Selective dissolution of halide perovskites as a step towards recycling solar cells

Abstract

Most research on perovskite solar cells has focused on improving power-conversion efficiency and stability. However, if one could refurbish perovskite solar cells, their stability might not be a critical issue. From the perspective of cost effectiveness, if failed, perovskite solar cells could be collected and recycled; reuse of their gold electrodes and transparent conducting glasses could reduce the price per watt of perovskite photovoltaic modules. Herein, we present a simple and effective method for removing the perovskite layer and reusing the mesoporous TiO₂ -coated transparent conducting glass substrate via selective dissolution. We find that the perovskite layer can be easily decomposed in polar aprotic solvents because of the reaction between polar aprotic solvents and Pb²⁺ cations. After 10 cycles of recycling, a mesoporous TiO₂ -coated transparent conducting glass substrate-based perovskite solar cell still shows a constant power-conversion efficiency, thereby demonstrating the possibility of recycling perovskite solar cells.

ARTICLE

Received 5 Dec 2015 | Accepted 25 Apr 2016 | Published 23 May 2016

Byeong Jo Kim1,*, Dong Hoe Kim2,*, Seung Lee Kwon1, So Yeon Park1, Zhen Li2, Kai Zhu2 & Hyun Suk Jung1

Most research on perovskite solar cells has focused on improving power-conversion efciency and stability. However, if one could refurbish perovskite solar cells, their stability might not be a critical issue. From the perspective of cost effectiveness, if failed, perovskite solar cells could be collected and recycled; reuse of their gold electrodes and transparent conducting glasses could reduce the price per watt of perovskite photovoltaic modules. Herein, we present a simple and effective method for removing the perovskite layer and reusing the mesoporous TiO2-coated transparent conducting glass substrate via selective dissolution. We nd that the perovskite layer can be easily decomposed in polar aprotic solvents because of the reaction between polar aprotic solvents and Pb2 cations. After 10 cycles of recycling, a mesoporous TiO2-coated transparent conducting glass substrate-based perovskite solar cell still shows a constant power-conversion efciency, thereby demonstrating the possibility of recycling perovskite solar cells.

DOI: 10.1038/ncomms11735 OPEN

Selective dissolution of halide perovskites as a step towards recycling solar cells

1 School of Advanced Materials Science and Engineering, Sungkyunkwan University, Suwon 16419, Korea. 2 Chemistry and Nanoscience Center, National Renewable Energy Laboratory, Golden, Colorado 80401, USA. * These authors contributed equally to this work. Correspondence and requests for materials should be addressed to H.S.J. (email: mailto:[email protected]

Web End [email protected] ).

NATURE COMMUNICATIONS | 7:11735 | DOI: 10.1038/ncomms11735 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 1

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms11735

The advent of perovskite solar cells (PSCs) has led to a change in the paradigm of emerging photovoltaic (PV) technology owing to the signicant increase in their

power-conversion efciency within the past 3 years13. Thus far, the best reported efciency, achieved by Seok and colleagues, is 22.1% (ref. 4). This efciency is comparable to that of other commercialized solar cells, such as multicrystalline Si (c-Si, 21.3%), cadmium telluride (CdTe, 22.1%) and copper indium gallium selenide (CIGS, 22.3%) solar cells4. However, to facilitate commercialization of these emerging PVs, various issues must be addressed, such as long-term stability, the need for Pb-free perovskite light-absorbing materials, the economic feasibility of the manufacturing process and an appropriate ecofriendly disposal process for Pb-contained waste. The price of a Si solar module ranges from $0.6 to $1.0 per Wp (ref. 5), which is signicantly higher than the target price of PSCs (lower than $0.2 per Wp, see ref. 6). The target price for PV modules required to achieve grid parity, as determined by the United States Department of Energy, is $0.5 per Wp; therefore, the target gure ($0.2 per Wp) constitutes a paradigm-changing price7.

The competitive price of PSCs results from the low material cost and non-vacuum fabrication process of perovskite light absorbers. However, the relatively high material costs for electron- and hole-transport materials, metallic electrode materials (including gold or silver) and transparent conducting glass (TCG) substrate materials render realization of the target price difcult. For example, at the laboratory scale, the unit price of spiro-MeOTAD (2,2,7,7-tetrakis[N,N-di(4-methoxyphenyl) amino]-9,9-spirobiuorene), a widely used hole-transport material, is $200 per g (Lumtec). Gold, as a metallic electrode material, and uorine-doped tin oxide (FTO) glass cost $40 per g and approximately $10 per m2, respectively. As such, these expensive materials account for 98% of the costs associated with small, 1 inch 1 inch square devices. The price of

hole-conducting materials is expected to decrease by virtue of new synthesis technologies and large-scale synthesis methods. However, a signicant reduction in the prices of gold and FTO glass seems unlikely; in fact, owing to limited mineral resources, their prices may vary signicantly depending on global demand.

The recycling of degraded PV modules has been considered as an alternative approach to making PVs economically viable. In particular, improvement of the long-term stability of PSCs is still needed to compete with other types of solar cells, such as c-Si, CdTe and CIGS solar cells, which are being used in commercial PV modules. However, considering the energy payback time (EPBT), which is the time required for a PV module to produce an amount of energy comparable to that consumed during its production, the EPBT of a perovskite PV module is 0.22 years, which is the shortest nominal EPBT compared with all other competitors such as c-Si and CdTe (ref. 8). Therefore, if it were possible to recycle degraded perovskite PV modules, the issues inhibiting the commercialization of perovskite PV modules would be solved, such as their long-term stability, the production of large amounts of Pb-containing waste, the high unit cost of PSCs and the retrieval and reuse of expensive components such as gold, silver and TCG. A shorter payback period of perovskite-based PV modules compared with that of Si-based PV modules could be realized via this process of recycling degraded PSCs.

Therefore, in this paper, we introduce a recycling or repair methodology for PSCs. In this system, trihalide perovskite materials, such as methyl-ammonium lead iodide (CH3NH3PbI3, abbreviated as MALI) as well as formamidinium lead iodide (HC(NH2)2PbI3, abbreviated as FALI) and methyl-ammonium lead bromide (CH3NH3PbBr3, abbreviated as MALBr) mixed halides (in this study, (FALI)0.85(MALBr)0.15 was used), are removed through a selective dissolution process. This process is

capable of dissolving PSCs such that the Au electrodes and the mp-TiO2-coated TCG substrates are separated through the dissolution of the perovskite layers. We nd that the recycling of PSCs can be realized based on the removal of trihalide perovskite materials and that a small amount of Pb residue has a negligible effect on the ability to reuse the mp-TiO2-coated TCG substrate. The selective dissolution mechanism is attributable to the breaking down of the PbI6 octahedral frame in the trihalide perovskite owing to the reaction between the partially positive Pb2 ions and a polar aprotic solvent. When the regenerated electron-transport-layer-coated FTO glass (referred to as the mp-TiO2-coated TCG substrate in this paper) is reused, the original power-conversion efciency (15%) is retained over 10 regeneration cycles. This study reveals the potential impact of easily exchangeable PSCs, similar to electric batteries, on the use of PSCs in various applications (for example, as portable power sources in wearable devices and internet-of-things devices).

ResultsRecycling procedure. Figure 1 shows the steps of the recycling process for PSCs. After the deposition and annealing of the electron-transport layer, which is composed (in general) of compact and mp-TiO2 layers, the MALI light absorber is coated on the substrate. The hole-transport layer (HTL) and gold electrode are then deposited. During recycling, the MALI perovskite and the HTL dissolve when the fabricated solar cell is immersed in a polar aprotic solvent such as dimethylformamide (DMF), g-butyrolactone (GBL) or dimethyl sulfoxide (DMSO).

The resultant solution contains the recyclable gold electrode and the mp-TiO2-coated TCG substrate (Supplementary Fig. 1); the recycling procedure is shown in Supplementary Movie 1. After rinsing and drying of the selectively dissolved mp-TiO2-coated

TCG substrate, the MALI layer, HTL and gold electrode are re-deposited to yield a recycled PSC.

The effect of the polar aprotic solvent on the removal of the MALI perovskite material was determined via eld-emission scanning electron microscopy (FESEM). Figure 2a,b show plane-view images of the mp-TiO2 layer coated on the TCG substrate and the MALI layer deposited on top of the mp-TiO2 layer, respectively. After selective dissolution processing of the MALI-coated substrate using DMF, the MALI layer is easily removed and only the mp-TiO2 layer is observed (Fig. 2c). We determined the efcacy of this recycling process by repeatedly recycling (in this work, we performed a total of 10 cycles) the mp-TiO2-coated TCG substrate of the PSC. The photocurrent densityvoltage (JV) curve of a 10-times-recycled device is similar to that of a PSC based on a new substrate (Fig. 2d and Supplementary Fig. 2). These results indicate that PSCs can be easily recycled.

Recycling mechanism. The selection of an appropriate solvent is essential for an effective recycling process. To investigate the solvent effect, we immersed PSCs in non-polar and protic/aprotic polar solvents. In the case of the non-polar solvents, only the Au/HTL was removed from the Au/HTL/MALI/mp-TiO2/TCG substrate, indicating that non-polar solvents selectively dissolve the spiro-MeOTAD layer (Table 1 and Fig. 3). The perovskite layer decomposed in both types of polar solvents, albeit with differing dissolution behaviours. This layer can dissolve in hydrogen-bonded polar protic solvents, such as ethanol or water. However, the dissolution kinetics is quite sluggish compared with the kinetics governing dissolution in a polar aprotic solvent. When the PSCs were immersed in the polar protic solvents, the perovskite layers became yellow in colour and, after more than

2 NATURE COMMUNICATIONS | 7:11735 | DOI: 10.1038/ncomms11735 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms11735 ARTICLE

a b

Perovskite

ETL

FTO

Deposition of

electron-

transport layer

Formation of

perovskite layer

Deposition of

hole-transport

layer

Deposition of

Au electrode

High-temp. annealing

Re-formation of

perovskite layer

Fabrication of PSCs

Rinsing and drying of

selectively dissolved

substrate

Selective

dissolution of

failed PSC

for recycling

Polar aprotic

solvent

Peeled-

metal electrode

Selective dissolution

of PSCs

Au

HTL

HTL

Figure 1 | Recycling process for PSCs. (a) Flow chart of the fabrication and recycling processes for PSCs. (b) Schematic illustration of the detailed process of recycling PSCs via selective dissolution. During soaking in a polar aprotic solvent, the deposited-metal electrode peels away from the device, leaving the clear electron-transport-layer-coated substrate behind, and the hole-transport layer and perovskite layer dissolve.

a b c

mp-TiO2/TCG MALI/mp-TiO2/TCG Cleaned mp-TiO2/TCG

New substrate PSC

Ten time-recycled PSC

d

Current density (mA cm2 )

24 20 16 12

8400.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1

Voltage (V)

Figure 2 | Features of the recycling process performed via selective dissolution. Plane-view FESEM images of (a) a new mesoporous TiO2 (mp-TiO2)/ TCG substrate, (b) a CH3NH3PbI3 (MALI)-layer-deposited mp-TiO2/TCG substrate and (c) a cleaned mp-TiO2/TCG substrate after the selective dissolution process. For the cleaning of the MALI/mp-TiO2/TCG substrate, DMF was used. Scale bars, 1 mm. (d) Current densityvoltage (JV) curve of a

PSC fabricated on a 10-times-recycled mp-TiO2/TCG substrate compared with that of a PSC fabricated on a new substrate.

24 h, eventually became transparent (Supplementary Fig. 3). This behaviour is attributable to the constituent organic cations (such as the methylamine group) of the perovskite light absorber, which easily dissolve in hydrogen-bonded solvents (for example, H2O, ethanol); the low solubility of metal halides (for example,

PbI2 and PbBr2) in polar protic solvents is indicative of the

relatively slow SN2 reaction compared with that in a polar aprotic solvent9,10. A polar aprotic solvent can dissolve the perovskite layer completely within a fairly short time after immersion. We tested the dissolution of PSCs in various other solvents, as shown in Supplementary Fig. 4. The results of these attempts were consistent with the results described above.

NATURE COMMUNICATIONS | 7:11735 | DOI: 10.1038/ncomms11735 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 3

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms11735

To determine whether the bonding between the metal halide and the aprotic solvent plays an important role in the recycling process, 13C nuclear magnetic resonance (NMR) measurements were performed at 500 MHz using DMF, 1 M PbI2 in DMF and 1 M CH3NH3PbI3 in DMF (the full range of the 500 MHz NMR results is shown in Supplementary Fig. 5a). As Fig. 4a,b show, the main NMR peaks of the PbI2DMF solution (red lines) were

shifted downwards compared with those of the DMF solution (blue lines). These results are consistent with the formation of metal halideDMF bonds; PbI2 bonds with the DMF molecule, resulting in the formation of a PbI2DMF structure via the PbO bond911. Because of the resonance structure of DMF (Supplementary Fig. 6), the partially negative oxygen atom in DMF can bond with partially positive atoms, such as Pb2 cations (Fig. 4c). Therefore, aprotic solvents (such as DMF) have the ability to form PbI2DMF compounds, as shown in Fig. 4d.

The chemical shifts of the 13C sp2 and 13C sp3 signals corresponding to the CH3NH3PbI3DMF solution (see Fig. 4a,b (black lines)) are similar to those for the PbI2DMF solution. These 13C-NMR results indicate that the Pb2 cations in the

CH3NH3PbI3 perovskite layer readily bond with the partially negative oxygen atoms in aprotic solvents (Fig. 4e). This bonding enables dissociation of the PbI6 octahedral frame that forms the skeleton of the organicinorganic perovskite material, thereby resulting in high solubility of the perovskite material in polar aprotic solvents.

As shown in Supplementary Movie 1, the used gold electrode can be easily collected during the recycling process. Supplementary Table 1 shows that the collected gold electrodes contained 0.41% lead impurities. The rening of metal collected from electronic and solution wastes is one of the most universal processes for the recycling of metal in industry12. Rened gold originating from industrial disposal represents 4% of the entire gold market in the industrial eld13. Thus, the gold gathered during the recycling process can be easily rened because of its low impurity concentration. Another important problem to be addressed is the removal of lead pollutants from the residual polar

Table 1 | Various solvents used for the selective dissolution

of the perovskite layer and their dipole moments.

Solvent Chemical formula Dipole moment (D)

Non-polar solventsToluene C6H5-CH3 0.375 Diethyl ether CH3-CH2-O-CH2-CH3 1.15 Dichoromethane CH2Cl2 1.60 Chlorobenzene C6H5-Cl 1.69

Polar protic solvents2-Propanol CH3-CH(-OH)-CH3 1.58 Ethanol CH3-CH2-OH 1.69 Methanol CH3-OH 1.70

Water H-O-H 1.85

Polar aprotic solventsAcetone CH3-C( O)-CH3 2.88

Dimethylformamide H-C( O)N(CH3)2 3.82

Acetonitrile CH3-C N 3.92

Dimethyl sulfoxide CH3-S( O)-CH3 3.96

g-Butyrolactone /-C2H4-O-C2H4-\ O 4.27

a Non-polar solvent

MALI/mp-TiO2

b Polar protic solvent

(MALI+PbI2)/mp-TiO2

Spiro-MeOTAD

c Polar aprotic solvent

mp-TiO2

Figure 3 | Features of PCSs dissolved in three different types of solvents. Cross-sectional FESEM and photographic images of dissolved PCSs obtained after 30 s of immersion of spiro-MeOTAD/MALI/mp-TiO2/TCG in (a) non-polar, (b) polar protic and (c) polar aprotic solvents. Diethyl ether, water and

DMF, respectively, were used as the representative solvents. Scale bars, 700 nm.

4 NATURE COMMUNICATIONS | 7:11735 | DOI: 10.1038/ncomms11735 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms11735 ARTICLE

a b

MALI in DMF

13C sp2 13C sp3

Pbl2 in DMF DMF

164 163

13C/ p.p.m. 13C/ p.p.m.

162 38 36 34 32 30

Pbl2 in DMF DMF

MALI in DMF

c

d

e

+

X

PbI2-DMF

MALI-DMF

+

+ +

X

+

+

X

Steric hindrance

Figure 4 | Dissolution mechanism of MALI in a polar aprotic solvent. (a) 13C sp2 and (b) 13C sp3 chemical shifts of PbI2DMF and MALI-DMF away from the corresponding signals for a pure DMF solution. (c) Scheme of the resonance structure of DMF, with a partially positive and negative polarity. Chemical bonding structures of the (d) PbI2 and (e) PbI6 frames of MALI with the partially negative oxygen of DMF. (Red spheres: oxygen, blue spheres: nitrogen, grey spheres: carbon, white spheres: hydrogen, yellow spheres: lead, violet spheres: iodine.)

aprotic solvent after the recycling process because of the attendant economic and environmental issues. Several methods of removing lead from lead-containing solution wastes have been investigated, such as electrodialysis14, solvent extraction15, ion exchange16 and biosorption17. In particular, hydroxyapatite (Ca10(PO4)6(OH)2, also called HAP), a biocompatible material present in human bone and teeth, shows a remarkable efciency for the absorption of Pb2 ions through ion exchange between

Ca2 and Pb2 (refs 18,19). We also demonstrated a lead-removal treatment for residual-lead-containing solution waste via a two-step process combining solvent extraction and ion exchange (Supplementary Fig. 7). In the rst step, the lead concentration in a waste solution of 0.05 M MALI in DMF, which was initially 2,017,791.3 p.p.b. (equivalent to mg kg 1), was decreased to 28,507.8 p.p.b. during the solvent extraction process. Then, after an ion exchange process with 50 mg HAP, the concentration of residual lead was decreased to 67.4 p.p.b., corresponding to a 99.99% reduction in lead concentration compared with that of the initial solution. Thus, these results show that the proposed two-step lead-removal process is effective in separating lead compounds from residual-lead-containing solution waste.

Surface analysis of a dissolved mp-TiO2-coated FTO substrate.

The optical transmittance spectra of an as-prepared mp-TiO2-

coated FTO substrate, a MALI/mp-TiO2/FTO substrate and a cleaned mp-TiO2/FTO substrate are shown in Fig. 5a. The transmittance of the MALI/mp-TiO2/FTO substrate is signicantly lower than that of the as-prepared mp-TiO2/FTO substrate, owing to the large absorption coefcient of the MALI layer. After the perovskite layer is removed, the transmittance is restored to that of the as-prepared substrate, indicating that perovskite-layer removal and the dissolution process do not affect the optical properties of the substrate. Figure 5b shows the X-ray diffraction patterns of the lms20. The peaks in the pattern of the DMF-cleaned mp-TiO2/FTO substrate correspond only to anatase TiO2 and

FTO, indicating that the perovskite layer was removed by the DMF solvent. The surface state of the cleaned substrate was determined from atomic depth proles obtained via secondary ion mass

spectrometry (SIMS) for each lm. As Fig. 5d shows, Pb and I were present in the MALI/mp-TiO2/FTO substrate. The X-ray diffraction pattern of the sample (Fig. 5b) indicates that these elements are components of the MALI perovskite material. These were detected even after the perovskite layer was removed, indicating that residual Pb2 and I ions (not detected via

X-ray diffraction) were present on the surface of the mp-TiO2 layer (Fig. 5e). Supplementary Fig. 8 shows the X-ray photoemission spectra of the lms. Consistent with the X-ray diffraction and SIMS data, the presence of Pb and I peaks in the signal from the pre-dissolution sample indicates that a perovskite layer formed on the MALI/mp-TiO2/FTO substrate.

The X-ray photoemission spectra of the cleaned mp-TiO2/FTO substrate also contain Pb and I peaks. Figure 5f,h and Fig. 5g,i show the X-ray photoemission spectra of the MALI/ mp-TiO2/FTO substrate and the cleaned mp-TiO2/FTO (inset)

substrate, respectively. The binding energies of Pb 4f (142.8 and 137.9 eV) and I 3d (630.6 and 619.1 eV), shown in Fig. 5f,h, respectively, indicate that the Pb and I ions were incorporated into the MALI perovskite structure21,22. The peaks in Fig. 5g,i corresponding to the binding energies of Pb and I ions remaining on the surface of the cleaned mp-TiO2/FTO substrate appear in positions that are typical of PbI2 (refs 23,24).

We also investigated the surface states of cleaned mp-TiO2/FTO substrates that had been selectively dissolved with various cleaning times (from 1 to 10 min) and numbers of recycling cycles (from one to ve cycles with 30 s of rinsing between each cycle) for complete removal of residual PbI2 and conrmation of the accumulation of residual PbI2, respectively. Supplementary Fig. 9a,b show that the intensities of the Pb 4f peak for cleaned mp-TiO2/FTO substrates treated with various cleaning times (from 1 to 10 min) and in multiple cleaning cycles (from one to four cycles) were of similar intensity regardless of the cleaning conditions. Furthermore, the Pb 4f peak of the ve-times-recycled mp-TiO2/FTO also showed a similar intensity to that of the once-recycled substrate (Supplementary Fig. 9c). It appears impossible to remove the residual PbI2 completely, but the amount of residual PbI2 on the TiO2 surface was negligible compared with that in the original MALI layer, and no accumulation of residual PbI2 occurred under repeated recycling.

NATURE COMMUNICATIONS | 7:11735 | DOI: 10.1038/ncomms11735 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 5

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms11735

a

b

^-FTO / #-anatase TiO2 / *-MALI

^

^

^

^

#

#

^

Cleaned mp-TiO2/FTO

MALI/mp-TiO2/FTO

As-prepared mp-TiO2/FTO

Transmittance (%)

100

80

60

40

20

*

Intensity (a.u.)

*

Cleaned mp-TiO2/FTO

MALI/mp-TiO2/FTO

As-prepared mp-TiO2/FTO

*

*

*

*

^

^

^

0 300

400 500

Wavelength (nm)

600 700 800 10 20 30 40

Two theta (degree)

c

d

e

TiO2 FTO

Ti

Sn

MALI/TiO2 FTO

TiO2 FTO

Intensity (cps)

105

104

103

102

101

100

Intensity (cps)

105

104

103

102

101

100

Ti

Sn

Pb

I

Intensity (cps)

105

104

103

102

101

100 0

Ti

Sn

Pb

I

100Sputter time (s) Sputter time (s)

150 200 250

0

50

0 100 300

200 400

50 100

Sputter time (s)

150 200 250

f

h

Pb 4f

I 3d

Intensity (a.u.)

Intensity (a.u.)

632 628 624 620 616

632 628 624 620 616

147 144 141Binding energy (eV) Binding energy (eV)

138 135

147 144 141 138 135

g

i

Pb 4f

I 3d

Intensity (a.u.)

147 144 141Binding energy (eV)

138 135

Intensity (a.u.)

Intensity (a.u.)

Intensity (a.u.)

632 628 624 620 616

Binding energy (eV)

Figure 5 | Optical and surface analysis before/after selective dissolution of the substrate. After the recycling process using a polar aprotic solvent, (a) the optical transmittance is preserved and (b) the X-ray diffraction patterns of the cleaned mp-TiO2/FTO substrate exhibit peaks corresponding to anatase TiO2 and FTO without MALI or PbI2 (^ with black: FTO, # with green: anatase TiO2, * with red: MALI). Results of secondary ion mass spectroscopy performed on (c) as-prepared mp-TiO2/FTO, (d) MALI/mp-TiO2/FTO and (e) cleaned mp-TiO2/FTO. The background schematics of the spectra represent the device structure; thus, the initial values of the spectra at 0 s of sputtering time represent each ion concentration at the surface of each substrate. The results for the cleaned mp-TiO2/FTO substrate reveal the presence of residual Pb2 and I- ions on the cleaned substrate. X-ray photoemission spectra of (f, g) Pb 4f and (h, i) I 3d on the substrate before and after selective dissolution, respectively. The binding energies of Pb 4f and I 3d correspond to Pb and I in the perovskite layer and in the form of residual PbI2 on the mp-TiO2/FTO and cleaned mp-TiO2/FTO substrates, respectively.

6 NATURE COMMUNICATIONS | 7:11735 | DOI: 10.1038/ncomms11735 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms11735 ARTICLE

Other groups have investigated the effect of non-stoichiometric MAI/PbI2 molar ratios (of PbI2 and CH3NH3I (MAI)) on perovskite crystal growth or crystallinity, leading to improved PV properties; molar ratios ranging from 0.83 to 1.2 were considered in those studies25,26. In addition, Cao et al. have reported that residual PbI2, which forms between the perovskite and the

TiO2, acts as a charge-recombination-hindering layer27. No such phenomenon was observed in the present study, in which the performance of the recycled PSCs remained stable, even in the presence of PbI2. Therefore, a polar aprotic solvent is capable of dissolving MALI perovskite materials within a rather short amount of time, although the PbI2 on the TiO2 surface is only partially dissolved. The residual PbI2 layer does not affect the PV properties of the regenerated PSC, thus enabling the recycling of these cells.

DiscussionThe nearly constant PV performance observed during the 10-times-repeated recycling of the mp-TiO2/FTO substrate (see Fig. 6) demonstrates the recyclability of PSCs using the proposed method. Furthermore, a recycled MALI-PSC, fabricated using a cleaned mp-TiO2/FTO substrate recovered from a degraded MALI-PSC through selective dissolution, exhibited a PV performance comparable to that of the MALI-PSC based on the fresh mp-TiO2/FTO substrate (Supplementary Fig. 10).

Four individual cells were measured for each type of device to investigate the inuence of the recycling process on long-term stability. The recycled MALI-PSC device showed stability comparable to that of the fresh MALI-PSC device (Supplementary Fig. 11 and Supplementary Table 2). To conrm the feasibility of the recycling process for various fabrication

a b

Normalized J sc

Normalized FF

1.21.11.00.90.80.7

0.6 1st

1.21.11.00.90.80.70.6

0.6

Normalized PCENormalized V oc

2nd 3rd

Number of recycling cycles Number of recycling cycles

4th 5th 10th 1st 2nd 3rd 4th 5th 10th

1st 2nd 3rd 4th 5th 10th 1st 2nd 3rd 4th 5th 10th

c

d

1.21.11.00.90.80.70.6

1.21.11.00.90.80.7

Number of recycling cycles

Number of recycling cycles

e 24

20

16

12

8 New Recycled

4

Current density (mA cm2 )

Two-step MALI PSC (FALI)0.85(MALBr)0.15 PSC

0 0.0 0.2 0.4 0.6 0.8 1.0

Voltage (V)

Figure 6 | Photovoltaic performance characteristics of recycled PSCs. Normalized trends in the (a) short-circuit current density, (b) open-circuit voltage, (c) ll factor and (d) efciency of MALI-based PSCs fabricated on 1st-, 2nd-, 3rd-, 4th-, 5th- and 10th-time recycled substrates compared with those of the MALI-PSC fabricated on the newly prepared substrate. (e) JV curves for two-step processed MALI-based PSCs (red spheres) and (FALI)0.85(MALBr)0.15-based PSCs (blue squares) on new (dotted lines and empty symbols) and once-recycled (solid lines and lled symbols)

mp-TiO2/TCG substrates. The error bars in ad indicate the s.d.s calculated from ve samples.

NATURE COMMUNICATIONS | 7:11735 | DOI: 10.1038/ncomms11735 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 7

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms11735

processes, such as a two-step process, and for different compositions, such as formamidinium- or bromide-containing perovskite, we fabricated two-step MALI- and (FALI)0.85

(MALBr)0.15-based PSCs using our selective dissolution recycling process. The results presented in Fig. 6e show that the selective dissolution process is also compatible with other perovskite fabrication processes and with PSCs based on different compositions. These results clearly demonstrate that our recycling process does not inuence the PV performance and that it is generally applicable to PSCs of any kind prepared with different perovskite deposition processes and compositions. We also found that the recycling process is applicable to planar-structured and exible PSCs (Supplementary Fig. 12).

These results may change the current paradigm in the eld of solar cells and their application. Although the stability of PSCs remains an important issue, the long-term stability of PSCs is not as essential as it is in the case of commercialized solar cells. If PSCs can be easily refurbished, then these cells can be used as batteries that could become available in consumer products. The combination of high efciency, recyclability and plastic-based exibility in PSC technology will shift the paradigm of utilization of perovskite-based PV devices.

Methods

Materials. All materials, except for 1-butanol (Tokyo Chemical Industry), were purchased from Sigma-Aldrich. All materials were used as received. Methyl-ammonium iodide (CH3NH3I, MAI) was synthesized using a previously reported method28. A solution of 57 wt.% hydriodic acid in water and a solution of 33 wt.% methylamine (CH3NH2) in absolute ethanol were stirred into 100 ml of ethanol at room temperature. The precipitate, methylamine iodide, was obtained using a rotary evaporator. Formamidinium iodide (HC(NH2)2I, FAI) was prepared using a previously reported method29, in which 25.2 g and 48 ml of formamidine acetate and hydriodic acid, respectively, were mixed into 250 ml of methanol and reacted at room temperature for 2 h. The resulting precipitate was collected using a rotary evaporator. The synthesized MAI and FAI were washed with diethyl ether and then recrystallized from ethanol and dried in a vacuum oven overnight.

Recycling process. The selective dissolution process for recycling mp-TiO2/TCG substrates that have been used in PSCs consisted of several simple steps. In the rst step, the trihalide perovskite layer and spiro-MeOTAD were removed by immersing the PSCs in a polar aprotic solvent, such as DMF, GBL or DMSO. The PSCs were then shaken for 30 s in solution. The cleaned mp-TiO2/TCG substrates were rinsed with deionized water and then dried on a hot plate. For optimal performance, the cleaned substrates were annealed at 500 C for 1 h. The gold (Au) material in the PSCs does not react with polar aprotic solvents, and, therefore, the gold that remained in the solutions after cleaning could be collected.

The removal of lead from a polar aprotic solvent used for the recycling process was performed via a two-step process combining solvent extraction with ion exchange. The prepared solution of 0.05 M MALI in DMF was poured into ether, a non-polar solvent. The precipitated lead compound was separated via centrifugation at 8,000 r.p.m. for 10 min. A laboratory-prepared HAP powder (between 10 and 50 mg) was added to the remaining solution to remove residual Pb2 ions via ion exchange (from Ca2 to Pb2 ). After the addition of the HAP powder, the solution was stirred at 200 r.p.m. for 3 h. After sufcient stirring, the

Pb2 -ion-adsorbed HAP powder was separated using a centrifuge at 8,000 r.p.m. for 10 min. The solution remaining after the completion of each step was analysed via inductively coupled plasma-mass spectrometry (Perkin-Elmer SCIEX, NexION). All solutions were analysed three times, and the results were averaged.

Synthesis of hydroxyapatite. HAP with a high surface area was synthesized using a hydrothermal method based on a previous report by Jiang et al.30. In a typical synthesis procedure, 2 mmol of CaCl2?2H2O (Sigma-Aldrich, Z99%) and 1 g of poly-L-aspartic acid sodium salt (Sigma-Aldrich, mol. wt. 5,00015,000) was dissolved in 20 ml of deionized water under 30 min of stirring (solution 1). Then,1.2 mmol of (NH4)2HPO4 (Sigma-Aldrich, Z98%) was dissolved in 15 ml of deionized water under vigorous stirring for 30 min (solution 2). After stirring, solution 2 was poured into solution 1. The mixed solution was sealed in a 50-ml Teon liner, placed in a stainless steel autoclave and maintained at 200 C for 12 h. The precipitated HAP was washed with distilled water and ethanol. After washing, the obtained HAP powder was dried in a freeze dryer for 12 h.

Device fabrication. FTO-coated glass substrates (Pilkington TEC15) were cleaned for 15 min with acetone, ethanol and DI water in an ultrasonic bath. A compact hole-blocking layer of TiO2 was spin-coated (at 3,000 r.p.m. for 20 s) on the

substrate, using 0.15 M titanium diisopropoxide bis(acetylacetonate) (75 wt.% in isopropanol) in anhydrous 1-butanol. This layer was then baked at 130 C for5 min. An B200-nm-thick layer of porous mp-TiO2 was prepared on top of the

TiO2 layer using a diluted TiO2 paste (EtOH:Dyesol 18NRT TiO2 paste in a ratio of4.5:1) and calcining at 500 C for 1 h. The resulting mp-TiO2/FTO substrate was immersed for 15 min in a 0.04 M TiCl4 solution at 70 C, rinsed with DI water and then annealed at 500 C for 30 min.

The prepared CH3NH3PbI3 solution (46 wt.% CH3NH3I and PbI2, at an equal molar ratio, in a mixture of GBL and DMSO (7:3 v/v)) was spin-coated at 1,000 and 5,000 r.p.m. for 10 and 30 s, respectively. During the 5,000-r.p.m. step, 1 ml of a toluene solution was dropped on the spinning CH3NH3PbI3 solution. The spin-coated substrate was then dried at 130 C for 20 min (ref. 31). Subsequently, 30 ml of a HTL solution (72 mg of spiro-MeOTAD, 28.8 ml of 4-tert-butylpyridine and 17.6 ml of Li-TFSI solution (720 mg of Li-TFSI in acetonitrile) in 1 ml of chlorobenzene) was formed via spin-coating at 4,000 r.p.m. for 30 s. The Au electrodes were deposited on the PSCs via thermal evaporation. The substrate size and the active device area were 2 2 and 0.14 cm2, respectively.

Device characterization. The surface morphology and cross-sectional structure were examined via FESEM (JSM-7600F, JEOL). 13C-NMR measurements were performed using a 500-MHz NMR spectrometer (Avance-500, Bruker). In addition, the transmittance spectra of the substrate and the perovskite-coated substrate were collected via UVvis/NIR spectrometry (Cary 5000, Agilent Technologies). The crystalline structure was determined using an X-ray diffractometer (New D8 Advanced, Bruker). Furthermore, compositional depth proles of the pristine, perovskite-coated and cleaned substrates were obtained via time-of-ight SIMS (ION-TOF). X-ray photoelectron spectroscopy (X-ray photoemission spectra, Sigma Probe, ThermoVG) was used to determine the chemical states of Pb and I on the surfaces of the substrates. The corresponding spectra were acquired using monochromatic A1-Ka radiation (100 W) and were calibrated with respect to the C 1-s level (at 284.5 eV). Moreover, the PV properties were measured using a solar simulator (Newport Oriel Solar 3A Class AAA, 64023A) equipped with a 450-W xenon lamp (Newport 6279NS) and a potentiostat (CHI 600D, CH Instruments). The light intensity was adjusted using a standard Si solar cell (Oriel, VLSI standards; 1 sun is equivalent to 100 mA cm 2).

Data availability. The authors declare that the data supporting the ndings of this study are available within the article and its Supplementary Information les.

References

1. Kim, H.-S. et al. Lead iodide perovskite sensitized all-solid-state submicron thin lm mesoscopic solar cell with efciency exceeding 9%. Sci. Rep. 2, 591 (2012).

2. Burschka, J. et al. Sequential deposition as a route to high-performance perovskite-sensitized solar cells. Nature 499, 316319 (2013).

3. Zhou, H. et al. Interface engineering of highly efcient perovskite solar cells. Science 345, 542546 (2014).

4. NREL, Best Research-Cell Efciencies http://www.nrel.gov/ncpv/images/efficiency_chart.jpg

Web End =http://www.nrel.gov/ncpv/images/ http://www.nrel.gov/ncpv/images/efficiency_chart.jpg

Web End =efciency_chart.jpg (2015).

5. Metz, A. et al.International Technology Roadmap for Photovoltaic (ITRPV.net) Results 2014 http://www.itrpv.net/Reports/Downloads/2015/

Web End =http://www.itrpv.net/Reports/Downloads/2015/ (2015).

6. Savage, L. Perovskite photovoltaics: hitting their stride. Opt. Photon. News 25, 2633 (2014).

7. U.S. Department of Energy. SunShot Vision Study. Report No. DOE/GO-102012-3037. http://www.osa-opn.org/opn/media/Images/PDF/2014/1114/26-33_Perovskite-Nov.pdf?ext=.pdf

Web End =http://www.osa-opn.org/opn/media/Images/PDF/2014/1114/26- http://www.osa-opn.org/opn/media/Images/PDF/2014/1114/26-33_Perovskite-Nov.pdf?ext=.pdf

Web End =33_Perovskite-Nov.pdf?ext=.pdf (Department of Energy, 2012).

8. Gong, J. et al. Perovskite photovoltaics: life-cycle assessment of energy and environmental impacts. Energy Environ. Sci. 8, 19531968 (2015).

9. Miller, J. et al. Dipolar aprotic solvents in bimoleculer aromatic nucleophilic substitution reactions. J. Am. Chem. Soc. 83, 117123 (1961).

10. Parker, A. J. The effects of solvation on the properties of anions in dipolar aprotic solvents. Q. Rev. Chem. Soc. 16, 163187 (1962).

11. Wakamiya, A. et al. Reproducible fabrication of efcient perovskite-based solar cells: x-ray crystallographic studies on the formation of CH3NH3PbI3 layers.

Chem. Lett. 43, 711713 (2014).12. Cui, J. et al. Metallurgical recovery of metals from electronic waste: a review.J. Hazard. Mater. 158, 228256 (2008).13. Hagelken, C. et al. Recycling of gold from electronics: cost-effective use through Design for Recycling. Gold Bull. 43, 209220 (2010).

14. Mohammadi, T. et al. Effect of operating parameters on Pb2 separation from wastewater using electrodialysis. Desalination 167, 379385 (2004).

15. Shah, D. B. et al. Lead removal from foundry waste by solvent extraction.J. Air Waste Manage. Assoc. 45, 150155 (1995).16. Ahmed, S. et al. The removal of cadmium and lead from aqueous solution by ion exchange with Na-Y zeolite. Sep. Purif. Technol. 13, 5764 (1998).

17. Axtell, N. R. et al. Lead and nickel removal using microspora and lemna minor. Bioresour. Technol. 89, 4148 (2003).

8 NATURE COMMUNICATIONS | 7:11735 | DOI: 10.1038/ncomms11735 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms11735 ARTICLE

18. Cui, L. et al. Mechanism of Pb(II) and methylene blue adsorption onto magnetic carbonate hydroxyapatite/graphene oxide. RSC Adv. 5, 97599770 (2015).

19. Lei, Y. et al. Bioinspired fabrication and lead adsorption property of nanohydroxyapatite/chitosan porous materials. RSC Adv. 5, 9878398795 (2015).

20. Kutes, Y. et al. Direct observation of ferroelectric domains in solution-processed CH 3NH3PbI3 perovskite thin lms. J. Phys. Chem. Lett. 5, 33353339 (2014).

21. Lindblad, R. et al. Electronic structure of TiO2/CH3NH3PbI3 perovskite solar cell interfaces. J. Phys. Chem. Lett. 5, 648653 (2014).

22. Conings, B. et al. Perovskite-based hybrid solar cells exceeding 10% efciency with high reproducibility using a thin lm sandwich approach. Adv. Mater. 26, 20412046 (2013).

23. Roy, A. et al. Spectroscopic studies on quantum dots of PbI2. Spectrochim. Acta A Mol. Biomol. Spectrsc. 48, 17791787 (1992).

24. Moulder, J. F. et al. Handbook of X-ray Photoelectron Spectroscopy: a Reference Book of Standard Spectra for Identication and Interpretation of XPS Data (Physical Electronics Division Perkin-Elmer Corp, 1995).

25. Roldn-Carmona, C. et al. High efciency methylammonium lead triiodide perovskite solar cells: the relevance of non-stoichiometric precursors. Energy Environ. Sci. 8, 35503556 (2015).

26. Yang, M. et al. Square-centimeter solution-processed planar CH3NH3PbI3 perovskite solar cells with efciency exceeding 15%. Adv. Mater. 27, 63636370 (2015).

27. Cao, D. H. et al. Remnant PbI2, an unforeseen necessity in high-efciency hybrid perovskite-based solar cells? APL Mater. 2, 091101091107 (2014).

28. Lee, M. M. et al. Efcient hybrid solar cells based on meso-superstructured organometal halide perovskites. Science 338, 643647 (2012).

29. Wozny, S. et al. Controlled humidity study on the formation of higher efciency formamidinium lead triiodide-based solar cells. Chem. Mater. 27, 48144820 (2015).

30. Jiang, S.-D. et al. Fabrication of hydroxyapatite hierarchical hollow microspheres and potential application in water treatment. J. Phys. Chem. C. 116, 44844492 (2012).

31. Jeon, N. J. et al. Solvent engineering for high-performance inorganicorganic hybrid perovskite solar cells. Nat. Mater. 13, 897903 (2014).

Acknowledgements

This work was nancially supported by the Global Frontier R&D Program of the

Center for Multiscale Energy System (NRF-2012M3A6A7054855). The work

was also supported by the National Research Foundation of Korea (NRF), funded by

the Ministry of Education, Science and Technology (NRF-2012M3A7B4049967 and

NRF-2014R1A4A1008474). The work at the National Renewable Energy Laboratory

was supported by the U.S. Department of Energy under Contract No.

DE-AC36-08-GO28308.

Author contributions

B.J.K. and D.H.K. performed the device fabrication, characterization and analysis. S.L.K.,

Z.L. and K.Z. prepared the samples and synthesized the perovskite materials. S.Y.P.

synthesized the hydroxyapatite and performed the lead-removal processing. All authors

discussed the results and prepared the manuscript. H.S.J. planned and supervised the

project.

Additional information

Supplementary Information accompanies this paper at http://www.nature.com/naturecommunications

Web End =http://www.nature.com/

http://www.nature.com/naturecommunications

Web End =naturecommunications

Competing nancial interests: The authors declare no competing nancial interests.

Reprints and permission information is available online at http://npg.nature.com/reprintsandpermissions

Web End =http://npg.nature.com/

http://npg.nature.com/reprintsandpermissions

Web End =reprintsandpermissions .

How to cite this article: Kim, B.J. et al. Selective dissolution of halide perovskites as a

step towards recycling solar cells. Nat. Commun. 7:11735 doi: 10.1038/ncomms11735

(2016).

This work is licensed under a Creative Commons Attribution 4.0

International License. The images or other third party material in this

article are included in the articles Creative Commons license, unless indicated otherwise

in the credit line; if the material is not included under the Creative Commons license,

users will need to obtain permission from the license holder to reproduce the material.

To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/

Web End =http://creativecommons.org/licenses/by/4.0/

NATURE COMMUNICATIONS | 7:11735 | DOI: 10.1038/ncomms11735 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 9