Principles of the XPS
Surface analysis by XPS is accomplished by irradiating a sample with monoenergetic soft x-rays and analyzing the energy of the detected electrons. Mg Ka (1253.6 eV) or Al K~ (1486.6 eV) x-rays are usually used. These photons have limited penetrating power in a solid on the order of 1-10 micrometers. They interact with atoms in the surface region, causing electrons to be emitted by the photoelectric effect. The emitted electrons have measured kinetic energies given by:
KE =hv – BE – Ii. (I)
where hv is the energy of the photon, BE is the binding energy of the atomic orbital from which the electron originates, and f, is the spectrometer work function.
The binding energy may be regarded as the energy difference between the initial and final states after the photoelectron has left the atom. Because there is a variety of possible final states of the ions from each type of atom, there is a corresponding variety of kinetic energies of the photoelectrons. Moreover, there is a different probability or cross-section for each final state. Relative binding energies and ionization cross-sections for an atom is shown schematically in Figure 1. The Fermi level corresponds 10 zero binding energy (by definition), and the depth beneath the Fermi level in the figure indicates the relative energy of the ion remaining after electron emission, or the binding energy of the electron. The line lengths indicate the relative probabilities of the various ionization processes. The p, d and f levels become split upon ionization, leading to vacancies in the
The spin-orbit splitting ratio is 1:2 for p levels, 2:3 for d levels and 3:4 for f levels. As an example, the spin-orbit splitting of the Si 2p is shown in Figure 2.
Because each element has a unique set of binding energies, XPS can be used to identify and determine the concentration of the elements in the surface. Variations in the elemental binding energies (the chemical shifts) arise from differences in the chemical potential and polarizability of compounds. These chemical shifts can be used to identify the chemical state of the materials being analyzed.
In addition to photoelectrons emitted in the photoelectric process, Auger electrons may be emitted because of relaxation of the excited ions remaining after photoemission. This Auger electron emission occurs roughly
seconds after the photoelectric event. The competing emission of a fluorescent x-ray photon is a minor process in this energy range. In the Auger process (Figure 3), an outer electron falls into the inner orbital vacancy, and a second electron is simultaneously emitted, carrying off the excess energy. The Auger electron possesses kinetic energy equal to the difference between the energy of the initial ion and the doubly charged formal ion and is independent of the mode of the initial ionization. Thus, photoionization normally leads to two emitted electrons – a photoelectron and an Auger electron. The sum of the kinetic energies of the electrons cannot exceed the energy of the ionizing photons. Probabilities of electron interaction with matter far exceed those of the photons, so while the path length of the photons is of the order of micrometers, that of the electrons is of the order of tens of angstroms. Thus, while ionization occurs to a depth of a few micrometers, only those electrons that originate within tens of angstroms below the solid surface can leave the surface without energy loss. These electrons which leave the surface without energy loss produce the peaks in the spectra and are the most useful. The electrons that undergo inelastic loss processes before emerging form the background. Calculations of the inelastic mean free paths of electrons in various materials are shown in Figure 4.
The electrons leaving the sample are detected by an electron spectrometer according to their kinetic energy. The analyzer is usually operated as an energy window, referred to as the pass energy, accepting only those electrons having an energy within the range of this window. To maintain a constant energy resolution, the pass energy is fixed. Incoming electrons are adjusted to the pass energy before entering the energy analyzer. Scanning for different energies is accomplished by applying a variable electrostatic field before the analyzer. This retardation voltage may be varied from zero up to and beyond the photon energy. Electrons are detected as discrete events, and the number of electrons for a given detection time and energy is stored and displayed.
- Preparing and Mounting Samples
In the majority of XPS applications, sample preparation and mounting is not critical. Typically, the sample is mechanically
attached to the specimen mount, and analysis is begun with the sample in the as-received condition. Additional sample preparation is discouraged in many cases because any preparation might modify the surface composition. For those samples where special preparation or mounting cannot be avoided, the following techniques are recommended.
- Removing Volatile Material
Ordinarily, volatile material is removed from the sample before analysis. In exceptional cases, when the volatile layer is of interest, lhe sample may be cooled for analysis. The cooling must be to a sufficiently low temperature to guarantee that the volatile element is not warmed to evaporation by any heat load that the analysis conditions may impact.
Removal of unwanted volatile materials is usually accomplished by long-term pumping in a separate vacuum system or by washing with a suitable solvent. Use freshly distilled solvent to avoid contamination by high boiling point impurities within the solvent. Choice of the solvent can be critical. Hexane or other light hydrocarbon
solvents are probably least likely to alter the surface, providing the solvent properties are satisfactory. Samples may also be washed efficiently in a Soxhlet extractor using a suitable solvent.
- Removing Nonvolatile Organic Contaminants
When the nature of an organic contaminant is not of interest or when a contaminant obscures underlying material that is of interest, the contaminant may be
removed with appropriate organic solvents. As with volatile materials, the choice of solvent can be critical.
- Surface Etching
Ion sputter-etching or other erosion techniques, such as the use of an oxygen plasma on organic materials may be used to remove surface contaminants. This technique is particularly useful when removing adventitious hydrocarbons from the sample or when the native oxides, formed by exposure to the atmosphere, are not of interest. Argon ion etching is commonly used to obtain information on composition as a function of the exposure time to ion etching. Calibration of the spuller rates can be used to convert spuner time to infonnation on depth into the specimen. Because sputtering may cause changes in the surface chemistry, identification of the changes in chemical stales with depth may not reflect the true composition.
Abrasion of a surface can be done without significant contamination by using a laboratory wipe, a cork, a file or a knife blade. This may cause local heating, and reaction with environmental gases may occur (e.g., oxidation in air and formation of nitrides in nitrogen). To prevent oxidation of more active materials, perform abrasion in an inell atmosphere such as a glove box. The abraded material should then be transferred to the ultra-high vacuum (UHV) chamber in a sealed vessel to preserve the clean surface.
5. Fracturing and Scraping
With proper equipment many materials can be fractured or scraped within the test chamber under UHV conditions. While this obviates contamination by reaction with atmospheric gases, attention must be given to unexpected results which might occur. Fracturing might occur along the grain boundaries which may not be representative of the bulk material. Scraping can cover hard material with soft material .C. Preparing and Mounting Samples
6. Grinding to Powder
If spectra characteristic of bulk composition is desired, samples may be ground to a powder in a mortar. Protection of the fresh surfaces from the atmosphere is required. When grinding samples, localized high temperatures can be produced, so grinding should be done slowly to minimize heat-induced chemical changes at the newly created surfaces. The mortar should be well cleaned before reuse.
7. Mounting Powders for Analysis
There are a number of methods which can be used to mount powders for analysis. Perhaps the most widely used method is dusting the powder onto a polymer-based adhesive tape with a camel-hair brush. The powder must be dusted across the surface carefully and lightly, with no wiping strokes. Some researchers shun organic tape bor UHV work, but others have successfully used certain types of tape in the 10(-10) Torr range. Alternative methods for mounting powders include pressing (the powder into indium or other soft foils, supporting the powder on a metallic mesh, pressing the powder into pellets or simply depositing (he powder by gravity. With the foil method, the powder is pressed between two pieces of pure foil. The pieces are then separated, and one of them is mounted for analysis. Success with this technique has been varied. Sometimes bare foil remains exposed and, if the sample is an insulator, regions of the powder can charge differently. Differential charging can also be a problem when a metallic mesh is used to suppon the powders. If a press is used to form the powder into a pellet of workable dimensions, a press with hard and extremely clean working surfaces should be used. Gravity can effectively hold some materials in place, particularly if a shallow well or depression is cuI in the surface of that sample mount. Allowing a liquid suspension of the powder to dry on the specimen holder is an effective way of producing a uniform layer. With these methods, care must be taken in pump-down to ensure that gas evolution does not disturb the sample. A throttled roughing valve is especially effective.
- Experimental Procedure
- Technique for Obtaining Spectra
All spectra in this handbook were obtained using a PHI Model 5600 MultiTechnique system. A schematic diagram of the apparatus (Figure 5) illustrates the relationship of major components, including the electron energy analyzer the x-ray source, and the ion gun used for sputter-etching. The Model J60 Electron Energy Analyzer incorporated into the 5600 is an SCA, and the input lens to the analyzer is an Omni Focus JII lens. The excitation sources used were a Model 10.550 x-ray. source with a Model 10-410 monochromator and a Model 04-548 dual-anode source which was used with a magnesium anode. All the spectra in the handbook were taken with the x-ray source operating at 400 W(15 kV – 27 rnA). The specimens were analyzed at an electron take-off angle of 70′, measured with respect to the surface plane. The monochromatic x-ray source is located perpendicular to the analyzer axis, and the standard x-ray source is located at 54.7′ relative to the analyzer axis. In the PHI Model 5600 MultiTechnique system, energy distribution, energy resolution and analysis area are all a function of the analyzer. For all the spectra in this handbook, the spectrometer was operated in a standard mode. The Omni Focus III lens was used to scan the spectrum while the SCA was operated at a constant pass energy. This resulted in constant resolution across the entire energy spectrum. The size of the analysis area was defined by the aperture selection of the Omni Focus III lens. Analyzer energy resolution was determined by the choice of pass energy and the selected
All of the spectra in this handbook were obtained using an 800 um diameter analysis area. All the spectra in this handbook were recorded and stored using the PHI Access data system. The instrument was calibrated daily, and the calibration was checked several times each day during data acquisition. The analyzer work function was determined assuming the binding energy of the Au 4f7 peak to be 84.0 eV. All survey spectra scans were taken at a pass energy of 58.7 eV. The narrow scans of strong lines were, in most
cases, just wide enough to encompass the peak(s) of interest and were obtained with a pass energy of 23.5 eV. A lower pass energy may show more structure for some materials. The narrow spectra were necessary to accurately determine the energy, shape and spin-orbit splitting of the strong lines. On insulating samples, a high-resolution spectrum was taken of the adventitious hydrocarbon on the surface of the sample to use as a reference for charge correction. The generally accepted binding energy for adventitious carbon is 284.8 eV. The samples analyzed to obtain the spectra in this handbook are standard materials of known composition. Metal foils and polycrystalline materials with large surface areas were mechanically fastened to the specimen mount. Powder samples were ground with a mortar and pestle to expose fresh surfaces and were dusted onto adhesive tape. Most elemental standards were sputter etched immediately prior to analysis to remove surfn
contamination. Most compounds, however, were ground or cleaved, and the freshly exposed surface was analyze without etching in order to avoid possible changes in surface chemistry. Ne, Xe and Kr were implanted in graphite and Ar into silicon via ion implantation to unknown concentrations prior to analysis.
2. Instrument Calibration
To ensure the accuracy of the data presented in this handbook, the instrument used to obtain the data was calibrated regularly throughout the data-gathering process. The best way to check calibration, and the method used here, is to record suitable lines from a known, conducting specimen. Typically, the Au 4f or Cu 2p3 and 3p lines are used. The lines should be recorded with a narrow sweep width in the range of 5-10 eV, an a pass energy of 23.5 eV or less (corresponding to the pass energy nominally used for high resolution scans) should be used.
Table 1. Reference Binding Energies (eV)
fran M. P. Seah Swf/fIlu/QaAnoJ. 14,488 (1989)
There is general agreement on accurate BE values of Cu, Au and Ag standard line energies. The values in Table I are recommended for clean Au, Ag and Cu:
Because the 2p3 and 3p photoelectron peak energies of Cu are widely separated in energy, measurement of these peak binding energies provides a quick and simple
means of checking the accuracy of the binding energy scale. Utilizing all the above standard energies establishes the linearity of the energy scale and its position, i.e., the location of the Fermi level.
- Programming Scans for an Unknown Sample
For a typical XPS investigation where the surface composition is unknown, a broad scan survey spectrum should be obtained first to identify the elements present. Once the elemental composition has been determined, narrower detailed scans of selected peaks can be used for a more comprehensive picture of the chemical composition.
This is the procedure that has been followed in compiling data for this handbook, even though specimen composition was known prior to analysis.
a. Survey Scans.
Most elements have major photoelectron peaks below 1100 eV, and a scan range from 1100-0 eV binding energy is usually sufficient to identify all detectable elements. The spectra in this handbook were recorded with a scan range of 1400-0 eV (Al excitation) or 1200-0 eV (Mg excitation) binding energy.
In an unknown sample, if specific elements are suspected at low concentrations, their standard spectra should be consulted before programming the survey scan. If the strongest line occurs above 1100 eV binding energy, the scan range can be morlificd accordingly. An analyzer pass energy of 187 eV, in conjunction with the appropriate aperture, is recommended for survey scans with the PHI Model 5600 MultiTechnique system. These settings result in adequate resolution for elemental
identification and produce very high signal intensities, minimizing data acquisition time and maximizing elemental delectability.
b. Detail Scans.
For purposes of chemical state identification, for quantitative analysis of minor components and for peak deconvolution or other mathematical manipulations of the data, detail scans must be obtained for precise peak location and for accurate registration of line shapes. There are some logical rules for this programming.
(I) Scans should be wide enough to encompass the background on both sides of the region of interest, yet with small enough step sizes to permit
determination of the exact peak position. Sufficient scanning must be done within the time limits of the analysis in order to obtain good counting statistics.
(2) Peaks from any species thought to be radiation- sensitive or transient should be run first. Otherwise, any convenient order may be chosen.
(3) No clear guidelines can be given on the maximum duration of data gathering on anyone sample. It should be recognized, however, that chemical states have vastly varying degrees of radiation sensitivity and that for anyone set of irradiation conditions, there exists for many samples a condition beyond which it is impractical to attempt gathering data.
(4) With the PHI Model 5600 MultiTechnique system, an analyzer pass energy of 23 eV is normally used for routine detail scans. Where higher energy resolution is needed, lower pass energies can be utilized. For example, the sputter-cleaned Si 2p on p. 56, taken at 23 eV pass energy, can be compared to the chemically etched Si 2p shown in Figure 2.
- Data Interpretation
- The Nature of the Spectrum
- General Features.
The spectrum is displayed as a plot of the number of electrons versus electron binding energy using a fixed, small energy interval. The position on the kinetic energy scale equal to the photon excitation energy minus the spectrometer work function corresponds 10 a binding energy of 0 eV with reference to the Fermi level (Equation I, p. 10). Therefore, a binding energy scale with 0 at that point arKI increasing to the left is customarily used. The spectra in this handbook are typical for the various elements. The well-defined peaks are due to electrons which have not suffered an inelastic energy loss emerging from the sample. Electrons that have lost energy xxxx the level of the background at binding energies higher than the peak energy. The background is continuous because the energy Joss processes are random and multiple. The background in the Mg Ka induced spectra is larger than the background in the monochromatic Al Ka induced spectra because of excitation by Bremsstrahlung radiation of the nonmonochromatic light.
The “noise” in the spectrum is not instrumental in origin but is the consequence of the collection of single electrons as counts randomly spaced in time. The standard
deviation for counts collected in any channel is equal to the square root of the counts so that the percent standard deviation is 100 counts. The signal-to-noise ratio is then proportional to the square root of the counting time. The background level upon which the peak is superimposed is a characteristic of the specimen, the excitation source and lhe transmission characteristics of the instrument.
b. Types of Lines. Several types of peaks are observed in XPS spectra. Some are fundamental to lhe technique’ and are always observed. Others are dependent upon the exact physical and chemical nature of the sample. A third type is the result of instrumental effects. The following describes the various spectral features that are likely to be encountered:
(1) Photoelectron Lines. The most intense photoelectron lines are relatively symmetrical and are typically the narrowest lines observed in the spectra. Photoelectron lines of pure metals can, however, exhibit considerable asymmetry due to coupling with conduction electrons. Peak width is a convolution of the natural line width (the lifetime of the “hole” resulting from the photoionization process), the width of the x-ray line which created the photclectron line and the Instrumental contribution to the observed line width. Less intense photoelectron lines at higher binding energies are usually wider by 1-4 eV than the lines at lower binding energies. All the photoelectron lines of insulating solids are of the order of 0.5 eV wider than photoelectron lines of conductors. The approximate hinding energies of all photoelectron lines detectable by AI or Mg radiation are cataloged in Appendices G and H.
(2) Auger Lines. These are groups of lines in rather complex patterns. There are four main Auger series observable in XPS. They arc the KLL, LMM, MNN and NOO series, identified by specifying the initial and final vacancies in the Auger transition. The KLL series, for example, includes those processes with an initial vacancy in the K shell and final double vacancy in the L shell. The symbol V (e.g., KVV) iodieates that the final vacancies are in valence levels. The KLL series has, theoretically, nine lines, and others have still more. Because Auger lines have kinetic energies which are independent of the ionizing radiation, they appear on a binding energy plot to be in different positions when ionizing photons of different energies (i.e., different x-ray sources) are used. Core-type Auger lines (with final vacancies deeper than the valence levels) usually have at least one Component of intensity similar to the most intense photoelectron line. Positions of the more prominent Auger components are catalogued alongwith the photoelectron peaks in Appendices G and H.
(3) X-ray Satellites. The x-ray emission spectrum from a nonmonochromatic source used for irradiation exhibits not only the characteristic x-ray but also some minor x-ray components at higher photon energies. For each photoelectron peak that results from the routinely used Mg and Al Kalpha x-ray photons, there is a family of minor peaks at
lower binding energies, with intensity and spacing characteristic of the x-ray anode material. The pattern of such satellites for Mg and Al is shown in Table 2. A resultant spectrum using Mg x-rays is shown in Figure 6.
(4) X-ray Ghost Lines. Occasionally, x-radiation from an element other than the x-ray source anode material impinges upon the sample, resulting in small peaks corresponding to the most intense spectral peaks but displaced by a characteristic energy interval. These lines may result from Mg impurity in the Al anode or vice versa, from the anode base structure, oxidation of the anode, or generation of x-ray photons in the Al foil x-ray window. On occasion, such lines can originate via generation of x-rays within the sample itself. This last possibility is rare because the probability of x-ray emission is low relative to Auger electron emission. Nevertheless, such minor lines can be puzzling. Table 3 indicates where such peaks are most likely to occur relative to the most intense photoelectron lines. Because such ghost lines rarely appear with nonmonochromatic x-ray sources and are not possible with monochromatic x-ray sources, .they should not be considered in line identification until all other possibilities are excluded.
(5) Shake-Up Lines.
Not all photoelectric processes are simple ones which lead to the formation of ions in the ground state. but there is a finite probability that the ion will be left in an excited state a few electron volts above the ground state. In this event, the kinetic energy of the emitted photoelectron is reduced, with the difference corresponding to the energy difference between the ground state and the excited state. This results in the formation of a satellite peak: a few electron volts lower in kinetic energy (higher in binding
energy) than the main peak. For example, the characteristic shake-up line for carbon in aromatic compounds, a shake-up process involving the energy of the xxxxx transition, is shown in Figure 7. In some cases, most often with paramagnetic compounds, the intensity of the shake-up satellite may approach that of the main line. More than one
satellite of a principal photoelectron line can also be observed, as shown in Figure 8. The occunence of such lines is sometimes also apparent in Auger spectral contours (Figure 9). The displacements and relative intensities of shake-up satellites can sometimes be useful in identifying the chemical stale of an element, as discussed in Section.
(6) Multiplet Splitting. Emission of an electron from a core level of an atom that itself has a spin (unpaired electrons in valence levels) can create a vacancy in two or more ways. The coupling of the new unpaired electron left after photoemission from an s-type orbital with another unpaired electrons in the atom can create an ion with several possible final state configurations and as many energies. This results in a photoelectron line which is split asymmetrically into several components like the ones shown in Figure 10.
Multiplet splitting also occurs in the ionization of p levels, but the result is more complex and subtle. In favorable cases, it results in an apparent slight increase in the spin doublet separation, evidenced in the separation of the 2p3 and 3p3 lines in first-row transition metals. and in the generation of a less easily noticed asymmetry in the line shape of the components. Often such effects on the p doublet are obscured by shake-up lines.
(7) Energy Loss Lines. With some materials, there is an enhanced probability for loss of a specific amount of energy due the interaction between the photoelectron and other electrons in the surface region of the sample (Figure 11), The energy loss phenomenon produces a distinct and rather sharp hump 20-25 eV above the binding energy of the
parent line. Under certain conditions of spectral display, energy loss lines can cause confusion. Such phenomena in insulators are rarely sharper than that shown in Figure II and are usually much more muted. They are different in each solid medium. With metals, the effect is often much more dramatic, as indicated by the loss lines for aluminum shown in Figure 12. Energy loss to the conduction electrons occurs in well-defined quanta characteristic of each metal. These plasmons arise from group oscillations of the conduction electrons. The photoelectron line, or the Auger line, is successively mirrored at intervals of higher binding energy with reduced intensity. The energy interval between the primary peak and the loss peak is called the plasmon energy. The so-called bulk plasmons are the more prominent of these lines. A second series, the surface plasmons, exists at energy intervals determined approximately by dividing the bulk plasmon energy by the square root of two. The effect is not easily observed in nonconductors, nor is it prominent in all conductors. Plasmon lines are especially prominent in the xxxxx metal spectra in Ihis handbook.
(8) Valence lines xxxx Bands.
Lines of low intensity occur in the low binding energy region of the spectrum between the Fermi level and 10-20 eV binding energy. These lines are produced by photoelectron emission from molecular orbitals and from solid state energy bands. Differences between insulators and conductors are especially noted by the absence or presence of electrons from conduction bands at the Fermi level. Valence bands may also be used to distinguish between materials where the core level XPS photoelectron lines arc quite similar in shape and position.
2, Line Identification
In general, interpretation of the XPS spectrum is most readily accomplished first by identifying the lines that are almost always present (specifically those of C and O), then by identifying major lines and associated weaker lines, and lastly by identifying the remaining weak lines. Most modem, commercially available instruments have peak identification algorithms within their data reduction packages. Poor signal-to-noise of the data or database limitations may require manual identification of some peaks. The following step-by-step procedure simplifies the data interpretation task and minimizes data ambiguities.
Step I, The C 1s, O 1s and 0 (KLL)
lines are usually prominent in any spectrum. Identify these lines first along with all derived x-ray satellites and energy loss envelopes.
Step 2. Identify other intense lines present in the spectrum, then label any related satellites and other less intense spectral lines associated with those elements. The energy positions of the less intense lines are noted in the line position table with the spectra. Keep in mind that some lines may be interfered with by more intense, overlapping lines from other elements, The most serious interferences by the C and O 1s lines, for example, are Ru 3d covered by C 1s, V 2p and Sb 3d by 0 1s, I (MNN) and Cr (LMM) by 0 (KLL), and Ru (MNN) by C(KLL).
Identify any remaining minor lines. In doing this, assume they are the most intense lines of an unknown element. If not, they should already have been identified in the previous steps. Again, keep in mind possible line interferences. Small lines that seem unidentifiable can be ghost lines. Check the conclusions by noting the spin doublets for p, d and f lines. They should have the right separation (cf. spin orbit splitting for individual elements and Appendices G and H) and should be in the correct intensity ratio. The ratio for p lines should be about 1:2, d lines 2:3 and f lines 3:4. P jines, especially 4p lines, may be less than 1:2.
3. Chemical State Identification
The identification of chemical states primarily depends: on the accurate determination of line energies. To determine line energies accurately, the voltage scale of the
instrument must be precisely calibrated (cf. Section 0.2., p. 15), a line with a narrow sweep range must be recorded with good statistics (of the order of several thousand counts-per-channel above background), and accurate correction must be made for stalic charge if the sample is an insulator.
a. Determining Static Charge on Insulators.
During analysis, insulating samples tend to acquire a steady state charge of as much as several volts. This steady state charge is a balance between electron loss from the
surface by emission and electron gain by conduction or by acquisition of slow or thermal electrons from the vacuum. The steady-slale charge, usually positive, can be minimized with an adjacent neutralizer or flood gun. It is often advantageous to do this to reduce differential charging and sharpen the spectral lines. A serious problem is exactly determining the extent of charging. Any positive charging retards outgoing electrons and tends to make the peaks appear at higher binding energies, whereas excessive charge compensation can make the peaks shift to lower binding energies. The following are four methods which are usually valid for charge correction on insulating samples:
(1) Measurement of the position of the C 1s line from adventitious hydrocarbon nearly always present on samples introduced from the laboratory environment or from the glove box. This line, on unsputtered metals such as Au or Cu, appears at 284.8 eV, so any shift from this value can be taken as a measure of the static charge. At this time, it is not known whether a reproducible line position exists for C 1s remaining on the surface after ion beam etching.
(2) The use of an internal standard, such as a hydrocarbon moiety of a polymer sample. For the study of supported catalysts or similar materials, one can adopt a suitable value for a constituent of the support and use that to interrelate binding energies of different samples. One must be certain that treatments of the various samples are not so different that the inherent binding energies of support constituents are changed.
(3) The use of a normally insulating sample so thin that it effectively does not insulate. This can be assumed if the spectrum of the underlying conductor appears in good intensity and if line positions are not affected by changes in electron flux from the charge neutralizer.
(4) For the study of insulating polymer films, binding energies of the C functional groups may also be determined by applying a small amount of poly(dimethyl siloxane) solution to the sample surface and charge reference to the Si 2p of the silicone (at about 102.3-102.5 eV).
Several precautions should be kept in mind. If the sample is heterogeneous on even a micrometer scale, particles of different materials can be charged to different extents,
and interpretation of the spectrum is complicated accordingly. One cannot physically mix a conducting standard like Au or graphite of micron dimensions with a powder and validly use the Au or graphite line in order to correct for static charge. Differential charging can be minimized to a great extent by using a flood source of low energy
- Photoelectron Line Chemical Shifts and Separations. An important advantage of XPS is its ability to obtain information on chemical stales from the variations in binding energies, or chemical shifts, of the photoelectron lines. While many attempts have been made to calculate chemical shifts and absolute binding energies, the factors involved (especially in the solid state) are imperfectly understood, and one must rely on experimental data from standard materials. The tables accompanying the spectra in this handbook record considerable data from the literature as well as data obtained specifically for this handbook. All literature data have been carefully evaluated to the instrumental calibration and static charge reference values given above and are, therefore, directly comparable. Because occasional line interferences do occur, it is sometimes necessary to use a line other than the most intense one in the spectrum. Chemical shifts of a minor line is within 0.2 eV of the chemical shift of the primary line. However, exceptional separations can occur in paramagnetic materials because of multiplet splitting. Separations of photoelectron lines can be determined approximately from the line position tables in Appendices G and H.
- Auger Line Chemical Shifts and the Auger Parameter. Core-type Auger lines (transitions ending with double vacancies below the valence levels) usually have at least one component that is narrow and intense,often nearly as intense as the strongest photoelectron line (cr. spec”, for F, Na, As, In, Te and Ph). There are four core Auger groups that can be generated by Mg or Al ,-mys: the KLL (Na, Mg); the LMM (Cu, Zo, Ga, Ge, As, Se); the MNN (Ag, Cd, 10, So, Sb, Te, I, Xc, Cs, Ba); and NOO (Th, U).. The MNN lines in the rare earths, while accessible, are very broad because of multiplet splitting and shake-up phenomena with most of the compounds. Valence-type Auger lines (final states with vacancies in valence levels) – such as those for 0 and F (KLL); Mo, Fe, Co and Ni (LMM); and Rn, Rh aod Pd (MNN) – can be intense and are, therefore, also useful. Chemical shifts occur with Auger lines as well as with photoelectron lines. The chemical shifts are different from those of the photoelectron lines, but they are often more pronounced. This can be very useful for identifying chemical states, especially in combination with photoelectron chemical shift data. If data for the various chemical Slates of an element are plotted with the binding energy of the photoelectron line on the abscissa and the kinetic energy of the Auger line on the
ordinate, a two-dimensional chemical slate plot can be obtained. Such plots are in Appendix A for F, Na, AI, Si, 5, Cu, ?n, As, Se, Ag, Cd, In, So and Te.
With chemical states displayed in two dimensions, the Auger parameter method becomes more powerful as a tool for identifying the chemical components than using photoelectron chemical shifts alone. In the (annat adopted for this handbook, the kinetic energy of the Auger line is plotted against the binding energy of the photoelectron line, with the latter plotted in the ·x direction (kinetic energy is still, implicitly, +x). The kinetic energy of the Auger electron, referred \0 the Fermi level, is easily calculated by subtracting from the photon energy the position of the Auger line on the binding energy scale. With this arrangement, each diagonal line represents all values of equal sums of Auger kinetic energy and photoelectron binding energy. The Auger parameter, a, is defined as,or as the difference in binding energy between the photoelectron and Auger lines. This difference can be accurately determined because static charge corrections cancel. With all kinetic and binding energies referenced to the Fermi level, and recalling that or the sum of the kinetic energy of the Auger line and the binding energy of lhe photoelectric line equals the .Auger parameter plus the photon energy. A plot showing
Auger kinetic energy versus photoelectron binding energy then becomes independent of the photon energy.
In general, polarizable materials, especially conductive materials, have a high Auger parameter, while insulating compounds have a lower Auger parameter.
- Chemical information from Satellite Lines and Peak Shapes
(I) Shakt-up Unts. These satellite lines have in· tensities and separations from the parent photoelectron line that are unique to each chemical state (Figure 8, p. 19). Some Auger lines also exhibit radical changes with chemical stale that reflect these processes (Figure 9, p. 20). With transition clements and fare earths, the absence of
shake-up satellites are usually characteristic of the elemental or diamagnetic slates. Prominent shakeup patterns typically occur with paramagnetic stales. Table 4 is a guide to some expected paramagnetic states.
(2) Multiplet Splitting.
On occasion, the multiplet splitling phenomenon can also be helpful in identifying chemical states. The 3s lines in the first series of transition metals, for example, exhibit
separations characteristic of each paramagnetic chemical state. The 3s line, however, is weak and therefore is not often useful analytically. The 2p doublet separation is also affected by multiplet splitting, and the lines are more intense. The effect becomes very evident with Co compounds where the separation varies up 10 eV. When first row
transition metal compounds are under study, it is useful to accurately record these line separations and make comparisons with model compounds.
- Data Interpretation are detected, and T is the detection efficiency for electrons emitted from the sample. From Equation 5: The denominator in Equation 6 can be defined as the atomic sensitivity factor, S. If we consider a strong line from each of two elements, then:
This expression may be used for all homogeneous samples if the ratio SdS2 is matrix-independent for all materials. It is certainly true that such quantities as (J and A. vary somewhat from material to material (especially A), but the ratio of each of the two quantities 01/02 and llA.2 remains nearly constant.
Thus, for any spectrometer, it is possible to develop a set of relative values of S for all of the elements. Multiple sets of values may be necessary for instruments with multiple
x-ray sources at different angles relative to the analyzer. A general expression for determining the atom fraction of any constituent in a sample, C1 , can be written as an
extension of Equation 7:
(3) Auger line Shape. Valence-type Auger transitions (onn final-state ions with vacancies in molecular orbitals. The distribution of the group of lines is strongly affected, therefore, by the nature of the molecular orbitals in the different chemical states. Although little has yet been tabulated on this subject, the spectroscopist should bear in mind the possible utility of Auger line shapes.
- Quantitative Analysis
-For many XPS investigations, it is important to determine the relative concentrations of the various constituents. Methods have been developed for quantifying the XPS measurement utilizing peak area and peak height sensitivity factors. The method which utilizes peak area sensitivity factors typically is the more accurate and is discussed below. This approach is satisfactory for quantitative work. For transition metal spectra with prominent shakeup lines, it is best to include the entire 2p region when measuring peak area. For a sample that is homogeneous in the analysis volume, the number of photoelectrons per second in a specific spectra peak is given by:where n is the number of atoms of the element per cm3 of the sample, f is the x-ray flux in photons/cm2-sec, xxxx is the photoelectric cross-section for the atomic orbital of interest in cm2, is an angular efficiency factor for the instrumental arrangement based on the angle between the photon path and detected electron, y js the efficiency in the photoelectric process for formation of photoelectrons of the normal photoelectron energy, A. is the mean free path of the photoelectrons in the sample, A is the area of the sample from which photoelectronsValues of S based on peak area measurements are indicated in Appendices E and F. The values of S in the
appendices are based on empirical data (CD. Wagner et al. Surf Interface Anal. 3, 211 (1981)) which have been corrected for the transmission function of the spectrometer. The values in the appendix are only valid for and should only be applied when the electron energy analyzer used has the transmission characteristics of the SCA supplied by Perkin-Elmer. An example of the application of Equation 8 to analysis of a sample of known composition, poly(tetrafluoroethylene), is shown in Figure 13.
The use of atomic sensitivity factors in the manner described will nonnally furnish semiquantitative results (within 10-20%), except in the following situations:
3. The technique cannot be applied rigorously to heterogeneous samples. It can be useful with heterogeneous samples in measuring the relative number of atoms detected, but one must be conscious that the microscopic character of the heterogeneous system influences the quantitative results. Moreover, an overlying contamination layer has the effect of diminishing the intensity of high binding energy peaks more than that of low binding energy peaks.
- Transition metals, especially of the first series, have widely varying and low values of xxxx about 0.8 eV. Thus, a value of determined on xxx chemical state for a transition
metal may not be valid for another chemical state. This effect can be minimized by including shake-up peaks in the area measurement.
- When peak interferences occur, alternative lines must sometimes be used. The ratios of spin doublets (except 4p) are rather uniform, and the weaker of the pair can
often be substituted. The spectra of the elements should be consulted, but caution must be exercised because the spectra of the elements themselves can be different from the spectra of their compounds.
- Occasionally, an x-ray satellite from an intense photoelectron line interferes with measurement of a weak component. A mathematical approach can then be used to subtract the x-ray satellite before the measurement. For quantitative work, check the spectrometer operation frequently to ensure that analyzer response is constant
and optimum. A useful test is the recording of the three widely spaced spectral lines from Cu. Measurement of the peak height in counts-per-second should be made on 2o volt wide scans of the 2p3. LMM Auger and 3p lines. Maintenance of such records makes it easy to notice if an instrument change occurs that would affect
- Determining Element Location Depth.
There are four methods of obtaining information on the depth of an element in the sample. The first two methods described below utilize the characteristics of the spectrum itself but provide limited infonnation. The third provides more detailed information but is attended by certain problems. The fourth utilizes measurements at two or more electron escape angles.
- The presence or absence of an energy loss peak or envelope indicates whether the emitting atoms are in the bulk or at the surface. Because electrons from surface atoms do not traverse the bulk, peaks from the surface atoms are symmetrical above level baselines on both sides, and the energy loss peak is absent. For a homogeneous sample, peaks from all elements will have similar inelastic loss
(2) Elements whose spectra exhibit photoelectron lines widely spaced in kinetic energy can be approximately located by noting the intensity ratio of the lines. In the energy range above approximately 100 eV, electrons moving through a solid with lower kinetic energy are attenuated more strongly than those with higher kinetic energy. Thus, for a surface species, the low kinetic energy component will be relatively stronger than the high kinetic energy component, compared to that observed in the pure material. The data for homogeneous bulk solids can be compared with intensity ratios observed on unknowns to determine qualitatively the distribution of the element! in the sample. Suitable elements include Na and Mg (1s); In, Ga, Ge and As (2p3 and 3d); and Cd, In, Sn, Sb, Te, I, Cs and Ba (3p, and 4d or 3d5/2 and 4d). When the element is in a bulk homogeneous layer beneath a thin contaminating layer, the characteristic intensity ratio is modified in the opposite direction. Thus, for a pair of lines from subsurface species, the low kinetic energy line will be attenuated more than the high kinetic energy line, distorting the characteristic intensity ratio. By observing such intensity ratios and comparing them with the pure bulk elements, it is possible to deduce whether the observed lines are from predominantly surface, subsurface or homogeneously distributed material.
(3) Depth profiling can be accomplished using controlled erosion of the surface by ion sputtering. Table 5 lists some data on sputter rates as a general guide. One can use this technique on organic materials, but few data are available for calibration. Chemical states are often changed by
the sputter technique, but useful information on elemental distribution can still be obtained.
Table 5. Relative Sputter Rate
Another useful method of controlled erosion, especially of organic materials, is reaction with oxygen atoms from a plasma. This technique may also change the chemical states in the affected surface. Further, because the elements differ in their rates of reaction with oxygen atoms, the rate of removal of surface materials will be sample dependent.
(4) In XPS studies, the sample-mounting angle is not usually critical, though it does have some effect on the spectra. Very shallow electron take-off angles accentuate lhe spectrum of any component segregated on the surface, whereas a sample mounted at an angle normal to the analyzer axis minimizes the contribution from such a component. This effect can be used to estimate the depth of layers on or in the surface. This effect is not limited to flat surfaces, because angular dependence is even observed with powders, though
the effects are muted. The spectrometer used to obtain the spectra presented in this handbook integrates the signal over only a narrow range of take-off angles. It is possible to change the angle between the plane of the sample surface and the angle of entrance to the analyzer. At 90″ with respect to the surface plane, the signal from the bulk is maximized
relative to that from the surface layer. At small angles, the signal from me surface becomes greatly enhanced, relative to that from me bulk. The location of an element can thus be deduced by noting how me magnitude of its spectral peaks changes with sample orientation in relation to those from other elements. The analysis depth may be estimated by d =lambda x sine, where d is the analysis depth of the overlayer, Lambda is the inelastic mean free path, and e is the take-off angle of the analyzed electrons.
Perkin-Elmer SCAs permit angle-dependent studies by simply varying the angle of the sample surface with respect to the input lens of the analyzer. The magnification of the lens detennines the half-angle acceptance of the analyzer. An example of the information that can be gained through the use of this capability is shown in Figure 14.
Data were obtained at normal (near 90′) and grazing (near 15′) take-off angles from a silicon sample with a thin silicon oxide overlayer. The observed intensity ratio of oxidized to elemental Si is much greater at the low take-off angle.
b. Surface Distribution.
Many current XPS systems have the capability to obtain data from areas as small as 100um in diameter. This relatively high lateral resolution allows for the acquisition of XPS maps which show both elemental and chemical state information.
- Insulating Domains on a Conductor. The occurrence of steady-state charging of an insulator during analysis sometimes has useful consequences. Microscopic insulating domains on a conductor reach their own steady-state charge, while the conductor remains at spectrometer potential. Thus, an element in the same chemical state in both phases will exhibit two peaks. If a change is made in the supply of low-energy electrons which stabilize the charge (as from the neutralizer filament) or if a bias is applied to the conductor, the spectral peaks from the insulating phase will move relative to those from the conducting phase. For such heterogeneous systems, this can be an extremely useful technique. It makes it possible to determine whether the elements that contribute to the overall spectrum arc in the conducting phase, the insulating phase or both.