Angus G. M.
Mathieson
abc,
Wesley M.
Dose
ad,
Hans-Georg
Steinrück
ef,
Christopher J.
Takacs
f,
Sascha
Feldmann
bg,
Raj
Pandya
b,
Alice J.
Merryweather
bd,
David
Mackanic
f,
Akshay
Rao
b,
Felix
Deschler
*bh and
Michael
De Volder
*a
aInstitute for Manufacturing, Department of Engineering, University of Cambridge, 17 Charles Babbage Rd, Cambridge CB3 0FS, UK. E-mail: mfld2@cam.ac.uk
bCavendish Laboratory, Department of Physics, University of Cambridge, 17 JJ Thomson Ave, Cambridge, CB3 0HE, UK
cCambridge Graphene Centre, Department of Engineering, University of Cambridge, 9 JJ Thomson Ave, Cambridge, CB3 0HE, UK
dYusuf Hamied Department of Chemistry, University of Cambridge, Lensfield Road, Cambridge, CB2 1EW, UK
eDepartment Chemie, Universität Paderborn, 33098 Paderborn, Germany
fSSRL Materials Science Division, SLAC National Accelerator Laboratory, Menlo Park, California 94025, USA
gRowland Institute, Harvard University, Cambridge, Massachusetts 02142, USA
hPhysikalisch-Chemisches Institut, Universität Heidelberg, Im Neuenheimer Feld 229, 69120 Heidelberg. E-mail: deschler@uni-heidelberg.de
First published on 6th September 2022
Doping halide perovskites (HPs) with extrinsic species, such as alkali metal ions, plays a critical, albeit often elusive role in optimising optoelectronic devices. Here, we use solid state lithium ion battery inspired devices with a polyethylene oxide-based polymer electrolyte to dope HPs controllably with lithium ions. We perform a suite of operando material analysis techniques while dynamically varying Li doping concentrations. We determine and quantify three doping regimes; a safe regime, with doping concentrations of <1020 cm−3 (2% Li:Pb mol%) in which the HP may be modified without detrimental effect to its structure; a minor decomposition regime, in which the HP is partially transformed but remains the dominant species; and a major decomposition regime in which the perovskite is superseded by new phases. We provide a mechanistic description of the processes mediating between each stage and find evidence for metallic Pb(0), LiBr and LiPbBr2 as final decomposition products. Combining results from synchrotron X-ray diffraction measurements with in situ photoluminescence and optical reflection microscopy studies, we distinguish the influences of free charge carriers and intercalated lithium independently. We find that the charge density is equally as important as the geometric considerations of the dopant species and thereby provide a quantitative framework upon which the future design of doped-perovskite energy devices should be based.
Broader contextHybrid perovskites are arguably one of the most promising material systems for the next generation of energy devices, with reports of record-breaking energy conversion efficiencies occurring on an almost yearly basis. However, unlike traditional semiconductors such as silicon and GaAs, the impact of external dopants in these materials remains poorly understood. There currently exist relatively few reliable methods with which to insert extrinsic species into the perovskite lattice and an experimental model describing the impact on the perovskite structure has remained somewhat elusive. This work represents a breakthrough in understanding what happens to hybrid perovskites when Li-ions are added into the lattice at dopant concentrations that span three orders of magnitude. This is achieved by a battery-inspired device architecture, which allows one to vary the dopant concentration in a perovskite semiconductor, while simultaneously facilitating operando synchrotron XRD and optical analysis. The findings of this work have general implications in both the energy conversion and energy storage communities by quantifying three doping thresholds in hybrid perovskites, within which the doping can be achieved with either reversible or irreversible changes to the perovskite structure. |
The perovskite class of materials now encompasses a great variety of structures and chemical compositions. Methylammonium (MA) lead iodide (CH3NH3PbI3) is perhaps the most commonly recognised composition and engenders the standard three-dimensional (3D) perovskite ABX3 structure (unit cell as shown inset in Fig. 1(a)). In this instance, A and B represent primary and secondary cations respectively with the A cation generally being a monovalent organic cation, such as MA or formadinium, and B a divalent inorganic cation – oftentimes lead, but also tin or germanium. X represents the halide anion species, typically iodide, bromide, chloride (I−, Br−, Cl−) or mixtures thereof. Compositional engineering allows for the design and synthesis of a plethora of materials, which despite sharing a common structure, can vary significantly in physical properties and applications. The ability to change the absorption coefficient, optical bandgap (over an extensive range of wavelengths extending beyond the visible),3,32–34 and improve stability against various environmental species such as water and oxygen35–37 through a simple alteration of the elemental composition is one of the characteristics of the perovskite portfolio making them so effective in a great variety of disciplines.
Furthermore, by inserting a primary A cation that is too large to fit directly within the bulk perovskite crystal structure, it is possible to form a quasi two-dimensional (2D) layered species of the form AlargeBX4, where Alarge denotes the larger cationic species such as butylammonium (BA).38,39 In doing so, upon reaching a truly 2D interlayer limit, electronic quantum confinement effects result in further modifications to the electronic structure and therefore optical bandgap of the material to tailor the characteristics of the material.34,40–42 Furthermore, due to the increased hydrophobicity of the longer organic cation, it is possible to increase the native stability of the material against atmospheric conditions.43,44
For optoelectronic and photovoltaic applications, the doping of external species into HPs has received considerable attention,45–47 but has to some extent remained experimentally elusive. It has not been possible to dope perovskites using the established techniques of the inorganic semiconductor industry – such as ion implantation in crystalline silicon48 – and has therefore required the development of new techniques. Some examples include ground-state molecular charge transfer – which has been shown as a viable technique to dope perovskite nanocrystals and quantum dots owing to their high surface area to volume ratio.49 However, although effective in these low dimensional class of perovskites, the technique is not suitable for thin film devices. Substitutional techniques, such as replacing the A site cation with EA+ (ethyl ammonium), Sc+, Ag+, etc. have been demonstrated as a viable alternative in the bulk materials.50 However, most substitutional techniques rely on in synthesis processes, which depending on the annealing and crystallisation conditions lead to a variable quality and homogeneity in the dopants.
In addition to the substitution of A or B ions in the HP, adding species that sit interstitially within the structure have recently been shown to engender interesting material effects. For example, the insertion of ions with an active spin, such as manganese, has been shown to create magnetically active HP semiconductors.51,52 The insertion of Li has also been proposed to improve the performance of LED devices by tailoring the energy levels at the interface between the charge insertion layers and the emitting perovskite layer, thus improving the electron insertion efficiency and reducing the turn-on voltage.14,53 The findings of this work aim to provide a quantitative framework within which these device performance enhancements may be harnessed.
Furthermore, the insertion of alkali metal ions into perovskites has been demonstrated to stabilise luminescence and reduce non-radiative losses when inserted in small quantities (0–10% by molar ratio Li:Pb) by quenching non-radiative trap states inherent to the material.46,54 We also demonstrated in earlier work that electrochemically doping MAPbBr3 in concentrations up to the order of 1019 cm−3, can result in an n-type doping of the material.55 This is due to the fact that each Li ion inserted using this technique must necessarily correspond to the insertion of an electron from the external battery circuit, which we have shown to be inserted into the conduction band of the HP.47,55 It is clear therefore, that the careful control of dopants in a HP thin film is of the utmost importance to improving the performance metrics of HP solar cell and LED devices – by mitigating the effects of inherent atmospheric dopants, but also tailoring HPs to a specific doping state for integration into full solar cell or LED device stacks. In this current work, we use a similar technique and increase the lithium doping concentration within the HP to levels far exceeding the previous study and in doing so, elucidate the resilience of HPs to external species and investigate the phase transformation processes that occur once the limit has been exceeded.
Reports on lithium insertion in perovskites also include the work of Cao et al.56 and Dawson et al.57 Cao et al. demonstrate how the incorporation of extrinsic alkali cations (of which Li is included) occupy interstitial sites, resulting in the suppression of halide migration within the perovskite layer. Dawson et al. demonstrate how Li ions can be incorporated interstitially into the perovskite layer, when used as an electrode in a lithium ion battery (LIB) device and use density functional theory to determine which conversion reactions are energetically favourable upon excessive electrochemical discharge. In the current work, we manage to provide experimental verification of the theoretically proposed mechanisms of Dawson et al.57 The work of Dawson et al. also calculates the possible structural distortions due to Li insertion of all three of the commonly studied HP halide chemistries, namely I, Br, and Cl based perovskites. A difference in the electrochemical behaviour of I− compared to Br−-based HPs can also be found in a previous study.31 The ionic nature of the bonding within the HP crystal structure renders it inherently sensitive to the insertion of external ions (compared to layered metal oxide materials employed in LIB electrodes for example), and only so many dopants may be inserted interstitially before the structure breaks apart.57 Previous research of HPs for LIBs has shown that lithiating HPs leads to a range of conversion and alloying reactions.27,57
In this work, Li is incorporated into the host material lattice in quantities ranging from 0.2–20 mol% and the effects of doing so, including the phase conversion reactions that take place at excessive concentrations are elucidated. To achieve this, we use an electrochemical cell resembling the architecture of an LIB, equipped with a window making the cells suitable for both operando X-ray diffraction (XRD) and optical measurements. The advantage of electrochemical doping is that the number of Li ions inserted into the HP can be calculated and controlled – simply by controlling the current flow through the cell and the time, since it is known that 1 Coulomb of charge measured through the external circuit represents 6.24 × 1018 Li ions added to the HP. However, side reactions can introduce errors and the electrolyte needed to transport the Li ions from the Li source (here a Li-metal foil) typically rely on polar solvents that tend to dissolve the HP over time. In this work however, we address this issue by using a polymer-based solid state electrolyte and demonstrate a stable HP/electrolyte interface – allowing the influence of the intercalating Li to be studied directly and without interference from dissolution.
We define and quantify three doping regimes: a safe regime, a minor decomposition regime, and a major decomposition regime. In addition to identifying each doping regime, we elucidate the mechanisms that mediate between them and determine the phase conversion processes responsible for the loss of perovskite structure at high dopant concentrations. This provides a framework and upper ceiling of attemptable doping concentrations for future perovskite device modification schemes. Finally, we utilise both the 3D and 2D perovskite compositions and mixtures thereof for generality and in order to isolate and probe a variety of crystallographic structures and ensure the universality of the results. The Ruddlesden Popper quasi 2D/3D mixed phase, of the form (BA)2(MA)n−1PbnX3n+1 (BA – butylammonium, MA – methylammonium, X – halide (I− and Br−)) with n = 4 (mixed phase) and n = ∞ (pure 3D) phases are studied and changes to the diffraction spectra at dynamically varying Li doping concentrations are shown. The loss in PL intensity is used to isolate the effects of the doping on the perovskite electronic energy structure – such as the introduction of trap states and majority charge carrier quenching – from the structural and phase conversion effects deduced from the XRD study. Finally, by looking at the loss in optical reflectivity from a perovskite crystallite, the lithium doping and phase conversion dynamics are characterised, demonstrating an outward-in procession of diminishing perovskite phase fraction.
To elucidate the underlying mechanisms arising from the insertion of Li into MAPbBr3 and frame them in a quantifiable manner, operando XRD analysis was used during three successive Li ion insertion/removal cycles. Fig. 1(d) shows the three cycles accordingly, traversing the potential plateau at 2.2 V (insertion) and 2.4 V (removal), while Fig. 2 shows the response of various species observed during the first cycle. The top row shown in blue corresponds to lithium insertion increasing in concentration from 0 cm−3 to 3.4 × 1020 cm−3. The final doping state achieved after the first insertion step (1.8 V vs. Li/Li+) corresponds to a ratio of Li:Pb of ∼7%, i.e. (Li)0.07(MAPbBr3). A full conversion between doping concentrations and compositional ratios is provided in ESI,† Section II. The bottom row of Fig. 2 (red data) corresponds to the subsequent Li removal and clearly shows different degrees of reversibility for the species tracked in this XRD study. (Motion gif image files are provided in the ESI† that illustrate this reversible, oscillatory behaviour of the peaks clearly).
The first species investigated is the MAPbBr3 itself and the behaviour of the XRD fingerprint is demonstrated in Fig. 2(a) and (b) which corresponds to the (020) and (101) reflections that overlap at 9.45° 2θ (wavelength λ = 0.9918 Å, 12.5 keV). The behaviour of this peak is representative of all indexed perovskite peaks (additional peaks provided in ESI,† Section III) and proceeds as follows. Initially, as Li is inserted up to a doping concentration of 1 × 1020 cm−3 (corresponding to a molar doping concentration of 2% Li:Pb) the peak intensity remains relatively stable – dropping by only 3%. This corresponds to the initial lithiation process,
MAPbBr3 + xLi + xe− → Lix:MAPbBr3 | (1) |
Following this initial insertion and continuing to a final doping concentration of 3.4 × 1020 cm−3 (Li0.07:MAPbBr3), the peak decreases in integrated intensity. The loss indicates that the inserted Li ions distort the perovskite structure, to the point where lattice order is reduced substantially in the material. As such, it is concluded that the prominent mechanisms are those that dramatically change the structure, including amorphization and conversions to non-perovskite compounds, rather than subtle modifications to the perovskite lattice structure, which may be expected if the Li ions were simply to sit interstitially within the lattice. It becomes clear, when the peaks that rise in place of the HP are studied, that the dominant process is indeed a conversion of the perovskite phase. A small diffraction signal attributable to HP is still observed after two subsequent Li insertion and removal processes, indicating that at these dopant concentrations, a small amount of the perovskite structure remains present. From this we also infer that performing multiple insertion and removal cycles results in a similar loss in diffraction peak intensity – indicating that it is not only the static doping concentration that causes the disruption of the perovskite crystal structure, but the very process of moving the lithium in and out. For the purpose of device modification therefore, we recommend the use of only a single insertion process when preparing the HP material for applications.
Next, we study the peaks that emerge as the perovskite peak intensity reduces. First, Fig. 2(c) and (d) show the behaviour of the (111) peak belonging to metallic lead (indexed phase: cubic Fmm) during the doping (c) and subsequent lithium removal (d) processes. (All lead peaks behave in this manner, as shown in ESI,† Section III.) The peak area increases from zero (in the fresh perovskite sample) to a maximum during the first lithium insertion process, involving the insertion of Li ions to a doping concentration of 3.5 × 1020 cm−3. In addition to the single peak fitting of Fig. 2, Rietveld refinements of the time series spectra are used to extract approximations to the relative phase fraction between the MAPbBr3 and Pb(0) at different doping concentrations. In doing so, we establish a relative phase fraction of perovskite:Pb(0)|0.3:0.7 at the end of the first insertion cycle. (cf. ESI,† Section IV for full details of the Rietveld refinement process). The formation of metallic lead during the discharge process is in accordance with ref. 25, 28 and 49. However, the reduction of Pb2+ to Pb(0) is typically reported to occur at potentials below 1.4 V vs. Li/Li+.58,59 In this work however, the onset of the reduction of Pb2+ clearly occurs during the first galvanostatic plateau at ∼2.2 V vs. Li/Li+. This seemingly early onset could be attributed to the different local Pb environment in a HP compared to, for example, a lead metal oxide, which exhibits the same reaction, Pb2+ → Pb(0) at 1.6 V vs. Li/Li+.60 Furthermore, previous studies mention the difficulty of detecting the exact onset of this reaction during ex situ studies – due to the propensity of the Pb2+ to react with ambient oxygen to form PbO before measurement.58 This effect is mitigated in this work by performing all measurements in situ, without exposing the electrodes to ambient oxygen. The use of a polymer electrolyte could also contribute to extending the overpotential of the reaction, shifting it to higher potentials than those reported previously.
The subsequent lithium removal cycle shown in Fig. 2(d), shows how the peak attributed to the metallic lead species drops accordingly, although not totally to zero. This result has two implications; first, the formation of metallic lead is partially irreversible in each cycle – resulting in a continuously increasing amount of the lead species with the number of cycles and the quanta of lithium doping. This is confirmed by phase fraction calculations extracted via Rietveld refinement (cf. ESI,† Section IV). These show a 100% Pb phase fraction relative to the perovskite after all three cycles are complete and the HP electrode left in the lithiated state. Second, since the XRD peak intensities of the perovskite itself decreases consistently with time, the general conversion of perovskite to lead is irreversible. This in turn has two concomitant and practical implications; first, that if one desires to use perovskites in an LIB device, for the purpose of energy storage, the irreversible nature of this first electrochemical process currently renders the reversible capacity of such a device to be extremely limited. It was possible to remove only 65% of the inserted Li ions from the structure during the first charge process, indicating that ∼35% is irreversibly locked up within the now formed species. This corresponds, in electrochemical terms, to a measured first cycle coulombic efficiency of 65% during the first cycle. Second, the insertion of excessive Li ions into a HP material causes initially the breakdown of the structure, including the reduction of the Pb2+ cation to Pb(0).
In addition to the formation of Pb(0), the growth of peaks corresponding to the species LiBr (indexed phase: hexagonal P63mc) are observed during the Li insertion process as shown in Fig. 2(e). This is in accordance with Dawson et al.57 who propose the following conversion pathway of the perovskite during electrochemical cycling:
MAPbBr3 + 2Li+ + 2e− → MABr + 2LiBr + Pb(0) | (2) |
Fig. 2(g) and (h) show the behaviour of the (001) peak attributed to PbBr2 (indexed phase: orthorhombic Pnma) during the first delithiation and subsequent second lithiation cycle. No peaks corresponding to this species are present in the fresh electrode and do not appear until the first Li removal cycle, indicating that this species is in fact formed from the perovskite conversion products. Once formed however, it reliably oscillates out of phase with the Pb(0) and LiBr species, indicating a reversible shuttling between these two aggregate states. This behaviour is summarised in Fig. 3(a) where the normalized integrated peak intensity of all species is compared during the complete three-cycle electrochemical regime. By combining the experimentally observed behaviour of these species, we determine that in addition to eqn (1), the following process takes place during the Li insertion:
MAPbBr3 + Li + e− → MABr + LiPbBr2 | (3) |
Fig. 3 Li doping induced decomposition of MAPbBr3. (a) Normalised XRD peak integrated intensity as a function of multiple Li insertion and removal cycles for each of the studied species. Blue regions correspond to Li insertion and red to Li removal. (b) Schematic representation of the MAPbBr3 phase conversion pathways and decomposition products. Blue arrows denote processes that occur during Li insertion and red (downwards-pointing) arrows during Li removal. Decomposition of MABr (orange) is assumed to occur regardless of applied current.62 Double headed arrows denote bidirectional processes. (c) Close examination of Li insertion effects on MAPbBr3. Peak corresponds to the (032) (main peak) and (040) (left hand shoulder) of MAPbBr3 and is representative of all indexed MAPbBr3 peaks during this process. (Wavelength λ = 0.9918 Å, 12.5 keV.) (d) Integrated peak intensity as a function of doping concentration. Green, blue and red regions correspond to three doping regimes: safe, minor decomposition and major decomposition respectively, which are also indicated by horizontal lines on the diffraction peak in (c). Beyond the safe doping regime, a rapid decrease in perovskite peak intensity is observed. |
Following the processes described by eqn (1)–(3) during Li insertion, the following processes proceed during the subsequent removal of the Li from the electrode, based on the response of the observed species’ XRD signals:
LiPbBr2 → Li+ + e− + PbBr2 | (4) |
Pb(0) → e− + Pb+ | (5) |
The integrated peak intensity for LiBr is observed to drop during Li removal cycles, which leads us to conclude that it likely decomposes chemically into its elemental constituent Li and Br2 by:
2LiBr ↔ 2Li + Br2 | (6) |
MABr ↔ HBr + CH3NH2 | (7) |
Having determined the implications of excessive Li doping in MAPbBr3, we now focus on smaller doping concentrations (<1020 cm−3), which are more typical and relevant for perovskite optoelectronic device applications. A closer examination of Fig. 2(a) and the integrated intensity trace of MAPbBr3 in Fig. 3(a) shows that the drop in perovskite diffraction signal does not proceed significantly until a doping concentration of ∼1 × 1020 cm−3 (corresponding to an elemental ratio of Li:Pb of 2%) is reached during the first insertion process. Fig. 3(c) shows this in greater detail where we break down the initial doping process into three regions: (1) safe, shown in green, (2) minor decomposition, shown in blue and (3) major decomposition, shown in red. Doping concentrations of up to ∼1 × 1020 cm−3 (corresponding to a Li:Pb ratio of 2%) are within the “safe” regime, whereby the crystal structure of the perovskite is little perturbed. At this doping concentration, we observe a drop in the integrated XRD peak intensity for the perovskite species of only 3% – measurements of which are shown in the green highlighted region of Fig. 3(d). We anticipate that this provides an experimental ceiling for any attempt at extrinsically doping a hybrid perovskite material. In situ doping methods, such as the demonstrated electrochemical doping, pushed beyond this limit are likely to initiate the decomposition of the perovskite material and therefore compromise the desired optoelectronic properties. Doping methods that take place ex situ, during the synthesis of the perovskite, by adding dopant salts to the precursor solution for example, are unlikely to result in the successful crystallisation of a single phase perovskite. For example, the work of Phung et al.66 shows how above doping concentrations of 0.1–1.0 mol% the chemical doping of Sr2+ or Mg2+ results in a surface phase segregation. Given that the ionic radii of the Mg2+ in the work of Phung et al. and the Li+ in our study are comparable (72 pm and 76 pm respectively), we propose that the fact that it is possible to insert twice as much Li+ is not due to geometric or volume based effects alone. That is, the phase segregation is not driven simply by the amount of space required for the dopants in the perovskite lattice. By considering the relative charge densities of the two ions, we conclude that the phase conversion/segregation is driven by a finite amount of charge density that the perovskite lattice can accommodate. Therefore, since the Mg2+ carries twice as much charge density as the Li+ – it follows logically that the ionic perovskite structure can only accommodate half as much Mg2+ as Li+ before it begins to decompose.66
Above this safe threshold of ∼2%, the peak intensity of the perovskite phase decreases rapidly with increasing lithiation. The effect is exemplified by the (032) and (040) reflection in Fig. 3(c) between the blue lines. Additionally, the effect on the total perovskite peak intensity with state of lithium doping concentration is shown in the blue highlighted area in Fig. 3(d). This region is denoted the minor decomposition regime between 1 × 1020 cm−3 (Li:Pb 2%) and 2.5 × 1020 cm−3 (Li:Pb 6%). At these doping concentrations, the perovskite phase remains the dominant phase (>50% calculated relative phase fraction compared to the products outlined in eqn (2) and (3)) but decreases sharply. The further addition of Li doping engenders significant drops in the perovskite diffraction signal and relative phase (from 98% to 50% phase fraction). This regime should be avoided when considering electrochemical doping as a method of doping perovskite materials for applications that require high levels of optoelectronic performance. (The modification to the optical properties will be considered in the relevant section below.) However, we note that the doping concentrations reached here, are already higher than those usually attempted for device applications, which are typically of the order of 1018 cm−3. The onset of this regime appears sharply at ∼2% Li:Pb mol%. Owing to the “soft” crystal structure characteristic of the HP class of materials,67 rather than decomposing linearly with added external species, the structure can withstand external species up to a specific loading threshold before suddenly dropping.
Finally, above these doping concentrations (from 2.5 × 1020 cm−3 to 4.5 × 1020 cm−3, or 6% to 9% Li:Pb ratio) the regime in which the perovskite no longer comprises the dominant species by relative phase fraction (<50%) is referred to as the majority decomposition regime. In this regime, the conversion processes, described by eqn (2) and (3), proceed so far that the signal from the perovskite decreases to <10% of its original intensity. Furthermore, Rietveld refinement of the spectra at this stage calculates a relative phase fraction between the perovskite and now formed lead to be 0.3:0.7 (ESI,† Fig. S6). We strongly caveat that in this regime the use of the word “doping” is no longer accurate since the onset of the conversion reactions in the previous regime demonstrate how the perovskite is now being converted to separate phases rather than being doped with interstitial lithium species. The (032) and (040) peak is shown to drop accordingly between the red lines in Fig. 3(c) and the overall effect on the perovskite reflection intensity is shown in the highlighted orange region in Fig. 3(d).
The n = 4 stoichiometric arrangement of the Ruddlesden Popper perovskite phase for the layered perovskites (BA)2(MA)3Pb4Br13 was investigated as an intermediate phase between the bulk-like perovskite MAPbBr3 discussed thus far and the quasi-2D (n = 1) phase34 (BA)2PbBr4 to establish generality for the HP decomposition reactions. The species can be considered as comprising four layers of MAPbBr3 intercalated between two longer BA cations. The galvanostatic cycling protocol was the same as for the MAPbBr3 above and comprised three discharge/charge cycles at a constant current of 1.5 μA (corresponding to 60 mA gHP−1). The corresponding potential/time and potential/capacity plots are shown in Fig. 4(a) and (b) respectively. Two regions of the diffractogram are shown in Fig. 4(c) and (d) that exhibit contributions from both intra- and inter-layer crystallographic planes. In both cases, the intra-MAPbBr3 layers (111) in (c) and (202) in (d) decrease in intensity rapidly during the first cycle shown and then remain negligible with subsequent cycles. This indicates that the same phase transformation processes of eqn (2) and (3) for the bulk perovskite hold in the quasi 2d/3d structure, with the decomposition of the MAPbBr3 layer occurring within the n = 4 perovskite layers. Interestingly, the inter-layer peaks (0 10 0) in (c) and (0 20 0) in (d) decrease in intensity much slower than the intra-layer peaks and are still observable after three complete cycles. Furthermore, the peak intensity recovers slightly (by ∼10% of relative intensity) during the charge (Li removal) processes. This is more evident when observing the combined normalised intensity shown inset in both (c) and (d) figures. The trace demonstrates a decreasing curve – representing the overall loss of the perovskite species with Li insertion and multiple cycles; however, with the superposition of an oscillatory signal caused by the partial recovery of the interlayer peak in each. This could indicate that despite the conversion of the MA3Pb4Br13 perovskite layers within the structure, that the inter-layer BA cation – being itself electrochemically inactive in this potential range between 1.8 V and 2.8 V vs. Li/Li+ – does not contribute electrochemically and does not host any inserted Li ions and thus does not exhibit such dramatic changes in the diffraction spectra due to deformation or conversion.
Since the crystal structure of the perovskite is one of the properties underpinning the optoelectronic character of the HP, the photoluminescence (PL) emission of the material is measured in situ during repeated Li ion insertion/removal cycles. (See ESI,† Section V for further experimental details.) First, one single insertion process is used to insert Li ions to a doping concentration of 8 × 1018 cm−3 (0.16% Li:Pb) before 20 complete cycles are executed between 1.8 V and 2.8 V vs. Li/Li+, inserting and removing a similar order of magnitude of Li ions. Finally, the device is electrochemically discharged to 0.1 V vs. Li/Li+ – inserting the maximum amount of Li ions to test what happens to the material in this extreme limit and an equivalent doping concentration of ∼1.1 × 1021 cm−3 (20% Li:Pb) is reached.
Fig. 5(a) shows the in situ PL characterisation of the MAPbBr3 electrode during these various stages of cyclic electrochemical doping. The characteristic emission of the perovskite species is used to infer the effect of lithium doping on the photoexcited charge carriers within the material and thus the changes to the energy structure. First, the strong emission shown in Fig. 5(a) in deep purple at ∼550 nm is characteristic of an unperturbed MAPbBr3 structure (cf.Fig. 5(b)). Upon electrochemical insertion of lithium ions to a representative doping concentration of 8 × 1018 cm−3 (0.16% Li:Pb), the PL emission intensity is observed to drop by ∼50% (1st insertion in Fig. 5(a)). Three concurrent effects contribute to this reduction in radiative emission. First, the introduction of trap-like energy states that lie within the bandgap of the perovskite semiconductor upon insertion of additional lithium species causes charges, that would otherwise recombine radiatively, to be trapped.48 Second, subtle changes to the perovskite phase crystal structure upon lithium doping are likely to affect the valence and conduction bands in position and curvature, diminishing the radiative recombination, while also allowing for additional lattice vibrations to couple to the optical excitations, resulting in increased non-radiative losses (Fig. 5(c)). Finally, the injection of one electron for every Li+ inserted in order to conserve charge neutrality results in an increase in the electronic doping concentration that is equal to the Li concentration. At the concentrations examined in this work of >1017 cm−3, the photoexcited holes will be rapidly quenched by the majority electrons thus reducing the charge carrier lifetime and PL emission.55 Given that we observe the preservation of >97% of the perovskite structure diffraction relative intensity at this level of doping, we can possibly infer that the main losses to the PL emission are not from the second mechanism described, but from the first and third – the introduction of trap states to the energy structure from the intercalated Li and from excited charge carrier quenching. This initial process is subsequently followed by twenty insertion and removal cycles, in order to determine the impact of repetitive cycling. As expected, the PL emission intensity drops due to the irreversible nature of the insertion and removal processes and the eventual conversion to non-emissive species, as described by eqn (1) and (2). The spectrum is labelled as “20× cycles” in Fig. 5(a) and shows a significantly reduced emission intensity from the perovskite species at 550 nm. Finally, for generality the cell is “deep cycled” whereby the electrode is fully lithiated by discharging to ∼0 V followed by a potential hold. This adds the maximum possible number of Li ions to the perovskite electrode and almost certainly results in additional side reactions, such as SEI formation, in addition to the phase conversions discussed before. The spectra from the cell after this step is labelled “end product” in Fig. 5(a) and shows a broad PL emission of very low intensity indicating a total loss of the emissive MAPbBr3 species.
Operando optical reflectivity (cf. ESI,† Section VI) was used to observe the changes to the 2d/3d perovskite (BA)2(MA)3Pb4Br13 directly and in real time with state of Li insertion.68Fig. 5(d) and (e) show reflection microscopy images obtained for a perovskite crystallite before and during the application of a constant current respectively. Subfigures (i) and (ii) refer in turn to reflected intensity images and differential frame images (see ESI,† Section VI for differential frame analysis method).68 Reflected intensity images show the raw reflected intensity of the perovskite crystallite and its surroundings and the differential images show the change in reflected intensity, at a given time, over a 40 s window. In this instance however, since the optical observations are limited to a localised region and indeed a single crystallite, we make no attempt to map the ensemble electrochemical state to that of the individually observed spot quantitatively. The ensemble electrochemical state (see ESI,† Fig. S10) is used as a guide as to where in the lithium insertion process the optically observable changes occur, without extrapolating to exact localised doping concentrations. Quantitative limitations aside, it is apparent that the square crystallite structure disappears as Li is added to the material from the electrochemical cycling (up to global Li concentrations of 3 × 1020 cm−3 (6% Li:Pb)). This implies that the crystallographic structure, responsible initially for the reflectivity image in Fig. 5(d)-i is changed beyond detection – as shown in Fig. 5(e)-i. This is attributed to the phase transformations described in eqn (2) and (3) as for the 3d MAPbBr3 analogue, since the products have neither the same cubic structure nor reflectivity spectrum as the initial perovskite.
The differential frame analysis is shown in Fig. 5(d)-ii and (e)-ii in which the dynamic changes occurring at the respective stationary frame emerge. Red regions correspond to areas of increasing reflectivity and blue areas to decreasing. An inwardly proceeding square of reflectivity loss (blue) is observed first at the edges of the crystallite before moving inwards. This confirms that the lithium ions originate from the electrolyte surrounding the crystallite and initially congregate at the outer edges. The inward motion corresponds to a homogenisation of the lithium ions through the perovskite structure, resulting in the initiation of the phase conversion reactions (eqn (2) and (3)) from the outside – in, finally resulting in the loss of the observable perovskite structure seen in (ei). For the reflectivity measurements, the material was discharged from OCP to 1.8 V vs. Li/Li+, with the current and time selected such that the final Li content was ∼6% Li:Pb – placing it on the border between the minor and major decomposition regimes described previously.
We demonstrate the ability to dope HPs with a high degree of precision and perform a suit of operando analyses of the material to study structural changes of HPs during doping. Synchrotron XRD measurements quantify three doping regimes; a safe regime, below doping concentrations of ∼1020 cm−3 (0–2% Li:Pb mol%) in which the HP may be doped without detrimental effects to its structure; a minor decomposition regime between doping concentrations of 1020 cm−3 and 2.5 × 1020 cm−3 (Li:Pb ratio of 2% to 6% respectively), in which the HP phase is partially transformed but remains the dominant species. Finally, we quantify a major decomposition regime (>2.5 × 1020 cm−3 corresponding to Lin:HP where n > 0.06) in which the perovskite is superseded in relative phase fraction by new conversion products including Pb(0), LiBr and LiPbBr2. By comparison to previous studies which use Mg2+ as the dopant species,66 we conclude that the dopant charge density is as important to stabilising the HP as the ionic radius of the species – explaining why twice as much Li+ as Mg2+ may be incorporated into the ionic structure. The results are confirmed by in situ photoluminescence spectroscopy which isolates the charge carrier doping effect from structural effects. Operando optical reflection microscopy shows the real-space motion of the conversion processes. In doing so, we provide an experimentally derived, quantitative doping framework with which future devices reliant upon the doping of perovskite thin films may be based.
The optical cell (EL-CELL, ECC-Opto-Std test cell) was mounted on an XYZ nano-positioner stack (Attocube, ECSx3030/AL/RT/NUM) with an overall travel range of 25 mm in all dimensions. The sample focus position was maintained via an active external focus stabilisation based on a calibrated line-reflection profile of a 980 nm reference laser, as described previously.68 Comprehensive details of the experimental setup and data acquisition are explained in ESI,† Section VI.
Footnote |
† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d2ee01754g |
This journal is © The Royal Society of Chemistry 2022 |