Toward a comprehensive understanding of solid-state core-level XPS linewidths: Experimental and theoretical studies on theSi 2pandO 1slinew

Abstract

High resolution X-ray Photoelectron Spectroscopy ͑XPS͒ core-level Si 2p and O 1s spectra of the nonconductors ␣-SiO 2 ͑quartz͒ at 120 and 300 K and vitreous SiO 2 at 300 K were obtained with a Kratos Axis Ultra XPS instrument ͑instrumental resolution of Ͻ0.4 eV͒ which incorporates a unique charge compensation system that minimizes differential charge broadening on nonconductors. The Si 2p and O 1s linewidths at 300 K ͑ϳ1.1 and ϳ1.2 eV, respectively͒ are similar for all silicates ͑and similar to previous thin film SiO 2 spectra obtained previously͒, showing that differential charging does not contribute significantly to our spectra. At 120 K, there is a small decrease ͑0.04 eV͒ in the Si 2p linewidth of ␣-SiO 2 , but no measurable decrease in O 1s linewidth. The O 1s lines are generally and distinctly asymmetric. We consider all possible sources of line broadening and show that final state vibrational broadening ͑FSVB͒ and phonon broadening are the major causes of the broad and asymmetric lines. Previous high resolution gas phase XPS studies have identified large FSVB contributions to the Si 2p spectra of SiCl 4 , SiF 4 , and Si͑OCH 3 ͒ 4 molecules, and this vibrational structure leads total Si 2p 3/2 linewidths of up to ϳ0.5 eV, even with individual peak linewidths of Ͻ0.1 eV. The Si atom of Si͑OCH 3 ͒ 4 is an excellent analog for Si in crystalline SiO 2 because the Si-O bond lengths and symmetric stretch frequencies are similar in both compounds. Similar vibrational contributions to the Si 2p and O 1s spectra of solid silicates are anticipated if the Si 2p and O 1s core-hole states produce similar changes to the Si-O bond length in both phases. To investigate the possibility, Car-Parrinello molecular dynamics calculations were performed and show that changes to Si-O bond lengths between ion and ground states ͑⌬r͒ for both Si 2p and O 1s hole states are similar for both crystalline SiO 2 and gaseous Si͑OCH 3 ͒ 4 . ⌬r are −0.04 Å for Si 2p and ϳ+0.05 Å for O 1s in both compounds. Indeed, the vibrational envelope from the Si 2p spectrum of Si͑OCH 3 ͒ 4 , broadened to our instrumental linewidth of 0.4 eV, accounts for the majority ͑ϳ0.8 eV͒ of the Si 2p 3/2 linewidth for crystalline SiO 2 ͑ϳ1.1 eV͒ with phonon broadening accounting for the remainder. The results provide excellent support for the tenet that final state vibrational splitting, as seen in the gas phase molecules, similarly affects the solid-state spectra. The calculations also indicate that the O 1s linewidths should be larger than the Si 2p linewidths, as observed in our spectra. FSVB should also lead to small peak asymmetries, as seen in the O 1s spectra. The contribution of phonon broadening to the linewidth is also evaluated and shown to be comparable to the FSVB contribution at 120 and 300 K but considerably smaller at very low temperatures.

Hard X-ray photoelectron spectra (HXPES) of bulk non-conductor vitreous SiO 2 : Minimum linewidths and surface chemical shifts

“…Second, using a thin metal films on the SiO 2 , we want to quantify the increase in escape depth for SiO 2 from 1486 eV to 5000 eV photon energies, and demonstrate that we can use the O 1s and Si 1s chemical shifts created by the interface Cr-subox to indirectly determine the unknown intensity of the expected [13,23,24], but not previously observed, surface peaks. Third, we will also show that the minimum Si 1s linewidths obtained are consistent with that expected from the final state vibrational broadening mechanism outlined previously for the Si 2p and O 1s linewidths [12][13][14][15][17][18][19]]. …”

“…Similar to the trend for the O 1s linewidth, the Si 1s linewidth increases at >4000 eV due to the increase in photon linewidth. The majority of the 0.35 eV linewidth difference between the Si 1s and Si 2p lines is due to the large difference in the two inherent lifetime linewidths: the Si 1s inherent width is 0.48 eV [33], compared to the Si 2p inherent linewidth of <0.1 eV [17,18,33]. The broadening mechanism should be very similar for both core levels, and the Si 1s final state vibrational effect should be similar to that for the Si 2p level.…”

“…[17]) compared to the total single peak linewidth of 0.4 eV for the Si 2p spectra. Using the correlation of total linewidth versus the single vibrational peak linewidth for the analogue Si (OCH 3 ) 4 ( Figure 5 in [17]), we obtain a total linewidth of 0.8 eV for the Si 2p level, and 1.0 eV for the Si 1s level. Thus, about 0.2 eV of the 0.35 eV increase in Si 1s linewidth is due to the much larger inherent Si 1s linewidth.…”

“…This is expected from the asymmetry of the final state vibrational envelope [13,15,16]. The constant BE, linewidth, and lineshape in XPS with changes in photon energy gives the technique an appreciable advantage over 17 O NMR, where the chemical shifts and linewidths vary with the magnetic field strength [30,31].…”

“…These problems have been overcome using modern charge compensation systems such as employed in the Kratos Axis Ultra XPS instruments [11][12][13][14]. For example, the O 1s linewidths from one O species on non-conducting silicate minerals and glasses using the Kratos XPS are 1.25 ± 0.1 eV (the minimum expected from theoretical considerations [17][18][19]), compared to O 1s linewidths of closer to 2 eV for instruments using conventional electron flood guns for charge compensation [13,16]. These much better linewidths are critical for optimum resolution of the BO and NBO peaks in silicates, and for accurate analysis of the BO, NBO and free oxide content of these glasses [12][13][14][15].…”