Data Collection Settings
define
Data Quality Results
Energy Resolution Settings, Flood Gun Settings, Sample Smoothness
directly define
Data Quality and Usefulness of the Information
Philosophy for Data Collection
Our philosophy is to collect spectra under analysis conditions that are practical, readily reproduced, and typically used in laboratories that use monochromatic X-ray sources and work under real world practical analysis conditions. We have assumed that the most XPS laboratories need practical reference spectra and will not spend the time or money to produce and to analyze pure, clean surfaces under ultimate energy resolution conditions.
However, we did spend extra time to collect data with above average signal to noise (S/N) ratios which reveal the presence of minor components that might otherwise be missed. In the production of these spectra we did not attempt to clean the surfaces of the native oxides or the insulators which would make charge referencing a difficult task. For practical reasons we used the C (1s) spectra from the naturally formed layer of adventitious hydrocarbons because that signal is the “de facto” standard for charge referencing insulating materials.
In the production of some spectra no attempt to produce a pure, clean surface, but some effort was made to produce surfaces with a minimum amount of natural surface contamination if needed. When ion etching was used to clean a material that contained more than one element, then ion etching was done with conditions that should minimize preferential sputtering. For the spectra of pure elements, the surface was strongly ion etched.
The elements have been analyzed under conditions that have maximized the accuracy of the binding energies. The binding energies for the pure elements are referenced to the reference energies recommended by the National Physical Laboratory in the UK. Please refer to section “Energy Scale Reference Energies and Calibration Details” for more details about calibration.
If the element is part of insulating chemical compound, then a C (1s) spectrum is also provided to allow the user to correct for sample charging. All narrow scan spectra are peak-fitted to reveal FWHM, peak asymmetry, and peak separation for spin-orbit pairs. The strong signals observed in the wide scan survey spectra are labeled and tabulated together with rough BE values of those strong signals. The details of the experimental protocol used to produce each these spectra are provided in the “Instrument and Analysis Details” section. All pure elements, except for Silicon and Selenium, were ion etched prior to analysis.
Data Collection Details
from
“The XPS Library” Website
Data Collection Process
and
Processing Protocols for Generating Correct Stoichiometry
in a Homogeneous Bulk Standard Material
used by The XPS Library to produce this Database
The results can then be used to establish unknown compositions for materials containing the same elements.
Assumptions for Using this Process
- Fresh HOMOGENOUS Bulk is exposed (see below) by fracture, cleave, scrape,
or grinding of sample using freshly cleaned scraping/cutting/grinding tools - Original Scofield SFs are correct
- Software uses unmodified, original Scofield SFs
- Carbon on surface is due only to normal adventitious (AMC) carbon
- Once sample bulk is produced, the sample will be loaded immediately into sample load lock (<5 minutes)
- Sample has NOT been exposed to any gases, siloxane, or metal coatings before receipt
- Analysis area has NOT been exposed to SEM electron or Ion beams
- X-ray source is Monochromatic
- Instrument is <20 years old
- Software provides Iterated Shirley Background fit
- Software removes transmission function effect
- X-ray to Lens angle is ~ 55⁰ to negate angular asymmetry effects
- Software provides IMFP correction or allows change of IMFP
- Pass energy does NOT saturate detector in 500-1300 eV range
- Detector system is a pulse counting A/D electron collector, not optical
- Detector system has a minimum of 5 channels
- Dead-time correction is turned OFF
- Electron take-off-angle relative to surface plane is >30 deg
- Lens-field-of-view area is NOT (or is only slightly) dependent on electron lens voltages
- X-ray spot is very near the center of the lens field-of-view
- Charge control gives FWHM for polypropylene <1.2 eV with no shoulders
- Sample is NOT degrading during the time of the analysis
- X-ray flux is NOT degrading sample during time of analysis
- Sample is NOT reacting with the gases in the analysis chamber
- Sample is NOT emitting solvent or water molecules into analysis chamber
- Vacuum of analysis chamber is <5 e-8 torr
- There are no spike noise peaks
- X-ray source is stable during analysis time
- Electron collection angle is between 20-60 degrees in width
- Magnetic lens, if present, is turned OFF, unless essential
- Software does NOT modify raw data except to remove TF
- Analysis of fresh exposed Teflon bulk gives a 66:33 atom% ratio with <5% error
- Using an Argon filled glove box (bag) for sample preparation is preferred, but not essential
Sample Preparation Methods used by The XPS Library for this Database
- Plastic sheet: Use a clean single edge razor to scrape the surface
- Plastic bead: Use tweezers or pliers to hold bead, and use a clean single edged razor to cut bead in half
- Fiber/hair: Tape fiber down, use a clean single edge razor to cut fiber
- Wafer: Scribe 1-2 mm line at edge, and then cleave using pressure from a scribe on the line or glass cleavers
- Lump: Mark outside with black or blue Sharpie pen, place in clean plastic bag, place bag on a metal sheet, hit with hammer
- Glass sheet: If sheet is 1-3 mm thick, scribe a line from edge to edge, and use glass cleavers to cleave the sheet
- Natural mineral: Mark outside with black or blue Sharpie pen place in clean plastic bag, place bag on a metal sheet, hit with hammer
- Metal sheet: Scrape surface with single edge razor blade, or scrape surface with a carbide or diamond tip
- Ceramic sheet: If sheet is 1-3 mm thick, scribe a line from edge to eedge, and use glass cleavers to break the sheet.
Alternatively, if sheet is >5 mm thick, mark outside with black or blue Sharpie pen, place in clean plastic bag, place bag on a metal sheet, hit with hammer
- Fine powder: Using a clean mortar & pestle, grind the powder enough to expose fresh bulk
- Granular pieces: Place inside clean plastic bag on a metal sheet, hit with hammer to form small grains, then try to grind in clean mortar & pestle
Data Collection Protocol
- Survey Window Size: -10 to 1100 eV (very rarely increase 1300 eV)
- Step Size for Survey: 7 to 1.0 eV per step
- Points/eV in full O (1s) peak: 1.3 pt/eV (provides 10 pts/8 eV endpt-endpt)
- Points/eV in full Fe (2p) peak: 1.3 pt/eV (provides 58 pts/46 eV endpt-endpt)
- Total Number of Channels (Points): 1,300 data points
- Dwell Time per Voltage Point: 50 msec per voltage (data) point
- Total Number of Scans: 2 scans for full survey
- X-ray Power: maximum
- Ion etching: none (scrape/cleave/grind are acceptable)
- Pass Energy: use maximum PE recommended by maker
- FWHM for clean Ag 3d5: should get ~1.8 – 2.0 eV using maximum PE
- FWHM for O (1s) for max PE should get ~2.0-2.2 eV using maximum PE
- Total Time for 2 scan Survey: <4 minutes for 2 scans (to avoid X-ray damage)
Data Processing Protocol Assuming Only Two Major Peak Types for Atom%s
There are two extreme situations:
- A single narrow symmetrical peak that is far from any related peaks from the same element (eg spin-orbit components, Auger signals), and is far from a peak from any other element present
- An overlapping spin-orbit coupled peak envelop (such as Fe 2p) that clearly has overlapping spin-orbit peaks or nearby shake-up structure (such as Cu2+2p)
- This protocol is designed for use on survey spectra only. Step size is usually 0.7-1.0
- This protocol uses iterated Shirley background method for peak area integration, and recommends peak area baseline ranges that are at usually 40-50 eV wide
- The recommended maximum range for any baseline range is an 80 eV spread
- This protocol should incorporate all significant Shake-up and Multiplet splitting signals.
- Do not use this protocol for high resolution spectra
- The Iterated Shirley baseline should never be allowed to cross over the baseline of the spectrum
Start Here
- Use (select) iterated Shirley type background method (do NOT use special modes) to integrate peaks
- Use (select) 5-10 iterations and 0.01-0.001 convergence
- Use (select) a 1.0 eV range or 5 data points for baseline (background) endpoint averaging
For Peak Type #1 – a single narrow symmetrical peak – that is well separated from any spin-orbit or other peak
- USING an O (1s) peak as an Example: Locate BE of O (1s) peak max (usually 530-532 eV)
- Add baseline start and endpoints that are very close to the nearly symmetrical peak being integrated (e.g. 5 eV below and 5 eV above the peak maximum counts)
- Place the two endpoints close to base of the peak where the background intensity is at a minimum.
- If possible, adjust (define) upper BE endpoint to be 8-15 eV above O 1s peak max BE (e.g. 538-545 eV).
- If however the Iterated Shirley baseline shows a bad fit (crosses the spectrum signal), then move the upper endpoint to lower BE toward the O 1s maximum until you locate a minimum intensity position
- Vertically expand (zoom) lower BE endpoint region to clearly see and check endpoints selected and noise
- If lower BE endpoint is located on a flat bottom region, then adjustment of lower endpoint position is finished. (Note: Placement of the lower BE endpoint is usually easy.)
- Now look at the upper BE endpoint.
- If upper (higher) BE endpoint is located on a flat bottom region, then adjustment of upper endpoint position is finished. In this case, stop. Peak integration of 1st peak is done.
- If, however, the upper endpoint is not located on a flat bottom region, then visually inspect the region of the spectrum that is 30-50 eV higher in BE
- Locate a low point in that extended 30-50 eV region, and move the upper endpoint to a point having lowest count
- Vertically expand (zoom) upper BE endpoint region to clearly see the upper endpoint selected and the noise.
- If upper BE endpoint is located on a flat bottom region, then you are finished. If not, then move upper endpoint to nearest point having lowest count.
- When upper BE endpoint is located at a flat bottom region, then stop. Peak integration is done.
For Peak Type #2 – eg an spin-orbit coupled peak envelop (such as Fe 2p) having overlapping spin-orbit peaks or nearby shake-up structure (such as Cu2p 2+)
- USING Fe (2p) as an Example: Locate BE of Fe (2p) peak max (usually 709-711 eV)
- Add baseline start point to be roughly 8-10 eV below the peak maximum counts (e.g. 695-688 eV in the case of Fe2p)
- If possible, adjust (define) upper BE endpoint to be 35-40 eV above spin-orbit lower peak max BE (e.g. 735-745 eV in the case of Fe2p)
- Vertically expand (zoom) upper BE endpoint region to clearly see endpoint selected and noise
- If upper BE endpoint is located on a flat bottom region, then you are finished.
- If not, move the upper endpoint toward higher BE to the nearest point having a valley or a minimum. Do not move more than 5-10 eV higher.
- When upper BE endpoint is finally located at a flat bottom region, then stop.
- If you have reached the maximum acceptable upper endpoint range (80 eV), then stop. You are finished.
Next Step – Generate Atom% values and Empirical Formula
- Generate atom% values by applying the t-RSF method using unmodified, original Scofield SFs and an IMFP exponent ~0.66
- If resulting atom% values match the expected theoretical atom% (ratio), then finished.
- If resulting atom% values differ by more than 10% from theory, then look at upper BE baseline endpoints. If there is no possibility of overlap with another signal, then try increasing upper BE endpoints by 5-10 eV until you locate a new minimum in the spectrum.
- Check expected ratio and atom% values again. If atom% values still deviate by more than 10% from theory, then the sample may have significant levels of contamination in the bulk or on the freshly exposed surface.
- If there is no bulk contamination, then consider the possibility that the sample might truly be a different chemical compound (homologue) and truly has different stoichiometry.
In case of Significant Overlap
- If the 80 eV peak area integration range for either of the 2 signals has a significant overlap with another XPS signal from the same element (e.g. Fe 3s overlaps upper BE range of Fe 3p), then join the peak area integration of those two peaks (e.g. Fe 3s-3p), increment the SF to reflect the addition, and increase the upper BE baseline endpoint to be 20-30 eV above the BE max of the higher BE, secondary XPS signal.
Cross-Check Results by Processing Alternate Minor Peak(s)
- To cross-check the atom% empirical ratio obtained, use an alternate secondary XPS peak to measure atom%, but use the same protocol defined above, eg for Fe in Fe 2O3, both the Fe (3p+3s) and O(2s) can be used
Data Quality is Defined by Data Collection Settings
Data Quality is Also Defined by
Electron Flood Gun Alignment and Settings
How do XPS Instruments actually measure XPS Spectra?
Instrument Parameters and Specifications
Analyzer Capabilities, Instrument Geometry, and Physical Design for SSI S-Probe XPS System
Energy Resolution Performance
Experimentally Observed Relation Between Energy Resolution and Measured FWHM from Reference Materials
for SSI S-Probe XPS System
| Element (XPS signal analyzed) | Resulting FWHM | Resolution Setting | Pass Energy | X-ray Spot Size |
| Silicon Wafer – Freshly Fractured Edge | ||||
| Si (2p3/2) crystal – fractured edge | 0.38 eV | 5 | 10 eV | 40 x 250 µ |
| Si (2p3/2) crystal – fractured edge | 0.43 eV | 1 | 25 eV | 80 x 350 µ |
| Gold Piece – Freshly Ion Etched – One Minute | ||||
| Au (4f7/2) foil – ion etched clean | 0.64 eV | 5 | 10 eV | 250 x 1000 µ |
| Au (4f7/2) foil – ion etched clean | 0.79 eV | 1 | 25 eV | 250 x 1000 µ |
| Au (4f7/2) foil – ion etched clean | 0.86 eV | 2 | 50 eV | 250 x 1000 µ |
| Au (4f7/2) foil – ion etched clean | 1.40 eV | 4 | 150 eV | 250 x 1000 µ |
| Silver Piece – Freshly Ion Etched – One Minute | ||||
| Ag (3d5/2) foil – ion etched clean | 0.42 eV | 5 | 10 eV | 40 x 250 µ |
| Ag (3d5/2) foil – ion etched clean | 0.64 eV | 1 | 25 eV | 40 x 250 µ |
| Ag (3d5/2) foil – ion etched clean | 0.75 eV | 2 | 50 eV | 40 x 250 µ |
| Ag (3d5/2) foil – ion etched clean | 1.00 eV | 3 | 100 eV | 40 x 250 µ |
| Ag (3d5/2) foil – ion etched clean | 1.30 eV | 4 | 150 eV | 40 x 250 µ |
| Copper Piece – Freshly Ion Etched – Two Minutes | ||||
| Cu (2p3/2) foil – ion etched clean | 0.85 eV | 5 | 10 eV | 250 x 1000 µ |
| Cu (2p3/2) foil – ion etched clean | 0.94 eV | 1 | 25 eV | 250 x 1000 µ |
| Cu (2p3/2) foil – ion etched clean | 1.06 eV | 2 | 50 eV | 250 x 1000 µ |
| Cu (2p3/2) foil – ion etched clean | 1.60 eV | 4 | 150 eV | 250 x 1000 µ |
| Smaller Spot Size | ||||
| Cu (2p3/2) foil – ion etched clean | 0.85 eV | 5 | 10 eV | 150 x 800 µ |
| Cu (2p3/2) foil – ion etched clean | 0.96 eV | 1 | 25 eV | 150 x 800 µ |
| Cu (2p3/2) foil – ion etched clean | 1.05 eV | 2 | 50 eV | 150 x 800 µ |
| Cu (3s) Check | ||||
| Cu (3s) foil – ion etched clean | 2.35 eV | 2 | 50 eV | 250 x 1000 µ |
Theoretical Analyzer Resolution versus Pass Energy Settings for SSI S-Probe XPS System
| Theoretical Analyzer Resolution |
Pass Energy Setting |
Effective Detector Width |
| 0.25 eV | 25 eV | 3.5 eV |
| 0.50 eV | 50 eV | 7.0 eV |
| 1.00 eV | 100 eV | 14.0 eV |
| 1.50 eV | 150 eV | 21.0 eV |
Energy Scale
Reference Energies and Calibration Details
From May 1986 to January 1993 (in Japan) for SSI X-Probe XPS System
Energy Scale Reference Energies: 932.47 eV for Cu (2p3/2) signal, 122.39 eV for Cu (3s) signal, 83.96 eV for Au (4f7/2) signal
Binding Energy Uncertainty: less than ±0.08 eV
Digital-to-Analog (DAC) Conversion Setting: 163.88
After January 1993 to 1999 (in Japan) for SSI S-Probe XPS System
Energy Scale Reference Energies: 932.67 <±0.05 eV for Cu (2p3/2) signal, 122.45 <±0.05 eV for Cu (3s) signal, 83.98 <±0.05 eV for Au (4f7/2) signal
Observed Reference Energy: 75.01 <±0.05 eV for Cu (3p3.2) signal Binding Energy Uncertainty: less than ±0.08 eV
Digital-to-Analog (DAC) Conversion Setting: 163.87
Note: NPL has recently revised reference energies to be 932.62 eV for Cu (2p3/2) and 83.96 eV for Au (4f7/2) for monochromatic systems using an electron take-off-angles of 45o
Reference Energies of Adventitious Hydrocarbon Contaminants
From May 1986 to January 1993 the electron binding energy of adventitious hydrocarbons was assumed to occur at 284.6 eV based on SSI and C. D. Wagner’s research and recommendations.
Publications by P. Swift (Surface and Interface Analysis 4, 47 (1982), S. Kohiki and K. Oki (J. Electron Spectrosc. Related Phenom. 33, 375-380 (1984), and G. Barth, R. Linder and C. E. Bryson, III (Surface and Interface Analysis 11, 307-311 (1988) have shown that the electron binding energy for various hydrocarbon contaminants and polymers is not necessarily a constant number.
Research by this author indicates that the electron binding energy for adventitious hydrocarbons lies somewhere between 284.4 and 287.0 eV depending on the underlying (oxide) materials. By taking a simple average of all available binding energies, the author has found that 285.0 eV is preferred for hydrocarbons on ion etched metals where the hydrocarbon is many hours old. For naturally-formed native oxides the preferred binding energy is 285.2 eV. Oxide based materials at the far left of the periodic element table (columns 1-4) tend to have higher values (285.2-287.0 eV, while most of the transition metal oxides center around 285.0 eV. Near the far right of the periodic table, the binding energy seems to rise to a 285.2-286.5 eV range (columns 12-14) when the native oxides of those elements are analyzed.
In routine practice, this author prefers to use the 285.0 eV number. Some potential factors that may cause this rather large range of electron binding energies for adventitious hydrocarbon contamination includes the dipole moment at the surface of the oxide material, which is expected to be much stronger than the dipole moment of a pure metal, and also, in the case of naturally formed native oxide films, the thickness of the native oxide, any physical or chemical treatments, the thickness of the adventitious hydrocarbon layer, and the type of instrument used to analyze the sample. The type of instrument being used may cause different shifts in the observed binding energy of the adventitious hydrocarbon contamination because the source may or may not generate different amounts of low energy secondary electrons from the window that protects the X-ray source. The heat from the source and contamination that degases from a just turned on source may also influence the observed binding energy. Electron flood guns and implanted ions may or may not influence the binding energy of semi-conductive materials.
Instrument Stability and Long Term Calibration for SSI XPS System
Initially each of the two SSI systems, that we have used, was calibrated 2-3 times per week because its ability to maintain accurate voltage settings was unknown. Once it was determined that the systems could maintain reliable voltage settings for 1-3 months, it was decided that good calibration could be maintained by checking and, if necessary, correcting the pass energies of the system on a 2-4 week basis. Each of the two SSI XPS instruments, that we have used, have been calibrated on a routine basis every 2-4 weeks by using SSI’s reference energies. By using this method over several years time, it was found that the maximum uncertainty (error in pass energies) was normally <±0.10 eV, but a few times rose to ±0.15 eV or less. In a very rare case, the uncertainty rose to 0.20 eV. Long term use of the SSI systems has shown that the DAC circuit does not change enough to be observed unless the room temperature changes by more than 10 degrees Centigrade. If the room temperature changes within a few hours time by more than 10 degrees or the temperature of the DAC chip is changed by more than 10 degrees, then a >0.1 eV shift, which is much smaller than the reliability of almost all literature BEs, can be observed. Variables, which seem to cause pass energy settings to change slightly, include building line-voltages, ion etching conditions, and the addition or removal of some electrical device.
Instrument Response Function &
Electron Counting Details
for SSI S-Probe XPS System
Traceability of Instrument Response Function (Transmission Function) for SSI S-Probe XPS System
Copper, gold and silver data obtained from the X-Probe system were submitted to Martin P. Seah at the NPL for a round robin test on transmission function; the results of which were published in Surface and Interface Analysis, p.243 (1993). In that publication, X-Probe data, which we contributed, were attributed to group #35. That paper reported that instrument has a Q(E) =E0.27 for Rex 4 pass energy (PE=150 V)and a Q(E) =E1.0 for the Res 2 pass energy (PE=50 V). If the NPL method is accepted as a “de-facto” standard, even though it is not an internationally recognized standard, then the transmission function and quantitation results of the S-Probe system are traceable to the “metrology spectrometer” at NPL.
Traceability of Relative Sensitivity Factors (RSFs) used for Quantitation for SSI S-Probe XPS System
Scofield’s theoretically calculated photo-ionization cross-sections are internationally used as the “de- facto” standard theoretical numbers, except in Russia and a few other places, where Band’s numbers are preferred but are almost identical to Scofield’s. The SSI system uses a very simple equation that modifies Scofield’s numbers to generate relative sensitivity factors that are used by the SSI software to calculate atom %s. That equation corrects for pass energy differences, transmission function differences, and inelastic mean free path versus kinetic energy dependency. The SSI system relies on Scofield numbers and that simple equation. Other instrument makers prefer to blend Scofield’s numbers and experimentally determined numbers.
Instrument Response Functions for the X-Probe XPS based on NPL Measurements
| Instrument Response Function: | Q(E)=E+0.27 for 150 eV PE |
| Instrument Response Function: | Q(E)=E+1.0 for 50 eV PE |
Signal/Background (S/BG) Ratios for Ion Etched Silver using a 250×1000 µ Spot* for SSI S-Probe XPS System
| Pass Energy | 25 eV | 50 eV | 100 eV | 150 eV |
| S/BG ratio** | >140 | >110 | >70 | >50 |
* Using a 90° electron take-off-angle and a smooth Ag°/Mylar film.
** The S/BG ratio is a simple numerical ratio of electrons counts at the peak maximum relative to the average electron counts observed at approximately 10 eV lower BE.
Electron Collection Lens Voltage Settings Available via Software under Instrument Calibration for SSI S-Probe XPS System
| Pass Energy (eV)* | 29.6-29.8 | 54.7-54.9 | 105.1-105.3 | 155.9-156.2 |
| Detector Widths (eV) | 3.743 | 7.486 | 14.954 | 22.297 |
| Sensitivity Exponent | -0.1 | 0.3 | 0.7 | 1.1 |
| V1 Offset (eV) | 30 | 55 | 105 | 155 |
| V1 Slope (eV) | 0.600 | 0.611 | 0.676 | 0.709 |
* These pass energies include corrections for instrument work function. True pass energies were set to 25, 50, 100, and 150 eV ±0.1 eV.
Quantitation Details and Choice of “Sensitivity Exponents” for SSI S-Probe XPS System
By default, the SSI software uses a 0.7 number as the sensitivity exponent factor for each pass energy setting which are used in an equation that modifies theoretically calculated atomic photo-ionization cross-sections (John H. Scofield, Ph.D.) to generate relative sensitivity factors that are valid for this XPS systems and which can be used to generate valid atomic percentages. The 0.7 value produces a ±10% accuracy in quantitative results for XPS signals obtained by using a 150 eV pass energy and occur within the 0-700 eV BE range. For signals that occur at higher BEs, it is generally necessary to change the sensitivity exponent factor to a 1.1 or higher value (1.4). To measure signals obtained by using other pass energies for quantitation, it is necessary to use other sensitivity exponent factors, if the user desires to maximize quantitative accuracy.
To determine useful sensitivity exponents, it is useful to use freshly ion etched poly-crystalline copper foil to test the validity of the sensitivity exponent for larger BE ranges and different pass energies. By integrating the peak areas of the Cu (2p1/2), Cu (2p3/2), Cu (3s), Cu (3p) and Cu (3d) signals with a modest amount of attention to baseline end points, it is possible to perform trial and error choices of the sensitivity exponents until a useful number is determined. Once a useful number has been entered into the computer software routine, then the software can generate “fictional” atomic percentages for each of the integrated copper signals which will generate 20 atom % values with a uncertainty of ±1-2 atom %. If the exponent factor is severely wrong then the atomic percentages will generate numbers such as 10%, 11%, 26%, 24%, and 29% or perhaps 31%, 28%, 14%, 13%, and 14%.
This trial-and-error approach may require 1-2 hours time and can be done on either wide scan data or more preferably narrow scan data for each of the 4-5 pass energies. This method, in effect, assumes that all five of the relative sensitivity factors for copper are reasonably correct. If wide scan data are used, then this method requires a little extra effort to avoid the satellites of the Cu (2p) signals.
This method, in effect, pretends that the pure copper sample is a standard material that is composed of 5 components which are present in 20 atomic % concentration. The objective is to change the sensitivity exponent until the software generates a 20 atom % result for each of the five copper signals. After useful sensitivity exponents are found, they are tested by analyzing freshly exposed bulk regions of crystalline materials such as SiO2, Al2O3, and NaCl. The high and low BE signals of the NaCl crystal are especially useful to test the validity of the sensitivity exponents.
As further checks, the freshly exposed bulk of common polymers (e.g. Mylar or PMMA) or a thin film of high purity silicone oil can also be analyzed. Teflon has repeatedly given a slightly larger than desirable error by comparison to the other materials listed above. For that reason Teflon seems to be a less desirable material to determine useful sensitivity exponent values.
Crude Tests of the Reliability of Relative Sensitivity Factors for SSI S-Probe XPS System
Crude testing of Scofield’s numbers are included in atomic percentage composition tables that give atomic percentages for only one element. This testing used the software’s automatic peak area integration software that is reasonably accurate. The results indicate that some of the relative sensitivity factors for some of the weaker signals are less reliable. If, however, all factors are taken into account, then Scofield’s numbers are reliable to a 95% accuracy level for truly homogeneous materials. These results are available when using the Spectral Data Processor software.
Effects of Poorly Focusing the Distance between the Sample and the Electron Lens
If the focus distance between the sample surface and the electron collection lens is poorly adjusted, then the number of electron counts drops very quickly. A 0.5 mm error in focus produces a >300% decrease in counts, but does not produce any observable error in binding energies, which is a common problem with many XPS instruments. A 0.1 mm error in focus produces a 15% decrease in peak area counts and is easily observed as a horizontal displacement in the static (un-scanned mode) XPS signal as observed on the standard CRT display of the detector response. Such a decrease in signal intensity generally urges the operator to correct the focus error so as to maximize the electron count rate. In this
manner, the operator has avoided any chance of obtaining false BE readings and has accurately reproduced a nearly absolute focus point which greatly increases the quantitative accuracy of any unknown sample. Experiments with the Bragg angle alignment of the crystal indicated that the maximum error due to an unusual bad alignment of the crystal would be <0.1 eV. To observe an error greater than 0.1 eV, the electron counts were found to decrease by >50%.
Traceability Details
The definition of traceability reported by Martin P. Seah and Cedric J. Powell in the J. Vac. Soc. Technol. Vol 8, p.736 (1990) publication is: “The property of a result of a measurement whereby it can be related to appropriate standards, generally international or national standards, through an unbroken chain of comparisons.” Based on this definition, the following correlations were envisioned.
Traceability of Reference Binding Energies (Energy Scale Calibration) for SSI S-Probe XPS System
At this time, there are no international standards for binding energies or reference energies. Numbers which are considered to be standard binding energies (BE), which would lead to traceability in BEs, include (a) those provided by Martin P. Seah at the National Physical Laboratory (NPL) in the United Kingdom (England), and (b) those provided by the ASTM in the USA “Standard Practice for Checking the Operating Characteristics of XPS Spectrometers” designated as “E 902-88”. Other nations also have similar national standards, which tend to imitate those set by the USA and the UK. Recently, many people in the world have been using NPL’s reference energies, which have become “de facto” standards but have not yet been accepted by the International Standards Organization (ISO).
There are still many workers and researchers using various numbers provided by the instrument makers. The author of this book was using Surface Science Instruments (SSI) Co. reference energies until December 1992 and then switched to NPL BEs in January 1993. SSI reference energies came from Hewlett-Packard (HP). SSI and HP both used high precision voltage meters from HP to calibrate their ESCA machines (i.e. X, M, and S-Probe and HP-5950 A-type and B-type, resp.). Hewlett Packard was the first company to offer a commercial ESCA system, which used reference energies developed in cooperation with Kai Siegbahn at Uppsala, who effectively developed ESCA into a useful science and received the Nobel Prize.
In a recent effort to improve the accuracy of BEs obtained from pure elements, the S-Probe pass energies were checked and corrected, if needed, almost every work-day for two months to obtain high precision and high accuracy BEs for the pure elements that are metals. This study used the NPL reference energies with Cu (2p3) at 932.67 eV with +/- 0.02 uncertainty and Au (4f7) 83.98 eV with +/-0.02 uncertainty by using 0.02 eV/pt. steps for the calibrations. To determine the “true” BE of each of the pure elements, which were scraped clean in air and then ion etched in vacuum, a 0.05 eV/pt. step was used. A repetitive ion etching (depth profile) style was used to collect wide scan, valence (Fermi edge) band, and narrow scans of the main signals for each metal at 50, 25 and 10 eV pass energies. Each repetitive experiment run lasted about 4 hours. Therefore, if NPL’s BE numbers are accepted as “de facto” international standards, then the ultimate traceability of BEs in this data collection can be related to NPL BE numbers for Cu (2p3) and Au (4f7). In a different, but similar manner, the BEs used to calibrate the S-Probe are traceable to Siegbahn’s work and the high precision, high voltage meters produced by the Hewlett Packard Company.
Traceability Transfer from Pure Metals to Non-conductive Binary Oxides
A question that should be posed is traceability to the oxide BEs. Traceability begins with NPL’s BEs for pure copper and gold as state above. Traceability then transfers to pure element BEs which are based on NPL reference BEs. Traceability then transfers to pure element BEs based on SSI’s reference BEs, and then the naturally formed native oxide data published in Volume 2 of our XPS Spectral Handbook series where BEs were measured from pure element signals and also the naturally formed native oxide signals.
Naturally formed native oxides typically have thin oxide films (10-80Å) which, in general, behave as good or true electrical conductors, which allows a direct measure of the true binding energy of many, but not all, binary oxides. To determine if traceability can indeed be transferred to true binary oxides, it was necessary to study the behavior of the naturally formed native oxides by applying various flood gun settings with the samples grounded and insulated. The results from that study can be used to transfer traceability to the experimentally observed BEs of pure binary oxides. The most difficult transfer of traceability occurs for the naturally formed native oxide systems. If the flood gun study was not done, then it is difficult to transfer traceability in a reliable manner from a conductive metal to one of its corresponding non-conductive binary oxides.
Traceability of Sample Purity
The purity of the commercially pure (99+%) binary oxides can be traced to Aldrich’s ICP or AA analyses performed by Aldrich. Copies of their results are included in the handbook at the beginning of each group of spectra. Similar data sheets were also obtained for samples bought from Cerac. A set of gold, copper, and silver samples, i.e. “Reference Metal Samples SCAA90” set, kit #367, was obtained from the NPL and used to test the instrument response function of the M-Probe system. Binding energies obtained from those gold, copper, and silver samples were identical to binding energies obtained from our commonplace gold, copper, and silver samples within the expected uncertainty of ±0.08 eV used for routine instrument calibration.
Organization of Spectral Data-Sets
Organization of Spectra
A set of spectra for a particular chemical is located by looking for the chemical formula abbreviation written in the upper right hand corner of each page. For the element called “aluminum (Al)” the user will find its chemical abbreviation “Al” in the upper right corner of the pages that belong to that set of data and spectra. The spectra are organized by using the chemical abbreviation. This means that spectra for “antimony (Sb)” can be found by looking for the chemical formula: “Sb”.
→ Periodic Table
Contents of Each Data-Set of Spectra
The spectra are presented exactly as printed by the Spectral Data Processor software which is provided in each XI SpecMaster Data-Base system. The first page of a set includes the “Detailed Surface Composition Table” which reports the peak assignments, binding energies, relative sensitivity factors, and Atom % abundance of each major signal contained in the wide scan survey spectrum for that chemical. In the title line of this first page the user will find the full chemical name along with other basic information about the chemical, such as Formula Weight, Chemical Abstract Services number, common name, and the Latin language name of the element if available.
The second page of each set is the wide scan survey spectrum with peak labels for each of the strong signals.
Detailed information about the operating capabilities of the SSI systems and the instrument and analysis conditions used to collect these data are presented in the “Instrument and Analysis Details” section of this book.
The remaining pages of each set are the high energy resolution narrow scan spectra which were obtained by measuring the strongest signals found in the wide scan survey spectrum. These spectra include detailed peak-fit results in a table and display the actual peak-fit results for each spectrum. The binding energies of insulating materials are reported in both raw and corrected form. Based on our research we have used a 285.0 eV value for the C (1s) signal of hydrocarbons for charge referencing spectra. The FWHM values for each peak of a high energy resolution spectrum is adjacent to the binding energy for that peak. The percentage numbers given for each peak is a relative percentage that is based on the intensity of that signal only (It is not an atom % value).
Reference Papers Describing Capabilities of X-Probe, M-Probe, and S-Probe XPS Systems
Robert L. Chaney, Surface and Interface Analysis, 10, 36-47 (1987)
Noel H. Turner, Surface and Interface Analysis, 18, 47-51 (1992)
M. P. Seah, Surface and Interface Analysis, 20, 243-266 (1993) [re: Response Function]
L.T. Weng et al, Surface and Interface Analysis, 20, 179-192 (1993) [re: Response Function]
L.T. Weng et al, Surface and Interface Analysis, 20, 193-205 (1993) [re: Response Function]
B. Vincent Crist, Surface Science Spectra, 1, 292-296 (1993)
B. Vincent Crist, Surface Science Spectra, 1, 376-380 (1993)
M. P. Seah, I.S. Gilmore, and G. Beamson, Surface and Interface Analysis, 26, 642-649 (1998)
Abbreviations Used
Due to the limited space provided to describe each sample in each electronic data-file, it was necessary to use various abbreviations. The abbreviations are:
scr = screen used for charge compensation
scrn = screen used for charge compensation
TOA = take-off-angle for the electrons
Aldr = Aldrich Chemical Co.
RMC = Rare Metallics Co.
SPP = Scientific Polymer Products Co. MS Co. = Metal Samples Company
FG = flood gun,
mesh = mesh-screen used for charge control,
1 mm=1 mm height used for the mesh-screen,
semi-con = semi-conductive behavior
conduc, = conductive behavior
Tech = technical grade purity,
pellet = sample pressed into pellet form by pellet press used to make Infrared KBr pellets,
plt = pellet
pel = pellet