Stability of Organic Cations in Solution-Processed CH3NH3PbI3 ...
Stability of Organic Cations in Solution-Processed CH3NH3PbI3 Perovskites: Formation of Modified Surface Layers A. Calloni,*,† A. Abate,‡ G. Bussetti,† G. Berti,† R. Yivlialin,† F. Ciccacci,† and L. Duò† †
Dipartimento di Fisica, Politecnico di Milano, Piazza Leonardo Da Vinci, 32, 20133, Milano, Italy
‡
Adolphe Merkle Institute, University of Fribourg, Ch. du. Musée 3, CH-1700 Fribourg, Switzerland
* corresponding author. E-mail: [email protected]
Figure S1 XPS Pb 4f (a), I 3d (b), N 1s (c), C 1s (d) and O 1s (e) profiles acquired from a representative as−received CH3NH3PbI3−xClx sample spin−coated on alumina. The spectra have been acquired at normal (continuous lines) and grazing (dashed lines) photoelectron emission, and are normalized to the Pb 4f spectral intensity. (f) Grazing−to−normal difference spectrum for the C 1s region. The two sets of spectra have been vertically offset for clarity. All spectra were analyzed by subtracting a mixed Shirley−linear type background (red line), and reproducing the photoemission features with synthetic line shapes (thin blue lines, individual components; thick blue line, envelope function). This spectral analysis is shown here only for the spectra acquired at normal emission and for the difference spectrum of panel (f). 1
Information on the line shape analysis performed on the photoemission spectra of Figure S1 The experimental data were processed by means of the automated routines available in the CasaXPS software.1 A mixed Shirley−linear type background was first subtracted. The mixing coefficient was optimized in order to account for the slope of the background at lower binding energies (BE) and, for the I 3d (Pb 4f) spectral region, in order to give the correct 3:2 (4:3) branching ratio between the intensities of the spin orbit−split peaks. The resulting spectra were reproduced with a sum of analytic profiles (GL, GaussianLorentzian functions). Each profile is composed as a weighted product of a Gaussian and a Lorentzian function, approximating a Voigt line shape.2 The weight of the Gaussian component varied from 30% for I 3d and Pb 4f to 70% for C 1s, O 1s and N 1s, in order to account for the (Gaussian) broadening due to disorder in the organic perovskite component and in (C− and O−containing) surface contaminants. For the Pb 4f and I 3d spectra, we could obtain satisfactory results by using symmetric GL functions. The calculated full width at half−maximum (fwhm) of the photoemission line shapes is about 1.1 eV (1.4 eV) for the Pb 4f (I 3d) spectrum, in good agreement with the result expected for an instrumental broadening of about 0.9 eV and a lifetime broadening of about 0.27 eV (0.56 eV) fwhm.3 A slight asymmetry is indeed detected on the low BE sides of the I 3d features accounting, however, for less than 1% of the total area of the main peaks. The N 1s spectrum is characterized by two features reproduced by two symmetric GL line shapes with a fwhm of about 1.6 eV. The origin of such features is thoroughly discussed in the manuscript. The interpretation of C 1s and O 1s line shapes is more complex. According to our experiment, a strong oxygen signal is detected only on the perovskite samples spin−coated on alumina, and is therefore related to oxide anions and contaminants like surface hydroxyls, water molecules and/or carboxyl groups, localized on the alumina surface. The reported binding energies for those compounds range from about 531 to 534 eV,4 compatible with the very broad (2.6 eV fwhm) and nearly symmetric line shape observed in the experiment. Similar to the O 1s case, the C 1s line shape shows a clear dependence on the perovskite substrate, due to the contribution of carbon contamination likely residing one the alumina surface. The signal characteristic of photoemission from surface species is isolated in Figure S1 by comparing the spectra acquired at grazing and normal photoelectron emission. Our procedure, in practice, leads to the subtraction of the signal due to the perovskite film from the C 1s line shape. The result (Figure S1f, black line) could be modelled as the sum of three features. The main one (about 80% of the total intensity) is related to aliphatic carbons (C−C or C−H), the second one (about 1.6 eV higher in energy) to alcohol functionalities (C−O) while the third one (3.8 eV higher in energy) is related to carboxyl groups (O−C=O).5 According to the results presented in the main text, the C 1s spectrum acquired on the evaporated samples (Figure 2b, “Si substrate”) is representative of photoemission from C species located within or at the surface of the perovskite film. No signal from oxide–related species is detected (consider, for instance, the almost nil O 1s intensity reported in Figure 2a, “Si substrate”). The C 1s line shape is instead characterized by a single broad peak (2.5 eV fwhm), probably encompassing the contribution from methyl carbons in various chemical environments (as expected from the presence of methylammonium ions and methylamine molecules). By mixing the two previous line shapes (contamination− and methyl−related, shown 2
in Figure S1d with dashed and continuous blue lines, respectively) it was possible to fit the C 1s spectra of spin−coated samples. The photoemission results related to a representative CH3NH3PbI3−xClx sample spin coated on alumina and subsequently submitted to in situ ultrahigh vacuum (UHV) annealing and sputtering are now reviewed. Strong modifications are observed with XPS in the Pb 4f line shape of the sputtered surface (Figure 3 in the main text and Figure S2 below), indicative of the formation of reduced Pb0 species, and in the relative intensity of the other photoemission features. According to Table S1, the subsequent application of sputtering and annealing treatments lowered the iodine content of the perovskite surface leading to the formation of a thin layer with a PbI2 stoichiometry. N and C atoms from the perovskite layer were almost completely sublimated during the annealing treatment, while carbon contamination was significantly reduced only by sputtering. The photoemission signal from O species, albeit reduced by the surface treatments, was always visible in the spectra, due to the unavoidable contribution of the mesoporous alumina substrate. Table 1S Intensity ratios for the main chemical species detected by XPS on the surface of a representative spin−coated CH3NH3PbI3−xClx sample, corrected by the elemental and orbital specific photoemission cross−sections. Numerical values are reported with an accuracy of ± 0.1. We also observed a large sample−to−sample variability in the concentration of C and O species, related to the contribution of surface contamination and to the presence of the alumina substrate. Surface condition
As-received
Annealed in UHV
Sputtered
⁄
2.9
2.1
2.1
⁄
1.3
0.2
~0
⁄
3.4
2.0
0.7
⁄
3.3
2.3
2.1
3
Figure S2 XPS Pb 4f profiles acquired from a spin-coated CH3NH3PbI3−xClx sample annealed for 1 h at 150 °C (red line) and sputtered after the annealing (green line), normalized to the intensity of the Pb2+ component. The line shape analysis of the spectrum acquired on the sputtered surface is shown with green lines (dashed line, Pb2+ component; dotted line, Pb0 component; dash−dotted line, envelope function). The result of the numerical analysis is validated by comparing the shape of the Pb0 component to the difference spectrum (black line) obtained by subtracting the spectrum from the annealed sample to the one of the sputtered sample. The spectra have been vertically offset for clarity.
Figure S3 UPS valence band of a representative as-received CH3NH3PbI3−xClx perovskite sample produced by spin−coating. Black (red) dots: spectrum acquired with He I (He II) photons. The spectra have been vertically offset for clarity. Black and red vertical bars mark the features associated with photoemission from N and C containing species and metal halides, respectively. The slight differences observed between the He I and He II spectra are mainly due to variations in the photoemission crosssections6 and in the inelastic mean free path of photoelectrons at the two photon energies.7 In the inset: a close−up of the He I spectrum, plotted with a logarithmic vertical scale,8 in the BE region close to the 4
Fermi energy (EF). The position of the valence band maximum is estimated according to the graphical method of ref 8. (1)
Fairley, N. CasaXPS http://www.casaxps.com.
(2)
Evans, S. Curve Synthesis and Optimization Procedures for X-Ray Photoelectron Spectroscopy. Surf. Interface Anal. 1991, 17, 85–93.
(3)
Campbell, J. L.; Papp, T. Widths of the Atomic K–N7 Levels. At. Data Nucl. Data Tables 2001, 77, 1–56.
(4)
Beamson, G.; Briggs, D. High Resolution XPS of Organic Polymers - The Scienta ESCA300 Database; John Wiley and Sons, Chichester, 1992.
(5)
McCafferty, E.; Wightman, J. P. Determination of the Concentration of Surface Hydroxyl Groups on Metal Oxide Films by a Quantitative XPS Method. Surf. Interface Anal. 1998, 26, 549–564.
(6)
Yeh, J. J.; Lindau, I. Atomic Subshell Photoionization Cross Sections and Asymmetry Parameters: 1 ≤ Z ≤ 103. At. Data Nucl. Data Tables 1985, 32, 1–155.
(7)
Tanuma, S.; Powell, C. J.; Penn, D. R. Calculations of Electron Inelastic Mean Free Paths. V. Data for 14 Organic Compounds over the 50-2000 eV Range. Surf. Interface Anal. 1994, 21, 165–176.
(8)
Schulz, P.; Edri, E.; Kirmayer, S.; Hodes, G.; Cahen, D.; Kahn, A. Interface Energetics in OrganoMetal Halide Perovskite-Based Photovoltaic Cells. Energy Environ. Sci. 2014, 7, 1377–1381.
Complete author list for references 34 and 43 in the main text (34)
Brambilla, A.; Calloni, A.; Aluicio-Sardui, E.; Berti, G.; Kan, Z.; Beaupré, S.; Leclerc, M.; Butt, H.J.; Floudas, G.; Keivanidis, P. E.; Duò L. X-Ray Photoemission Spectroscopy Study of Vertical Phase Separation in F8BT:PDI/ITO Films for Photovoltaic Applications. In SPIE NanoScience + Engineering; Banerji, N., Hayes, S. C., Silva, C., Eds.; International Society for Optics and Photonics: Bellingham, WA, 2014; pp 91650C/1−91650C/9.
(43)
Kim, H.-S.; Lee, C.-R.; Im, J.-H.; Lee, K.-B.; Moehl, T.; Marchioro, A.; Moon, S.-J.; Humphry-Baker, R.; Yum, J.-H.; Moser, J. E.; Grätzel, M.; Park, N.-G. Lead Iodide Perovskite Sensitized All-SolidState Submicron Thin Film Mesoscopic Solar Cell with Efficiency Exceeding 9%. Sci. Rep. 2012, 2, 591/1−591/7.
5
Dipartimento di Fisica, Politecnico di Milano, Piazza Leonardo Da Vinci, 32, 20133, Milano, Italy
‡
Adolphe Merkle Institute, University of Fribourg, Ch. du. Musée 3, CH-1700 Fribourg, Switzerland
* corresponding author. E-mail: [email protected]
Figure S1 XPS Pb 4f (a), I 3d (b), N 1s (c), C 1s (d) and O 1s (e) profiles acquired from a representative as−received CH3NH3PbI3−xClx sample spin−coated on alumina. The spectra have been acquired at normal (continuous lines) and grazing (dashed lines) photoelectron emission, and are normalized to the Pb 4f spectral intensity. (f) Grazing−to−normal difference spectrum for the C 1s region. The two sets of spectra have been vertically offset for clarity. All spectra were analyzed by subtracting a mixed Shirley−linear type background (red line), and reproducing the photoemission features with synthetic line shapes (thin blue lines, individual components; thick blue line, envelope function). This spectral analysis is shown here only for the spectra acquired at normal emission and for the difference spectrum of panel (f). 1
Information on the line shape analysis performed on the photoemission spectra of Figure S1 The experimental data were processed by means of the automated routines available in the CasaXPS software.1 A mixed Shirley−linear type background was first subtracted. The mixing coefficient was optimized in order to account for the slope of the background at lower binding energies (BE) and, for the I 3d (Pb 4f) spectral region, in order to give the correct 3:2 (4:3) branching ratio between the intensities of the spin orbit−split peaks. The resulting spectra were reproduced with a sum of analytic profiles (GL, GaussianLorentzian functions). Each profile is composed as a weighted product of a Gaussian and a Lorentzian function, approximating a Voigt line shape.2 The weight of the Gaussian component varied from 30% for I 3d and Pb 4f to 70% for C 1s, O 1s and N 1s, in order to account for the (Gaussian) broadening due to disorder in the organic perovskite component and in (C− and O−containing) surface contaminants. For the Pb 4f and I 3d spectra, we could obtain satisfactory results by using symmetric GL functions. The calculated full width at half−maximum (fwhm) of the photoemission line shapes is about 1.1 eV (1.4 eV) for the Pb 4f (I 3d) spectrum, in good agreement with the result expected for an instrumental broadening of about 0.9 eV and a lifetime broadening of about 0.27 eV (0.56 eV) fwhm.3 A slight asymmetry is indeed detected on the low BE sides of the I 3d features accounting, however, for less than 1% of the total area of the main peaks. The N 1s spectrum is characterized by two features reproduced by two symmetric GL line shapes with a fwhm of about 1.6 eV. The origin of such features is thoroughly discussed in the manuscript. The interpretation of C 1s and O 1s line shapes is more complex. According to our experiment, a strong oxygen signal is detected only on the perovskite samples spin−coated on alumina, and is therefore related to oxide anions and contaminants like surface hydroxyls, water molecules and/or carboxyl groups, localized on the alumina surface. The reported binding energies for those compounds range from about 531 to 534 eV,4 compatible with the very broad (2.6 eV fwhm) and nearly symmetric line shape observed in the experiment. Similar to the O 1s case, the C 1s line shape shows a clear dependence on the perovskite substrate, due to the contribution of carbon contamination likely residing one the alumina surface. The signal characteristic of photoemission from surface species is isolated in Figure S1 by comparing the spectra acquired at grazing and normal photoelectron emission. Our procedure, in practice, leads to the subtraction of the signal due to the perovskite film from the C 1s line shape. The result (Figure S1f, black line) could be modelled as the sum of three features. The main one (about 80% of the total intensity) is related to aliphatic carbons (C−C or C−H), the second one (about 1.6 eV higher in energy) to alcohol functionalities (C−O) while the third one (3.8 eV higher in energy) is related to carboxyl groups (O−C=O).5 According to the results presented in the main text, the C 1s spectrum acquired on the evaporated samples (Figure 2b, “Si substrate”) is representative of photoemission from C species located within or at the surface of the perovskite film. No signal from oxide–related species is detected (consider, for instance, the almost nil O 1s intensity reported in Figure 2a, “Si substrate”). The C 1s line shape is instead characterized by a single broad peak (2.5 eV fwhm), probably encompassing the contribution from methyl carbons in various chemical environments (as expected from the presence of methylammonium ions and methylamine molecules). By mixing the two previous line shapes (contamination− and methyl−related, shown 2
in Figure S1d with dashed and continuous blue lines, respectively) it was possible to fit the C 1s spectra of spin−coated samples. The photoemission results related to a representative CH3NH3PbI3−xClx sample spin coated on alumina and subsequently submitted to in situ ultrahigh vacuum (UHV) annealing and sputtering are now reviewed. Strong modifications are observed with XPS in the Pb 4f line shape of the sputtered surface (Figure 3 in the main text and Figure S2 below), indicative of the formation of reduced Pb0 species, and in the relative intensity of the other photoemission features. According to Table S1, the subsequent application of sputtering and annealing treatments lowered the iodine content of the perovskite surface leading to the formation of a thin layer with a PbI2 stoichiometry. N and C atoms from the perovskite layer were almost completely sublimated during the annealing treatment, while carbon contamination was significantly reduced only by sputtering. The photoemission signal from O species, albeit reduced by the surface treatments, was always visible in the spectra, due to the unavoidable contribution of the mesoporous alumina substrate. Table 1S Intensity ratios for the main chemical species detected by XPS on the surface of a representative spin−coated CH3NH3PbI3−xClx sample, corrected by the elemental and orbital specific photoemission cross−sections. Numerical values are reported with an accuracy of ± 0.1. We also observed a large sample−to−sample variability in the concentration of C and O species, related to the contribution of surface contamination and to the presence of the alumina substrate. Surface condition
As-received
Annealed in UHV
Sputtered
⁄
2.9
2.1
2.1
⁄
1.3
0.2
~0
⁄
3.4
2.0
0.7
⁄
3.3
2.3
2.1
3
Figure S2 XPS Pb 4f profiles acquired from a spin-coated CH3NH3PbI3−xClx sample annealed for 1 h at 150 °C (red line) and sputtered after the annealing (green line), normalized to the intensity of the Pb2+ component. The line shape analysis of the spectrum acquired on the sputtered surface is shown with green lines (dashed line, Pb2+ component; dotted line, Pb0 component; dash−dotted line, envelope function). The result of the numerical analysis is validated by comparing the shape of the Pb0 component to the difference spectrum (black line) obtained by subtracting the spectrum from the annealed sample to the one of the sputtered sample. The spectra have been vertically offset for clarity.
Figure S3 UPS valence band of a representative as-received CH3NH3PbI3−xClx perovskite sample produced by spin−coating. Black (red) dots: spectrum acquired with He I (He II) photons. The spectra have been vertically offset for clarity. Black and red vertical bars mark the features associated with photoemission from N and C containing species and metal halides, respectively. The slight differences observed between the He I and He II spectra are mainly due to variations in the photoemission crosssections6 and in the inelastic mean free path of photoelectrons at the two photon energies.7 In the inset: a close−up of the He I spectrum, plotted with a logarithmic vertical scale,8 in the BE region close to the 4
Fermi energy (EF). The position of the valence band maximum is estimated according to the graphical method of ref 8. (1)
Fairley, N. CasaXPS http://www.casaxps.com.
(2)
Evans, S. Curve Synthesis and Optimization Procedures for X-Ray Photoelectron Spectroscopy. Surf. Interface Anal. 1991, 17, 85–93.
(3)
Campbell, J. L.; Papp, T. Widths of the Atomic K–N7 Levels. At. Data Nucl. Data Tables 2001, 77, 1–56.
(4)
Beamson, G.; Briggs, D. High Resolution XPS of Organic Polymers - The Scienta ESCA300 Database; John Wiley and Sons, Chichester, 1992.
(5)
McCafferty, E.; Wightman, J. P. Determination of the Concentration of Surface Hydroxyl Groups on Metal Oxide Films by a Quantitative XPS Method. Surf. Interface Anal. 1998, 26, 549–564.
(6)
Yeh, J. J.; Lindau, I. Atomic Subshell Photoionization Cross Sections and Asymmetry Parameters: 1 ≤ Z ≤ 103. At. Data Nucl. Data Tables 1985, 32, 1–155.
(7)
Tanuma, S.; Powell, C. J.; Penn, D. R. Calculations of Electron Inelastic Mean Free Paths. V. Data for 14 Organic Compounds over the 50-2000 eV Range. Surf. Interface Anal. 1994, 21, 165–176.
(8)
Schulz, P.; Edri, E.; Kirmayer, S.; Hodes, G.; Cahen, D.; Kahn, A. Interface Energetics in OrganoMetal Halide Perovskite-Based Photovoltaic Cells. Energy Environ. Sci. 2014, 7, 1377–1381.
Complete author list for references 34 and 43 in the main text (34)
Brambilla, A.; Calloni, A.; Aluicio-Sardui, E.; Berti, G.; Kan, Z.; Beaupré, S.; Leclerc, M.; Butt, H.J.; Floudas, G.; Keivanidis, P. E.; Duò L. X-Ray Photoemission Spectroscopy Study of Vertical Phase Separation in F8BT:PDI/ITO Films for Photovoltaic Applications. In SPIE NanoScience + Engineering; Banerji, N., Hayes, S. C., Silva, C., Eds.; International Society for Optics and Photonics: Bellingham, WA, 2014; pp 91650C/1−91650C/9.
(43)
Kim, H.-S.; Lee, C.-R.; Im, J.-H.; Lee, K.-B.; Moehl, T.; Marchioro, A.; Moon, S.-J.; Humphry-Baker, R.; Yum, J.-H.; Moser, J. E.; Grätzel, M.; Park, N.-G. Lead Iodide Perovskite Sensitized All-SolidState Submicron Thin Film Mesoscopic Solar Cell with Efficiency Exceeding 9%. Sci. Rep. 2012, 2, 591/1−591/7.
5
