleylammonium. Of course, the opposite is also valid. When the signal of the antisymmetric peak is larger than the signal of the ammonium peak (from CsPbI3 to
CsPbCl2Br), then the indication suggest that the major binding is between the perovskite
cell and the carboxylate group of the Oleate.
The physical reason behind the halide dependence is the size of the unit cell which renders the density of the aliphatic chains to be sterically limited. As the unit cell is decreasing in size (from CsPbl3 to CsPbCl2Br), its “free space” between the Pb atoms is also decreasing.
These suggest two consequences, the first is that the steric hindrance for Oleate ligands in binding to surface Pb is increasing. The second is that the symmetric and asymmetric peaks at (1402 and 1528)cm-1 can be slightly shifted depending on the halide. These
variations in the peak position and intensity of the peaks are happening due to differences in the lead carboxylate bonding angle and variation in the Pb-O bond distance[22]. Moreover,
some lead atoms might “share” one carboxylate, resulting in a different coordination and hence slightly different binding energy[22]. However, for all the halides the difference
between the symmetric and antisymmetric COO- stretching mode is 1
150 170v cm−
which indicates that the Oleate coordination is either chelating or bridging bidentate coordination[22]. Peters et al. claims that low wavelength peak of the v
sym vibration and the
higher wavelength peak vantisym belong to the bridging bidentate carboxylic coordination,
while the higher wavelength peak of the vsym and the low wavelength peak of vantisym belong
to the chelating coordination. In this study the difference of the wavenumber between the vsym and vantisym depending on the halide is too small indicating that there is only one
coordination, the bridging bidentate coordination.
6 Conclusions and remarks
First of all, as far as the purification process (washing procedure) is concerned, two remarks are absolutely clear. The first one which concerns the Raman measurements is that
(
70
4
)
% of the ODE is being removed from our samples during the first washing step of the purification process. This proves that this process is indeed successful at the removal of the ODE. The second remark is that OLA and OA are also getting removed during this purification process. It is well known that perovskite nanocrystals which are processed with the washing procedure described at chapter 2.3 lose their stability and hence their high PLQY as the steps of the washing procedure are taking place. The main25
focus for future applications is to achieve the best possible optimization between high PLQY and maximum purity, depending on the application. In addition, during the washing steps of the CSPbCl2Br series, by tracing the washed out cis and trans isomers we have an
indication that more and more OLA is getting washed out in comparison with OA. As OA and OLA are identical molecules the washing procedure should not affect the washed out ratio between the two molecules. This indicates that the difference is coming due to the bonding with the nanocrystal’s surface.
Furthermore, FTIR analysis on the CsPbBr3 series indicates that OA is getting washed out
which is the opposite result compared with the Raman measurements for the CSPbCl2Br
series. However, it needs to be noted that this is another series of halide perovskite nanocrystals. This result indicates that ligands bond with the perovskite nanocrystal with different bonding energies depending on the halide. The FTIR measurements for different halide series reveal two results. The first one is that as the perovskite lattice spacing is decreasing (from CsPbI3 to CsPbCl2Br) the major binding is between the perovskite cell’s
surface and the carboxylate group of the Oleate. This also can be the reason of the different washed out material (ligand) between the CsPbCl2Br and the CsPbBr3 series.
Secondly, as far as the bending motifs of the different halide series are concerned, the halide size does not seem to affect the binding motif between the nanocrystal’s surface and the Oleate. All the different halide series seem to bind with the bridging bidentate coordination.
More general, some suggestions may help the future research. The first one is to conduct the same experiment inside a glove box or at least to do the sample preparation inside it. This means that a water vapor and a carbon dioxide free environment will minimize any environmental impact. The second suggestion is to create the same experiment with new samples as some of them were already in the end of their lifetime (see figure 2.3.3) and hence the results can be questioned. Finally, in order to maximize the colloidal stability of the nanocrystals, some recommendations are either to change the whole purification process or to search for alternative processors-solvents.
26
7 Acknowledgements
First and above all, I would like to thank Elizabeth von Hauff for being my supervisor and accept me as a member in her team. Furthermore, I would like to express my gratitude and my thanks to Simon Boehme for sharing his expertise and helping me with the practical part of this research. Also, I would like to express my thanks to the group of Tom Gregorkiewicz at the University of Amsterdam and to Leyre Gómez Navascués for providing some of the measured samples. Finally, I would like to thank the group of Liberato Manna from the Istituto Italiano di Tecnologia, Genova, Italy, for their support, with a special thanks to Jence Mulder for the synthesis of the nanoparticles used in this study.
27
8 References
[1]Loredana Protesescu,Sergii Yakunin,Maryna I. Bodnarchuk, Franziska Krieg, Riccarda Caputo, Christopher H. Hendon, Ruo Xi Yang, Aron Walsh, Maksym V. Kovalenko. 2015 Nanocrystals of Cesium Lead Halide Perovskites (CsPbX3, X = Cl, Br, and I): Novel Optoelectronic Materials Showing Bright Emission with Wide Color Gamut. Nano Lett. 2015, 15, 3692−3696 DOI:10.1021/nl5048779
[2] Linzhong Wu, Qixuan Zhong, Di Yang, Min Chen, Huicheng Hu, Qi Pan, Haiyu Liu, Muhan Cao, Yong Xu, Baoquan Sun, Qiao Zhang. 2017. Improving the Stability and Size Tunability of Cesium Lead Halide Perovskite Nanocrystals Using Trioctylphosphine Oxide as the Capping Ligand. DOI:10.1021/acs.langmuir.7b02963 Langmuir 2017, 33, 12689−12696
[3] Maksym V. Kovalenko, Loredana Protesescu, Maryna I. Bodnarchuk. 2017. Properties and potential optoelectronic applications of lead halide perovskite nanocrystals. Science 358 (6364), 745-750. DOI: 10.1126/science.aam7093
[4]Quinten A. Akkerman, Gabriele Rainò, Maksym V. Kovalenko , Liberato Manna. 2018. Genesis, challenges and opportunities for colloidal lead halide perovskite nanocrystals. https://doi.org/10.1038/s41563-018-0018-4
[5] Binbin Luo, Sara Bonabi Naghadeh, Jin Z. Zhang. Lead Halide Perovskite Nanocrystals:Stability,Surface Passivation, and Structural Control. ChemNanoMat 2017, 3,456– 465. DOI :10.1002/cnma.201700056
[6] Wang, S.; Jiang, Y.; Juarez-Perez, E. J.; Ono, L. K.; Qi, Y. Accelerated degradation of methylammonium lead iodide perovskites induced by exposure to iodine vapour. Nat. Energy 2016, 2, 16195.
[7] Li, X.; Wang, Y.; Sun, H.; Zeng, H. Amino-Mediated Anchoring Perovskite Quantum Dots for Stable and Low-Threshold Random Lasing. Adv. Mater. 2017, 29, 1701185.
[8] Martin A. Green, Anita Ho-Baillie, Henry J. Snaith. 2014. The emergence of perovskite solar cells. Nature photonics. DOI: 10.1038/NPHOTON.2014.134
[9] J. De Roo, M. Ibanez, P. Geiregat, G. Nedelcu, W. Walravens, J. Maes, J. C. Martins, I. Van Driessche, M. V. Kovalenko, and Z. Hens. 2016. Highly dynamic ligand binding and light absorption coe-cient of cesium lead bromide perovskite nanocrystals. ACS nano, 10(2):20712081.
[10] M. V. Kovalenko, L. Protesescu, and M. I. Bodnarchuk. 2017. Properties and potential optoelectronic applications of lead halide perovskite nanocrystals. Science, 358(6364): 745750. [11] L. Gomez, J. Lin, C. De Weerd, L. Poirier, S. C. Boehme, E. Von Hau, Y. Fujiwara, K. Suenaga, and T. Gregorkiewicz. 2018. Extraordinary interfacial stitching between single all-inorganic perovskite nanocrystals. ACS applied materials & interfaces, 10(6): 59845991.
[12] Ewen Smith. Geoffrey Dent. Modern Raman Spectroscopy – A Practical Approach. 2005. QD96.R34S58
28
[13]Jason Lupoi. Erica Gjersing. Mark F. Davis. 2015. Evaluating Lignocellulosic Biomass, Its Derivatives, and Downstream Products with Raman Spectroscopy. Frontiers in Bioengineering and Biotechnology. DOI: 10.3389/fbioe.2015.00050
[14]Fundamentals of FOURIER TRANSFORM INFRARED SPECTROSCOPY.2011. Brian C. Smith. [15] P. Larkin, Infrared and Raman spectroscopy (Elsevier, ed.1, 2011), pp. 1 - 12.
[16] Jeremy Ong. 2010. Investigations of Light with a Michelson Interferometer Applied & Engineering Physics, Cornell University
[17] R. A. Nyquist, R. O. Kagel, Infrared Spectra of Inorganic Compounds, (Elsevier, ed.1, 1971), pp 1 - 3.
[18] Dmitry Baranov, Michael J. Lynch, Anna C. Curtis, Alexa R. Carollo, Callum R. Douglass, Alina M. Mateo-Tejada, David M. Jonas. 2018. Purification of Oleylamine for Materials Synthesis and Spectroscopic Diagnostics for trans Isomer. DOI:10.1021/acs.chemmater.8b04198
[19] Laurianne Robinet and Marie-Claude Corbeil. 2001. The Characterization of Metal Soaps. STUDIES IN CONSERVATION 48 (2003) PAGES 23–40
[20] Lance M. Wheeler, Nicholas C. Anderson, Taylor S. Bliss, Matthew P. Hautzinger, Nathan R. Neale. 2018 Dynamic Evolution of 2D Layers within Perovskite Nanocrystals via Salt Pair Extraction and Reinsertion. DOI:10.1021/acs.jpcc.8b01164
[21] Li-Qiang Xie, Tai-Yang Zhang, Liang Chen, Nanjie Guo, Yu Wang, Guo-Kun Liu, Jia-Rui Wang, Jian-Zhang Zhou, Jia-Wei Yan,a Yi-Xin Zhao, Bing-Wei Mao, Zhong-Qun Tiana. Organic– inorganic interactions of single crystalline organolead halide perovskites studied by Raman spectroscopy. Phys.Chem.Chem.Phys., 2016, 18, 18112
[22] Joep L. Peters, Jur de Wit, Daniël Vanmaekelbergh. 2019. Sizing Curve, Absorption Coefficient, Surface Chemistry and Aliphatic Chain Structure of PbTe Nanocrystals. Chem. Mater. DOI: 10.1021/acs.chemmater.8b05050
29
Appendix 1
One example of the fitting using Origin and hence quantify the area of each peak is the illustrated by figure ap1. Where the black line correspond to the measured data and in this case is for the sample of pure OLA, the brown line corresponds to the baseline, the red line corresponds to the contribution of the cis-isomer, the green line corresponds to the contribution of the trans-isomer and finally the purple line corresponds to the fit.
1620 1640 1660 1680 1700 0,0 0,2 0,4 0,6 Int en sity a. u. Wavenumber cm-1 data baseline cis trans fit
Figure ap1: Example of fitting by Origin for the sample of pure OLA.
It is clear that the data (black line) and the fitting (purple line) do not match 100%. As explained as section 3, which is the methodology,
Equation 1: 𝐼𝑠𝑎𝑚𝑝𝑙𝑒,𝑟𝑒𝑎𝑙 = 𝐼𝑠𝑎𝑚𝑝𝑙𝑒,𝑟𝑎𝑤− 𝛼 ∙ 𝐼𝑠𝑢𝑏,𝑟𝑎𝑤− (1 − 𝛼) ∙ 𝐼𝑜𝑏𝑗,𝑟𝑎𝑤
is used in order to extract the real part of the measurement. The parameter a, which is handmade is optimized so as the impact of the baseline to be minimized. However, the fitting is not always matching 100% with the data. This is also the reason that the error bar has been introduced. The aim is to obtain a clearer inside on the actual real part of the measurement as well as defining the introduced human made impact. For example, some measurements have huge error bars which have almost the same value as the measurements. This can have double meaning, either the measurement is indeed not the best it can be (broken sample) or the fitting is not the best it can be. However, the data which have been chosen to be reported at that thesis is the best data available.
30
Appendix 2
Figure ap2 shows the intensity of the trans/(cis+trans) isomers for different molecular concentrations of OA and OLA. From the figure, it is clear that as the molecular concentration goes from 100% OA to 100% OLA the relative intensity is increasing. This trend can be used as a marker for the concentration of the cis and trans isomers in the samples. This information is supplementary to figures 5.1.3 and 5.1.4. The results agree to each other. 100% OA 25% OLA 75% OA 50% OLA 50% OA 75% OLA 25% OA 100% OLA 0,00 0,02 0,04 0,06 0,08 0,10 0,12 0,14 0,16 0,18 0,20 Inten s it y of tr an s /( c is+ tr an s ) iso me rs a.u. Molecular concetration
31
From figure ap2.2.a it can be seen that as the washing procedure is evolving the figure has the opposite same trend as figure ap2.1. Following the same assumption which is that the cis and the trans isomers bond with the surface of the nanocrystal on the same way. This is an indication that there is a higher concentration of OA compared to OLA into the sample. This result is the same with the main analysis. However, product 1 (raw synthesis) does not seem to follow the same trend. The reason is that during the fitting process, as product 1 has major amount of ODE into the sample the peak of ODE interfered with the peaks of the cis and trans isomers leading to miscalculation. This is also stated by the huge error bars for the product 1 and supernatant 1 at figure 5.2.4(in the main analysis). In addition, figure 2.2.b should have the opposite trend as figure 2.2.a as the washed out materials need to be inside the supernatants of the same washing step. Indeed, this seems to be the case, as the washing procedure is evolving from step 1 to step 3, the supernatants seem to have higher concentration of OLA than OA.
P1 PW1 PW2 PW3 0,00 0,05 0,10 0,15 0,20 0,25 0,30 0,35 0,40 0,45 Int en sity of tr an s/(cis+ tra ns) isom er s a. u. Washing steps CsPbCl2Br a. S1 SW1 SW2 SW3 0,00 0,05 0,10 0,15 0,20 0,25 Int en sity of tr an s/(cis+ tra ns) isom er s a. u. Washing steps CsPbCl2Br b.
Figure ap2.2: a, Intensity of trans/(cis+trans) isomers throughout the washing procedure for the precipitates (green bars) and b, for the supernatants(blue bars).
Finally, apart from the measurements which concern the intensity of the cis/trans isomers for the product 1(raw synthesis) and supernatant 1 of the CsPbCl2Br series, all the other
indications agree to each other leaving less uncertainty about the final results. The reason which was stated above for the product 1 is the same for the supernatant 1.
