Abstract Previous measurements of the X-ray absorption spectra of PbCl 2 at the

Transcription

1 1 On the difficulty of reproducibly measuring PbCl 2 X-ray absorption spectra Craig P. Schwartz 1, David Prendergast 2* 1 Stanford Synchrotron Radiation Laboratory, SLAC National Accelerator Lab, Menlo Park CA, Molecular Foundry, Materials Sciences Division, Lawrence Berkeley National Laboratory, Berkeley CA, dgprendergast@lbl.gov Abstract Previous measurements of the X-ray absorption spectra of PbCl 2 at the chlorine K-edge have shown significant variation between different studies. Herein, using first principles simulations of XAS we show that the observed spectral variations are due to the generation of Cl 2 gas and depletion of chlorine from PbCl 2, consistent with what is observed during ultraviolent absorption for the same compound. We note that Cl 2 gas generation can also be initiated using higher resonant X-ray energies, including Pb X-ray absorption edges. While this casts doubt on previous interpretations of certain measurements, it does indicate a means of generating chlorine gas during in situ experiments by passing high energy x-rays through a hard x-ray transparent medium and onto PbCl 2. Introduction Near edge X-ray absorption fine structure (NEXAFS) spectroscopy is an element-specific probe of local electronic structure accessible via X-ray induced excitations from occupied core electronic orbitals to available unoccupied valence orbitals. 1 The technique can provide local structural information for noncrystalline systems and has been used to study a variety of solids, ranging from proteins to polymers to simple salts. 2-4 For the case of PbCl 2, a simple inorganic salt with a well-

2 2 defined crystalline phase under ambient conditions, spectral variation should be minimal between samples. 5-9 The published lead (II) chloride NEXAFS spectra at the Cl K-edge show significant variation, which cannot be explained via static or dynamical structural or conformational variations, based upon the expected, simple, orthorhombic Pnma crystal structure. 10, 11 Herein, we will show using first principles calculations that the observed experimental spectra are due to a combination of PbCl 2 and chlorine gas created by the x-rays. 12 We do this by simulating both PbCl 2 and chlorine gas form first principles and noting that the observed experimental spectra appear to be a combination of the two species. It is well established in the literature that PbCl 2 decomposes with loss of chlorine gas upon exposure to UV light above ~4.5 ev, 13 but the importance of radiation damage during the NEXAFS measurements was neglected. This may become increasingly important because PbCl 2 could potentially be used as a model system to study lead-based halide perovskites, currently being explored as hole conductors in third generation solar cells. 14 Given the use of Cl 2 gas in processes such as etching and the long penetration depth of X-rays, we note that PbCl 2 has potential use as a method of generating Cl 2 gas at buried interfaces in materials. 15, 16 Methods Experimental The Cl K edge NEXAFS measurements were taken from the literature. 10, 11 The literature spectra as published were aligned differently. This is corrected here by aligning all spectra to the primary peak of KCl at 2825eV. Theory All X-ray calculations were made using orbital-occupancy constrained density functional theory (DFT) 17, 18 within the XCH (X-ray core-hole) formalism,

3 3 detailed elsewhere. 19 The electronic structure of a core-excited chlorine atom was represented using a pseuodpotential generated with the following atomic electronic configuration: 1s 1 2s 2 2p 6 3s 2 3p 6. Structural models were based on the experimental crystal structures. 20 To test sensitivity to optically induced defects, Frenkel defects, believed to be the dominant native form of defects in PbCl2, were added to specific structural models. 13 Relaxed equilibrium structures were generated from these starting points by energy minimization within DFT using the PBE functional. 22 DFT electronic structure calculations were performed using Quantum ESPRESSO. 13, 21 The effect of finite temperature (thermal motion of the nuclei) on the X-ray absorption spectrum was added by sampling first-principles molecular dynamics trajectories for structural models using the VASP code. 23 A structural model for chloride-depleted PbCl2 (termed a colloid in the work of Plekhankov) 13 was generated by removing a large number of Cl atoms from a crystal supercell and then performing a structural relaxation to a local minimum of the total energy. Representative structures are shown in Figure 1. Spectral calculations were aligned based on the experimental reference system (KCl at 2825 ev) and a theoretical atomic alignment scheme. 24 Calculated spectra were numerically broadened using a Gaussian convolution of 1.0 ev FWHM. Results A comparison of two different experimental spectra (red, blue) with a variety of calculations is shown in Figure 2a. 10, 11 The experiment consists of a peak at ~ ev and a second peak at ~ ev, although the first peak intensity changed greatly between experiments. A well-separated peak is found at ~2827 ev and a significantly broader peak at ~2838 ev. All of the simulated PbCl 2 spectra,

4 4 regardless of the crystal structures (yellow-pnma, orange-pnam) and including the effect of finite temperature (light blue) or defects (green) all look very similar and lack the first peak seen in experiment. To ensure the quality of the sample, PbCl 4 (black) was investigated but it has a peak that is too low in energy while the spectrum of Cl 2 (purple) has a transition at the correct energy. Thus, of the calculated spectra, only Cl 2 has the correct transition energy. When the Cl 2 is imbedded inside a PbCl 2 crystal (purple thin dashes), the sharpness of the highenergy features disappears due to hybridization of the high energy excited states with available states of the surrounding solid. The experimental peak intensity associated with Cl 2 changes, consistent with a change in concentration of Cl 2. 10, 11 We thus must assign the first peak to Cl 2 contamination. This is consistent with what has been observed by other experimental techniques. 13 Based on previous light scattering experiments, lead-rich colloids of 1-10 nm in size form following exposure to ionizing radiation in the near surface layer. For this reason, a model spectrum was calculated from Cl depleted PbCl 2 (black dashed). It leads to a peak position between those of Cl 2 and PbCl 2. Shown in Figure 2b are the experiment (red, blue) 10, 11 compared to a linear combination of the calculated spectrum of PbCl 2, embedded Cl 2 and the calculated lead colloid (black, green). With respect to the molecular Cl 2 spectrum, we note the following: (1) In isolation (simulating the gas phase) the high-energy spectral features are far too sharp (even with 300K sampling) by comparison with existing gas phase Cl 2 measurements. This may possibly be due to using an insufficiently large supercell in our plane-wave periodic calculations or an inadequate treatment

5 5 of excitations above the ionization potential. (2) For model systems which place one Cl 2 molecule within a void in the PbCl 2 crystal, we find that the low energy absorption onset peak shifts slightly to higher energies due to interactions with the surrounding crystal, but we regard this effect as negligible, given that at separations as small as 2Å the main transition energy is still within.01 ev of the isolated case. (3) For Cl 2 embedded in PbCl 2, the high energy spectral features are entirely blurred out, which, is consistent with experiment in the combination spectrum (PbCl 2 :Cl 2 :colloid). Discussion It seems clear now that the previously measured spectra exhibit absorption intensity consistent with at least 15% contribution from Cl 2 gas. It is important to note that this does not mean that the entire sample comprises 15% Cl 2. Absorption intensity from Cl 2 could originate from several sources: (1) gas phase Cl 2 in the vacuum chamber (produced by the measurement itself); (2) emergent Cl 2 at the sample surface; and (3) Cl 2 generated in the sub-surface but trapped inside the crystal. Additionally, the Cl 2 signal could be artificially enhanced over that of the native PbCl 2 due to the formation of a (proposed) lead-rich region near the surface which would strongly attenuate x-rays and prevent them from reaching the intact PbCl 2 region. The scenario proposed here is consistent with previous work on the mechanism of Cl 2 generation upon UV light exposure. In the present case, X-ray exposure can lead to chemical processes driven by inelastic scattering, UV fluorescence or X-ray induced electron emission. 13 Following core ionization, it is also possible that the excited-state forces drive the ionized chlorine atom in the

6 6 direction of a neighboring chlorine atom, perhaps initiating the process that leads to Cl 2 formation. The effect of having chlorine gas in the spectrum of lead (II) chloride has strong implications on previous work. 10, 11, 25 In particular, previous work on fly ash at the Chlorine K-edge which uses comparisons to a contaminated PbCl 2 to determine the amount of PbCl 2 present in the spectrum cannot be interpreted as simply as the authors had stated. 10, 11 Furthermore, it is quite possible that some of the photogenerated Cl 2 is then reacting again, which would affect the other components of the flyash sample, contaminating those results. In particular, it is difficult to determine if PbCl 2 is actually an adjuster in the amount of generation of aromaticchlorides as the authors stated or rather Cl 2 generated by the measurement is an adjuster in the generation of aromatic-chlorides. Even when the measurements are not on the chlorine edge, the risk of contamination is quite real and has been seen frequently historically. 13 Given previous results, one should expect that chlorine gas would be formed upon exposure to photons at energies above its UV onset energy, which means that some of the PbCl 2 data at the L-edge is potentially contaminated and something akin to a combination of Pb and PbCl 2 will be measured. Because this can be caused by any phenomena above 5 ev including inelastic scattering, reabsorption of fluorescence and electron scattering, that this processes will occur to some extent at the lead L- edge. We note that some amount of Pb would be consistent with the observed PbCl 2 spectra shown in the literature. 26 The amount cannot be determined at present on the Pb L-edge due to the available literature being of insufficient resolution, but if

7 7 chlorine gas is present the risk of further reactions contaminating samples must be considered and may make comparisons to previous work difficult. 27 At present, we do not see a simple way around this problem in photoabsorption measurements, as the light itself will induce the formation of chlorine gas. A lower photon flux would mitigate the problem, but one would have to carefully raster a very homogeneous sample in order to solve the problem. However, we note that this is a potentially useful reaction to generate chlorine gas in-situ. Due to the relatively long penetration depth of x-rays at these energies (tens of microns for low-z materials), this reaction could be used to generate chlorine gas beneath a buried interface. 28 Given the use of chlorine in etching and driving certain reactions, clever experiments may well find a use for this reaction. 15, 16 Conclusion The reason for variations across different experiments in the measured Cl K-edge NEXAFS spectra of PbCl 2 is due to generation of Cl 2 gas following x-ray exposure. We estimate the observed spectrum is composed of at least ~15% chlorine gas, and these findings make the interpretation of the amount of PbCl 2 in NEXAFS measurements difficult. We expect that some Cl 2 will be produced at all x-ray energies, including the Pb L-edges. Due to the long penetration depth of x-rays, this can be useful as a means to generate chlorine gas in-situ. Acknowledgements - The Molecular Foundry are National User Facilities operated by the University of California Berkeley for the US Department of Energy, grant DE- AC02-05CH11231 respectively. The Molecular Foundry portion of this work was performed as part of a user proposal. CPS was supported by EERE Bridge project

8 The calculations were performed at the National Energy Research Scientific Computing Center.

9 9 Figures Figure 1

10 10

11 Figure 1 Representative structures of PbCl 2, PbCl 4 and a lead rich colloid used in this paper. The colloid has the structure of Pb 4 Cl and is intended to be indicative of a potential colloid structure. Chlorine atoms are green, lead are black. 11

12 12 Figure 2 a) b) Figure 2 a) Comparison of simulated spectrum of PbCl 4 (black), Cl 2 at 0K (purple solid), 300K (purple broad dashes) and embedded in PbCl 2 (purple thin dashes), the two ambient pressure crystal structures of PbCl 2 (yellow-pnma and orange-pnam respectively), the common crystal structure with Frenkel defects (green), a PbCl 2 crystal at 300K (light blue), a ~1nm lead rich colloid (black dashed) and experimental spectra from the literature (dark blue red). All of the PbCl 2 structures are very similar and lack the first peak found in experiment. b) A comparison of experimental spectrum (red and blue) with the embedded Cl 2 and lead rich colloid added to the PbCl 2 spectrum at 300K.

13 13 References 1. Stöhr, J., NEXAFS Spectroscopy. Springer: Hudson, E.; Moler, E.; Zheng, Y.; Kellar, S.; Heimann, P.; Hussain, Z.; Shirley, D. A. Phys. Rev. B 1994, 49, (6), Kawatsuki, N.; Inada, Y.; Goto, K.; Kondo, M.; Haruyama, Y.; Matsui, S. Mol. Cryst. Liquid Cryst. 2014, 601, (1), Milosavljevic, A. R.; Canon, F.; Nicolas, C.; Miron, C.; Nahon, L.; Giuliani, A. J. Phys. Chem. Lett. 2012, 3, (9), Zubavichus, Y.; Shaporenko, A.; Grunze, M.; Zharnikov, M. J. Phys. Chem. A 2005, 109, (32), Zubavichus, Y.; Shaporenko, A.; Grunze, M.; Zharnikov, M. J. Phys. Chem. B 2006, 110, (7), Zubavichus, Y.; Shaporenko, A.; Grunze, M.; Zharnikov, M. J. Phys. Chem. B 2007, 111, (33), Zubavichus, Y.; Shaporenko, A.; Korolkov, V.; Grunze, M.; Zharnikov, M. J. Phys. Chem. B 2008, 112, (44), Zubavichus, Y.; Zharnikov, M.; Schaporenko, A.; Grunze, M. J. Electron Spectrosc. Relat. Phenom. 2004, 134, (1), Fujimori, T.; Fujinaga, Y.; Takaoka, M. Environ. Sci. Technol. 2010, 44, (19), Zhu, F.; Takaoka, M.; Shiota, K.; Oshita, K.; Kitajima, Y. Environ. Sci. Technol. 2008, 42, (11), Fronzoni, G.; Stener, M.; Decleva, P. PCCP Phys. Chem. Chem. Phys. 1999, 1, (7), Plekhanov, V. G. Prog. Mater. Sci. 2004, 49, (6), Unger, E. L.; Bowring, A. R.; Tassone, C. J.; Pool, V. L.; Gold-Parker, A.; Cheacharoen, R.; Stone, K. H.; Hoke, E. T.; Toney, M. F.; McGehee, M. D. Chem. Mat. 2014, 26, (24), Bell, F. H.; Joubert, O. J. Vac. Sci. Technol. B 1996, 14, (6), Zhang, L.; Sun, Y. Y.; Ma, Y. Q.; Zhao, J. R.; Feng, Y.; Yin, J. H. Polym. Bull. 2009, 63, (3), Hohenberg, P.; Kohn, W. Phys. Rev. B 1964, 136, (3B), B Kohn, W.; Sham, L. J. Phys. Rev. 1965, 140, (4A), 1133-&. 19. Prendergast, D.; Galli, G. Phys. Rev. Lett. 2006, 96, (21), Bergerhoff, G.; Hundt, R.; Sievers, R.; Brown, I. D. J. Chem. Inf. Comput. Sci. 1983, 23, (2), Giannozzi, P.; Baroni, S.; Bonini, N.; Calandra, M.; Car, R.; Cavazzoni, C.; Ceresoli, D.; Chiarotti, G. L.; Cococcioni, M.; Dabo, I.; Dal Corso, A.; de Gironcoli, S.; Fabris, S.; Fratesi, G.; Gebauer, R.; Gerstmann, U.; Gougoussis, C.; Kokalj, A.; Lazzeri, M.; Martin-Samos, L.; Marzari, N.; Mauri, F.; Mazzarello, R.; Paolini, S.; Pasquarello, A.; Paulatto, L.; Sbraccia, C.; Scandolo, S.; Sclauzero, G.; Seitsonen, A. P.; Smogunov, A.; Umari, P.; Wentzcovitch, R. M. J. Phys.-Condes. Matter 2009, 21, (39), 19.

14 22. Perdew, J. P.; Burke, K.; Ernzerhof, M. Phys. Rev. Lett. 1996, 77, (18), Kresse, G.; Hafner, J. Phys. Rev. B 1993, 47, (1), England, A. H.; Duffin, A. M.; Schwartz, C. P.; Uejio, J. S.; Prendergast, D.; Saykally, R. J. Chem. Phys. Lett. 2011, 514, (4-6), Fujimori, T.; Tanino, Y.; Takaoka, M. Environ. Sci. Technol. 2013, 47, (5), Yu, Y. H.; Tyliszczak, T.; Hitchcock, A. P. J. Phys. Chem. Solids 1990, 51, (5), Fujimori, T.; Tanino, Y.; Takaoka, M. Environ. Sci. Technol. 2014, 48, (1), Henke, B. L.; Gullikson, E. M.; Davis, J. C. Atomic Data and Nuclear Data Tables 1993, 54, (2),

15

16

17