Ytterbium Doping of Cesium Lead Halide Perovskites Nanocrystals

(1)

Universiteit van Amsterdam

Master Physics, Track Advanced Matter and Energy Physics

Ytterbium Doping of Cesium Lead

Halide Perovskite Nanocrystals

Master Thesis

Author:

D. C. Laan

Supervisor:

Prof. Dr. T. Gregorkiewicz

Second Reader:

Prof. Dr. P. Schall

Daily Supervisor:

M. van der Laan MSc.

(2)

Abstract

Fully inorganic lead halide perovskite nanocrystals show bright photoluminescence and are in-teresting materials for optoelectronic and photovoltaic applications. Doping ytterbium into the nanocrystals makes it theoretically possible to measure high photoluminescence quantum yields of above 100%. With photoluminescence quantum yields above 100% more photons are emitted than absorbed. This creates the possibility to exceed the Shockley-Queisser limit. In this thesis different

doping procedures and their optical characterizations will be discussed. Incorporation of Yb3+inside

the pores of mesoporous SiO2together with CsPbCl3nanocrystals might be a novel method of Yb

3+

sensitization.

This thesis is part of the completion of the master Physics (track Advanced Matter and Energy Physics) at the University of Amsterdam. Work was performed between September 2018 and June 2019 at the Van der Waals-Zeeman institute for a total of 60 ECTS. The results were publicly presented in presence of all supervisors and other interested on June 14, 2018 at the University of Amsterdam.

(3)

1 Introduction

1.1 General Introduction

In 2004 Richard Smalley, Nobel laureate in Chemistry, formulated the terawatt challenge [1]. It was first stated at his presentation on Symposium X—Frontiers of Materials Research, then edited and typed out in an article in Materials Matters. Smalley formulates the ten biggest concerns for human-ity, where energy is the first on this prioritized list. He argues that in order to provide enough energy for the entire world population in 2050 we will have to generate 60 terawatts around the globe. The energy problem is linked with other problems on this list, including water, food and environment. To generate enough energy we will need to find the ‘new oil’. At the time the challenge was formu-lated only 0.5 % of the generated energy was solar, wind and geothermal energy. The point of this challenge is that not only do we need to generate more energy than we already do, we also need to find a different way. A way which would not burden the environment as much as we do now with our energy generation. Photovoltaic devices could be the solution to this problem. As a physics student you learn that you would only need to fill a small part of the Sahara with silicon solar cells to generate the amount of energy that would be needed for the entire world population. The problem with this idea is energy transport and energy storage. If we would transport the energy from this one energy generation point to locations where it is needed, it would already be gone by the time it has arrived at its destination due to energy losses. So for now it seems like we need to generate the energy locally. In figure [2] the efficiency of the best photovoltaic research cells of different materials is shown. The result displayed is the efficiency of the device that has the highest perfor-mance of its class of materials. The efficiencies are tested and confirmed by independent institutions.

2000 1995 Cell Ef ficiency (%) 1990 1985 1980 1975 12 8 4 0 16 20 24 28 32 36 2005 2010 2015 2020 40 44 48

52Best Research-Cell Efficiencies

(Rev . 01-03-2019) Single-Junction GaAs Single crystal Concentrator Thin-film crystal Thin-film crystal Multijunction Cells (2-terminal, monolithic)

LM = lattice matched MM = metamorphic IMM = inverted, metamorphic

Three-junction (concentrator) Two-junction (concentrator) Three-junction (non-concentrator) Two-junction (non-concentrator) Four-junction or more (concentrator) Four-junction or more (non-concentrator)

Crystalline Si Cells Single crystal (concentrator) Multicrystalline Silicon heterostructures (HIT) Single crystal (non-concentrator)

Thin-Film Technologies CIGS CdTe Amorphous Si:H (stabilized) CIGS (concentrator)

Emerging PV

Dye-sensitized cells Organic cells (various types) Perovskite/Si tandem (monolithic) Organic tandem cells Perovskite cells (not stabilized)

Inorganic cells (CZTSSe) Quantum dot cells (various types)

NREL (ZnO/PbS-QD)

U.Toronto (PbS-QD)

MIT U.TorontoU.Toronto

NREL Univ. of Queensland 16.6% IBM 12.6% 11.9% EPFL EPFL EPFL

EPFL Sharp NIMS

Sharp UCLA-Sumitomo UCLA Heliatek Heliatek UCLA Sumi-tomo U. Dresden 11.5% Siemens Groningen U. Linz _{U. Linz} NREL / Konarka U. Linz Plextronics Konarka Mitsubishi HKUST Phillips 66 UCLA Raynergy Tek of Taiwan SCUT-CSU ICCAS Konarka Solarmer 15.6% IBM IBM IBM Oxford PV Oxford PV EPFL Stanford/ASU 28.0% EPFL KRICT KRICT _EPFL KRICT/UNISTKRICT ISCASISCAS 23.7% UNSW / Eurosolare UNSW Georgia

Tech GeorgiaTech

Georgia Tech SolarexSolarex FhG-ISE Trina FhG-ISE FhG-ISE 22.3%

RCA RCARCARCA RCA

Solarex ARCO UniSolar RCA RCA UniSolar UniSolar (aSi/ncSi/ncSi) AIST AIST LG UniSolar 14.0% Matsushita Monosolar Kodak

Kodak KodakKodak

AMETEK Photon Energy

U. So. Florida

First Solar First Solar First Solar First Solar GE GE Matsushita NREL NREL 22.1% 22.9% U.of Maine U.of Maine Boeing Boeing Boeing Boeing Boeing

ARCO ARCO Boeing

Euro-CIS

NREL NREL NREL

EMPA (Flex poly)

NREL NREL NREL

ZSW ZSW SolarFron NREL NREL NREL NREL Solibro ZSW SolarFron 23.3% NREL (14x) NREL (15.4x)NREL (14.7x) Solexel Solexel U. Stuttgart U. Stuttgart FhG-ISEISFH 21.2%

Sanyo Sanyo Sanyo

Sanyo Sanyo Panasonic

Panasonic Panasonic Kaneka Kaneka 26.6% 27.6% SunPower (96x) Stanford (140x) Amonix (92x) UNSW UNSW SunPower (large-area) FhG-ISE ISFH UNSW UNSW UNSW UNSW ARCO RCA Mobil Solar Sandia UNSW UNSW UNSW Spire Stanford Westing-house 26.1% Radboud U. Alta

Alta Alta Alta Devices29.1%

Varian (216x) Varian (205x) FhG-ISE (117x) LG NREL (258x) FhG-ISE (232x) 30.5% 27.8% IBM (T.J. Watson Research Center) Kopin Radboud Univ. FhG-ISE LG LG NREL Varian Boeing-Spectrolab (5-J) NREL (6-J) 39.2% 46.0% Soitec (4-J, 319x) FhG-ISE/ Soitec Soitec (4-J, 297x) NREL NREL (4-J, 327x) Alta AltaNREL (MM)LG NREL Varian NREL 32.8%

NREL EnergyJapan

Spire No. Carolina State U. Varian IES-UPM (1026x) NREL (467x) FhG-ISE NREL (38.1x) 35.5% NREL (IMM) Sharp (IMM) Sharp (IMM) Sharp (IMM) Spectrolab 37.9% NREL/ Spectrolab Spectrolab Spectrolab Boeing-Spectrolab Boeing-Spectrolab Boeing-Spectrolab Boeing-Spectrolab (MM, 240x) Boeing-Spectrolab (MM,179x) NREL (IMM)

NREL (IMM, 325.7x)NREL

FhG-ISE (MM, 454x) SpireSemicon(MM, 406x) SolarJunc (LM, 418x) SolarJunc (LM, 942x) Sharp (IMM, 302x) Spectrolab (MM, 299x) Boeing-Spectrolab (LM, 364x) 44.4%

Figure 1: Best research-cell solar cell efficiencies by NREL.

In figure 1 both perovskite cells and perovskite/silicon cells are shown. Cells which only use perovskites are not yet stabilized and therefore not shown. The tandem cells of perovskites and silicon show an efficiency of 28.0 %. The perovskite top cells do not score the highest efficiency, in comparison with other materials on the chart, but have other interesting properties. Perovksites have outstanding absorption properties combined with high carrier mobilities. Also it is possible to

(5)

produce a thin solar cell from perovskites that is flexible, making it interesting for applications such as portable solar cells and installing it on shades of windows. In 1960 William Shockley and Hans Queisser calculated the maximum light to electric power conversion energy η of a single junction solar cell [3]. An important assumption that they made is there can be only on electron-hole pair resulting from an incoming photon. The limit was recalculated in 2016 from the then known solar spectrum. The result is a maximum efficiency of 33.16 % for a single-junction solar cell with a band gap of 1.34 eV [4]. Exceeding this limit can be realized if we make it possible to create more than one electron-hole pair per incoming photon.

1.2 Inorganic Cesium Lead Halide Perovskites (CsPbX

3

, with X = Cl,

Br, I)

In 1839 Gustav Rose discovered CaTiO3 and its structure. The structure is become known as the

perovskite structure named after the Russian L. A. Perovski and shown in figure 2. The structure

is described by ABX₃(in figure 2 AMX₃) with X either an oxygen, carbon, nitrogen or halogen.

Figure 2: Figure from: Ziyong Cheng and Jun Lin. The perovskite structure described with AMX₃

in the figure, otherwise ABX₃. On the left the perovskite in cubic structure is displayed. On the

right the extended network structure connected by corner-shared octahedra. [5]

Perovskites generally have a cubic structure, but can get an octahedral distortion [6]. The form of the structure can be estimated by calculating the Goldschmidt tolerance factor:

t = √ra+ rx

2(rb+ rx)

(1)

Here ra, rb and rxare the effective ionic radii for A, B and X ions of the perovskite respectively [6].

When 0.8 < t < 1, a perfect cubic structure is expected, for t < 0.8 the octahedral distortion can be observed. The Goldschmidt factor also is an indication for the stability of a perovskite, where it is supposed to be stable for with a cubic structure. The structure of a perovskite is a factor in its

optoelectronic properties [6]. We will look at CsPbX3with X = Cl–1, Br–1, I–1and we can calculate

(6)

instance by doping, the tolerance factor might change due to the other ionic radii of the introduced ions.

In 2015 inorganic cesium lead halide perovskites (CsPbX3, with X = Cl, Br, I) nanocrystals (NCs)

were first synthesized by Protesescu et al. showing photoluminescence, exhibiting strong absorption and high photoluminescence quantum yield [7]. In figure 3 these properties are displayed.

Figure 3: Figure taken from Protesescu et al. [7]. (a) and (b) show the indication of tunable band

gaps by showing the emission from CsPbX3NCs with varying halide and combining different halides

emission is observed at different wavelengths. (c) shows the absorption and emission for different

perovskites. (d) shows the time-resolved photoluminescence decay of CsPbX3NCs.

In figure 3a the colloidal solution in toluene is held under a UV lamp (λ = 365 nm), showing

different coloured emission. Figure 3b and 3c display the tunablity of the band gap of CsPbX3

over the entire visible spectrum, by showing emission from different wavelengths for different halide

composition. The band gap can also be changed by varying the size of the CsPbX3 NCs due to

quantum-size effects. In figure 3d radiative lifetime between 1 − 29 ns combined with high photolu-minescence quantum yield is indicated [7].

1.3 Optical Doping of Semiconductors with Rare Earth Ions

Rare-earths are characterised by partially filled 4f shells which are shielded from external fields by

5s2and 5p5electrons [8]. Luminescence from rare-earth (RE) ions has developed interest over a

cou-ple of decades [8]. The sharp luminescence bands of lanthanides (a group within the RE materials, which only excludes scandium and yttrium) have been known since the beginning of the 20th century [6]. In RE ions the 4f shell is shielded which makes them difficult to excite. Another complicating factor for direct excitation of RE ions is their small absorption cross section. RE ions can be used as dopants (intentional introduction of impurities into a material [9]) and can then be excited via

the host. Er3+doped materials have been used in optoelectronics and telecommunication, because

the emission of erbium (around 1535 nm) lies within the so called ”ultra low-loss window” of silica

fibres [8]. Yb3+ has only one excited state and it is possible to bring the Yb3+ion from the2_F

7/2

ground state to the 2_F

(7)

1.4 Sensitization of Rare Earth Emitters by Semiconductor

Nanocrystals

Now that we know that it is possible to dope traditional semiconductors with RE ions, the next step for us is to look at sensitization of RE emitters via semiconductor NCs hosts. In order to accomplish this, the effects of doping semiconductor NCs should be investigated. In 2005 Erwin et al. made predictions about the success of doping nanocrystals with manganese. The result of their paper was that it is possible to dope several semiconductor NCs [9]. The doping of NCs can be interesting for solar cell applications [9]. Semiconductor NCs employ the interesting property of quantum confinement, making it possible to tune the band gap of the NCs by altering the size of the NCs. If the NC is used as a host in order to excite RE ions this quantum confinement effect can be important to make excitation of the RE possible. The combination of strong absorption and size tunability of this absorption makes semiconductor NCs compelling for RE doping. Doping NCs with REs makes it possible to get an emission spectrum from visible to near-infrared (NIR) light [10]. There is a great similarity between different RE ions, which means that if doping is effective with one of the REs, other REs can, probably, also be used [11]. Previously reported Yb doped NCs have reported low photoluminescence quantum yield of below 10 %, which makes practical applications difficult [12].

1.5 Lanthanide Doping of Inorganic Lead Halide Perovskites

CsPbX3 NCs have demonstrated excellent optical properties, but it is difficult to achieve

white-light and multicolor emission for practical applications [13]. Just as with NC semiconductors it is

possible to introduce RE ions into CsPbX3 creating a wide emission spectrum [13]. This Emission

spectrum will make CsPbX3 NCs more interesting for practical applications. In 2017 Pan et al.

showed successful doping of various lanthanides into CsPbCl3NCs [13]. Interestingly, the result for

photoluminescence quantum yield shows minor differences for all lanthanides doped in CsPbCl3NCs,

except for ytterbium. The photoluminescence quantum yield of ytterbium doped CsPbCl3 NCs is

very high and above 100 % [13, 14, 15]. For the other lanthanides the photoluminescence quantum yield is below 100 % [13]. Pan et al. used a hot-injection method to dope a series of

lanthanide-ions in CsPbCl₃[13]. Besides emission from the lanthanides, an excitonic photoluminescence peak

corresponding with CsPbCl₃is observed. The center of the excitonic photoluminescence peak shifts

with the introduction of lanthanides [13]. With the introduction of the lanthanides the size of the NCs will change due to lattice contraction of doped NCs [13]. As the size of the NCs decrease the band gap will gradually increase because of the quantum confinement effect. [13] Lanthanide doped

CsPbCl3 still exhibit the perovskite structure. Based on measurements on Eu+3 doped CsPbCl3

NCs Pan et al. conclude that the lanthanides most likely take the place of lead in the perovskite structure [13].

The exceeding of photoluminescence quantum yield of 100 % of Yb doped CsPbCl3lead to questions

about the physical background of how the energy transfer happens between the perovskite host and

the ytterbium ions. Milstein et al. measured the excitonic peak of undoped CsPbCl3and compared

it with Yb doped CsPbCl3and observed that the excitonic photoluminescence of undoped samples

is very bright, but almost disappears in the doped sample. The Yb3+:CsPbCl3 NCs show a clear

NIR photoluminescence peak at 990 nm [14]. The intensity of the Yb photoluminescence peak of

Yb3+:CsPbCl3 NCs increases with increasing amount of Yb in the NCs [14]. However, the lifetime

of the excited state Yb3+ 2F5/2 is largely independent of the concentration of Yb3+and the halide

composition [14]. After the observations on Yb3+:CsPbCl3 NCs, Kroupa et al. used a two-step

(8)

demonstrate the same Yb3+emission peaks, where the ytterbium is excited via the perovskite [12]. The fact that it exhibits the same high photoluminescence quantum yield and peak corresponding

with the2_F

7/2 to2F5/2transition is an indication that this does not come from effects exclusive to

NCs, but probably is a bulk phenomenon [12]. This result also shows that high photoluminescence quantum yield is almost independent of the size of the NCs. Another result from the measurements

on thin film Yb3+:CsPb(Cl1–xBrx)3 perovskites is facile power saturation of Yb

3+

emission which will constitute in a big challenge for practical applications [12].

To reduce the power saturation in Yb3+:CsPb(Cl1–xBrx)3 perovskites three solutions have been

proposed [16]:

• Shorten the Yb3+_lifetime

• Decrease the photo-excitation rate per Yb3+

• Reduce the cross-relaxation rate

The last of the suggestions needs some clarification. In their paper Erickson et al. developed a model giving microscopical insights on why these NCs saturate. [16] They found that excited NCs transferred energy to already excited Yb which prevented quantum-cutting.[16] It is found that this cross-relaxation competes with the quantum-cutting, making it one of the reasons of saturation.[16] Important for all these solutions is that the high photoluminescence quantum yield and clear

yt-terbium emission of Yb3+:CsPb(Cl1–xBrx)3perovskites are preserved. An overview with very brief

summaries from the most important articles on Ytterbium doping of cesium lead halide perovskites can be found in the appendix.

1.6 Photoluminescence Quantum Yield

Photoluminescence quantum yield (PLQY) has been mentioned several times already. Before we

move on to discuss the origin of the high PLQY of Yb3+:CsPb(Cl1–xBrx)3perovskites, PLQY should

be discussed in more detail. The PLQY of a system is defined as the ratio between photons emitted

and photons absorbed: Φ = _{#photons absorbed}#photons emitted. Not every absorbed photon will result in an emitted

photon. Essentially, the PLQY describes the efficiency of a system of converting excitation light into fluorescence. Although it is possible to emit more photons than there are absorbed, conservation of energy still holds. If a high-energy photon is absorbed, it can be possible for some systems to emit two lower energy photons. The sum of the energy of these emitted photons must be lower than or equal to the energy of the absorbed photon.

1.7 Proposed Energy Transfer of Ytterbium Doped Cesium Lead Halide

Perovskites

In the previous section developments on Yb3+:CsPb(Cl1–xBrx)3 perovskites are discussed but one

important part is omitted: the energy transfer between the perovskite host and ytterbium. High PLQY has been measured, but the origin has to be discussed. Theory has not yet been conclusive

about the way the energy transfer within Yb3+:CsPb(Cl_1–xBr_x)₃ perovskites works. However, in

various articles similar transfer mechanisms have been proposed [14, 12, 15, 13]. First of all it is

important to state the difference between Yb3+and the other lanthanides. Pan et al. measured the

PLQY for all lanthanides doped in CsPbCl3 NCs and found that only when Yb3+ is introduced a

(9)

the excitonic emission of CsPbCl3which lies around 410 nm. This should correspond with a band

gap of ∼ 3.0 eV. Ytterbium is the only used lanthanide where the transition from the ground state to the excited state is less than 1.5 eV, this transition is around 1.3 eV. This means that energy conservation holds when one incoming photon will be absorbed by the perovskite, which excites

Yb3+and results in two outgoing photons with an energy of ∼ 1.3 eV. Pan et al. proposed two

dif-ferent possible pathways for the quantum cutting mechanism in Yb3+:CsPbCl3which are displayed

in figure 4 [13].

By excitation with a wavelength of 365 nm an electron gets excited from the valence band to the

conduction band. Due to excitonic recombination blue photons are emitted, which excite Yb3+

[13]. However, the blue photons are not detected in any other study on Yb3+:CsPbCl3. The other

pathway that is proposed makes use of the defects in the perovskite NC host coming from the

in-troduction of Yb3+. These defects are reported in several articles and possibly come from the fact

that Yb3+ comes in the place of Pb2+ which requires charge compensation [14, 13]. The excited

electrons can transmit from the conduction band to the defect state and transfer energy to a Yb3+

ion, generating a2_F

7/2to2F5/2transition. From the defect state the electron can further recombine

with the valence band, exciting another Yb3+ ion, generating another 2_F

7/2 to 2F5/2 transition.

This sensitization of Yb3+ via the perovskite host also has a thermal component. Increasing the

temperature shows an increasing Yb photoluminescence intensity indicating that some aspect of the sensitization is thermally assisted, however even at 5 K there still is Yb emission [14].

Figure 4: Figure taken from Pan et al. The proposed energy transfer between ytterbium and CsPbCl3

schematically described.[13]

(10)

cell absorbs the energy of a photon at the band gap. Every photon with an energy above the band gap will be absorbed (at least for calculating the limit), but the excess energy will be lost due to

thermalization. Ytterbium doped CsPbCl3could be used as a solar cell down converter, because it

can create two photons from one absorbed photon [17]. The emitted photons will have an energy

around 1.3 eV, the difference between the 2_F

7/2 ground state and the 2F5/2 excited state. This

energy matches the band gap of silicon (1.1 eV) very well [12]. With the down converting of the incoming high-energy photons to two photons at a lower energy close to the band gap of silicon, the Shockley-Queisser limit can be broken [12].

1.8 Experiments

In this thesis three performed experiments will be discussed. The first experiment looks at the effects

on the intensity of Yb3+ emission when the halides are changed in CsPb(Cl1–xBrx)3by varying x.

The band gap of the perovskite NC will alter with varying x. This could have an effect on the transfer

between the perovskite host and the doped Yb3+, which could result in decreasing intensity of Yb3+

emission. The second experiment follows a synthesis protocol for doping Yb3+in CsPbCl3proposed

by Mir et al. [18]. The third experiment revolves around the incorporation of Yb3+inside the pores

of a SiO2 matrix together with CsPbCl3 and CsPbBr3 NCs. This method has already proven to

be a stabilizing method for undoped cesium lead halide perovskite NCs [19]. Before discussing the experiments, the results we expect to see according to the literature should be briefly discussed. The experimental set-ups can be found in the appendix.

Photoluminescence and Absorption

Undoped and Yb3+ doped CsPbX3 perovskite NCs will absorb photons with energies above their

band gap. An absorption spectrum can be measured indicating which photons will be absorbed. A small difference in band gap can be observed in the absorption spectrum between undoped and

doped CsPbX3 perovskite NCs. The photoluminescence (PL) is measured by exciting the sample

and monitoring the resulting emission. With our samples we want to have an excitation wavelength that has an energy above the band gap of the perovskite NCs. The idea is then to observe a PL peak from the ytterbium. An excitonic peak could also be observed which will be around the band gap of the perovskite host. Eventually the PLQY of a PL active sample can be measured. The PLQY, together with the absorption and PL will give us a characterization of our sample. An example is given in figure 5.

(11)

��=��% 300 500 600 700 800 wavelength(nm) 0.2 0.4 0.6 0.8 1.0 absorption

Figure 5: Characterization of CsPbI3 NCs in toluene. For this characterization PL, PLQY and

absorption measurements are performed. The results can be presented in one figure. The absorption start around 700 nm and at the same wavelength the PL peak is positioned. The PL is represented in red with arbitrary units (a.u.). The absorbance is represented by the blue line.

(12)

2 Experiment: Halide Exchange

2.1 Introduction

Figure 6: Figure from Akkerman et al. [20]. From parent NC CsPbBr3 halide exchange has been

employed creating different compositions of halides in the perovskite structure. Going from a pure CsPbBr NC to the chloride variant the band gap will increase. Emission from the band gap is shown in the figure, showing the tunability over a wide range.

From figure 6 we see that different combinations of halides in the perovskite structure show emission

at different wavelengths [20]. The band gap of CsPbCl3NCs is around 2.95eV and the band gap of

CsPbI3NCs is around 1.8 eV. If a sample is made purely of NCs with one halide the size distribution

of the individual crystals will determine the centre of the resulting PL peak. Bigger crystals will result in a smaller band gap due to the quantum size effects. With mixing of the halides the band

gap is tunable over the entire region between pure CsPbCl3and CsPbI3, this is shown in figure 6

[20]. In the introduction the energy transfer between Yb3+ and CsPbCl3is discussed. One of the

constraints for this energy transfer is that the band gap of the host perovskite NC should be at least

twice the energy transition of2F7/2 to2F5/2 in order to generate a PLQY of above 100%. For pure

CsPbX3NCs only the variant with X = Cl fulfils this requirement. However a combination of Br and

Cl as halide in the perovskite structure also has a band gap that can comply with this constraint. At

the time the experiment was performed it was not yet clear how Yb3+excitation via other CsPbX3

perovskites NCs than X = Cl works. A decrease in the PL intensity coming from Yb3+is expected

when the band gap energy of CsPb(Cl_1–xBr_x)₃goes below 2.6 eV. During the experiments Kroupa et

al. reported the decrease of Yb3+PL intensity if the band gap energy of perovskite host goes below

the critical band gap energy [12]. The result can be seen in figure 7. From this figure we observe that the NIR PL intensity almost goes to zero. This is interesting because for other lanthanides, without

the quantum-cutting process, clear PL is observed [7]. This almost annihilation of Yb3+emission is

explained by the fact that there is no other efficient transfer mechanism for Yb₃ sensitization in

CsPbX3[12]. The energy mismatch is too large to make energy transfer by multiphonon emission

likely [12]. Resonant and near-resonant energy transfer is also not possible because Yb3+ does not

(13)

Figure 7: Figure from Milstein et al. [15] from parent NC CsPbCl3 halide exchange shows the

tunability of cesium lead halide perovskites.

Another reason, besides the investigation of the energy transfer, for using halide exchange is that

it could be used to create doped samples with another halide. Yb3+ doping in CsPbCl3perovskite

NCs have been well established, but other halides seem to run in more difficulties for doping with

Yb3+. If it is possible to start with pure CsPbCl3NCs that has already been doped, let the halide

exchange happen, and eventually observe Yb3+ emission we can create Yb3+:CsPb(Cl1–xBrx)3NC

samples.

2.2 Experiment

We want to slowly alter the halide starting from pure Yb3+:CsPbCl3monitoring the intensity of the

ytterbium peak and the centre of the excitonic peak. The centre of the excitonic peak will give us a strong indication of where the band gap of our perovskite NCs will be. From previous experiments

we know what the expected band gaps are for different halide compositions in non-doped CsPbX3.

However, due to the introduction of ytterbium the band gap of NCs change. We wanted to use the

procedure proposed by Akkerman et al. for conventional CsPbCl3halide exchange for Yb

3+

:CsPbCl3

NCs. Adding a fixed amount of ODA-Br solution to our sample and check if there were changes in the PL spectrum. The salt did not dissolve well enough, which made the amount of added salt difficult to keep fixed. Eventually, not the solution but the salt itself was added to the sample in order to make sure there was at least something other than solvent added. Measuring absorption will create a more accurate picture of the band gap. Eventually, if it is clear NIR emission is observed, PLQY should also be measured, in order to create a dependency of the PLQY on the band gap of

Yb3+:CsPb(Cl1–xBrx)3 NCs. For the PL measurements in this experiments the Horiba Fluorolog

(14)

2.3 Results

In figure 8 the results of the PL measurements are shown. We observe that there is NIR emission even if the halide changes. This gives an indication that it would be possible to extent this up to the critical band gap. Further altering the halide would have been possible, but the experiment was terminated because of the results already published by Milstein et al.

Figure 8: Emission intensity of Yb3+:CsPb(Cl_1–xBr_x)₃NCs samples. On the top on the x-axis the

excitonic wavelength is displayed. On the bottom this is translated to the expected band gap. We

observe Yb3+ emission for different excitonic emissions. A decrease in intensity is probably not

caused by decreasing band gap, because it is not close enough to the critical band gap of 2.6 eV. All results come from measurements with the same set-up and settings.

2.4 Discussion and Conclusion

In our results we observe a shift in excitonic peak, while Yb3+ emission is still observed. This

means halide exchange in Yb3+:CsPbCl3NC samples is possible, which is confirmed by Kroupa et

(15)

visible far above the critical band gap. This is probably caused by dilution of the sample due to

adding the salts. From other results we know that with decreasing concentration of present Yb3+,

NIR emission will decrease. Eventually to make the halide exchange happen, the amount of solvent was at least doubled, making decrease in NIR intensity plausible. After the result by Milstein

et al. it was not necessary to pursue confirmation of their result. We expect the Yb3+ emission

intensity to rapidly decrease crossing the critical band gap of 2.6eV. However the result that the

PL almost disappears is not necessarily expected. Other lanthanides have been doped in CsPbCl3

and are excited via the perovskite host [13]. Kroupa et al. state that the PL disappears because of energy mismatch and the abundance of upper f − f or charge-transfer excited states at relevant

energies [12]. However Mir et al. show emission for Yb3+:CsPbI3NCs and Yb

3+

:CsPbBr3NCs, see

figure 9 [18]. The emission is significantly lower for CsPbBr₃NCs and CsPbI₃NCs compared with

CsPbCl₃ NCs, but not zero. These NCs have all been doped with the same protocol. It should

be interesting to investigate the resulting energy transferx , when the quantum-cutting process is impossible. This transfer could possibly happen simultaneously with quantum-cutting, although being a slower process [milsteinquantum]. The dependence of the PLQY on the band gap of

Yb3+:CsPb(Cl1–xBrx)3NCs has been measured and does not significantly change up to the critical

band gap [12]. This experiment has been terminated following the result of Kroupa et al. However

(16)

3 Experiment: Post-Synthesis Protocol

3.1 Introduction

For the synthesis of ytterbium doped CsP bX3several methods have been developed. Milstein et al.

synthesized Yb3+:CsPbCl₃ using a hot-injection synthesis method creating NCs with high PLQY

of above 100 % [14]. Within our own group ytterbium doped NCs have been synthesized showing below 100 % PLQY [21]. The synthesis method used previously in our own group is first described

for Yb3+:CsPbCl3in the article by Milstein et al. [14]. The hot-injection method described in the

article is first used in the synthesis of other perovskite based doping: Cs2AgBiX6 (with X = Cl,

Br) elpasolite NCs [14]. There are two reasons to look for different synthesis methods for ytterbium

doping of CsPbX3 (with X = Cl, Br, I). First we want to create samples with PLQY of above 100

% as several papers have already reported [14]. The observed PLQY above 100 % is one of the

important properties of Yb3+:CsPbCl3NCs and before continuing the research on the properties of

Yb3+:CsPbCl3 NCs we want to make sure our samples are comparable with the already reported

samples created by other groups. Secondly, a more practical reason is that an easier synthesis procedure would ensure more research possibilities, making more scientific discoveries possible. In

2018 Mir et al. proposed a post-synthesis procedure of Mn and Yb doping of CsPbX3(X = Cl,Br,

I) [18]. The ytterbium doped samples are characterized without a result of the PLQY. Interesting is to see whether it is possible to use this new post-synthesis protocol to dope perovskite NCs with ytterbium and break the 100 % PLQY barrier.

In the article by Mir et al. manganese doping in CsPbX3is also proposed, showing emission around

590 nm [18]. We are solely looking at ytterbium doping of CsPbX3it is however also possible to dope

manganese with the same protocol. The results of the photoluminescence of the created samples by Mir et al. are promising and displayed in figure 9.

Figure 9: Result from Mir et al. [18]. The figure shows the concentration Yb3+inside the perovskite

NCs. PL is displayed of the perovskite NC host for doped and undoped samples. Varying the halide and percentage Yb results in a different intensity of the Yb peak. Interesting to see is that there also is a Yb peak with CsPbI, which we so far have not observed.

(17)

3.2 Methods

Chemicals

Methyl acetate (MeOAC anhydrous, 99,5 %, Sigma-Aldrich), Ytterbium(III) nitrate pentahydrate

(Yb(NO3)3· 5 H2O, 99.9 %, Sigma-Aldrich), hexane (Sigma-Aldrich), toluene (ACS reagent, ≥ 99.5

%, Sigma-Aldrich), all chemicals are used as ordered with no further purification. For chemicals

used to synthesize CsPbCl3NCs, see the appendix.

Synthesis

We are following the synthesis procedure for Yb:CsPbCl3from the article by Mir et al. [18]. Because

we want to maximize our PLQY, we want to vary the amount of perovskite NCs and Yb precursor. The synthesis of used NCs can be found in the appendix. The following steps are taken to dope the

CsPbCl3NCs with Yb:

1. Dissolve Yb(NO3)3· 5 H2O in a solution of methyl-acetate MeOAc and toluene in ratio 1:3. For

this experiment 3 ml toluene and 9 ml MeOAc was used for the precursor. Let this stir to a homogeneous mix to finalize the precursor.

2. Distribute undoped CsPbCl3NCs in different vials.

3. Add Yb(NO3)3· 5 H2O precursor to the vials. Vary the amount of precursor to ensure different

amount of Yb doping. The amount of Yb(NO3)3· 5 H2O in the precursor is varied. The amount

of precursor added is alos varied which could result in eventual overlap of added ytterbium. Stir the precursor and NCs for 1 minute.

4. Wash the product NCs with MeOAc as anti-solvent in order to precipitate the doped NCs. 5. Centrifuge at 6000 rpm for 5 minutes.

6. The wet pellet obtained after centrifugation should be our doped NCs. Redisperse the wet pellet in hexane.

For the samples the amount of undoped CsPbCl3NCs, amount of precursor and amount of Yb(NO3)3·

5 H2O in the precursor are varied. The successful doping was obtained using 6ml of a precursor with

0.150g Yb(NO3)3· 5 H2O dissolved in 3 ml MeOAc and 9 ml toluene. In order to let the wet pellet

redisperse the sample was held in a sonic bath for 30 minutes. Experiments

After the doping of the NCs we need to characterize the new samples. We follow our general

characterization procedure, where we want to know the following properties: absorption, PL and PLQY. For this measurement the Horiba Fluorolog was used to create a PL spectrum.

3.3 Results

From all attempts only one sample was successfully redispersed and showed PL. In figure 10 emission from this sample is presented. The sample is measured in the Horiba Fluorolog. From the figures

we see a fairly broad peak coming from Yb3+. The peak has three distinct peaks probably coming

(18)

causing a slight shift in emission.

Figure 10: Emission intensity CsPbCl3 is shown in the figure on the top. Emission intensity from

Yb3+ is shown in the figure on the bottom. Lanthanide emission is characterized by the narrow

peak, here however we notice a fairly broad peak. This probably comes from hyperfine splitting or close defect states. In the figure on the top we see an intense peak from the xenon lamp and around

415 nm excitonic emission is observed corresponding with the band gap of CsPbCl₃perovskite NCs

Looking at the contour plot displayed in figure 11 we can get the strong indication that Yb3+is

indeed excited via the perovskite host. We observe a decrease when nearing the band gap of CsPbCl3

NCs. Further insight should be given by making a cross section at an emission wavelength around 980 nm. A PLQY measurement has been attempted and seems to result in a PLQY of around 15%.

(19)

This result should be evaluated with some considerations. The measurement was performed later than the PL measurement where already some part of the sample was evaporated.

Figure 11: Contour plot for Yb3+:CsPbCl3. The emission wavelength (nm) and excitation

wave-length (nm) are shown on the x-axis and y-axis respectively. The decay in intensity is clearly visible

when the excitation wavelength goes above 400 nm. This is an indication Yb3+ is exited via the

perovskite host.

3.4 Discussion & Conclusion

From the results we can conclude that it is possible to dope CsPbCl₃ perovskite NCs using the

proposed post-synthesis protocol by Mir et al. [18]. The resulting PL shown in figure 10 however is different from the result from Mir et al.. Our results clearly shows distinct peaks coming from

hyperfine splitting or close defect states. This could be an indication of incorporation of Yb3+within

the perovskite structure. However, absorption has not been measured due to the quick precipitation of the sample. The measured absorption spectrum was almost identical to the absorption spectrum of the hexane reference sample. This can be explained by precipitation of all NCs, or by a very low absorption spectrum which should result in a high PLQY. The PLQY was measured to be 15%, which is low. Another problem is the quick evaporation. Proper closing of the cuvettes might be a solution but even then the sample evaporates quickly. The PLQY is significantly lower than the high PLQYs reported in literature. The success rate of the doping procedure also provides a challenge for further employment of this procedure. Up to now only one sample was successfully doped. To redisperse the resulting NCs in hexane remained a problem. To move on with this procedure these issues should be solved and an optimum for the amount of added ytterbium could be found. This experiment should be seen as a basis for further experiments. So if this procedure should be optimized, depends mainly on the question whether research on these kind of doped NCs should be continued.

(20)

4 Experiment: Encapsulation of Lead Inorganic Halide

Per-ovskites in SiO

₂

with Ytterbium

4.1 Introduction

In the previous experiments CsPbX3NCs were doped with ytterbium and redispersed in a solvent.

In this chapter we will look at ytterbium doping of CsPbX3 in a mesoporous silica matrix. The

colloidal cesium lead halide perovskites are not stable enough to be used on a commercial basis [22]. To increase the stability the growth of perovskite NCs within the pores of mesoporous matrixes was proposed [19]. The growth of undoped perovskite NCs within a silica matrix is a ligand-free synthesis method [19]. Dirin et al. shows that the infiltration of precursor solution of lead halide

perovskite NCs within the pores of mesoporous SiO2followed by drying, results in a sample with

bright PL and quantum efficiencies above 50 % [19]. The resulting PL is dependent on the size of the pores [19]. With increasing pore size a red shift is observed [19]. The change in PL spectra comes from the difference in band gaps from the different sized perovskite NCs. The difference in sizes of the NCs then comes from the difference in pore sizes. Bigger pore sizes have room for bigger NCs, hence the difference in resulting PL spectra. The pore size distribution can be chosen beforehand and therefore be used to influence the resulting NC size distribution. A different method

for stabilizing perovskite NCs is outer shell coating [22]. Liao et al. researched CsPbBr₃

nanocrys-tal/MO₂ (M = Si, Ti, Sn composites) [22]. More insight is needed because previously conducted

experiments were done with electric insulators as coating materials, making it difficult to use in optoelectronic applications [22]. Both the encapsulation and the mesoporous silica matrix method improve the stability of the perovskite NCs [sio2liao , 19]. Several composites are explored and

SiO₂comes out as the most promising encapsulating material [22]. Besides CsPbBr₃CsPbBr₃ xClx

is encapsulated with SiO2. Encapsulated CsPbBr3 xClx also shows a PL peak corresponding with

the band gap [22]. We want to look at the possibility of introducing ytterbium in the silica matrix and encapsulating ytterbium doped NCs. Introducing ytterbium in the silica matrix together with perovskite NCs is not a doping method, the ytterbium is not incorperated within the perovskite NCs. However, it is possible that with the ytterbium in the same pore as our perovskite NCs en-ergy transfer could still happen. Besides an interesting new method of combining cesium lead halide perovskite NCs and ytterbium it could also be used as a stabilizing method of Yb doped perovskites.

4.2 Methods

Chemicals

Cesium(I) carbonate (Cs2CO3, 99.9 %, Sigma-Aldrich), octadecene (ODE, 90 %, Sigma-Aldrich),

oleic acid (OA, 90 %, Sigma-Aldrich), oleylamine (OLA, 80–90 %, Acros), Lead(II) bromide (PbBr2,

98 %, Sigma-Aldrich), Lead(II) iodide (PbI2, 98 %, Sigma-Aldrich), Lead(II) chloride (PbCl2, 98

%, Sigma-Aldrich), hexane (Sigma-Aldrich), toluene (ACS reagent, ≥99.5 %, Sigma-Aldrich),

meso-porous silica gel (SiO2, high purity grade, pore size 15 nm, 200-425 mesh, Sigma-Aldrich), methanol

(CH3OH, Sigma-Aldrich), N-methylformamide (MFA, 99 %, Sigma-Aldrich), methyl acetate

(an-hydrous, 99.5 %, Sigma-Aldrich), trioctylphosphine (TOP, 97 %, Sigma-Aldrich), ytterbium(III)

acetate Yb(AC)3, 99.95 %, Aldrich). Cesium(I) bromide (CsBr, anhydrous, 99 %,

(21)

Synthesis

The used CsPbCl3 and CsPbBr3 NCs are synthesized as described in the appendix. The idea is

to see whether it is possible to excite ytterbium via CsPbBr3 or CsPbCl3 if it was added in the

pores of mesoporous silica. We want to see if this is possible via physical adsorption. Synthesis was

performed by a bachelor student coming from Brazil for a short project: Beatriz Mouri˜no.

1. Add 0.76g SiO₂(with a pore size of 15nm) 1.25 g of Yb (III) hydrated acetate (3.7 mmol) and

23 mL of methanol/water 1:1 to a flask

2. Close the flask and stir it for about 20h at room temperature 3. Filtrate the sample

4. Wash the sample using methanol and water

5. Let the sample dry for 24h at 60◦C

6. 50 mg of each mesoporous silica sample, with and without Ytterbium (reference) are dried for

20 h at 150◦C

7. 800 µL of previously synthesized CsPbBr3 NCs or CsPbCl3 NCs were incorporated to the

material in point 6 by adsoprtion.

For CsPbCl₃the same synthesis procedure was followed, this synthesis was performed by dr. L.

G´omez.

Measurements

For the previous experiments the sample NCs were dissolved in either hexane or toluene and could be measured in a cuvette. Because the resulting sample is a powder a different container is needed. The sample holder is shown in figure 12.

Figure 12: Sample holder for ytterbium with either CsPbBr3or CsPbCl3in the pores of mesoporous

(22)

An important difference between this sample holder and a cuvette is that the holder itself does not transmit light. All the incoming and outgoing light should go through the cover glass. This means that absorption and PLQY can not be measured using this sample holder. It could be possible to do this using for instance very narrow cuvettes, which at least should make it possible to measure the PLQY of the sample.

For the sample with CsPbBr3a laser is used to excite the NCs, where for the sample with CsPbCl3

a xenon lamp is used as excitation source. The reason behind this difference is simply that the

set-up with the xenon lamp was broken at the time that the measurement of the CsPbBr3 sample

was performed.

4.3 Results

In figure 13 the result from the PL measurement on a sample with CsPbBr3as perovskite NC. This

sample was prepared in situ. No emission from Yb3+ is observed. In figure 14 the same

measure-ment is performed with exactly the same set-up and configurations on CsPbBr3in SiO2, but without

any Yb3+as reference. Again a clear peak coming from the perovskite is observed, but clearly no

emission from where we would expect ytterbium emission as this is the reference sample with no ytterbium. The difference in resulting intensity can be explained by the fact that the laser had to be pointed at the sample. This was done with adjusting mirrors. The laser can be better adjusted

sometimes and more directly point inside the sample. Yb3+is more likely to get excited via CsPbCl3

perovskite NCs in comparison with CsPbBr₃[12]. For this reason we switched to CsPbCl₃together

with Yb3+.

Figure 13: Photoluminescence of CsPbBr₃with Yb3+in SiO₂under laser excitation of wavelength

450 nm. The narrow peak at 450 nm corresponds with the laser. The broad peak with center around

525 nm comes from CsPbBr₃NCs. No emission from Yb is observed.

In figure 15 the resulting PL of excitation at 365 nm of CsPbCl3with Yb

3+

(23)

Figure 14: Photoluminescence of CsPbBr3without Yb 3+

in SiO2under laser excitation of wavelength

450 nm. The narrow peak at 450 nm corresponds with the laser. The broad peak with center around

525 nm comes from CsPbBr3NCs. No emission from Yb is observed.

(24)

Figure 15: Photoluminescence of CsPbCl3 with Yb 3+

in SiO2 under excitation of xenon lamp.

Excitation wavelength at 365 nm. Emission peak from Yb3+visible at 980 nm.

From this result it is clear that ytterbium is excited. The contour plot, displayed in figure 16,

can tell us more about the way Yb3+ is excited. In figure 16 we observe that intensity abruptly

drops if excitation light goes beyond 350 nm. Interestingly, the emission is still observed at 980 nm with excitation light above 450 nm. In figure 18 a cross-section of figure 16 is displayed. This shows

that Yb3+ emission happens when the excitation light goes above the band gap of CsPbCl3NCs,

(25)

Figure 16: Contour plot of CsPbCl₃with Yb3+in SiO₂. On the x-axis the excitation wavelength (nm) is displayed and on the y-axis emission wavelength (nm). The result tells us where the ytterbium PL peak has the highest intensity and where the intensity starts to decay. We see that that the intensity decreases when excitation light goes above 350 nm. This is earlier than expected. The

Yb3+emission continues even beyond 450 nm.

We can also compare the intensities of these ytterbium PL peaks for different excitation wave-lengths. This comes from another cross section of figure 16. We see that there indeed is a significant difference in intensity if we excite with wavelengths above 350 nm. The point where the decay

in Yb3+ emission starts is difficult to directly pin-point. If we compare the start of the decay in

intensity of Yb3+ emission from figure 16 with the contour plot from figure 11 we observe that the

(26)

Figure 17: Cross section of figure 16 with excitation light of 452 nm and 324 nm. An Yb3+emission

peak is observed for both excitation wavelengths. Yb3+ emission intensity is clearly higher for

excitation above the band gap. There is however still Yb3+emission with excitation energies below

the band gap.

Figure 18: Cross section of figure 16 with excitation light of 452 nm and 324 nm. On the x-axis

the emission wavelength is displayed in nm. On the y-axis the counts in a.u is displayed. A Yb3+

emission peak can be observed for both excitations. Yb3+ emission intensity is significantly lower

for excitation of 452 nm but is still observable. It is not yet clear how Yb3+is excited at excitation

wavelength of 452 nm.

4.4 Conclusion

(27)

of the PL peak coming from Yb3+when the energy of the excitation light goes below the band gap

energy of CsPbCl3NCs we can conclude that Yb3+is, at least mainly, excited via the perovskite NCs.

However Yb3+emission is still observed when the energies of the excitation light is below the band

gap of the NCs. It is not clear how this excitation happens. Looking at the the shape and comparing that with the result coming from the sample from the post-synthesis protocol sample a difference is observed. The results from incorporated NCs in a silica matrix do not show the distinct peaks observed in figure 10. An explanation for these distinct peaks comes from the hyperfine splitting

which could result from the incorporation of Yb3+in the perovskite structure. This means that the

fact that we do not see these distinct peaks might be an indication that Yb3+ is not incorporated

in the structure, but is close to the NCs. It is difficult to give a conclusive answer to the question whether ytterbium is outside the NCs or incorporated. To answer that question further research should be developed, where it will still be a difficult question to answer, without being able to look directly at the resulting structure.. We do note however that with the ytterbium in or out

the perovskite structure some transfer mechanism happens between Yb3+ and CsPbCl3. No peak

corresponding with Yb3+is observed if it is put in the pores of SiO2together with CsPbBr3. Mir et

al. has shown that it is possible to excite Yb3+via a CsPbBr3NC host, without being incorporated

inside a silica matrix. Possibly the circumstances for transfer between Yb3+ and CsPbBr3 NCs

within SiO2 is not optimized yet, so we can not yet conclude that it is impossible to make this

excitation happen.

5 Summary and Outlook

5.1 Summary

This thesis started with a literature study on ytterbium doped cesium lead halide perovskite NCs. Three different experiments with these NCs have been described. In the previous year several articles on this subject were published which answered several questions we asked ourselves at the beginning of the project. The results we were looking for in the halide exchange experiment was published a month after starting our own experiment. In the introduction the general proposed energy transfer mechanism in ytterbium doped perovskite NCs is discussed which is still not completely proven.

Mir et al. doped CsPbI3 NCs with Yb3+ and got a signal at the expected wavelength for Yb3+

emission [18]. This result is shown in figure 9 and seems to contradict the proposed energy trans-fer mechanism. Milstein et al. have shown that with the excitonic peak around 480 nm the Yb peaks intensity rapidly decreases [12]. With the excitonic peak above 520 nm the Yb peak vanishes

[12]. This result contradicts with the Yb peak in the CsPbI3NC in the study of Mir et al. This

is potentially interesting because it might tell us more about the energy transfer mechanism. This transfer mechanism is something we wanted to know more about and the post-synthesis protocol was developed in order to support this research. This protocol itself will not tell us more about the

energy transfer, but creating a quick and easy way to dope CsPbCl3NCs with ytterbium would make

it possible to generate many experiments with these NCs. These experiments can then be focussed on the energy transfer. In the end no Yb peak was visible with two different photoluminescence set-ups. This can be explained by the fact that after centrifugation the wet pellet did not redisperse in hexane, making it impossible to measure the sample. Both emission from the perovskite as from

Yb3+was not observed. However at the end of the project a clear Yb3+emission peak was visible.

This means that this protocol has the potential also for our group to be used in the future to create

ytterbium doped CsPbCl3NCs. There is still a need for optimization however in order to ensure that

the success rate of the protocol rises. Another project was started with the mesoporous SiO2matrix

(28)

the NCs. An ytterbium emission peak is measured using CsPbCl3as NCs. The measurements that

are done with ytterbium doped CsPbBr3NCs show a bright perovskite PL peak, but no ytterbium

emission. The improved stability is already proven for cesium lead halide perovskite NCs inside the

pores of SiO2without ytterbium[19].

Looking at the results from Kroupa et al., which are discussed in the chapter on the halide

exchange experiment, PL intensity in the NIR region drops to zero for CsPbBr3hosts [12]. Although

it is possible to transfer energy to various lanthanides using CsPbX3 as hosts, the combination of

Yb3+ and CsPbI3 and CsPbBr3 seems to be challenging for NIR emission [12]. From the result

of the post-synthesis protocol we get a strong indication that Yb3+ is indeed inside the perovskite

structure, an idea for further research is putting these doped NCs inside the pores of the silica matrix. Checking beforehand if the NCs still show the distinct peaks and comparing this result with

the peak when the NCs are inside the pores. If already doped CsPbCl3NCs inside the matrix still

show the distinct peaks it might be a strong indication that Yb3+is not included in the perovskite

structure in our experiment. Furthermore a method for measuring PLQY and absorption for the

sample should be developed in order to create a clearer picture of the physical properties of Yb3+

together with CsPbCl3NCs inside SiO2.

5.2 Outlook

The main reason of proposing a novel synthesis method for introducing Yb in perovskite NCs was to find out more about this interesting class of materials. Although already in the previous year several papers with answers on question we had at the beginning of the project have been published, there will always be new opportunities and possibilities for research. The fact that there is a contradiction between two papers creates an interesting possibility for further research. Before this research can be done it is important to get Yb doped NCs with PLQY exceeding 100 % in order to compare it with other studies and making it an interesting material for photovoltaic devices. In our NCs lead is still used. Lead has one big problem: it is poisonous. Because we are talking about a material that shows promising characteristics to be used as a down converter of a solar cell it is important to stabilize and not use poisonous materials. Bismuth based halide perovskites are suggested to be used this way. If it would be possible to combine Yb doped Bismuth based perovskites with for

instance SiO2mesoporous materials an interesting new material arises. This would make it possible

(29)

6 Samenvatting

In 2004 werd door Richard Smalley, Nobelprijswinnaar in de scheikunde, een uitdaging geformuleerd: de terawatt challenge. Deze uitdaging houdt in dat in 2050 60 terawatt aan energie geproduceerd moet worden om de gehele mensheid te voorzien. Belangrijk hierbij is de manier waarop de en-ergie opgewekt wordt. De effecten van klimaatverandering worden steeds duidelijker, waardoor de noodzaak groeit om op een milieuvriendelijke manier energie op te wekken. Gelukkig voor ons is er een grote energiebron in de lucht: de zon. Om energie hiervan te kunnen gebruiken moeten manieren

gevonden worden om deze energie te converteren. Met de effici¨entie van huidige zonnecellen zouden

we slechts een deel van de Sahara hoeven te vullen om genoeg energie op te wekken om te voldoen aan de terrawatt challenge. Echter is het probleem hier dat er nog geen manier is om energie te

transporteren of op te slaan die effici¨ent genoeg is om hiermee ook daadwerkelijk de mensheid in

zijn energiebehoefte te voorzien. Energie zal dus, tot de uitvinding van supergeleiding op

kamertem-peratuur, lokaal opgewekt moeten worden. Zonnecellen die op dit moment op commerci¨ele schaal

geproduceerd worden, gebruiken slechts een deel van het zonnespectrum. In landen dichtbij de eve-naar met veel zon is dit geen probleem. Echter, als we verder van de eveeve-naar afgaan moeten we zuiniger met de hoevelheid zon omgaan; we willen zo veel mogelijk van het zonnespectrum gaan gebruiken. De fotonen die worden opgevangen hebben een verschillende energie. Niet alle energie van deze fotonen kan worden omgezet, een groot deel gaat verloren aan warmte. Dit heeft te maken met de zogenaamde band gap (in het Nederlands: ”bandkloof” of ”verboden zone”). Deze band gap bepaalt hoeveel energie een foton op kan leveren, alle energie boven deze band gap gaat verloren.

Bij sommige materialen is het mogelijk om ´e´en foton met hoge energie te veranderen in twee fotonen

met lagere energi¨en, waardoor er uiteindelijk minder energie verloren gaat. Dit is wat er gebeurd

als we het element ytterbium introduceren in een perovskiet materiaal: CsPbX₃. Het gaat hier om

pervoskiete nanokristallen. Het voordeel van perovskiete materialen is dat ze veel voorkomen en goedkoop zijn om te produceren. Een ander voordeel is dat het mogelijk is een zeer dunne zonnecel te maken die vouwbaar is. Dit kan gebruikt worden voor zonnecelen op zonneschermen of bijvoor-beeld een draagbare zonnecel. In mijn onderzoek ben ik gaan kijken naar een andere manier om

ytterbium te intdouceren in CSPbX3gebaseerd op een artikel gepubliceerd aan het begin van mijn

onderzoek. Als we ytterbium en CsPbX3 nanokristallen combineren is het mogelijk om ytterbium

te exciteren via de CsPbX3 nanokristallen. Deze procedure is uiteindelijk succesvol gebleken. Een

nieuwe methode voor de combinatie van deze materialen is door de ytterbium niet op te nemen in

de structuur van de nanokristallen maar het samen met CsPbX3 nanokristallen op te nemen in de

pori¨en van een ander materiaal: SiO2. We hebben aangetoond dat het mogelijk is om ook op deze

manier de ytterbium te exciteren via de nanokristallen. In de toekomst zou het samen opnemen van deze materialen ook kunnen worden gebruikt met andere stoffen. Onze nanokristallen bevatten de giftige stof lood. Een mogelijkheid is om lood te vervangen met een ander element: bismut. Dit element werd vroeger zelfs soms verward met lood omdat het veel overeenkomstige eigenschappen heeft. Echter is bismut ongevaarlijk en wordt het bijvoorbeeld in cosmetische producten gebruikt. Het is nog onduidelijk of het mogelijk is om ytterbium te exciteren via een perovskiet nanokristal met bismut op de plek van lood. Het is wel een zeer interessant onderwerp voor vervolgonderzoek.

(30)

7 Appendix

In this appendix the used experimental set-ups and the synthesis of our perovskite NCs will be dis-cussed. The theoretical background of the measurements are already discussed in the introduction.

7.1 Synthesis of CsPbX

₃

Nanocrystals

Synthesis of CsPbX3 with X = Cl, Br, I. Synthesis is performed as reported before by our group

and strictly follows the synthesis procedure proposed by the Kovalenko group. [21, 7]. Chemicals

octadecene (ODE, 90%,Sigma-Aldrich), oleic acid (OA, 90%, Sigma-Aldrich), oleylamine (OLA,

80–90%, Acros), Lead(II) bromide (PbBr₂ , 98%, Sigma-Aldrich), Lead(II) chloride (PbCl₂, 98%,

Sigma-Aldrich), hexane (Sigma-Aldrich), toluene (ACS reagent, ≥99.5%, Sigma-Aldrich) Synthesis Procedure

Synthesis was performed by dr. L. Gom´ez and follows the synthesis as described in the supplementary

information from the article by Protesescu et al. [7].

5mL ODE and 0.188 mmol PbX2, with in our case X = Br or Cl, were loaded into 25ml 3-neck

flask and dried under vacuum for 1h at 120◦C. Dried oleylamine (0.5ml) and dried OA (5ml) were

injected at 120◦C under N2. After the PbX2salt is completely dissolved, the temperature was raised

to 140 − 200◦_{C and prepared Cs-oleate solution was quickly injected and, 5s later, the reaction}

mixture was cooled by an ice-water bath. To completely solubilize PbCl₂ a higher temperature is

needed: 150◦C. [7].

The supernatent can be dissolved in a chosen solvent, in our case for the post-synthesis protocol hexane and for the halide exchange experiment in toluene.

7.2 Experimentals

The samples are measured in quartz cuvettes. Except for the experiment with mesoporous SiO2

where we used holders as shown in figure 12. Photoluminescence

For measuring PL we want to excite the sample and measure the emission. A practical problem is that the excitation light intensity will be higher than the emission light intensity. This means that it is possible that we are not able to see the emission light in our data. Therefore the emission is

measured under an angle of 90◦. Emission can be measured with the PLQY set-up, but for emission

spectra the Horriba Fluorolog is best to be used. Because this was broken down for a significant amount of time a second set-up for measuring PL was created. The second set-up made use of laser for excitation, making it possible to excite with a higher intensity. The downside of the PL set-up with the laser is that a measurement with different excitation wavelengths is more difficult compared with the Fluorolog.

(31)

Figure 19: On the left a schematic overview of the PL set-up with the Horiba Fluorolog is displayed. A xenon lamp is used as a light source which enter the excitation monochromator. The excitation monochromator select the excitation light which then excites the sample. The emission light goes to the emission monochromator under a right angle relative to the excitation light to prevent direct excitation light on the emission monochromator. The emission monochromator selects the chosen wavelengths which than can be detected by the desired detector. On the right the other PL set-up is displayed. This set-up uses a laser as light source. The chosen wavelength comes out of the laser and travels via mirrors and through a filter to the sample holder. Here a sample is placed which is excited under a right angle. The emission light travels through a fiber-optic cable to an emission monochromator. The desired wavelengths are selected and can then be detected by the detector.

Quantum yield

The PLQY is defined as: Φ = _{#photons absorbed}#photons emitted. It is therefore important to acquire all the absorbed

and emitted photons. If we look at the PL measurements, some of the emitted and absorbed

photons will not be measured due to scattering. To include all the scattered photons in our data, an integrated sphere is used. The integrated sphere’s walls are made of highly reflective materials, ensuring the continuous scattering of photons. Eventually only a small fraction of all photons will be measured, it is however a good representation of the distribution of all photons coming from: emission, excitation, reflection and scattering. The outcome of the measurement is a number of counts for every wavelength. integrating the curve gives us the number of photons emitted and absorbed: P LQY = Nem Nabs = R Isample(λ) − Iref(λ)dλ R Eref(λ) − Esample(λ)dλ Absorption

To measure the absorbed photons we have to measure the amount of photons that disappear after going through the sample. The samples are dissolved in either hexane or toluene, which both also absorb. Therefore we compare a sample with the NCs with a cuvette filled only with the solvent. For the absorption measurements a Lamba 950 UV/Vis Spectrophotometer from PerkinElmer was used.

(32)

Figure 20: On the left is a schematic overview of the used quantum yield set-up. A xenon lamp is used as excitation source, when it goes through an excitation monochromator which selects the used excitation wavelength. The light then hits the sample inside the integrated sphere which is displayed in more detail on the right. An emission monochromator selects the chosen wavelengths which the detector then detects. All light travels via fiber-optic cables between the different devices. When the light enters the integrating sphere both the excitation and emission light is reflected several times via the walls of the integrated sphere resulting in uniformly distributed light over the area of the integrated sphere. Eventually only a small fraction of the light exits the sphere via the output.

This device measures the transmittance. With the results from the measurements the absorbance can be calculated. Scattering and reflection are not taken into account.

7.3 Overview Important Articles on Yb Doping of CsPbX

3

In this part of the appendix a brief overview will be given of published articles on the Yb doping of

CsPbX3. Only the important results will be represented here, for further details the articles should

be consulted. The articles are chronologically sorted.

Doping Lanthanide into Perovskite Nanocrystals: Highly Improved and Expanded Op-tical Properties

Published on: November 28, 2017. In Nano Lett. 2017, 17 8005-8011 DOI: https://doi.org/10.1021/acs.nanolett.7b04575

Lanthanide doping in perovskite NCs is possible via a hot-injection method. Lanthanides show emission at excitation wavelength of 365 nm. Excitation of lanthanides goes via perovskite NCs. Excitonic emission goes up with atomic number of the lanthanide. PLQY of above 100 % is possible for Yb doped perovskites.

(33)

Picosecond Quantum Cutting Generates Photoluminescence Quantum Yields over 100

% in Ytterbium-Doped CsPbCl3

Published on: May 10, 2018. In Nano. Lett. 2018, 18, 3792-3799 DOI: https://doi.org/10.1021/acs.nanolett.8b01066

Low excitonic emission in Yb:CsPbCl₃NCs with high PLQY (170 %). Increasing the amount of

Yb will increase the intensity of the PL peak of Yb. Energy transfer to Yb3+ from the perovskite

NC happens on an ultrafast timescale of 1 ps.

Quantum-Cutting Ytterbium-Doped CsPb(Cl1–xBrx)3Perovskite Thin Films with

Pho-toluminescence Quantum Yields over 190%

Published on: September 11, 2018. In ACS Energy Lett. 2018, 3, 2390-2395. DOI: https://doi.org/10.1021/acsenergylett.8b01528

PLQY over 190 % measured in thin films of CsPb(Cl_1–xBr_x)₃ showing that high PLQY and

quantum-cutting is an intrinsic effect of the material and not a NC effect. Yb emission drops with

excitonic peaks above 520 nm. PLQY of CsPb(Cl1–xBrx)3 decreases with increasing fluence, with

PLQY below 100 % as the fluence goes above 1015_cm−2_s−1_.

Anion Exchange and the Quantum-Cutting Energy Threshold in Ytterbium-Doped

CsPb(Cl1–xBrx)3Perovskite Nanocrystals

Published on: January 29, 2019. Nano Lett. 2019, 19, 3, 1931-1937 DOI: https://doi.org/10.1021/acs.nanolett.8b05104

Anion exchange is used to investigate the energy transfer between Ytterbium and CsPb(Cl_1–xBr_x)₃.

As in the article on perovskite thin films a rapid decrease in Yb PL intensity is noted when exciton

PL wavelength goes to two time the energy transition Ef–f. Research is done with perovskite NCs.

The absorption range is tunable over a large enough range for practical applications as solar cells.

Postsynthesis Doping of Mn and Yb into CsPbX3(X = Cl, Br, I) Perovskite

Nanocrys-tals for Downconversion Emission

Published on: October 25, 2018. In Chem. Mater. 2018, 30, 8170-8178 DOI: https://doi.org/10.1021/acs.chemmater.8b03066

New protocol for doping Yb3+into CsPbX₃(X = Cl, Br, I). Article focuses on doping of Mn2+

into CsPbX3. Measures a Yb PL peak for Yb:CsPbI3. Mentions nothing about PLQY measurements.

Photoluminescence Saturation in Quantum-Cutting Yb3+-Doped CsPb(Cl1–xBrx)3

Per-ovskite Nanocrystals: Implications for Solar Downconversion

Published on: March 19, 2019. In J. Phys. Chem. C 2019, 123, 12474-12484 DOI: https:

//doi.org/10.1021/acs.jpcc.9b01296 Yb3+-Doped CsPb(Cl1–xBrx)3NCs saturate under modes

photoexcitation fluences which is problematic for practical applications. Investigates the reasons behind these saturation and suggests strategies for solving this problem. One is demonstrated in the

(34)

to achieve microscopic insights in order to develop a picture on why saturation happens in these materials.

Ytterbium Doping of Cesium Lead Halide Perovskites Nanocrystals

Universiteit van Amsterdam