Improved surface analysis method and apparatus

Abstract

Improved surface analysis methods and apparatus are provided. [Solution] An improved surface analysis method is a method for generating an X-ray photoelectron spectrum of a sample, the method including the steps of generating a plurality of different oxidation states on the surface of the sample by exposing the surface of the sample to an agent configured to change the oxidation state of the surface of the sample, placing the sample in an X-ray photoelectron spectroscopy device, obtaining an X-ray photoelectron spectrum for each of a plurality of oxidation states of the surface of the sample, and identifying materials within the sample by analyzing the plurality of spectra.

Description

[Technical Field]

[0001] Concerning the chemical and physical analysis of materials using X-ray photoelectron spectroscopy. [Background technology]

[0002] In analytical laboratories, it is frequently necessary to chemically characterize small amounts of sample. Techniques such as X-ray photoelectron spectroscopy (XPS) (Non-Patent Document 1) and secondary ion mass spectroscopy (SIMS) can provide excellent chemical characterization of the top few nanometers of a surface. See an example of an XPS instrument in Figure 1 and a view through the viewport in Figure 2. A schematic cross-section of a typical commercial XPS instrument is shown in Figure 15. Throughout this specification, references to X-ray photoelectron (XPS) spectra include X-ray induced Auger peaks that appear in the same spectrum from an XPS instrument. The invention presented here should be equally valuable for elucidating the chemical composition of these Auger peaks strictly as photoelectron peaks. The terms "sample" and "specimen" are used interchangeably throughout.

[0003] In XPS, energy resolution is typically limited in laboratory-based instruments (compared to, for example, synchrotrons). Over the past two decades, there has been a trend toward monochromated X-ray sources in XPS, so that whereas simple anode X-ray sources became popular in the 1980s, nearly all new XPS instruments now include such monochromators to improve energy resolution, albeit at a high cost. The reason analysts need higher energy resolution is to distinguish between different chemical states in core-level peaks in a sample. In a recent survey on the LinkedIn® social network, the results of which are shown in Table 1, XPS users ranked peak fitting (i.e., resolving different chemical states within an XPS peak) as their biggest problem with XPS.

[0004] [Table 1]

[0005] Improving energy resolution helps minimize such peak overlap in core-level peaks, and technical improvements have been made over the years to make this easier in commercial instruments. However, it is often unavoidable. To improve energy resolution in a hemispherical electron energy analyzer, one must choose a larger analyzer (which is expensive) and / or move to a lower pass energy in that analyzer, which can reduce count rates and result in noisy spectra. Much of the art of operating an XPS system lies in selecting the correct parameters of pass energy (eV), spectral channel spacing (eV), and channel dwell time (sec, s) to give a spectrum that has the necessary resolution to resolve the peaks but also exhibits a good spectral signal-to-noise ratio.

[0006] Nevertheless, peaks often fail to resolve, which is a major problem in the surface analysis community. This problem has been discussed in detail in recent literature, for example, by Baer et al. (Non-Patent Literature 2). Based on an extensive review of papers published in 1949, Major et al. (Non-Patent Literature 3) state that "more than 65% of papers showing XPS spectra also show some degree of fit, which was the cause of the majority of errors." This is one of the most serious problems faced by a fairly large international analytical community. XPS allows for erroneous peak attribute and intensity measurements unless peak parameters and linkages are skillfully selected by the analyst. Major et al. (Non-Patent Literature 4) state that "high-resolution spectra of transition metals are among the most difficult spectra to peak fit. In fact, it is common to see transition metal peak fit with oversimplification of peak assignment to individual chemistrys, followed by misinterpretation of the results." Furthermore, transition metals are technically very important, with elements such as Hf and Ta being extremely important in semiconductor surface analysis, and a wide range of these elements being important in battery technology. Figure 21 shows the oxidation states of transition metals (quoted from Wikipedia). Black circles indicate common oxidation states, and white circles indicate possible but unlikely states. This figure clearly shows that transition metals can be found in many oxidation states in many possible compounds, and the core-level spectra from these transition metal elements are overlaid on XPS spectra in a way that is very confusing for most analysts except the most skilled.

[0007] While many analysts try their best with reasonable techniques, peak fitting to such poorly decomposed peaks is often criticized in the literature as erroneous and misleading.

[0008] The usual cause of multiple unresolved peaks in narrow XPS spectra is that the surface contains multiple oxidation states of one or more elements or compounds. In principle, it is possible to tabulate the binding energies of all chemical states of an element. This is a good first guide, and analysts often turn to the NIST database (Non-Patent Document 5) when dealing with a particular element. However, using this method exclusively presents real challenges. 1. Some oxidation states of some elements are not in the database. 2. It may be uncertain that the binding energy has been obtained over a long period of time by many different people using many different instruments and their respective binding energy calibration procedures. 3. A small error in the binding energy of a peak (e.g., one that remains unresolved between two other peaks in a spectrum) can result in a large error in the relative intensities of these three peaks. Even a 0.1 eV error in binding energy, the absolute limit of what is currently achievable even with repeated and expensive calibration steps, can result in large uncertainties in the relative intensities of adjacent unresolved peaks in a spectrum.

[0009] Therefore, some kind of "internal" reference or method is very useful. In other words, some way to change the relative intensities of different unresolved peaks away from their nominal values. This can sometimes be achieved by tilting the sample, which can be achieved if different chemical states exist at different depths. Often, the surface of the sample is fairly uniform, so they do not occur at different depths. In addition, tilting is often not easy with modern instruments that have large sample holders (Non-Patent Document 6). Tilting can also change the inelastic background beneath the peaks, which adds further complexity to peak fitting.

[0010] One technique that may prove useful is to lightly or briefly sputter the surface with energetic ions (most XPS systems have an ion gun installed to allow sputter depth profiling). This introduces damage, often removing some oxidizing species close to the surface or chemically reducing the oxidation states of other species. It can be useful to compare narrow-scan spectra containing overlapping chemical states before and after this brief sputtering. However, this has limitations: operating the sputter gun for extended periods will remove all chemical states from the surface, leaving behind a spectrum of damaged bulk material. This is easily accomplished because these guns are designed to remove material quickly. Worse yet, if little is chemically reduced at the surface, further reduction by sputtering will barely change the spectrum and convey little about the peaks that are present (oxidizing is desired instead).

[0011] Nevertheless, brief sputtering is often useful and can be performed quickly in the XPS analysis chamber itself. In a sense, it is the opposite process to that of the present invention, so that the two can be usefully combined, and indeed spectra from both can be usefully analyzed within the same principal component analysis (PCA), singular value decomposition (SVD), or nonnegative matrix factorization (NMF) methods.

[0012] Ironically, as the speed of commercially available XPS systems increases, more spectra are generated each year, but there are only a limited number of skilled personnel to interpret them. The cost of XPS analysis is shifting from the cost of instrument time to the cost of interpretation (which often includes peak fitting). One feature of the invention described herein is to use the extra instrument time (which is inexpensive) to reduce the analyst time (which is expensive) spent on interpretation (by generating more spectra, which removes the ambiguity that exists when only one is recorded), improving the reliability of the results and the ability to demonstrate to readers of analytical reports and scientific publications that the correct conclusions have been reached.

[0013] Oxidizing the surface of a sample using UV light and ozone can provide more information about the surface chemistry (specifically, XPS peak shapes) than can be obtained from the "as-received" spectrum alone. For many types of samples, this works very well. For some, especially those where the material within the XPS sampling depth is already at or near its highest oxidation state, exposure to UV and ozone can barely change their oxidation state. Therefore, in this invention, we describe both oxidation and reduction as surface options for the sample, so that even XPS spectra from such initially oxidized samples can be resolved by the multivariate methods described above.

[0014] The UV-induced reduction of copper oxide in a hydrogen atmosphere has been demonstrated using XPS (Non-Patent Document 9), but for a different purpose than the present invention and within a timescale suitable for application as an analytical method for a range of sample types, in this case approximately 10 times faster in the presence of hydrogen than 254 nm UV exposure alone. The UV-induced reduction of graphene oxide has been demonstrated for a different purpose than the present invention (Non-Patent Document 10), but nevertheless suggests that UV-induced reduction of sample surfaces may be useful in XPS of carbonaceous materials in addition to transition metals.

[0015] There is evidence that UV exposure can cause anomalous oxides to form that replace other elements (i.e., those not normally expected even after long periods of time in air at room temperature or high temperatures) rather than simply following a linear path between oxidation and reduction (Non-Patent Literature 11) (for example, for Ni, or for semiconductor surfaces (Non-Patent Literature 12)), which is extremely useful for elucidating peak structures in XPS.

[0016] US Pat. No. 5,699,499 describes an X-ray photoelectron spectroscopy system and method for surface analysis. US Pat. No. 5,629,493 describes the use of photoelectron spectroscopy to determine layer thickness. US Patent No. 5,629,999 describes an instrument for surface analysis using scanning and high resolution X-ray photoelectron spectroscopy and imaging. [Prior art documents] [Patent documents]

[0017] [Patent Document 1] U.S. Patent No. 7875857 [Patent Document 2] U.S. Patent No. 7420163 [Patent Document 3] U.S. Patent No. 5315113 [Non-patent literature]

[0018] [Non-Patent Document 1] Fred A Stevie and Carrie L Donley, Introduction to x-ray photoelectron spectroscopy, J. Va. Sci. Technol. A38, 063204(2020) doi:10.1116 / 6.0000412 [Non-patent document 2] DR Baer et al., Practical guides for x-ray photoelectron spectroscopy: First steps in planning, conducting and reporting XPS measurements, J. Va. Sci. Technol. A37(3), May / June 2019, 031401-1 [Non-patent document 3] GH Major et al., Evaluation of the frequency and nature of erroneous X-ray photoelectron spectroscopy in scientific literature, J. Va. Sci. Technol. A38, 061204(2020) doi:10.1116 / 6.0000685 [Non-patent document 4] GH Major et al., Practical Guide for curve fitting in x-ray photoelectron spectroscopy, J. Vac. Sci. Technol. A38, 061203(2020); doi:10.1116 / 6.0000377. [Non-Patent Document 5] https: / / srdata.nist.gov / xps / [Non-patent document 6] F Stevie et al., J. Vac. Sci. Technol. A38, 063202(2020);doi:10.1116 / 6.0000421 [Non-Patent Document 7] Kevin M. McEvoy, Michel J. Genet and Christine Dupont, Principal Component Analysis: A Versatile Method for Processing and Investigation of XPS Spectra, September 2008, Analytical Chemistry 80(19):7226-38 [Non-patent document 8] https: / / en.wikipedia.org / wiki / Singular_value_decomposition [Non-Patent Document 9] TH Fleisch, GJ Mains, Reduction of copper oxide by UV radiation and atomic hydrogen studied by XPS, Applications of Surface Science, Volume 10, Issue 1, 1982, pp. 51-62, ISSN 0378-5963, https: / / doi.org / 10.1016 / 0378-5963(82)90134-9 [Non-Patent Document 10] Graeme Williams, Brian Seger, and Prashiant V. Kamat, TiO2-graphene nanocomposite materials. UV-induced photocatalytic reduction of graphene oxide, ACS Nano 2008, 2, 7, 1487-1491 https: / / doi.org / 10.1021 / nn800251f [Non-Patent Document 11] Deng, Sh., Lu, H., and Li, DY: Effects of UV irradiation on the corrosion behavior of electrodeposited Ni and Cu nanocrystalline foils. Sci Rep 10, 3049(2020). https: / / doi.org / 10.1038 / s41598-020-59420-6 [Non-Patent Document 12] Yit Lung Khung, Siti Hawa Ngalim, Andrea Scaccabarozzi and Dario Narducci, Beilstein J Nanotechnol. 2015; 6:19-26. [Non-Patent Document 13] Waymouth, John (1971). The Discharge Lamp. Cambridge, MA:MIT Press. ISBN 978-0-262-23048-3 [Non-Patent Document 14] H Amandusson, L.-G Ekedahl, H Dannetun, Hydrogen permeation through surface-modified Pd and PdAg membranes, Journal of Membrane Science, Volume 193, Issue 1, 2001, Pages 35-47, ISSN 0376-7388, https: / / doi.org / 10.1016 / S0376-7388(01)00414-8 [Non-Patent Document 15] AG Knapton, Palladium Alloys for Hydrogen Diffusion Membranes, Platinum Metals Rev., 1977, 21, (2)p44-50 [Non-Patent Document 16] Xueni (Denton Sun) et al., Sustained enhancement of adsorption desulfurization on TiO_2 after one-time ex-situ UV treatment, Fuel 193 (2017) pp. 95-100 [Non-Patent Document 17] John R. Vig, “UV / ozone cleaning of surfaces”, Journal of Vacuum Science & Technology A 3, 1027-1034(1985) https: / / doi.org / 10.1116 / 1.573115 [Non-Patent Document 18] Photodissociation of ozone in the Hartley band. Theoretical analysis, J. Chem. Phys. 123, 074305(2005); https: / / doi.org / 10.1063 / 1.2001650, Z.-W. Qu, H. Zhu, S. Yu. Grebenshchikov and R. Schinke [Non-Patent Document 19] Shangwei Huang et al., 2020 J. Electrochem. Soc. 167 090538 [Non-Patent Document 20] Jeong, BJ and Jo, YN, A Study on the Self-Discharge Behavior of Zinc-Air Batteries with CuO Additives. Appl. Sci. 2021, 11, 11675. https: / / doi.org / 10.3390 / app112411675 [Non-Patent Document 21] For example, Varta Type V 150 H2 MF. https: / / www.varta-ag.com / en / industry / product-solutions / hydrogen [Non-Patent Document 22] Peter J. Cumpson, MP Seah and SJ Spencer, Simple Procedure for Precise Peak Maximum Estimation for Energy Calibration in AES and XPS, September 1996, Surface and Interface Analysis 24(10):687-694 DOI:10.1002 / (SICI)1096-9918(19960930)24:103.0.CO;2-Q Summary of the Invention [Problem to be solved by the invention]

[0019] The present invention has been made in view of the above, and an object of the present invention is to provide an improved method and apparatus for surface analysis. [Means for solving the problem]

[0020] According to a first aspect of the present invention, there is provided a method for producing an X-ray photoelectron spectrum of a sample. The steps include: exposing the surface of the sample to a substance configured to change the oxidation state of the sample's surface, thereby generating a plurality of different oxidation states on the surface of the sample; placing the sample in an X-ray photoelectron spectrometer; obtaining an X-ray photoelectron spectrum for each of a plurality of oxidation states of the surface of the sample; The X-ray photoelectron spectra obtained for each of the multiple oxidation states on the surface of the sample are compared with the others, and component peaks are numerically extracted from each of the obtained X-ray photoelectron spectra using a computer. and identifying substances within the sample by detecting the presence of a substance in the sample.

[0021] Preferably, the sample is exposed multiple times in succession to an active substance configured to change the oxidation state of the sample's surface, wherein in each subsequent exposure of the sample to the active substance, the oxidation state of the sample's surface changes relative to the oxidation state of the sample's surface resulting from the prior exposure to the active substance configured to change the oxidation state of the sample's surface. The sample may be divided into multiple subsamples, each having a subsample surface, and different oxidation states of the subsample surface are generated for each subsample. The agent configured to change the oxidation state of the surface of the specimen may be a gaseous agent. The agent configured to change the oxidation state of the surface of the specimen may be one or more of ultraviolet light, ozone, and hydrogen. Advantageously, ultraviolet light is provided by at least one ultraviolet (UV) lamp, and the UV light emitted from at least one UV lamp is directed towards the surface of the sample. The UV light emitted from the at least one UV lamp may be in the wavelength range of 200 nm to 300 nm.

[0022] Ozone is necessary for the oxidation of the sample surface. Oxidation is faster with any UV in the 200-300 nm wavelength range. For many sample materials, reduction of the sample surface can be achieved slowly with UV in the 200-300 nm wavelength range alone when performed in a vacuum, or about 10 times faster in the presence of hydrogen. Preferably, the UV lamp is a mercury lamp. Ozone may be supplied by an ozone generator that generates ozone gas at a concentration in the range of 0.01 to 20 ppm in the gas surrounding the sample. This method may include a step of controlling the degree of change in the oxidation state of the sample surface by controlling one or more of the following: the exposure time of the sample surface to the reagent, the concentration of the reagent, and the wavelength and / or frequency of the reagent. The step of identifying substances in a sample by analyzing multiple spectra may include performing multivariate analysis, such as principal component analysis or non-negative matrix factorization.

[0023] According to a second aspect of the present invention, an apparatus for capturing an X-ray photoelectron spectrum (XPS) is provided, configured to perform the process of the first aspect of the present invention. The equipment for capturing X-ray photoelectron spectra (XPS) is Sample holder and a source of the agent configured to change the oxidation state of a surface of a sample held in the sample holder; means for controlling the exposure of the surface of the sample to the agent configured to alter the oxidation state of the surface; and an X-ray photoelectron spectrometer capable of recording multiple XPS spectra, one for each oxidation state of the surface of the sample.

[0024] The apparatus may further comprise a data processor configured to perform principal component analysis. Preferably, the sample holder of the device is housed within a housing. The agent configured to change the oxidation state of the surface of the specimen may be a gaseous agent. The agent configured to change the oxidation state of the surface of the specimen may be one or more of ultraviolet light, ozone, and hydrogen. Ultraviolet light may be provided by at least one ultraviolet (UV) lamp, and the UV light emitted from at least one UV lamp is directed towards the surface of the sample. Preferably, the UV light emitted from the at least one UV lamp is in the wavelength range of 200 nm to 300 nm.

[0025] The at least one UV lamp may be a mercury vapor lamp. The apparatus may further comprise an ozone generator configured to emit ozone around the sample located in the sample holder. The ozone generator may be a UV lamp emitting in the wavelength range of 100 nm to 300 nm. Advantageously, the UV lamp of the ozone generator is capable of emitting ultraviolet light of 185 nm and / or 254 nm. Advantageously, the UV lamp of the ozone generator is a mercury vapor lamp.

[0026] The apparatus may further comprise a hydrogen source configured to release hydrogen around the sample in the sample holder. Advantageously, the hydrogen source is at least one zinc-air battery. Preferably, the zinc-air battery operates in the absence of oxygen, for example in a partial vacuum. The sample holder may be adapted to hold a plurality of sub-samples, each sub-sample having a surface with a different oxidation state, and the X-ray photoelectron spectrometer is configured to record an XPS spectrum for each of the sub-samples. [Brief explanation of the drawings]

[0027] [Figure 1] Diagram of a typical commercial XPS instrument. Note the stainless steel ultra-high vacuum (UHV) chamber and ports. The computer that runs the system is not shown here. [Figure 2] FIG. 2 is a typical view through a viewport into a system similar to that shown in FIG. 1. [Figure 3] Computer-aided design of one type of commercially available sample holder and a matching disk-shaped sample "stub." [Figure 4] A diagram of another commercially available sample holder, this time from a Thermo K Alpha XPS instrument. In normal operation, this sample block is transported through the instrument in ultra-high vacuum (UHV) to the analysis position. [Figure 5]This is a schematic diagram of one embodiment of the UV / ozone or UV / hydrogen exposure apparatus of the present invention. [Figure 6] FIG. 1 is a schematic diagram showing how oxidizing species are generated and react with the surface of a sample (substrate) during UV / ozone cleaning. [Figure 7] 1 illustrates a general procedure for using the spectral acquisition method of the UV / ozone option of the present invention. [Figure 8] 10. UV / ozone exposure of carbonaceous contaminants on surfaces - XPS spectra with increasing exposure duration. [Figure 9] A composite spectrum with three identical component peaks at binding energies of 1.5, 2.5 and 3.5 eV is shown. [Figure 10] (a) to (f) show composite spectra for increasing numbers of UV / ozone exposure steps. [Figure 11] Figure 10 shows the results of singular value decomposition (SVD) of the spectrum. [Figure 12] Similar to Figure 9, but showing three composite component peaks with reduced energy separation. These are 2 eV, 2.5 eV, and 3 eV, but without knowing this, it would be difficult to determine how many peaks were actually below this envelope and what energies or widths they might have. [Figure 13] 13 shows a composite spectrum based on a model of the three closely separated peaks shown in FIG. 12. [Figure 14] Figure 14 shows the results of singular value decomposition (SVD) applied to the spectrum shown in Figure 13. The numerical labels indicate the center of each peak, as measured by fitting a parabola to five values ​​around the maximum. [Figure 15] FIG. 1 illustrates a typical commercially available XPS instrument configuration for use with the present invention. [Figure 16] One configuration of this experiment is illustrated, in which the sample is transferred in air from the XPS system to a housing (1610) containing a UV / ozone or UV generating lamp and hydrogen (1620). [Figure 17]Figure 17 shows another configuration of the present invention in which a UV / ozone containing enclosure is integrated into the entry lock of an XPS system, which requires a UV transparent window above the entry lock and a backfill gas cylinder (1700) to contain oxygen or an oxygen-containing gas mixture (e.g., air). [Figure 18] FIG. 1 shows a typical UV lamp type GTL3. [Figure 19] Figure 1 shows a typical high-power LED UV emitter with an emission wavelength of approximately 270 nm, which is sold for water disinfection in pools and bathtubs. [Figure 20] The Hartley absorption band of ozone is shown. Note that the 254 nm emission from a low-pressure mercury vapor lamp is near the peak of this absorption feature in the ozone spectrum. [Figure 21] The available oxidation states for elements with increasing atomic numbers are shown. Filled circles represent common oxidation states, while open circles represent uncommon oxidation states. [Figure 22a] The XPS spectra in the Ti2p region for potential marker particles (EPMPs) deposited on two different insulating polymers (a) and (b) are shown. Note that although these spectra were obtained in the same (Thermo k-Alpha model) instrument separated by a few minutes, the larger Ti2p3 / 2 peaks have slightly different apparent energies relative to the instrument energy scale, due to the establishment of slightly different charge equilibrium potentials in the two cases. [Figure 22b] The XPS spectra in the Ti2p region for potential marker particles (EPMPs) deposited on two different insulating polymers (a) and (b) are shown. Note that although these spectra were obtained in the same (Thermo k-Alpha model) instrument separated by a few minutes, the larger Ti2p3 / 2 peaks have slightly different apparent energies relative to the instrument energy scale, due to the establishment of slightly different charge equilibrium potentials in the two cases. [Figure 23]This is a schematic diagram showing how potential marker particles (EPMPs) are deposited by ion beam sputtering. [Figure 24] A general procedure for using the spectral acquisition method of the present invention to reduce the surface area of ​​a sample is shown. Here, the "hydrogen-containing chamber" is the same as the sample enclosure discussed herein when filled with hydrogen. [Figure 25] A composite spectrum with three identical component peaks at binding energies of 1.5, 2.5 and 3.5 eV is shown. [Figure 26] The synthesized spectra obtained by increasing the number of UV / hydrogen exposure steps from (a) to (f) are shown. [Figure 27] Figure 26 shows the results of singular value decomposition (SVD) of the spectrum. [Figure 28] Similar to Figure 25, but showing three composite component peaks with reduced energy separation. These are 2 eV, 2.5 eV, and 3 eV, but without knowing this, it is difficult to determine how many peaks there actually were under this envelope and what energies or widths they might have. [Figure 29] Figure 28 shows the synthesized spectrum based on the model of three closely separated peaks. [Figure 30] Figure 29 shows the results of singular value decomposition (SVD) applied to the spectrum shown. The numerical labels indicate the centers of each peak, as measured by fitting a parabola to five values ​​around the maximum value. [Figure 31] This document outlines two hydrogen-generating "button" cell batteries from Varta. [Figure 32a] 7 shows one embodiment of a hydrogen production device. In (a), the switch is open and no hydrogen is being produced by the button cell 720 or passing through the sealed palladium / palladium alloy tube 730. When the switch is closed (as shown in (b)) under the control of the programmable logic controller 750, a current predetermined by the value of resistor R flows through the cell, causing hydrogen production. [Figure 32b] An embodiment of a hydrogen production apparatus is shown. Under the control of a programmable logic controller 750, when the switch is closed (as shown in (b)), a current predetermined by the value of the resistor R flows through the cell, causing hydrogen production. The hydrogen permeates the sealed Pd / Pd alloy tube. [Figure 33] One possible simple embodiment B of the cell housing is shown. [Figure 34a] A schematic diagram of one possible embodiment of cell housing type B (with a pressure relief valve incorporated) is shown. The space labeled "internal housing" is the sample housing, and 930 is the cell housing. The idle state is shown in this diagram. [Figure 34b] A schematic diagram of one possible embodiment of cell housing type B (with a pressure relief valve incorporated) is shown. The space labeled "internal housing" is the sample housing, and 930 is the cell housing. The hydrogen generation state is shown in this diagram. [Figure 35] This figure shows an embodiment in which a sample is divided into sub-samples. [Figure 36] Another configuration of the present invention is shown, in which a housing containing a UV generating lamp and hydrogen is integrated with the entry lock of the XPS system. In particular, an arbitrary hydrogen generating cell is shown here (1780). DETAILED DESCRIPTION OF THE INVENTION

[0028] The improved surface analysis method and apparatus according to embodiments of the present invention will be described below with reference to the drawings. In multiple embodiments, substantially identical components are denoted by the same reference numerals, and their descriptions are omitted. The drawings show preferred embodiments of the present invention and are presented as examples.

[0029] (First embodiment) A first embodiment of the present invention will be described with reference to Figures 1 to 36.

[0030] The present invention increases the reliability and accuracy of XPS peak fitting by chemically modifying the surface being analyzed by either oxidation or reduction. Oxidation uses ultraviolet light and / or ozone in the presence of an oxygen-containing gas (e.g., laboratory air), thereby changing the proportion of different chemical states at the surface, for example, by increasing the proportion of highly oxidized states. Reduction uses exposure to ultraviolet light, optionally with hydrogen gas. By comparing XPS spectra recorded before and after this step (and optionally two or more such reduction / oxidation steps), component peaks in the spectra are numerically extracted in silico using, for example, multivariate statistical methods (in some embodiments, principal component analysis (PCA), nonnegative matrix factorization (NMF), or singular value decomposition (SVD)). This UV / ozone exposure / hydrogen exposure / XPS spectrum acquisition cycle is performed using the same XPS settings over a short period of time, so that drift in the XPS energy scale is negligible.

[0031] The device of the present invention has the following features. 1. An enclosure constructed of materials (e.g., metal, glass) that are not easily attacked by UV and ozone. It has a door or lid that is easy to open and close, and is mostly (though not necessarily completely) airtight when closed. The door or lid, when open, allows for the insertion of a sample holder. 2. In some embodiments, a type of sample holder designed to hold common sample stub types used in electron microscopy and surface analysis. 3. Within the enclosure, or directed inward from the outside through a UV-transmitting window, are one or more UV light sources, preferably at least one of which is capable of emitting significant radiation at wavelengths short enough to produce ozone in air at room temperature and atmospheric pressure. In some embodiments, these are mercury vapor lamps (13), and in other embodiments, short-wave light-emitting diodes (LEDs), or a combination of the two. 4. An electronic circuit that switches on the light source(s) for a predetermined period of time, or until a predetermined ozone concentration is reached, or until a predetermined exposure of the sample holder to UV and / or ozone is reached. 5. The ozone and / or UV generated within the enclosure are at a level sufficient to chemically modify the sample surface, and as a result, the envelope of chemical states seen in the XPS spectrum is altered by this exposure, but it is low enough that elements (even carbon) are not completely removed. 6. Optionally, sensors for measuring UV and / or ozone concentrations within the housing. These allow UV and ozone levels to be reported to the user, and as a result, in some embodiments, repeatable and reproducible exposure of the sample to UV and ozone is possible, even under closed-loop (e.g., proportional-integral-derivative or PID) control. 7. Optionally, use potential marker particles (EPMPs) for accurate charge reference when the sample is not a good conductor. 8. An X-ray photoelectron spectroscopy (XPS) instrument capable of recording XPS narrow-scan spectra of the type commonly used for XPS peak fitting. 9. The obtained XPS spectra are processed by a computer using multivariate methods such as singular value decomposition, non-negative matrix factorization, or principal component analysis to identify spectral components that change together.

[0032] The enclosure may be completely separate from the vacuum system or may form part of it (e.g., the entry lock of an XPS system so that the sample block never leaves the automated sample handling system of the XPS instrument). Ultraviolet (UV) and ozone generation is achieved using a miniature mercury lamp. Optionally, ozone is augmented from an external electric ozone generator. Ozone is generated in situ from diatomic oxygen in air by irradiating it with very short wavelength UV light, in one embodiment, 185 nm radiation from a mercury lamp. Destroying the ozone and opening the vessel is achieved by irradiating the contained air with longer wavelength UV, in one embodiment, 254 nm radiation from a mercury lamp (the shorter 185 nm radiation is blocked by a glass envelope or filter). Figure 5 shows a schematic representation of one embodiment of this portion of the invention.

[0033] In Figure 5, a battery or main power supply (630) supplies energy to one of two UV lamps (640) labeled A and B. In this embodiment, both lamps are mercury vapor lamps. Both emit UV energy at both 185 nm and 254 nm. Lamp B has an optical filter (610) covering it so that only the longer of these two wavelengths reaches the air surrounding the sample. A programmable timer controls which lamp receives power (in either case). In either case, a “ballast” component (620) is required to manage the lamp voltage and current through a tolerance range when the lamp starts operating (for most discharge and fluorescent lamps). Often, a high voltage is applied first to establish the discharge, and then a lower voltage and current are applied to maintain the discharge. In this embodiment, the programmable timer powers lamp A (irradiating the sample with UV light and forming ozone around it), then turns off lamp A and turns on lamp B (from which only 254 nm radiation can reach the space around the sample and decompose the remaining ozone), and finally turns off both lamps. This allows all ozone to be rapidly removed from the housing and the sample to be quickly and safely placed in the XPS analysis chamber. In some embodiments, the sample is placed on a slowly rotating stage to homogenize exposure to UV and ozone.

[0034] <Reducing the surface layer> Another aspect of the present invention increases the reliability and accuracy of XPS peak fitting by chemically modifying the surface to be analyzed using ultraviolet and / or hydrogen gas, thereby changing the proportion of different chemical states on the surface, for example, by reducing the proportion of highly oxidized states. By comparing the XPS spectra recorded before and after this UV and / or hydrogen exposure step (and optionally one or more such redox steps), component peaks in the spectra are numerically extracted in a computer using, for example, multivariate statistics (in some embodiments, principal component analysis (PCA), non-negative matrix factorization (NMF), or singular value decomposition (SVD)). This UV / hydrogen exposure / XPS spectrum acquisition cycle is performed over a short period (typically less than one day) using the same XPS settings so that drift in the XPS energy scale is negligible.

[0035] The device of the present invention has the following features. 1. A sample enclosure made of a material (e.g., suitable metal, glass) that is not easily attacked by UV and hydrogen. It has a door or lid that is easy to open and close, and is mostly airtight (though not necessarily completely) when the door or lid is closed. When the door or lid is open, it allows for the insertion of a sample holder. 2. In some embodiments, a type of sample holder designed to hold common sample stub types used in electron microscopy and surface analysis. 3. Inside the housing, or outside the sample housing but directed inward through an optical window, are one or more UV light sources, preferably at least one of which can emit a considerable amount of radiation at wavelengths short enough to help photocatalyze the reduction of the sample surface in the presence of hydrogen gas. In some embodiments, these are mercury lamps (Non-Patent Literature 13), or other types of discharge lamps such as xenon lamps, and in others, short-wave light-emitting diodes (LEDs), or a combination thereof. 4. Optionally, 10 -3 Less than millibars, preferably 10 -6Oxygen is removed from the sample housing by pumping the air out of it to reach a pressure in the sub-millibar range. Alternatively, the sample housing may be purged with an inert gas such as nitrogen or argon. 5. Hydrogen gas is optionally introduced into the sample enclosure from one or more hydrogen gas generation cells as described below, or optionally from an external hydrogen cylinder. 6. Optionally, a zinc-air (or similar) cell (forming a battery) may be fixed within a hydrogen-permeable cell housing (Non-Patent Literature 14) (e.g., a palladium or palladium alloy tube) which allows hydrogen to exit the battery and pass through the walls of the hydrogen-permeable cell housing, but prevents other species (such as water vapor) from doing so. Optionally, the permeable cell housing may be heated, for example, by passing an electric current through it to induce Joule heating (Non-Patent Literature 15), increasing the rate of hydrogen diffusion through its walls into the main space within the sample housing containing the sample. Instead of a hydrogen-permeable cell housing, an impermeable cell housing and a pressure relief valve configuration may be used as described below. 7. Optionally, an electronic circuit that switches on current through said zinc-air or similar metal-air battery for a predetermined time, or until a predetermined total charge has passed, or until a predetermined hydrogen concentration, or a predetermined exposure of the sample holder to UV and / or hydrogen has been reached. 8. The hydrogen and / or UV generated within the sample enclosure is at levels sufficient to chemically modify the surface of the sample, such that more reduced chemical states become more prevalent at the surface, thus changing the envelope of chemical states seen in the XPS spectrum. 9. Optionally, sensors for measuring UV and / or hydrogen concentrations within the sample housing. These allow UV and hydrogen levels to be reported to the user, and as a result, in some embodiments, repeatable and reproducible exposure of the sample to UV and hydrogen is possible, even under closed-loop (e.g., proportional-integral-derivative PID) control. 10. Optionally, use electric potential marker particles (EPMPs) for accurate charge referencing if the sample is not a good conductor. 11. X-ray photoelectron spectroscopy (XPS) instrument capable of recording XPS narrow scan spectra of the type typically used for XPS peak fitting. 12. Computational processing of the obtained XPS spectra using multivariate methods such as singular value decomposition (SVD), non-negative matrix factorization (NMF), or principal component analysis (PCA) to identify spectral components that change together.

[0036] The sample housing may be completely isolated from the vacuum system or may form part of it (for example, an entry lock for the XPS system to ensure that the sample block never leaves the automated sample handling system of the XPS instrument). In one embodiment, ultraviolet (UV) light is generated using a small mercury lamp(s) and (in one embodiment) a zinc-air (or similar metal-air) cell(s). Optionally, hydrogen is increased from an external hydrogen source or cylinder. One advantage of using zinc-air cells instead of hydrogen cylinders is that only the very small amount of hydrogen required is delivered in an electrically controlled manner, reducing the safety concerns that can arise when handling larger quantities of hydrogen. Most XPS facilities do not keep hydrogen cylinders nearby (although some do). Therefore, by using zinc-air cells or other metal-air cells, the cost of safely delivering hydrogen to the sample surface is significantly reduced.

[0037] FIG. 36 shows a schematic representation of one embodiment of a portion of the present invention. Figures 32a and 32b show an embodiment in which hydrogen is supplied by a zinc-air cell. In FIG. 32, the hydrogen generation cell 720 is controlled by the timer or PLC using a switch or relay, such that when the switch is "on" and current passes through the cell via the current limiting resistor R, the cell generates hydrogen gas at a known pre-planned rate determined by the value of the resistor R, as shown in FIG. 32b. A typical value for this resistance is 100 - 300 ohms, although this value is not so important; the important quantity is the total charge that can be passed, which determines the total amount of hydrogen released.

[0038] <Comparison with the prior art> XPS devices have sometimes included a UV source for many years, but this was (a) to enable ultraviolet photoelectron spectroscopy (UPS) to be performed in a vacuum, not for surface modification in air as in the present invention, or (b) to modify a particular surface by a particular chemical method (e.g., Sun et al.) (Non-Patent Document 16) using a particular type of UV (hydrogen-free), for example, long wavelength UV of about 365 nm. The purpose of this initial research was not to modify the chemical state of the surface for the purpose of identifying chemical states as described in the present invention, but to study the particular chemical reactions induced by a particular UV exposure on a particular sample material.

[0039] <UV / ozone cleaning device: Why the present invention is different from these and how differences arise for different purposes.> UV / ozone cleaning devices have been used for decades, for example, concentrated on the use of mercury lamps, as popularized by the research of J R Vig (Non-Patent Document 17). Low-pressure mercury lamps have two main emissions at 185 nm and 254 nm in the UV. The UV rays at 185 nm decompose oxygen molecules and synthesize ozone O3 in situ. The UV rays at 254 nm decompose ozone and produce high-energy O *It generates (reactive oxygen species). These highly oxidizing species interact with carbonaceous contaminants on the surface (in fact, anything on the surface that can be oxidized). Finally, in combination with direct UV exposure (which strongly interacts with the C=O moiety via a Norrish-type chemical process), the organic species are oxidized and / or decomposed into volatile compounds, mainly CO2, which diffuse from the surface. This process is schematically shown in Figure 6.

[0040] Commercially available UV / ozone cleaners are high-power devices designed to remove all carbonaceous contaminants as quickly as possible. They typically do not need to measure or report UV or ozone levels; instead, they are simply designed to provide very high levels of both to rapidly remove the contaminants. If chemical modification of the surface is attempted (e.g., to reveal chemical shifts in XPS) rather than complete removal of the carbonaceous contaminants, it is too easy to "overshoot" the design objective of the UV / ozone cleaner and remove it all. There are additional design differences between commercial UV / ozone cleaners and the present invention, motivated by different objectives. Commercial UV / ozone cleaners are typically designed for use with large objects, such as silicon wafers 200 mm or larger in diameter. This requires less power, and as a result, core-level peaks are modified, though perhaps not completely removed, from the XPS spectrum over several increasingly aggressive oxidation steps. Also, the sample space is smaller, allowing the device to be placed near the XPS sample entry lock for rapid exposure and then the sample returned to the XPS system vacuum, minimizing exposure time to atmospheric contaminants. While UV / ozone cleaning only works on carbonaceous contamination (because it leaves the surface as a gaseous oxide), the present invention aims to use UV and / or ozone to oxidize any sample material, leaving it for subsequent XPS analysis. Also, measuring UV intensity and ozone concentration helps ensure reproducibility of measurements at different locations, and therefore built-in UV and ozone monitors are useful in XPS peak fitting applications; they are optional but very useful as part of the present invention.

[0041] With all this being said, the present inventors have in the past successfully modified UV / ozone cleaning devices to generate spectra of samples exposed to UV / ozone. Typically, this has been done by modifying the device, disabling it in some way (e.g., pressing an "emergency stop" button after a few seconds to avoid excessive UV / ozone exposure), or by disassembling it to remove components (e.g., lamps), and then placing those components in different enclosures. In fact, much of the work leading to the present invention was done by modifying commercially available UV / ozone units to accomplish purposes for which they were not intended.

[0042] <Procedure for using the present invention (oxidation)> Figure 7 shows a flowchart of a method for performing spectral acquisition using the present invention. In some cases, the user will have enough information to pre-set UV / ozone exposure levels / times, providing a series of exposure steps of a predetermined length, without applying the questions in this flowchart. If not, the computer can, for example, perform a PCA analysis of the spectra acquired up to that point and recommend increasing exposure in the next step to make observed differences in the spectra more likely.

[0043] Figure 8 shows the results of using UV / ozone exposure and XPS to identify which chemical states correspond to which (sometimes unresolved in energy) peaks in the XPS spectrum of a carbonaceous layer on a metal. These narrow scan spectra show the region around the C 1s peak. As a result of this analysis, smoothing lines were retroactively introduced. Note that these spectra were acquired on the same instrument within a few hours, so the energy scale and energy resolution can be considered stable.

[0044] Therefore, UV / ozone exposure alters the height of peak components, but they do not shift on the energy scale. Whatever new peak heights UV / ozone exposure causes, the fact that they change allows them to be extracted from such a set of spectra. For example, in Figure 8(a), the different oxidation states of carbon are initially almost decomposed. However, (b), (c), and (d) show that the intensity of these states is changing, with (d) showing a new carbide state. This is C * If the spectrum is not one of the sets of spectra (a), (b), (c), and (d) that have approximately the same binding energy as the -H peak and whose peak intensity is changing, then C * -H (which is more general) can be mistaken for this. By the end of this process, the binding energies and strengths of the four original states found in (a) are clearly determined.

[0045] Figures 9 to 14 show numerical simulation results demonstrating the effectiveness of the present invention. Figure 9 shows a composite spectrum extracted from the actual data shown in Figure 8, consisting of three peaks separated by 1 eV. Since these can be resolved fairly easily with modern spectrometers, looking at the envelope spectrum (the continuous line in Figure 9), it is clear that there are at least three peaks here. Note that the C 1s peak in XPS is typically on a fairly flat background, so it includes 20% of the peak height in background, which is constant with respect to energy.

[0046] Figure 10 shows simulated spectra after zero iterations (a), one iteration (b), two iterations (c), etc. of the procedure shown in Figure 7. Thus, each of 10(b) through 10(f) shows the effect of more UV / ozone exposure than the previous one. As we progress through this series, some states (i.e., peaks) decrease in intensity more rapidly than others. It does not matter whether the state that is more rapidly eliminated is high or low binding energy; simply, the atoms that give rise to a peak in one spectrum may reappear in a different chemical state in the next spectrum (or may leave entirely as a highly oxidized gas such as CO, for example).

[0047] Figure 11 shows the results of numerical processing of the simulated spectrum shown in Figure 10. Figure 11(a) is a reminder of the peaks and their envelopes shown in Figure 9. Figure 11(b) shows the initial spectrum generated from this model, with counts on the vertical axis and Poisson noise added. Figures 11(c) and 11(d) show the second and third components extracted from the set shown in Figure 10 using singular value decomposition (SVD). The first component is not very useful, as it simply resembles the average of the spectrum shown in Figure 10. The second and third components (and in other real cases, higher components as well) show significant peak structure. Here, we inverted the negative parts of these components and plotted them as solid lines; the positive parts are plotted as dotted lines. SVD tells us that in Figure 11(c), a UV / ozone-induced process is shown that removes intensity from the peak at approximately 3.56 eV and adds it to the peak at 1.51 eV. These numerical labels for the peak energies are calculated by fitting a parabola to five points around the highest point of the peak. This process, of course, is a function of UV / ozone exposure, but the gist is that this component reveals two of the peaks that make up the initial spectrum (b) and gives the energies of these peaks fairly accurately (within 0.06 eV) of their true values. The third component, shown in Figure 11(d), shows another peak at approximately 2.53 eV (within 0.03 eV of where the actual peak is), again one of the previously identified peaks at 1.53 eV, close enough to the previously identified 1.51 eV to ensure it is the same chemical state. Thus, Figures 9-11 demonstrate that the UV / ozone effect of the present invention allows for the unique identification of the number of states and their energies (and a fairly good estimate of their widths).

[0048] The peaks shown in Figure 9 are already fairly well separated, so it can be argued that this is an easy problem. They overlap, but three distinct peaks are visible, one of which, even visually, might estimate their energies to be 1.5, 2.5, and 3.5 eV. Now, consider the case shown in Figure 12, where the separation of the peaks is reduced by assigning energies of 2 eV, 2.5 eV, and 3 eV to the peaks. These overlap to such an extent that the envelope (the continuous line in Figure 12) has only a single maximum. An inexperienced analyst may well attempt to fit this curve to a single peak or a few peaks with varying energies and intensities. Many alternative models fit reasonably well in a statistical sense, but have no chemical significance whatsoever. This is the source of much of the error in published peak fitting, as mentioned in the background section.

[0049] Figure 13 shows how UV / ozone cleaning can affect the spectrum, as done in Figure 10 for well-separated peaks. The spectrum in Figure 13 is very confusing for those new to XPS, and it has been observed that inexperienced analysts interpret this type of spectrum as a single-peak binding energy shift, a charging effect (even for conductive surfaces), or an instrument problem such as voltage instability in the instrument. However, SVD applied to the dataset in Figure 14 provides a useful and robust answer. The initial spectrum has peaks at approximately 2.07 eV, 3.07 eV, and 2.54 eV, all very close to the real values ​​of 2, 3, and 2.5 eV. Given these energy values, conventional XPS peak fitting becomes easier, and this initial data for such fitting is very useful, removing the ambiguity that can confuse those new to XPS.

[0050] <Procedure for use of the present invention (reduction)> Figure 24 shows the general procedure for using the spectral acquisition method of the present invention, where the "hydrogen-containing chamber" is identical to the sample enclosure discussed in the text when filled with hydrogen. Figure 24 shows a flowchart of a method for performing spectral acquisition using the present invention. In some cases, the user will have enough information to pre-set UV / hydrogen exposure levels / times, providing a series of exposure steps of a predetermined length, without applying the questions in this flowchart. If not, the computer can, for example, perform a PCA analysis of the spectra acquired up to that point and recommend an increase in exposure in the next step to make observed differences in the spectra more likely.

[0051] Figures 25-30 show numerical simulation results demonstrating the effectiveness of the present invention. Figure 25 shows a composite spectrum consisting of three peaks spaced 1 eV apart. These can be resolved fairly easily with modern spectrometers, so looking at the envelope spectrum (the continuous line in Figure 25), it is clear that there are at least three peaks. Note that the C 1s peak in XPS typically lies on a fairly flat background, so it includes a background of 20% of the peak height, which is constant with energy.

[0052] Figure 26 shows simulated spectra after zero iterations (a), one iteration (b), two iterations (c), etc. of the procedure shown in Figure 24. Thus, each of Figures 26(b)-26(f) shows the effect of more UV / hydrogen exposure than the previous one. As we progress through this series, some states (i.e., peaks) decrease in intensity more rapidly than others. It doesn't matter whether the state that is more quickly removed is high or low binding energy; simply, the atoms that give rise to a peak in one spectrum may reappear in a different chemical state in the next spectrum (or, for example, drop off completely after reaction with hydrogen to produce HO).

[0053] Figure 27 shows the results of numerical processing of the simulated spectrum shown in Figure 26. Figure 27(a) is a reminder of the peaks and their envelopes shown in Figure 25. Figure 27(b) shows the initial spectrum generated from this model, with counts on the y-axis and Poisson noise added. Figures 27(c) and 27(d) show the second and third components extracted from the set shown in Figure 26 by using singular value decomposition (SVD). The first component is not very useful as it simply resembles the average of the spectrum shown in Figure 26. The second and third components (and in other real cases, higher components as well) show important peak structures. Here, the negative parts of these components are inverted and plotted as solid lines. The positive parts are plotted as dotted lines. SVD tells us in Figure 27(c) that it shows a UV / hydrogen-induced process that removes intensity from the peak at approximately 3.56 eV and adds intensity to the peak at 1.51 eV. These numerical labels are calculated by fitting a parabola to five points around the peak's highest point. This process is, of course, one aspect of UV / hydrogen exposure, but the point is that this component reveals two of the peaks that make up the initial spectrum (b) and gives the energies of these peaks fairly accurately (within 0.06 eV) of their true values. The third component shown in Figure 27(d) shows another peak at approximately 2.53 eV (within 0.03 eV where the actual peak is located), and again shows one of the previously identified peaks at 1.53 eV, which is close enough to the previously identified 1.51 eV to ensure it is the same chemical state. Thus, Figures 25–27 demonstrate that the UV / hydrogen effect of the present invention allows for the unique identification (and fairly good estimation of their widths) of the number of states and their energies.

[0054] The peaks shown in Figure 25 are already fairly well separated, so it can be argued that this is an easy problem. They overlap, but three distinct peaks are visible, one might even estimate their energies to be 1.5, 2.5, and 3.5 eV by sight alone. Now consider the case shown in Figure 28, where the separation of the peaks is reduced by assigning energies of 2 eV, 2.5 eV, and 3 eV to the peaks. These overlap to such an extent that the envelope (the continuous line in Figure 28) has only a single maximum. An inexperienced analyst may well attempt to fit this curve to a single peak or a few peaks with varying energies and intensities. Many alternative models fit reasonably well in a statistical sense, but have no chemical significance whatsoever. This is the source of much of the error in published peak fitting mentioned above in the background section.

[0055] Figure 29 shows how UV / hydrogen exposure can affect the spectrum, as Figure 26 did for well-separated peaks. The spectrum in Figure 29 is very confusing for newcomers to XPS, and it has been observed that inexperienced analysts interpret this type of spectrum as a single-peak binding energy shift, a charging effect (even for conductive surfaces), or an instrument problem such as voltage instability in the instrument. However, SVD applied to the dataset in Figure 29 provides a useful and robust answer, as shown in Figure 30. The initial spectrum has peaks at approximately 2.07 eV, 3.07 eV, and 2.54 eV, all very close to the real values ​​of 2, 3, and 2.5 eV. Given these energy values, conventional XPS peak fitting becomes easy, and this initial data for such fitting is very useful, removing the ambiguity that can confuse newcomers to XPS.

[0056] <Additional ion beam sputtering of the sample> Optional ion beam sputtering of the sample surface before and / or after UV / ozone / hydrogen exposure can provide additional spectra useful for inclusion in PCA or machine learning datasets. This is because a very short sputtering treatment with monoatomic argon ions, presumably at low kinetic energies (100–1,000 eV), is highly reducing in the sense of removing oxygen but also damaging in the sense of creating chemical states rare in the as-received material. Some samples may already be highly oxidized in their as-received state, proving the need for analytical information. Further oxidation (by UV / ozone) barely alters narrow-scan XPS spectra. However, photosputtering before and / or after such oxidation can provide broader spectra that allow for better spectral definition of the chemical states present, for example, by using PCA.

[0057] <Possible embodiments and configurations (oxidation of surface layer)> Figure 15 shows a schematic vertical cross section of a typical commercial XPS instrument. It contains a nominally ultra-high vacuum (UHV) analysis chamber (1500) with a hemispherical electron energy analyzer (1510). Pumps (1505) maintain the vacuum in various parts of the system. Valves (1520) are opened and closed to allow the sample to enter the analysis chamber through an entry lock (1525). A transfer arm (1535) is used to move the sample between the analysis chamber and the entry lock. When a sample is withdrawn from the system, the entry lock is returned to atmospheric pressure by admitting gas (typically nitrogen) from a cylinder (1515). The entry lock typically has a transparent glass window (1530).

[0058] Figure 16 shows one configuration of the present invention in which the XPS system and the enclosure (as described above) are separate but in close proximity. Samples are transferred in air from the XPS system to the enclosure (1610) containing the UV / ozone generating lamps (1620) and then transferred back again after UV / ozone exposure, completing the iterative loop shown in Figure 7. These transfers could be automated using a small air-side robotic arm or similar, but are most likely performed manually by an operator.

[0059] Figure 17 shows another configuration of the present invention in which the enclosure containing the UV / ozone generating lamp is integrated into the entry lock of an XPS system. This requires a UV-transmitting window on the entry lock instead of the usual glass window (1530), and the backfill gas cylinder (1700) must contain oxygen or an oxygen-containing gas mixture (e.g., dry air) rather than pure nitrogen. UV passes through the UV-transmitting window and generates ozone inside the entry lock itself.

[0060] When the sample exits the entry lock, it can be exposed to UV and ozone and then returned to the analysis chamber for the next spectral acquisition, as described in the flowchart of FIG. 7. This configuration best utilizes automated sample handling, lamp, and valve control. For example, the entire sequence described in FIG. 7 can be run as an automated sequence under computer control and without the need for a human operator. For example, it could be run overnight, making efficient use of instrument time that would otherwise be difficult to utilize. In the morning, the operator returns and finds the entire set of spectra of the type shown in FIGS. 10 and 13, with the results of the calculations of the type shown in FIGS. 11 and 14 already performed (because the calculations do not require operator knowledge or intervention).

[0061] Regarding possible UV lamps usable for this application, good results were obtained with the small "GTL3" UV lamp of the type shown in Figure 18, although many other models would likely function similarly well. These lamps were operated with a single 33-ohm ballast resistor and a 24V supply voltage (AC provides a longer lamp life, but both AC and DC work). During operation, these lamps consume approximately 3W at 10V (the remainder is dropped off by the ballast resistor). Both ozone-emitting and non-ozone-emitting versions of this lamp are available (these have different glass formulations for transmitting or blocking 185nm radiation, respectively), and as a result, these can be used as either or both of lamps A and B in Figure 5. They have E17 standard screw-in bases and therefore easily fit into confined spaces. The Sankyo Electric GTL3 is a 3W lamp with a UV output of 0.16W. Typical lifespan is specified by the manufacturer as 2000 hours. This lamp has an E17 screw-in base on a clear T7 tube measuring 20mm in diameter and 63mm in length. This item is manufactured in Japan and is also generally available from Ushio, Fisher Scientific, Eiko, Hikari, and American Ultraviolet under part numbers GTL3W, PO300-0350, 29-258-23, GRM0036, and 3000022, among others.

[0062] These GTL3 lamps are often used for sterilization applications, such as disinfectants for washing machines or toothbrushes. They are very inexpensive, typically costing less than $10. They are not very electrically efficient, especially when used with a 33-ohm stabilizer, but this is not a practical issue for this application. They do not have sufficient radiant power for UV / ozone cleaning of surfaces. Instead, larger mercury grid lamps are typically used for this purpose. However, as mentioned above, for the applicant's applications where a gentle and gradual oxidation or reduction of a surface is desired, these GTL3 lamps only need to be placed within about 10 cm of the sample. According to this specification, these lamps emit 450 × 10⁻¹⁶ ray at 254 nm, measured at a distance of 3 cm from the bulb. -6 W / cm 2 Although the inventors have not found any specifications for shorter wavelength emissions, it is undoubtedly 185 nm, and ozone is produced by these lamps.

[0063] An alternative lamp to lamp B, which has a longer wavelength, is a short-wave light-emitting diode (LED), as shown in Figure 19. Currently, these are available at wavelengths up to about 270 nm and therefore cannot be used as ozone-generating lamps, but can be used as ozone-depleting lamps or to irradiate samples in the presence of hydrogen. Referring to Figure 20, which shows the Hartley absorption band of ozone, we can see that Hg vapor emission at 254 nm is close to the maximum absorption of ozone (and therefore rapidly converts ozone back to diatomic oxygen), but at 270 nm, the LED emitter is not as rapid and therefore, for similar photon intensity, is about twice as efficient in terms of time, but is more efficient in terms of power used and possibly lamp lifetime.

[0064] The operating distance or range (from lamp to sample) in this application is a problem that is very difficult to determine theoretically. This is because the two competing wavelengths emitted by the UV lamp, 185 nm and 254 nm, generate and destroy ozone, respectively (Non-Patent Literature 18), and as a result, the ozone concentration decreases non-linearly with distance from the lamp. For complex housing shapes, this is best determined experimentally using ozone and UV measuring devices, and these measurements are expected to be particularly applicable to specific housings and models of UV lamps.

[0065] <Possible embodiments and configurations (reduction of surface layer) Hydrogen emission element > Many laboratories that operate XPS instruments have high-purity hydrogen gas available; others do not. In either case, when working with large quantities of hydrogen gas, even when the amounts actually used (as in this application) are very small, the cost of implementing safety procedures is often high.

[0066] Therefore, optionally, and in some embodiments, hydrogen gas produced in situ within the sample housing by a zinc-air or similar cell(s) is utilized. These may be commercially available "button" batteries sold for devices such as hearing aids, which replaced common mercury batteries 20 years ago. In fact, variations of such cells are commercially available for the purpose of hydrogen production.

[0067] Thus, to provide a simple source of high-purity hydrogen gas, optionally, the zinc-air, or other metal-air type battery (which shall be understood to mean one cell or multiple cells) is placed within a sample enclosure, with external control over the current passing through the battery. When a resistive load is applied to the zinc-air battery without access to oxygen, the zinc-air battery generates hydrogen gas at a fairly controllable rate (Non-Patent Document 20) (Non-Patent Document 19). In one embodiment, this may be achieved by having an external switch and resistor in series across the battery, so that switching on causes a resistor-limited current to pass through the battery.

[0068] This type of zinc-air battery is known to generate a small amount of hydrogen gas approximately proportional to the total charge that passes through it. This allows the hydrogen gas to be released at a pressure about 10° higher than the pressure of other reactants (e.g., potentially oxidizing species such as oxygen and water) within the sample housing. -3 This allows partial pressures of up to 1000 mb or more to be delivered to the area surrounding the surface being analyzed.

[0069] Zinc-containing cells (such as those manufactured by Varta) specially designed for hydrogen generation can be used (21) and are actually modified forms of zinc-air batteries that are commercially available as precision hydrogen generators. An important consideration is that in this application only small amounts of hydrogen are required, filling the small volume of the sample housing at pressures much lower than atmospheric pressure, resulting in a maximum hydrogen flow rate of perhaps 150 cm over its lifetime. 3 The idea is that a zinc-air cell producing 600 cm over its lifetime is sufficient. Four such cells in a battery will produce 600 cm 3 The amount of H2 can be supplied, perhaps sufficient for over 500 reduction cycles of exposing the sample to low-pressure H2 under UV light, before needing to be replaced.

[0070] Figure 31 shows a product overview of two battery cell products from Varta specifically designed to produce hydrogen. Because the cell contains an aqueous electrolyte, it cannot be used without being enclosed within a vacuum chamber (sample enclosure) such as the entry lock of an XPS instrument. This evaporates under dry conditions, and vacuum is a very dry environment. Therefore, the cell must be enclosed within a container (cell enclosure) within the sample enclosure that allows hydrogen gas to be released when needed, but maintains at least a partial pressure of water at the operating temperature of the XPS instrument, e.g., approximately 18 mmHg at 20 °C. There are at least two possible embodiments to achieve this cell housing.

[0071] Possible Cell Enclosure Embodiment A. A sealed tube around the cells is made from a hydrogen-permeable (but water-impermeable) material such as palladium or a palladium alloy. Figure 3 shows a schematic of this arrangement, and in any embodiment, four such cells are formed into a battery within a sealed palladium (or palladium alloy) tube that allows hydrogen to permeate therethrough.

[0072] Alternatively, in possible cell enclosure embodiment B, the sealed tube around the cell is connected to the main volume of the entry lock or other sample enclosure via a normally closed pressure relief valve that opens when the internal pressure (caused by hydrogen produced by the cell) exceeds a predetermined value above the vapor pressure of water at that temperature. For example, a spring-loaded safety valve set to open when the pressure inside the cell enclosure rises more than 0.2 atm above the external pressure inside the sample enclosure. Some water vapor escapes each time the valve opens, but in its normally closed state, the cell does not dry out during the hours between uses.

[0073] FIG. 32 shows a schematic representation of one possible embodiment A of the permeable hydrogen-releasing element of the present invention, with FIG. 32(a) showing it when deselected and FIG. 32(b) showing it when selected to activate and release hydrogen.

[0074] Figure 33 shows a schematic diagram of a simple embodiment type B of the cell housing. In Figure 33, the weight selected for the tip of the ball 1020 in the funnel 1030 allows the pressure inside the cell housing 1010 to exceed the pressure of the saturated water vapor amount of water at the operating temperature of the device (typically room temperature or slightly higher), even if the sample housing pressure (which should not be confused with the cell housing) is a vacuum. This ensures that the cell 720 does not dry out. By momentarily pressing the push button, the capacitor is discharged, and as a result, when the button is released, current flows through the cell until the capacitor is fully charged, generating hydrogen and releasing a fixed amount of hydrogen predetermined by the selection of capacitance value C. The ball 1020 is then momentarily displaced because the pressure of hydrogen inside the cell housing 1010 is higher than the pressure inside the sample housing around it, and a small fixed predetermined amount of pure hydrogen is released into the area around the sample inside the sample housing for reduction of the sample surface or reduction by UV.

[0075] Figure 34 schematically shows a slightly more advanced possible embodiment of cell housing type B in both (a) the dormant state and (b) the hydrogen generation state. Here, the switch controlling hydrogen generation may be a relay or similar switch that is normally open and under the control of a programmable logic controller (PLC) 750, as shown in (a). The pressure around cell 720 within cell housing 930 is higher than the inside of sample housing (which should not be confused with cell housing 930) as a result of a pressure relief valve formed by spring 960, adjustment screw 970, and “poppet” 950. As shown in (b), when the PLC closes the switch, a DC current (limited by resistor R to limit the hydrogen generation rate) passes through cell 720. The cell releases hydrogen until this pressure on the poppet is sufficient to overcome the force of spring 960, and the hydrogen escapes into sample housing. In this embodiment, computer code executed by the PLC may select different durations for the switch closure, thereby releasing different amounts of hydrogen into the region surrounding the sample for subsequent reduction of the surface of the sample or reduction by UV.

[0076] Figure 15 shows a schematic vertical cross-section of a typical commercial XPS instrument. It contains a nominally ultra-high vacuum (UHV) analysis chamber (1500) equipped with a hemispherical electron energy analyzer (1510). A pump (1505) maintains the vacuum in various parts of the system. A valve (1520) opens and closes to allow the sample to enter the analysis chamber through an entry lock (1525). A transfer arm (1535) is used to move the sample between the analysis chamber and the entry lock. When the sample is withdrawn from the system, the entry lock is returned to atmospheric pressure by admitting gas (typically nitrogen or dry air) from a cylinder (1515). The entry lock typically has a transparent glass window (1530).

[0077] Figure 16 shows one configuration of the present invention in which the XPS system and sample enclosure (described in the paragraph above) are separate but in close proximity. The sample is transferred in air from the XPS system to the sample enclosure (1610), which contains the UV-generating lamp (1620), and then transferred back again after UV / hydrogen exposure. Hydrogen is supplied to the sample enclosure 1610 from a cylinder or other "pipe" supply, or from a hydrogen release cell according to one of the embodiments described above and in Figures 10, 11, or 12. These transfers can be automated using a small air-side robotic arm or similar, but are most likely performed manually by an operator.

[0078] Figure 36 shows another embodiment of the invention, in which a sample enclosure containing a UV-generating lamp is integrated into the entry lock of an XPS system. This requires a UV-transparent window on the entry lock (the location of this window is 1530 in Figure 13). A backfill gas cylinder (1700) may be used to supply hydrogen to the vicinity of the sample, or hydrogen release cells within their own cell enclosure 1780 may provide it (electrical connections and safety valves not shown). UV light passes through the UV-transparent window and into the entry lock itself.

[0079] When the sample is in the entry lock, it is therefore exposed to UV and hydrogen and can be returned to the analysis chamber for the next spectrum acquisition, as described in the flowchart above. This configuration best utilizes automated sample handling, lamp and valve control. For example, the entire sequence can be run under computer (and / or PLC) control and without the need for a human operator. For example, it can be run overnight, making effective use of instrument time that would otherwise be difficult to utilize effectively. In the morning, the operator returns to obtain the entire set of spectra.

[0080] Regarding possible UV lamps usable for this application, good results were obtained with the small "GTL3" UV lamp of the type shown in Figure 16, but many other models would likely work just as well. These lamps were operated with a single 33-ohm stabilizing resistor and a 24V supply voltage (AC provides a longer lamp life, but both AC and DC work). During operation, these lamps consume approximately 3W at 10V (the remainder is dropped off by the ballast resistor). They have E17 standard screw-in bases and therefore fit easily into confined spaces. The Sankyo Electric GTL3 is a 3W lamp with a UV output of 0.16W. Typical lifespan is specified by the manufacturer as 2000 hours. This lamp has an E17 screw-in base on a clear T7 tube measuring 20mm in diameter and 63mm in length. This item is manufactured in Japan and is also generally available from Ushio, Fisher Scientific, Eiko, Hikari, and American Ultraviolet under part numbers GTL3W, PO300-0350, 29-258-23, GRM0036, and 3000022, among many others.

[0081] According to this specification, these lamps measured at a distance of 3 cm from the bulb at 254 nm, with a wavelength of 450 × 10⁻¹⁰. -6 W / cm 2 It emits ultraviolet light of the above intensity. These GTL3 lamps are often used for sterilization applications, such as in washing machines or toothbrush sterilizers. They are very inexpensive, typically costing less than $10. They are not very electrically efficient, especially when used with a 33-ohm stabilizing resistor, but this is not actually a problem in this application. In fact, the waste heat from these lamps could be used to heat a hydrogen permeation cylinder or membrane to increase its permeability to hydrogen. An alternative to the UV lamp is the short-wave light-emitting diode (LED). Currently, these are available at wavelengths up to about 270 nm.

[0082] <Closed-loop control> Using a signal from a hydrogen or pressure sensor close to the sample, the current through a zinc-air (or similar) battery can be switched on and off to automatically control the hydrogen level to the required value.

[0083] <Interpretation of the results of PCA, NMF or SVD> When a set of spectra (after gradually exposing the sample to increasing amounts of UV / ozone or UV / hydrogen) is processed via a PCA or SVD algorithm, this can be regarded as one step in extracting the spectra of the pure components from that data. For example, the following can be done by applying the constraints that such a combination of components has. 1. There are no or very small negative features. 2. The overall curvature of the spectrum is minimized.

[0084] The spectra of the pure chemical components (typically some oxides of metals, for example) generated and removed during UV / ozone treatment can be obtained automatically. Not forcing any or very little negative features results in non-negative matrix factorization or NMF.

[0085] Furthermore, more useful data exists. Each of these main components can be considered to represent a chemical process that occurs under UV / ozone or UV / hydrogen exposure. For example, polymers are related to the bond energy CC of hydrocarbons. * -C has a negative peak, and C * A component with a positive peak at the -OH bond energy can be observed. This can be interpreted as representing a chemical oxidation reaction. However, the rate of such a reaction depends, first and foremost, on the chemical environment of the carbon. In samples with different molecules (e.g., O or N as part of a polymer backbone), it can be predicted that the rate of this reaction will differ, and therefore the sequence of the UV / ozone-treated sample spectra will differ, and therefore the type (or order) of the main components will differ. Thus, this gives us access to more information that is not possible with a single XPS spectrum alone. For example, it may be possible to distinguish molecules using SMILES codes. CCCCNCCCC From those with SMILES codes CCCC(N)CCCC

[0086] This is difficult or impossible from observing the chemical shift of the spectrum before UV / ozone exposure. Machine learning algorithms such as neural network models or "deep learning" are particularly useful when applied to sample spectra exposed stepwise to UV / ozone and / or UV / hydrogen, as described herein, even though the same algorithms reveal little when applied to spectra from treated samples only (and not UV / ozone).

[0087] <Closed-loop control> Using signals from an ozone sensor placed near the sample, the ozone level can be automatically controlled to maintain the desired value by switching lamps A (ozone generation) and B (ozone depletion) on, as shown in Figure 5. This can be conveniently done using a light absorption or chemical ozone sensor and a proportional-integral-derivative (PID) controller that switches the lamps on or off (or modulates their power output or duty cycle, if LED devices can be used). Controlling the ozone level allows for some reduction in the duration of later, longer ozone exposure processes by increasing the ozone level.

[0088] <Non-conductive sample> Non-conductive samples present a special problem, and to overcome this problem, the inventors have developed a special method. In conventional XPS instrument operation, a "flood gun" is typically used to obtain good spectra from non-conductive samples. This gun "irradiates" the sample with low-energy electrons, and sometimes ions, and as a result, the charge accumulated on the sample surface is neutralized by these charged species. Once charge equilibrium is achieved, the photoelectrons that leave the surface leave a positive charge, which is then neutralized by charged particles emitted by the flood gun. This charge equilibrium does not necessarily return the surface to exactly earth potential, but it stabilizes the potential close to earth potential. The stability of the surface potential is crucial for spectrum acquisition. XPS operators quickly learn to recognize when charge equilibrium stability has not been achieved, as the XPS spectrum peaks are indistinct across the energy range, and sometimes do not even appear as peaks.

[0089] The problem with non-conductive samples when applying the UV / ozone or UV / hydrogen exposure methods described here is that successive exposures alter the surface chemistry, resulting in very slight changes in the charge equilibrium potential, meaning that peaks from sample components will appear to shift.

[0090] To eliminate the problem of charge equilibrium shifts, we have successfully used ion-beam sputtered particles deposited on the sample surface, sometimes called electrical potential marker particles (EPMPs). These consist of materials that can be sputtered (by ion beam sputtering using an ion gun built into almost all XPS systems) to cover a small portion (perhaps 1-5%) of the sample surface with small particle clusters. This is done once before the subsequent UV / ozone or UV / hydrogen exposure steps. These particles appear in all XPS spectra of the sample surface. The chemical composition of the EPMPs is chosen to provide sharp peaks that can be used to mark changes in surface potential.

[0091] Ion beam sputtering is possible for a wide range of materials, so in principle, many different materials could be used to provide EPMPs. Furthermore, in this application, the amount of material sputtered is small, so there is no real cost constraint, and even the most expensive noble metals are acceptable if they perform well. Therefore, we must look carefully at the chemistry of the elements to see which will perform best as EPMPs.

[0092] Clearly, it is undesirable to select a material that oxidizes with increased exposure to ozone, especially if the oxide peaks are poorly resolved and, as a result, the monitored EPMP XPS peaks appear to shift in position as oxidation progresses, making Cu, for example, a poor choice.

[0093] Ideally, element X used as a sputter target to provide EPMPs should have a single, constant oxidation state during successive UV / ozone or UV / hydrogen irradiation. This can be achieved by ensuring the following:

[0094] 1. X is readily oxidized to its highest oxidation state, XOy. 2.X is so noble that it does not oxidize at all and remains in the metallic state as EPMPs even when exposed to ozone. 3. X has XPS peaks close to the C 1s and O 1s peaks, which are typically the most important for the analysis of most insulators. The energy proximity means easier capture in fewer XPS scans and reduced chance of energy scale drift causing significant errors.

[0095] Figure 21 is helpful. To form the best possible EPMPs, noble metals, even gold, are not sufficiently "noble." When exposed to ozone, there is a measurable and obvious shift in the Au 4f peak due to surface gold oxide (and sometimes nitrides). Instead, we should focus on elements that are in little or no oxidation state and readily oxidize. Sc, Ti, Ni, Zn, Y, Cd, Lu, and Hf are reasonable candidates. Some are difficult to justify handling for safety reasons (e.g., Cd). Ease of handling is also useful, and therefore, widely available, easily manipulated, and malleable foils are advantageous.

[0096] In this application, we achieved good success using titanium as a sputtering target for EPMP deposition. Titanium satisfies criterion 1 above in that it is readily oxidized to its highest oxidation state. Therefore, when attempting to obtain an XPS spectrum of Ti metal, it is typically hindered by the appearance of surface oxides, even under "ultra-high vacuum" conditions where very low concentrations of oxygen-containing species (typically water) are present. In fact, Ti is used in sublimation pumps in XPS precisely for this reason, namely its affinity for capturing oxygen-containing species and its high adhesion coefficient. Ti foil is widely available in various thicknesses and is easily formed into the correct shape by bending. The Ti 2p peak, the strongest XPS peak of Ti, is located approximately midway between the C 1s and O 1s energies, which is ideal.

[0097] Ti 2p3 / 2 The peak positions of XPS peaks can be accurately determined by a number of methods. We used a polynomial fitting method developed in the 1990s that works very well (22). All spectra recorded from samples after a fixed UV / ozone or UV / hydrogen exposure are then shifted to ensure accurate EPMP peak alignment, including interpolation for integer shifted channels.

[0098] <Deposition of EPMPs on the sample surface> A small coupon of foil (in one embodiment, titanium, as described above) is bent to have an internal angle of approximately 110°, as shown in Figure 23. A beam of ions (typically argon, but may also be clusters or monoatoms, as desired) is used to first clean the foil sputter target. This ion gun, or another ion gun, is used to focus monoatomic argon ions onto the target foil surface, sputtering atoms from the foil onto the sample. This sputter deposition step is shown schematically (in cross section) in Figure 23, in which particles that will go on to form EPMPs are sputtered from the target 2310 (typically Ti metal foil) onto the insulating sample 2320 to be analyzed.

[0099] Typically, sputter deposition requires an ion gun operating time of 1-2 minutes. Ti 2p 3 / 2 The peak should be at least 5% of the most intense peak in the spectrum. If this is not achieved (as measured by a wide scan spectrum of the sample), it can of course go back to sputtering from more target 2310 than ever before.

[0100] Figure 35 shows a sample with multiple subsamples. 1 ~s 9 35 shows an embodiment of the invention in which the sample material is divided into sub-samples. This can be useful when the sample material is particularly uniform, such as a wafer of semiconductor material. The oxidation state of each sub-sample varies by a different amount. While each sample can be analyzed individually, it is possible to analyze multiple sub-samples simultaneously, as shown in FIG. 35.

[0101] As described above, the present invention is not limited to the above-described embodiment, and can be embodied in various forms without departing from the spirit and scope of the present invention.

Claims

1. 1. A method for producing an X-ray photoelectron spectrum of a sample, comprising: generating a plurality of different oxidation states on the surface of the sample by exposing the surface of the sample to an agent configured to change the oxidation state of the surface of the sample; placing the sample in an X-ray photoelectron spectroscopy instrument; obtaining an X-ray photoelectron spectrum for each of a plurality of oxidation states of the surface of the sample; comparing the X-ray photoelectron spectrum obtained for each of a plurality of oxidation states of the surface of the sample with the others and numerically extracting component peaks from each of the obtained X-ray photoelectron spectra by computer to identify materials within the sample; A surface analysis method comprising:

2. 2. The surface analysis method of claim 1, wherein the sample is exposed multiple times in succession to the agent configured to alter the oxidation state of the surface of the sample, and in each subsequent exposure of the sample to the agent, the oxidation state of the surface of the sample is changed relative to the oxidation state of the surface of the sample resulting from a prior exposure to the agent configured to alter the oxidation state of the surface of the sample.

3. 2. The surface analysis method of claim 1, wherein the sample is divided into a plurality of sub-samples, each sub-sample having a surface, and a different oxidation state of the surface of the sub-sample is generated for each sub-sample.

4. 10. The surface analysis method of claim 1, wherein the agent configured to change the oxidation state of the surface of the sample is a gaseous agent.

5. 2. The surface analysis method of claim 1, wherein the agent configured to change the oxidation state of the surface of the sample comprises one or more of ultraviolet light, ozone, and hydrogen.

6. 6. The surface analysis method of claim 5, wherein ultraviolet light is provided by at least one ultraviolet (UV) lamp, and UV light emitted from the at least one UV lamp is directed towards the surface of the sample.

7. 7. The surface analysis method according to claim 6, wherein the UV light emitted from said at least one UV lamp is in the wavelength range of 200 nm to 300 nm.

8. 7. The surface analysis method according to claim 6, wherein the UV lamp is a mercury vapor lamp.

9. 6. The surface analysis method according to claim 5, wherein ozone is supplied by an ozone generator that generates ozone gas at a concentration in the range of 0.01 to 20 ppm in the gas surrounding the sample.

10. 2. The surface analysis method of claim 1, comprising controlling the degree of change in oxidation state of the surface of the sample by controlling one or more of the exposure time of the surface of the sample to the agent, the concentration of the agent, and the wavelength and / or frequency of the agent.

11. 2. The surface analysis method of claim 1, wherein the step of identifying materials in the sample by analyzing the plurality of spectra comprises performing a multivariate analysis including principal component analysis or non-negative matrix factorization.

12. 12. An apparatus for capturing X-ray photoelectron spectra (XPS) configured to carry out the method of any one of claims 1 to 11, comprising: a sample holder; a source of the agent configured to change the oxidation state of a surface of a sample held in the sample holder; a means for controlling the exposure of the surface of the sample to the agent configured to alter the oxidation state of the surface; and an X-ray photoelectron spectrometer capable of recording multiple XPS spectra, one for each oxidation state of the surface of the sample.

13. 13. The surface analysis instrument of claim 12, further comprising a data processor configured to perform principal component analysis or non-negative matrix factorization.

14. 13. A surface analysis instrument according to claim 12, wherein the sample holder is housed within a housing.

15. 13. The surface analysis device of claim 12, wherein the agent configured to change the oxidation state of the surface of the sample is a gaseous agent.

16. 13. The device of claim 12, wherein the source of the agent configured to change the oxidation state of the surface of the sample is one or more of ultraviolet light, ozone, and hydrogen.

17. A surface analysis apparatus as described in claim 16, wherein ultraviolet light is provided by at least one ultraviolet (UV) lamp, and UV light emitted from at least one of the UV lamps is directed toward the surface of the sample.

18. 18. A surface analysis instrument according to claim 17, wherein the UV light emitted from the at least one UV lamp is in the wavelength range of 200 nm to 300 nm.

19. 18. A surface analysis instrument according to claim 17, wherein at least one of the UV lamps is a mercury vapor lamp.

20. 17. A surface analysis instrument according to claim 16, further comprising an ozone generator configured to emit ozone around a sample located in the sample holder.

21. 21. A surface analysis instrument according to claim 20, wherein the ozone generator is at least one UV lamp emitting in the region of 100 to 300 nm in the air surrounding the sample.

22. 22. A surface analysis instrument according to claim 21, wherein at least one of the UV lamps emits UV at 185 nm and / or 254 nm.

23. 17. A surface analysis instrument according to claim 16, further comprising a hydrogen source configured to release hydrogen around the sample in the sample holder.

24. 24. The surface analytical instrument of claim 23, wherein the hydrogen source is at least one zinc-air cell.

25. 13. The surface analysis instrument of claim 12, wherein the sample holder is adapted to hold a plurality of sub-samples, each sub-sample having a surface with a different oxidation state, and the X-ray photoelectron spectrometer is configured to record an XPS spectrum for each of the sub-samples.

Images