Electron Flood Gun Optimization & Alignment
Artist Depiction of
Optimum Distribution of Flood Gun Electrons on a
Flat Insulator in a DREAM World
REAL WORLD Distribution of Flood Gun Electrons
on a Flat Insulator
PAPER by B. Vincent Crist
Electron Flood Gun Alignment & Optimization
XPS Guide for Insulators:
Electron Flood Gun Alignment & Optimization,
Six (6) Types of Charging Problems
Current day XPS instrument makers have made significant advances in charge compensation systems over the last 20 years that make it easier to analyze insulators, but samples still have many differences in chemistry, dielectric properties, sizes, surface roughness, etc. that force instrument operators to tweak flood gun settings if they want or need to obtain high quality chemical state spectra that provide the most information. This guide teaches which flood gun variables to check, and how to optimize electron flood gun settings by presenting high energy resolution, chemical state spectra that show the result of using a poorly aligned flood gun on modern XPS instruments equipped with a monochromatic Aluminum Kalpha X-ray source. This guide is focused on the XPS measurement of insulators – non-conductive metal oxides and polymers. This guide shows that by measuring commonly available polymers (Poly-propylene and PET) or ceramic materials (SiO2 and Al2O3), that the operator can easily characterize the good and bad effects of XY position settings, and other settings provided by modern electron flood gun systems. This guide includes many original, never-before published XPS peak FWHM that will greatly assist peak-fitting efforts. This guide reveals a direct correlation between electron count-rate and best charge control settings. This guide discusses sample and instrument issues that affect surface charging; and explains how to check the quality of charge control by measuring the FWHM and BE of C (1s) or O (1s) spectra produced from the sample currently being analyzed. A list of other charge control methods is provided; along with advice and the author’s recommended method (ARM). The recent availability of extensive databases of actual XPS spectra that are cost-free is extremely beneficial to users who need real world examples of high quality chemical state spectra to guide their in-house efforts to collect high quality spectra and to interpret valuable information from the peak-fits of those spectra.
This publication is focused on the XPS analysis of insulators such as: non-conductive metal oxides and polymers. More than 70% of all materials analyzed by XPS are insulators which require charge compensation (control) when using a monochromatic X-ray source. This paper provides guidance about the many aspects of surface charging that occurs when analyzing insulating materials by using monochromatic Aluminum X-rays. The major goal of this guide is to help the reader to produce high quality XPS chemical state spectra from insulating materials by teaching the instrument operator and the data analyst how to obtain high energy resolution spectra with excellent peak-shapes that have little or no differential charging that broaden FWHM or produce unexpected shoulders.
Actual spectra having significant distortions are presented so the instrument operator has some idea about the shape of bad spectra. To simplify this learning, we also present example good and bad spectra from simple pure materials (Poly-propylene, SiO2, Al2O3, PET) that are commonly available in most laboratories. These insulators produce simple symmetrical chemical state peak-shapes that are easy to process when surface charging is fully charge controlled. This guide discusses the sample, surface, and instrument variables that affect the production of a uniform, low voltage, negative electrical field which in turn influences the peak-shapes of high energy resolution, chemical state XPS spectra.
The example spectra provided in this report were measured by using Aluminum Kalpha monochromatic X-rays and different types of modern and older flood gun systems. To obtain useful monochromatic XPS spectra from an insulating material, the instrument operator needs to control/ neutralize/ compensate that positive charge. The equipment, commonly used to control/ neutralize/ compensate the positive charge at the surface of the sample, is a beam of low voltage electrons (0.1-20 eV) produced by an electron gun that is installed inside the XPS analysis chamber (see Figure 1). The high energy resolution, chemical state XPS spectra presented in this report were taken from the The XPS Library Spectra-Base collection of >70,000 monochromatic XPS reference spectra.2-3 In the process of collecting these spectra, the author saved various sets chemical state spectra that produced abnormal peak-shapes.
FIG. 1. Artistic depiction of Idealized Uniform Charge Control. Sample is thick insulating film. Flood gun electrons fly straight. Surface has a uniform electrical field. Emitted photoelectrons fly straight with original KE values. Flood gun neutralizes all valence-hole charges.
The beam of low voltage flood gun electrons, which is static (not rastered over the analysis area), can be a focused or defocused beam. Even though very modern electron guns produce an electron beam that is focused, most users refer to focused or flood type electron beams as a “flood gun (FG)” – a term that we use in this guide. This charge control method is often called “charge compensation” or “charge neutralization”, which are different from “charge control”, also known as surface potential control.4
In modern XPS systems, the electron gun is fixed inside the analysis chamber, and is supplied with digital controls built into the instrument operating software that allows the operator to control the XY position and the other operating variables of the electron beam. Many modern charge compensation systems also include a built-in source of low voltage argon ions (+1 to +30 eV) that simultaneously irradiates the X-ray analysis area while that same area is irradiated with low voltage electrons from the flood gun. When properly adjusted, the overlapping flood of low voltage Argon ion and low voltage electrons produces a well-controlled surface charge (potential). One modern XPS instrument maker (Kratos) provides a repeller plate system that enhances the charge compensation produced by a co-axial source of FG electrons without the use of argon ions.
The voltage of the positive charge from the X-ray induced core-holes within the surface analysis area will range between the Aluminum X-ray energy (+1486 eV, upper limit) and valence-hole energies (+ 1 to +25 eV, lower limit). Because most materials have Carbon as a surface contaminant or it is part of the bulk, the positive charge at the surface is due, in part, to the emission of the C (1s) electron which is equal to the binding energy (BE) of the C (1s) electron, +285 eV. The loss of photoelectrons from Oxygen atoms produces a corresponding +530 eV positive core hole in the atom. The loss of photoelectrons from various electron orbitals of other elements in the sample produce positive core-holes and valence-holes with positive voltages that can range from +1 to +1,486 eV. The resulting XYZ distribution, and density of the positive charges trapped within the surface can be as deep as 1-5 microns, which corresponds to the depth of penetration of Aluminum X-rays. In comparison, photoelectrons that are measured, can only escape from the top 0-10 nm of the surface. Photoelectrons produced below the top 0-10 nm depth range do not produce intense XPS signals. The deep photoelectrons that are deeper than 10 nm contribute to some of the background signal, but most deep photoelectrons are trapped and never escape out of the bulk into the vacuum of the instrument.
TABLE 1. Electrical, physical, optical, and chemical properties that distort XPS spectra
- Variations in chemical, electrical, and physical properties of the surface being analyzed,
- Dielectric nature of the material in the top 20 nm,
- Flux and shape of the focused X-ray beam,
- Depth distribution of the positive valence-holes (charges)
- Depth distribution of the positive core-holes (charges)
- Quality of the charge control/compensation process,
- Electrical grounding of the sample stage,
- Differential charging that exists.
FIG. 2. Example of well Controlled Charging of Poly-propylene. Peak is symmetrical. FWHM is <1.2 eV. No Distortions. No sloping tails. No shoulder. Electrical field (black mesh) is flat due to a 90% transmission nickel mesh-screen sitting <1 mm above surface.
TABLE 2. Contents of this publication includes the following.
- Normal symmetrical peak-shapes of the C (1s), O (1s), Si (2p) and Al (2p) spectra produced by pure polymers, SiO2, and Al2O3,
- Normal peak-shapes of spin-orbit pairs (e.g., Cr (2p), Zr (3d), Cd (3d) Si (2p), K (2p), Br (3d) Y (3d), S (2p)),
- Different types of peak-shape distortion caused by uncontrolled charging (aka differential charging),
- Effects of changing XY positions of the flood gun for a pure hydrocarbon polymer (Poly-propylene, C (1s)),
- Effects of adjusting the flood gun beam voltage, emission current, extractor voltage, and focus,
- List of typical FWHM for C (1s) and O (1s), and metal oxides that help you to decide if you have differential charging,
- List of charging problems and potential causes,
- Guidance on sample preparation,
- Other methods that can be used to enhance charge compensation
To minimize or eliminate this positive (+) surface charging of an insulator, the operator turns on a charge compensation or charge control system (electron flood gun). Very soon after the electron flood gun is turned on, the X-ray irradiated surface area of the insulating sample will soon develop a small negative charge (-1 to -5 eV) if the electron flux (emission current) is sufficient, and the direction of the electron beam directly overlaps the X-ray irradiated area of the surface.
When the surface charging of the analysis area is uniform, balanced, and steady it is considered to have “Controlled Charging”. An example of controlled charging is shown as Figure 2. If that surface area has charging that is not uniform, not balanced, or unsteady, then that surface area has “Uncontrolled Charging,” also known as “Differential Charging”. Examples of these two types of charging are presented later. The shape and usefulness of chemical state XPS spectra produced from the top 10-12 nm of the surface of insulating samples are affected by variations in chemical, electrical, and physical properties of the surface being analyzed and the type of “charging” that exists.
The various chemical and electrical properties of the surface volume being analyzed can cause distortions in the resulting negative (-) electrical field in the surface and just above the physical surface protruding into the vacuum outside of the sample surface. Distortion of the uniformity of the electric field within the surface and above the analysis area can distort the shape of high energy resolution, chemical state spectra. These distortions can appear as anomalous shoulders at low BE side, sloping charge tails at the low BE side, false doublets, or peaks with broad Full Width at Half Maximum (FWHM). Charging can and often does produce peaks with unexpected binding energies, which, in turn, degrades the reliability of the binding energies, BEs, used to make chemical state assignments.1
This guide discusses the sample, surface, and instrument variables that affect the production of a uniform, low voltage, negative electrical field which in turn influences the peak-shapes of high energy resolution, chemical state XPS spectra.
Because industry and academia regularly analyzed many different types of materials by XPS the instrument operator needs to understand some of the charging effects that exist. The operator must learn how to effectively use a charge compensation system (e.g. electron flood gun) to collect reliable5 and meaningful chemical state spectra from: polymer films, paint films, semi-conductors, topological materials, catalysts, unknown contaminations, ceramics, paper, dental implants, medical implants, 3D printed plastics, non-conductive metal oxides, hydroxides, carbonates, sulfides, sulfates, minerals, glasses, glass coatings, laser parts, cosmetics, medicines in tablets, and more.
Each new sample from a different material has a unique surface behavior that is different from the last surface, so the operator needs to learn and practice how to optimize the electron flood gun and how to recognize or determine if the current sample is suffering from unexpected differential charging. This guide addresses those variables and shows that the XY position of the flood gun is the main variable.
Practical real world experience teaches each operator that the BE, FWHM, and expected Gaussian shape of the O (1s) and the C (1s) signals are the main indicators of the quality of charge control for the current sample. A single, nearly symmetrical O (1s) peak-shape is available from many different materials. The shape of the O (1s) signal can be discerned by using the un-scanned mode spectra, Snap-shotTM mode spectra, or normal scanned mode spectra. The optimum beam voltage to use depends on the flood gun design and the surface, but the usual range is between -0.1 to -4.0 eV.
Materials that produce a single symmetrical O (1s) peak are useful to help the operator to adjust flood gun settings, recognize differential charging, control charge, and deal with charging problems. In the process of collecting spectra from these types of insulating materials, the author collected and saved various sets chemical state spectra that produced abnormal peak-shapes. The high energy resolution, chemical state O (1s), C (1s), Si (2p) and other XPS spectra presented in this report were taken from the The XPS Library Spectra-Base collection of >70,000 monochromatic XPS reference spectra.2-3.
SURFACE CHARGING, FLOOD GUN ELECTRON BEAM, UNCONTROLLED CHARGING,
CONTROLLED CHARGING, CHARGING EFFECTS
Surface charging (+ or -) is a phenomenon that is normally associated with true insulators, but, in fact, surface charging also occurs for semiconductors, very thin insulating films on metals, native oxides, side-by-side strips made of conductors and insulators, and conductors that are deliberately isolated from grounding. As a result, >70% of all materials analyzed by XPS can or do suffer from various levels of surface charging.
Surface charging is the accumulation of electrical (voltage potential) charge on the sample surface and or within the sample near the surface. Surface charging starts when particle or photon bombardment strikes a surface, and that surface retains either a negative or positive charge. Surface charging is often assumed to be uniformly distributed in XYZ with bulk charging being a similar phenomenon deep within a sample.
For XPS, surface charging begins when a monochromatic beam of Aluminum Kalpha X-rays irradiates the surface of a material causing photoelectrons to be emitted leaving behind atoms that have positive (+) core-holes within the top 10 nm of the surface. The lifetime of those core-holes is extremely short, typically 10–16 to 10–18 seconds because the atom with its core-holes undergoes extremely fast intra-atomic processes that include electrons cascading from filled lower energy orbitals (mid-level or valence-level) into empty higher energy (core) holes, in turn producing mid-level (mid-core) holes. See Figures in Supplementary Information section.
Normal materials are electrically neutral, but when monochromatic X-rays excite the atoms of any material a positive surface charge develops. The uniformity of the resulting positive charges across the surface area being analyzed can be uniform or non-uniform. This uniformity or non-uniformity directly affects the operator’s efforts to produce high quality XPS spectra. In some cases, controlled charging is easy to achieve, but in other cases, the surface suffers from uncontrolled differential charging. The concepts and features of “controlled charging” and “uncontrolled differential charging” are introduced in this section. See Figure 3.
In the case of a properly grounded conductive material, the resulting positive core-holes are instantaneously neutralized by electrons that exist in the valence band (Fermi sea), free electrons that are 0.0 to -0.1eV above the Fermi level (~25C), and electrons supplied through an electrical ground attached to the instrument. In this case, there is no net charge, so there is absolutely no need to use a flood gun. As a result, there is absolutely no need to correct any BEs.
In contrast, when an insulating material is irradiated by a monochromatic X-ray source, the surface area of that insulator will instantly develop a net positive sample charge due to the loss of the photoelectrons that left behind core-holes and valence-holes with positive charges ranging from +1 eV to as large as +1486 eV. The positive core-hole charges force the atom to undergo various intra-atomic cascades of electrons and other phenomena to neutralize the core-hole charges. In effect, this process forces the positive holes to migrate from the inner core of the atom out to the valence band where the net effective positive charge will range from +1 to +30 eV. When a properly adjusted beam of flood gun electrons overlaps that X-ray irradiated area, then that beam of low voltage electrons will quickly neutralize the low voltage positive valence-holes and develop a small net negative charge (-0.1 to -5 eV) within the top 10-12 nm of the surface. .
The buildup of a net negative (-) charge on the surface of an insulator is because insulating materials behave as air-based capacitors (batteries), that collect a pool of the flood gun electrons on the surface of the sample. That pool of electrons can be scattered over an area of the surface that is several times larger than the area irradiated by the X-ray beam.
Based on real-world experience and design testing, the XPS instrument makers design the flood gun electron beam diameter to be 2-5X larger (< 2mm) than the X-ray irradiated area. For that reason, the negatively charged surface area is 2-5X larger (<2 mm) than the X-ray irradiated area. If we could readily measure beyond that negatively charged area, then we will find a neutral, electron free surface area, unless that area had been previously exposed to the X-ray beam or the electron beam. This phenomenon reflects the nature of the insulator to behave as a capacitor.
Both types of surface charging (+ or -) produce various electrical field phenomena and problems that distort the shape of the chemical state spectra and cause chemical state signals to appear at erroneous or misleading BEs, which, in turn, degrades the reliability1 of the BEs used to make chemical state assignments.
The localized micron-sized variations in the chemical, electrical, and physical properties (e.g. dielectric nature, roughness…) of the analysis area, produce variations in the net negative charge at the surface that are non-uniformly distributed in XY. Highly polar chemical compounds produce a special type of differential charging. Highly polar compounds have strong surface dipole moments that can project, in the Z axis, a positive field several hundred nanometers above the surface.10 This dipole moment produces unexpected increases in BEs because the surface dipole moments produce preferential retardation of the KE of the photoelectrons. We will report on this phenomenon in a later publication.
FIG. 3. Schematic examples of: (a) uncontrolled differential charging and (b) controlled charging across a net negative surface
Surface charging can be either controlled or un-controlled. Un-controlled differential charging is more commonly known as Differential Charging (Figure 3a) which produces one or more types of distortion in XPS peaks. Controlled Charging (Figure 3b) produces distortion free XPS peaks. A simple schematic in Figure 3 displays the basic difference between controlled charging and un-controlled differential charging.
Differential charging normally occurs horizontally, but it can also be a vertical phenomenon which is discussed in more detail later.
FIG. 4. Artistic depiction of idealized uniform controlled charging. Sample is a thick insulating film. Flood gun electrons fly straight. Surface has a uniform electrical field. Emitted photoelectrons fly straight with original KE values. Flood gun neutralizes all valence-hole charges.
Flood gun beam of low voltage electrons
As stated above, a low voltage beam of electrons is commonly used to control the surface charging that occurs when a monochromatic beam of X-rays produces positive core-holes and valence-holes within the surface of insulating materials.
Under optimum conditions, the beam of low voltage electrons emitted from the flood gun will fly in a straight line path and land on the surface of the insulator. (See Figure 4). By design, the surface area covered by these low voltage electrons is roughly 2-5X larger (<2 mm) than the area irradiated by the X-ray beam. The electron flux (emission current) in that flood gun beam is normally sufficient to allow the valence-holes to be neutralized fast enough that a slight excess of electrons remains on the surface to produce a small negative electrical field, which may or may not be uniformly distributed across the surface. To be useful, this charge-balance system needs to retain a slight excess of negative charge on the surface.
If the surface of all insulating samples was perfectly smooth, and the dielectric nature of the surface was uniform, and the X-ray beam width was an orthogonal step-function, and the surface was perfectly clean, then it is possible to produce a charge-balance that is uniform. In this near perfect situation, the spectra would have no distortions, peak-fitting would be easy, and the measured BEs would be easy to correct. See Figure 2.
However, in the real world, real world samples produce non-uniform electrical fields within the surface and just above the surface. An excessive flood of low voltage electrons can replicate the non-uniform electrical field of the surface. The resulting irregular non-uniform field causes emitted photoelectrons to be deflected, slightly accelerated, or to have various KEs as they escape the surface and fly toward the electron collection lens. The irregular non-uniform field also causes incoming flood gun electrons to be deflected or to lose some KE. An artistic depiction of this situation is shown in Figure 5.
FIG. 5. Artistic Depiction of Uncontrolled Differential Charging. Sample is a thick insulating film. Flood gun electrons flight paths are irregular due to irregular surface voltage. Emitted photoelectrons flight paths are irregular due to irregular surface voltage.
Using a standard flood gun system, the C (1s) peak would have a sloping charge tail at lower BE or maybe a shoulder. The FWHM is >2 eV.
In Sections C and D, we describe the differences between “uncontrolled differential charging: and “controlled charging”. These sections include Tables and Example Spectra to compare and contrast the key features of both types of surface charging.
Uncontrolled differential charging
The most common type of surface charging problem is un-controlled “Differential Charging.” Differential charging means that there is a non-uniform distribution of charge across the surface of the area being irradiated and above the surface as depicted in Figure 5. Differential charging can be horizontal (across the surface) or vertical (into the Z aspect of the sample). Differential charging is usually due to an irregular distribution of negative voltages (or in rare cases, positive voltages) over the area being irradiated. A summary of the causes of differential charging is given in Table 3.
TABLE 3. Main causes of un-controlled differential charging
- Flood gun is mis-aligned – not optimized,
- Flood gun voltage or emission current – not optimized,
- Differences in X-ray flux due to the Gaussian-shape of the X-ray beam,
- Non-uniform electrical, chemical, or physical nature of the surface area being analyzed,
- Irregular electrical grounding of the sample, the sample stage, or the instrument itself,
- Bottom of sample has irregular conductivity,
- Due to vertical differential charging, the main causes are surface dipole moments and irregular capacitor-like behavior of different surface layers.
Distortion of the uniformity of the electric field on a surface, also known as Differential Charging, can occur above or within the surface of the analysis area. Differential charging distorts the true shape of the peak(s) in a high energy resolution, chemical state, spectrum. Differential charging often produces anomalous shoulders or sloping charge tails at low BE side for each high resolution spectrum obtained from the same sample. An example of this problem is shown here for the C (1s) spectrum obtained from a dirty piece of acrylate adhesive (Figure 6).
Differential charging can also produce false doublets, and peak broadening with FWHM 3-4X larger than the true FWHM. Differential charging can produce unexpected shifts in binding energies which, in turn, degrade the reliability of the binding energies used to make chemical state assignments. A list of the possible features caused by uncontrolled differential charging is presented in Table 4 below. Examples of spectra having differential charging features are presented throughout the remainder of this publication. These spectra will have labels that highlight the problem type or feature that distorts the peak-shapes in those spectra.
TABLE 4. Possible features of uncontrolled differential charging visible in chemical state spectra.
- Obvious “sloping charge tails” spread over a 2-6 eV range on lower BE side of each high resolution spectrum,
- Shoulders (10-30% intensity) on lower BE side of all high resolution spectra that might be a different chemical state,
- Peak broadening – FWHM >2x larger than typical (the normal range of FWHM for main XPS peaks is: 1.0 – 1.5 eV),
- Weak broadening of true peak-shape with BE shifting,
- Distortion of Peak-shapes due to Flood Gun Effects on Non-conductive Native Oxides – Grounded versus Insulated – aka Vertical Differential Charging
A typical example of “uncontrolled differential charging” is shown in Figure 6. The sample is a Poly-acrylate adhesive on the back side of a paper label. The chemical state spectrum is from the C (1s) signal. This spectrum shows the data that was initially measured. By looking at this spectrum we suspected there was a charging problem. The peak-shape features that indicate there is a problem are: (a) the sloping charge tail on the low BE side, and (b) the non-Gaussian shape of the main carbon peak on the low BE side. Because we had analyzed this adhesive many times before and had a “reference spectrum” we also knew that the peak at 278 eV was not real (a ghost peak), and we knew that the valley at 288 eV should be deeper. This is the advantage of having high quality XPS reference spectra2. Unfortunately, we sometimes do not have reference spectra so we must learn to recognize peak-shape features that indicate the probable presence of uncontrolled differential charging. This guide is designed to teach the instrument operator and the data analyst how to recognize differential charging,
FIG. 6. This C (1s) spectrum from an acrylate adhesive on the backside of a paper label shows a sloping charge tail and a small peak at the low BE side of the C (1s) spectrum which are typical of un-controlled differential charging.
In addition to the shoulders and sloping tails that might appear when a surface is suffering uncontrolled charging, other differential charging features include peak broadening that can be small but sometimes produces FWHM that are >3-4X the normal FWHM. Such large broadening is easy to notice, but weak broadening is easily overlooked unless you have experimented with your flood gun system and know how to optimize the flood gun settings. By analyzing simple commonplace materials having well-known FWHM the operator will learn which flood gun setting directly affects their samples. In this guide, we list several commonplace materials that have a single chemical state, and one commonplace material (a water bottle) that has 3 well known chemical states (PET).
Later in this guide, there are numerous example spectra that help you to see and to learn which flood gun settings are important to study and learn to control.
Differential charging, when it exists, makes it exceedingly difficult to generate useful and reliable5 chemical state spectra which are key to producing useful information. In certain cases, differential charging will also affect the shape of survey spectra. Sometimes the effect on a survey spectrum is small, and we do not see an obvious effect on the survey spectrum. In such a case, we may not see charging problems until we collect chemical state spectra.
The effects of strong differential charging on survey spectra are shown in Figures 7 and 8. In this guide, we focus on differential charging that affects high energy resolution, chemical state spectra, because they are expected to provide chemical state information.
Differential charging usually does not affect the quantitative results measured from the survey spectrum unless the effect is extreme. Differential charging problems have, however, likely produced errors in the BEs listed in the free NIST SRD-20 XPS database of BEs. This on-line database has thousands of BEs derived from insulators, but this type of problem makes it difficult to trust the reliability of the NIST BEs to be used to make reliable5 chemical state assignments from insulating materials. To make reliable chemical state assignments the data analyst needs reliable monochromatic spectra2-3 from the same or related chemical compounds, and FWHM if available.
FIG. 7. These survey spectra show controlled charging and uncontrolled (differential) charging from Teflon when the electron flood gun was ON, turned OFF, then immediately run again. Blue colored line shows data with flood gun turned ON. Red colored line is due to flood gun turned OFF.
FIG. 8. These survey spectra show controlled charging and uncontrolled (differential) charging from silica glass. The electron flood gun was ON, turned OFF, then immediately run again. Red colored line shows data with flood gun turned ON. Blue colored line is due to flood gun turned OFF.
Charging has been studied in detail, but it is still an active topic and problem12-20. The C (1s) XPS spectrum shown in Figure 9 is from a simple non-conductive insulator, Poly-propylene (-CH2-). The measured spectrum has a single, fully symmetrical peak. Poly-propylene is a pure hydrocarbon polymer that has just one chemical (state) type of carbon that has sp3 hybridization. For this reason, the C (1s) signal from clean, pure Poly-propylene is expected to appear as a single symmetrical peak.
FIG. 9. Prime example of peak-shape having controlled charging. Gaussian shaped peak, no sloping charge tail, no shoulders. Narrow FWHM (<1.2 eV). Peak-shape is symmetrical.
This spectrum displays a C (1s) chemical state signal, measured from Poly-propylene, that is free from any obvious distortions. It is a prime example of controlled charging. There are no shoulders. There is one single peak. There are no sloping tails on the low BE side. The peak-shape is Gaussian.
The FWHM of the C (1s) peak is ~1.1 eV, which is typical for C (1s) peaks from a clean, pure organic material when the sample has “controlled charging”. The features observed for spectra having “controlled charging” are listed in the Table 5 below.
FIG. 10. Peak-fit of a complex XPS chemical state spectra from an insulator (PET) having controlled charging. Gaussian shaped curves, No sloping tail, No shoulders. Narrow FWHM (<1.0 eV).
TABLE 5. Features of XPS peaks having controlled charging for insulating samples.
- Isolated single peak is symmetrical, Gaussian shape,
- For contamination type, adventitious hydrocarbon, the C (1s) peak should be <1.6 eV (FWHM is typically 1.4 to 1.6 eV for adventitious hydrocarbon),
- For pure organics, the C (1s) FWHM ranges from ~1.0 to 1.4 eV,
- Peak-shapes and FWHM do not change after the Flood Gun voltage is increased from 0.1 to 2.0 eV,
- Shoulders, tails, or small peaks at higher BE, are real and due to higher oxidation states,
- BE of C (1s) hydrocarbon peak is ~285 eV or slightly smaller. (Never greater than 285.5 eV, or smaller than 284.5 eV) Carbide BEs range between 281 and 284 eV,
- When you have 2 or more XPS signals (e.g., C 1s and O 1s) and only one signal shows a shoulder at lower BE, then that shoulder is a real species. (If all high resolutions signals have similar shoulders on low BE side, then they are due to differential charging.).
The C (1s) chemical state spectrum shown in Figure 10 shows a C (1s) complex chemical state spectrum measured from freshly scraped film of Poly-ethylene terephthalate (PET). By comparison to a published reference spectra for PET this C (1s) spectrum has the same four (4) peaks and is free from any obvious distortions. The FWHM of the C (1s) peak produced by the ester (COOR) type carbon is ~0.9 eV which is typical for C (1s) peaks from a clean, sample of PET. There are no obvious shoulders or sloping tails on the high BE side of this spectrum which means that we have produced a useful C (1s) spectrum from PET. Because the FWHM of the ester type carbon is <1.0 eV, we consider this spectrum to be free from differential charging, which means that this surface charging of this sample is “well” controlled. To keep terms consistent, we say that this sample has “controlled charging”.
PET has become the reference material that is recommended to check on the charge control performance of the flood gun system sold with brand new XPS instrument. The key feature that defines that charge control performance is the FWHM of the ester peak. To pass instrument performance tests, the FWHM from the C (1s) ester peak at ~289 eV BE must be <0.9 eV.
PET is readily available because it is used to make various plastic bottles, including soft drink bottles and water sold in bottles. PET can be cut with a clean razor blade to expose fresh clean bulk or scraped with a clean razor blade. PET can be safely cleaned with various alcohol solvents
The C (1s) peak-shape of PET is complex and has 3 well resolved peaks which is why it is useful to check on charge control performance. But to learn about the basics of charge control and how to operate a flood gun, we use pure Poly-propylene which is available as plastic cups. Look for the initials “PP” on the bottom of the cup. Poly-propylene is not transparent. It is translucent.
The features observed for spectra having “controlled charging” are listed in the Table 5.
Charging effects on XPS peak-shapes, FWHM, and BEs
Surface charging effects on XPS peak-shapes are affected by: (a) flood gun mis-alignment, (b) flood gun voltage or emission current too low or too high, (c) non-uniform Gaussian shaped flux intensity of the focused X-ray beam, (d) non-uniform electrical, chemical, or physical nature of the surface area being analyzed, (e) irregular electrical grounding of the sample, the sample stage, or the instrument itself, or (f) strong surface dipole moments in highly polar compounds (see Table 3).
These interactions produce “surface charging effects”, which are due to the electrical potential in or on the surface that are produced by particle or photon bombardment. The total charge on a sample can be either negative or positive depending on the overall density of the charges.
Examples of charging effects on XPS peak-shapes, FWHM, and BEs are presented in Figures 11. A sloping charge tail is highlighted by an arrow in Figure 11. This sloping charge tail is due to mis-alignment of the flood gun beam. This mis-alignment caused the C (1s) peak to move to higher BE and lose electron counts.
FIG. 11. C (1s) signal from Poly-propylene sheet having “Controlled Charging” versus “Un-controlled (Differential) Charging” due to differences in XY positions of the electron flood gun.
The positive valence-hole charges that produce this type of surface charging can range from +1 to +10 eV which, after the flood gun is turned on, becomes a surface charge that is negative and ranges from -0.1 to -10 eV.
In the following five (5) sections we discuss:
- Simple peak-shapes from insulating surfaces.
- Complex peak-shapes from insulating surfaces.
- Peak-fits of complex peak-shapes of insulators.
- Undesired peak-shape features caused by differential charging.
- Advantage of using a charge-control mesh-screen system.
- Simple peak-shapes from insulating surfaces having controlled charging.
The XPS spectra shown as Figures 12 and 13 are from the non-conductive insulators, Poly-propylene and manmade SiO2. The XPS signals produced by these materials have only one chemical state for each XPS signal. When surface charging is controlled, then these peak-shapes should be a single peak, fully symmetrical, having Gaussian shape, and no sloping tails which is the shape seen in all three (3) chemical state spectra – Figures 12 and 13.
These measured spectra have peak-shapes that are symmetrical and have no charging tail because the charging was fully controlled, so we say that these samples have “controlled charging”.
The features observed for spectra that have controlled charging are listed in the Table 6 below.
FIG. 12. Peak-shape of a properly charge controlled XPS signal. This C (1s) spectrum from clean Poly-propylene is symmetrical. It does not have a sloping charge tail or shoulder on the low BE side which is characteristic of differential charging. The FWHM for Poly-propylene C (1s) peak is ~1 eV.
FIG. 13. O (1s) and Si (2p) spectra from pure SiO2 (silica) freshly cleaved to expose the bulk. The symmetry and absence of tails or shoulders on the lower BE side of both (a) and (b) spectra is typical of a material that is fully charge controlled in the surface area being analyzed.
TABLE 6. Features of spectra having controlled charging.
- Single, isolated (1s) peaks from Li, Be, B, C, N, O and F are fully symmetrical, and Gaussian shaped for controlled charging case,
- Spectra from spin-orbit coupled signals (e.g., 2p, 3d, 4f) have smooth Gaussian shaped slopes at low BE side,
- FWHM of the main peaks (e.g., main metal peak or O (1s) on the same sample) range between 1.0 to 1.5 eV,
- FWHM of adventitious hydrocarbon C (1s) peak on the same sample should range between 1.2 to 1.6 eV,
- For polymers, the hydrocarbon type C (1s) FWHM should range between 0.9 to 1.2 eV,
- Peak-shapes and FWHM do not change when Flood Gun voltage is increased from 0.1 to 2.0 eV,
- BE of C (1s) hydrocarbon peak max is 285 eV or smaller. (Never greater than 285.2 eV, except due to vertical charging),
- Shoulders, tails, or small peaks on higher BE side of 1s, 2p, 3d or 4f spectra are real, and are due to higher oxidation states,
- When you have 2 or more XPS signals (e.g., C and O) and only one shows a shoulder at lower BE, then that shoulder is a real species. (If all high resolutions signals have similar shoulders on low BE side, then they are due to differential charging.).
Complex peak-shapes from insulating surfaces that have controlled charging
The high energy resolution, chemical state spectra shown in Figure 14a-b have (3d) and (2p) spin-orbit couplings that have and are good examples of “controlled charging” as noted by the absence of shoulders or sloping tails on the lower BE side of the more intense peak and the smooth Gaussian shaped curve on the low BE sides of each set of signals.
Figures 14a-b have a smooth Gaussian shaped curves on both high and low BE sides of the 3d peak-shape that indicate good charge control. Spectrum in Figure 14b is more complex but, very importantly, show Gaussian shaped curves at the low BE side, which is also free from any sloping tails or shoulders. Allowing us to label these spectra as having “controlled charging”.
The two 3d peaks in Figure 14b have shoulders on the higher BE side of both 3d spin-orbit peaks which are due to the 2nd chemical state present in that sample. The two peaks in Figure 14b have smooth Gaussian shaped curves at the low BE side that indicate good charge control.
FIG. 14. These spectra are more complicated because they show spin-orbit coupled peaks, peaks with a second chemical state, and peaks have multiple splittings (M.S.). The Zr (3d) spectrum (a) from ZrO2, and Cd (3d) spectrum (b) from native oxide of Cd metal show excellent charge control characteristics. Narrow FWHM, smooth Gaussian shaped curve on low BE side indicating no differential charging.
So, what is the definition of a “native oxide”? Native oxides are thin oxides of pure metals or metal alloys that have formed naturally at room temperature and pressure while sitting in the normal lab air or in a work area for many weeks or months. The main native oxides that form are assumed to be the most thermodynamically stable metal oxide.
After analyzing a set of >40 native oxides of pure elements, the thickness of native oxides was found to range from 1 to 5 nm with 2-4 nm of adventitious carbon on top. A few elements formed thicker oxide films and some formed oxide that include carbonate films that were more than 20 nm thick, but most elements formed native oxides that are <5 nm thick. An interesting observation from using XPS to analyze these native oxides is that >80% of them behaved as though the metal oxide layer was conductive. Based on our experience much more work is needed to understand the surface physics of native oxides.
Undesired peak-shape features caused by uncontrolled differential charging
Distortion of the uniformity of the electric field at the surface, known as Differential Charging, can occur above or within the surface of the analysis area of an insulator during XPS analysis. Differential charging can be horizontal (across the surface) or vertical (into the Z aspect of the sample).
Differential charging is due to a mixture of effects described in Table 3. Differential charging when using a monochromatic X-ray source and an electron flood gun is mainly due to an irregular distribution of negative voltages over the area being irradiated that can reduce the voltage, the flux, or deflect flood gun electrons. See Figures 3 and 5 for artistic depictions.
As a result, differential charging distorts the true shape of the peak(s) in a high energy resolution, chemical state spectrum which makes peak-fitting difficult. Differential charging often produces anomalous “sloping charge tails” or “shoulders” at the low BE side of each high resolution spectrum obtained from the same sample. When sloping charge tails appear on all high energy spectra from the same sample, those sloping tails are due to differential charging. An example of this effect is shown in Figure 15a-c for 3 chemical state spectra obtained from an irregular, crumbling pellet of Y2O3 powder. After collecting these spectra, a fresh sample was prepared, and the flood gun needs tested.
TABLE 7. Observable features caused by uncontrolled differential charging.
- Sloping charge tails spread over a 2-6 eV range on lower BE side of each high resolution spectrum in that data-set,
- Shoulders (10-30% intensity) on lower BE side of all high resolution spectra that were collected at the same time,
- Peak broadening – FWHM >2-4x indicates differential charging. (in contrast to Normal FWHM range: 1.0 – 1.5 eV),
- Weak broadening of the FWHM of true peak-shape with or without charge-induced shifting. Difficult to see,
- C (1s) BE of adventitious hydrocarbons shifts differentially relative to other BEs in same sample.
Differential charging can produce peak broadening with a FWHM 3-4X larger than the true FWHM. Differential charging can and does produce unexpected shifts in binding energies toward higher BEs, which degrades the reliability of the binding energies, BEs, used to make chemical state assignments (see Figure 15a-c). A list of the observable features of uncontrolled differential charging is presented in Table 7
XPS spectra shown in Figure 15 are from a non-conductive insulator that have features due to distortions of the true peak-shape. These distortions can be described as (1) a non-symmetrical peak-shape with a non-Gaussian shaped curve on low BE side, (2) a very broad FWHM, and (3) a sloping charge tail on the low BE side of the peak. These three (3) features are typical of a surface having differential charging.
FIG. 15. Examples of uncontrolled differential charging from the pellet of Y2O3. (a) C (1s) spectrum, (b) O (1s) spectrum, and (c) Y (3d) chemical state spectra from Y2O3 powder pressed onto double side tape with flood gun set to 4 eV. BE positions look reasonable when comparing them to handbooks and numerical BE databases, but there are obvious sloping charge tails on the low BE sides of all 3 high resolution spectra which is a strong indicator of uncontrolled differential charging. The O (1s) and Y (3d) spectra show a peak and a shoulder.
If we compare these experimental spectra to reference spectra from Y2O3 in a database of actual spectra, then we know that both the O (1s) and Y (3d) spectra have two sharp peaks with a valley between them.
FIG. 16. Uncontrolled differential charging from a crumbling pellet of pure Y2O3 powder showing a very broad, nearly symmetrical peak for the Y (3d) signal. Analyst might try to fit this with 3-4 peaks.
When data analysts try to make a peak-fit of the spectrum shown in Figure 16 they might try one broad peak which would have a FWHM ~6 eV. Is this too small or too large? What is normal? Other analysts might try to use 2 peaks; using one peak for the shoulder and one for the main peak. They would probably use different FWHM values for each peak. Even so, the resulting FWHM would be >3 eV for each of those 2 peaks. This is 2X larger than the average, normal FWHM listed in Table 13, which is provided to help peak-fitting.
When data analysts do peak-fitting, it is common practice to use FWHM that are 10-15% smaller than the expected FWHM of most peaks, and to add a new chemical state peak at 1-1.5 eV steps. If the shape of the peak looks like the peak in Figure 16, then the analyst might use 5-7 synthetic peaks to fit this peak by using a FWHM in the “normal” range (1.0 to 1.5 eV). This is an example of the problems that can occur due to differential charging. This is one of the reasons to practice aligning the XY positions of the flood gun using a single peak such as the C (1s) from Poly-propylene, and peak-fitting the resulting C (1s) spectra.
The normal range for FWHM for a normal O (1s) peak is between 1.0 and 1.5 eV as listed in Table 13. However, the O (1s) peaks due to hydroxides or carbonates typically have FWHM that range between 1.4 to 1.8 eV. If you have any peak with a FWHM >2.0, then you should add a second peak and reduce the FWHM for both peaks to be <1.6 eV. (A rule of thumb.) The FWHM of metal peaks are more challenging due to the possible presence of multiplet splittings and shake-up structure.
As an end to this section, we suggest the following: Each operator, analyst, scientist, manager, and professor needs to know what is “normal” for the FWHM of most insulators because >70% of all samples analyzed by XPS are insulators. Everyone needs to keep in mind that a peak that is due to different chemical state often occurs just 1-1.5 eV higher than the first peak in the series. Everyone should look for a website that provides free access to high quality monochromatic XPS spectra having good charge control with and without peakfits.
Advantage of using charge-control mesh-screen with a flood gun
The charge-control mesh-screen technology was invented and patented by Chuck Bryson in 1986.4 This mesh-screen provides excellent charge control of insulators by producing a flat, uniform electrical field just above the surface of the sample with the electron flood gun is turned on. Optimum height from the surface ranges from 0.5 to 1.0 mm. An artistic depiction of the flat uniform electrical field is shown in Figure 17.
FIG. 17. Artistic depiction of controlled charging that has a flat uniform electrical field with flood gun ON. The flat electrical field is due to the presence of a grounded, 90% transparent metal mesh-screen (colored as dark gray sheet) sitting 0.5 to 1.0 mm above the surface of the sample. Sample is Poly-propylene. The C (1s) peak is symmetrical. No sloping charge tail. No shoulder. FWHM <1.2 eV.
FIG. 18. Handmade charge-control mesh-screen. 25×25 mm Aluminum square (0.8 mm thick) with 90% transparent Nickel metal mesh taped down by hand using Aluminum tape to one side of square. Holes allow screws to be used if clips are not available. The mesh-screen itself is not perfectly flat. To produce a flat electric field, the Aluminum needs to be grounded. Depending on sample thickness, other mesh-screens or spacers were built using 0.5, 1.0, 1.5, 2.0 mm thicknesses.
The handmade charge-control mesh-screen (shown in Figure 18) is easy to machine and assemble. In place of the metal square, we have used large diameter thin brass or SS washers. When you use this mesh-screen you need different heights (spacers) because samples have different thicknesses. The holes in the metal square allow the user to use screws or clips. Be certain to ground the metal square. Warning: Because the mesh-screen is thin and flimsy, it is easy to tear, which is why you may decide to make several as spares.
By using the simple handmade version of the mesh-screen (shown in Fig 18), the author measured the O (1s) FWHM from 40+ metal oxides that ranged from 1.0 to 1.4 eV; and also thousands of other materials. The corresponding metal oxide FWHM for the metal signal gave the same range. See Table 13.
FIG. 19. The difference in energy resolution between these two C (1s) spectra from PET is due to the use of the charge-control mesh-screen. The mesh-screen was ~0.8 mm above the surface of the PET film. The instrument was a PHI 5802 XPS equipped with a monochromatic Al X-ray source and translatable flood gun.
The overlay of C (1s) spectra from PET (Mylar) in Figure 19 shows that the charge-control mesh-screen removes low levels of horizontal differential charging that is difficult to know that it might be present. These two C (1s) spectra were made using a PHI 5802 XPS system that is equipped with a monochromatic X-ray source. Without the mesh, the C (1s) spectrum looked useful. With the mesh, using the same XY position of the flood gun, all three C (1s) peaks have smaller FWHM.
The charge-control mesh-screen has been used with older monochromatic XPS instruments, newer instruments that use a magnetic lens, and modern XPS instruments including the Thermo K-alpha. When using the mesh-screen on a Thermo K-alpha the flow of ionized Argon needs to be stopped. With the magnetic lens system we decreased the voltage of collection lens #1. When the magnetic lens was turned on the Nickel mesh raised up slightly.
TABLE 8. Differential charging effects – Problem Types #1-6.
- Obvious sloping charge tails spread over 2-6 eV range on lower BE side of each high resolution spectrum,
- Shoulders (10-30% intensity) on lower BE side of all high resolution spectra that might be a different chemical state,
- Severe peak broadening (distortion) – FWHM >3-4x larger than typical (normal range of FWHM for main peaks: 1.0 – 1.5 eV),
- Weak broadening of true peak-shape with charge induced shifting,
- Vertical differential charging of native oxides – grounded versus insulated,
- Peak broadening, sloping tails, and BE shifts due to overly large sample size where the ground is >20 mm away
EXAMPLES OF SIX (6) TYPES OF DIFFERENTIAL CHARGING PROBLEMS
This section provides real-world examples (chemical state spectra) of differential charging for those scientists and engineers who are learning how to control the differential charging of insulators and need examples. The following example spectra were obtained from real-world samples because differential charging. To better understand what differential charging looks like, we provide the same spectrum with un-controlled differential charging and with controlled charging in Figure 20a-b. From the exact same sample, Figure 20 shows a C (1s) spectrum having uncontrolled charging as Figure 20a and a C (1s) spectrum having controlled charging as Figure 20b.
Problem Type #1
Sloping charge tail that spreads over a 2-6 eV range on low BE side (Figure 20a)
FIG. 20. The adhesive surface of a paper label was analyzed (a) without a charge-control mesh-screen and then (b) with a charge-control mesh-screen keeping the flood gun at the same conditions. After a charge-control mesh-screen was installed ~0.8 mm above the surface, the shapes of the C (1s) and O (1s) changed from being distorted to normal peak-shapes having no sloping charge tails on the low BE side.
Problem Type #2
Shoulders (10-30% intensity of adjacent peak) on lower BE side (Figures 21a and 21b)
When differential charging occurs, it is often visible on the low BE side of all the chemical state spectra for a given sample. It can appear as a sloping tail as shown for Problem Type #1. Or, in this case, the differential charging can appear as a new peaks or shoulders on the low BE side which suggests the presence of a new chemical state.
In this example, 21 the Zn (2p3) and O (1s) spectra from a single crystal of ZnO (0001) were compared to the same spectra from nanoparticles of ZnO (mounted on Indium foil) that were produced using different levels of pH. The peak-shapes in the spectra shown as Figures 30a and 30b are complicated and were not expected based on reference spectra from the pure ZnO crystal also shown.
ZnO has a bandgap of 3.3 eV so it is expected to be non-conductive, which means that the instrument operator had to turn on the flood gun to eliminate surface charging. In this case, the Zn (2p3) spectra were expected to appear as symmetrical peaks with only one component because the Zn (2p3) signal from ZnO has no chemical shift from pure Zn metal which means that ZnO and Zn metal suffer a direct overlap. Depending on the final chemistry of the reactions, the O (1s) spectra was expected to have either a single peak or more likely two peaks, one due to the oxide and one due to a hydroxide so seeing 2 peaks was not unexpected.
From the final set of XPS spectra (Figures 21c and 21d) it is obvious that the electrical field within these nanoparticles is complicated when the samples are grounded, and the flood gun is turned on. With the flood gun turned on, the different local electrical environments of the nanoparticles produced a new, unexpected peak for the Zn (2p3) signal, that appears as a shoulder on the low BE side of the true peak. It is a distinct intense shoulder, not a sloping tail.
Due to the unexpected presence of the extra Zn (2p3) signal in Figure 21b, the operator decided to cross-check the chemistry by floating the samples on a non-conductor instead of standard mounting of pressing nanoparticles onto Indium foil. By floating the sample, the operator learned that the unexpected shoulders were a differential charging artefact. (Actually, a vertical differential charging artefact.)
The Zn (2p3) and O (1s) spectra (21c and 21d) on the right side of Figure 21 show the peaks that were expected based on prior experience. The take-home lesson from this set of spectra, is that the analyst must be ready to cross-check the results and analyze samples that are truly insulated in case the physical or chemical structure of the material produces some unexpected peaks.
FIG. 21. Nanoparticles of ZnO were exposed to different levels of pH to change their conductivity. The different nanoparticles were pressed onto Indium metal foil and then analyzed by XPS. Because the nanoparticles had developed levels of conductivity they caused vertical difference charging shoulders (figures 21a and 21b) highlighted by arrows pointing at the shoulder that is a charging artefact. After mounting the nanoparticles onto non-conductive adhesive tape, the differential charging was replaced by true insulation producing one electrical potential for each sample.21
Problem Type #3
Severe peak broadening (distortion) – FWHM >3 eV (Figures 22a-22c)
The 3 spectra shown here in Figure 22 from C (1s), O (1s) and Y (3d) signals were collected from the same sample. These spectra (figures 22a-c) are examples of severe peak broadening caused by differential charging. The beam voltage of the flood gun for this sample was initially set to 2 eV when the first data set was collected. After removing the sample from the instrument the samples was found to have crumbled exposing a much rougher surface for analysis.
After the initial data was collected, the measured BE positions were reasonable when compared to the NIST SRD20 database of numerical BEs. Based on an XPS reference spectrum of pure Yttrium metal, the Y (3d) peak of the Y2O3 material should have two peaks separated by a valley, but the raw spectrum for the Y (3d) signal of Y2O3 appeared as a single wide peak. Why? The C (1s) and O (1s) spectra also show broad peak and a broad peak with a shoulder. The C (1s) actually looks like 2 peaks of equal intensity, which is different from the usual C (1s) spectrum from adventitious carbon. Note: It is very useful to remember the typical shape of the C (1s) of adventitious carbon.
When we compare these experimental spectra with actual reference spectra (Figures 22d-f) from Y2O3 from a database of actual spectra, then we found that both the O (1s) and Y (3d) peaks should appear as two peaks with an obvious valley between them.
FIG. 22. Examples of uncontrolled differential charging from the Y2O3. The C (1s), O (1s), and Y (3d) chemical state spectra (Figures 22a-c) from 99.99% pure Y2O3 powder pressed onto double sided tape. The BE positions look reasonable when comparing them to the PHI handbook and NIST numerical BE database, but there is an obvious sloping tail on the low BE sides of all 3 spectra which is an indicator of uncontrolled differential charging.
By comparing these spectra to reference grade spectra2 (Fig 31d-e) from as-received 99.99% pure Y2O3 pressed into a pellet and covered with the charge-control mesh-screen we understand how strong the differential charging can be. The right hand side of each peak has a nice Gaussian shape which is typical of controlled charging. (The reference spectra2 are from the XPS International SpecMaster database of >70,000 spectra.)
Problem Type #4
Weak broadening of true peak-shape (Figures 23a, b)
In this example, PET was analyzed by using a PHI 5802 XPS with a translatable flood gun. After optimizing the flood gun voltage to 2 eV, the C (1s) spectrum from PET (in Figure 23a) gave a BE at ~283.5 eV which matches the 2 eV offset from the flood gun voltage. Based on the reference peak-shape for C (1s) spectrum of PET (figure 10), the C (1s) peak-shape looked Gaussian shaped, but the valley between the C-C peak and the C-O peak was missing when looking at the reference C (1s) spectrum .
After placing a handmade charge-control mesh-screen ~0.8 mm above the surface of the sample, a new C (1s) spectrum revealed the missing valley. The BEs shifted 2 eV lower which indicates better charge control due to the mesh-screen. Visually, we found peaks with smaller FWHM. The overlay in Figure 23b reveals the improvement in peak-shapes. If we had tried to peak-fit the original C (1s) spectrum, then we might have added an extra synthetic peak to explain the total peak envelop or we might have used FWHM that are broader than normal. The improved charge control peak-shape provided by the charge-control mesh-screen result helped us to avoid mis-interpreting this spectrum.
FIG.23. FWHM of C (1s) from PET appears at normal BE values but FWHM is slightly broader and peak valley is missing. This effect was revealed by placing a charge-control mesh-screen above the same sample and collecting the same spectra. Figures 23a shows some of the effect of using the mesh-screen. After shifting BEs to align, Figure 23b clearly reveals the improved charge control provided by the charge-control mesh-screen.
Problem Type #5
Vertical differential charging of native oxides – grounded versus insulated (Figure 24)2
Most naturally formed, native oxides are thin, having oxide layers ranging from 10 to 50 angstroms thickness. These thin layers are usually colorless to the human eye. Some metals form relatively thick layers of native oxides or carbonates that can spall off as a form of corrosion.
In general, thin films of many metal oxides behave conductively when they are grounded to the sample stage. In powder form, metal oxides that behave conductively are brown or black in color, but some are green or red.
In powder form, non-conductive metal oxides are white, or pale yellow. Unfortunately, we cannot see any color from native oxides because they are so thin.
Thin layers of various non-conductive metal oxides on metals can suffer from vertical differential charging which becomes obvious when a flood gun beam of electrons is applied at various voltages. See Figure 24a and 24b below.
Most naturally formed non-conductive native oxides are thin enough that Fermi level electrons (at 295 K) from the underlying metal can and do migrate (tunnel) along grain boundaries to neutralize the valence-holes (not core-holes) formed by XPS. For many native oxides there is no obvious need to use a flood gun beam of electrons to neutralize non-conductive native oxides as long as the sample is grounded.
However, various non-conductive native oxides can and do produce adventitious hydrocarbon C (1s) BEs that range from 285.5 to 286.5 eV, with the metal oxide BEs being 0.5 to 1.2 eV higher than the BEs reported in various handbooks and the NIST SRD-20 database of numerical BEs. Why? Based on our research the answer involves the presence of strong surface dipole moments produced by highly polar metal oxides, which is the topic of our next paper.
When instrument operators find larger than expected BEs for the C (1s) and the main metal oxide peak for a thin native oxide of a metal, those operators then imagine that the larger BEs are due to a charging effect and so turn on the flood gun to compensate the charging. When they try a small flood gun beam voltage, there might be almost no change in BEs. However, when they use larger voltages (5, 10, or 15 eV), they find significant BE shifts for the metal oxide peak, the oxygen peak, and the C (1s) peak (Figure 24a and 24b).
In this figure we see peak broadening, and sloping charge tails for the C (1s), O (1s), and metal oxide peak, but the pure metal peak does not shift. Why? The pure metal peak does not shift because the metal is properly grounded due to contact with the sample stage that is properly grounded. The pure metal peak BE is true. Only the upper layers of the surface (the carbon and the metal oxide layers) are suffering vertical differential charging.
What produces this type of vertical differential charging and how do we deal with it? The vertical differential charging can be attributed to 2 factors. One factor is the vertical charge gradient that develops as soon as we turn on the flood gun. The charge gradient begins at the metal-metal oxide interface which is the start point for a natural dipole moment. The dipole moment due to the interface produces a charge gradient that extends upward into the adventitious carbon layer. The positive end of the dipole moment retards (decreases) the KE of the emitted C (1s) photoelectron, in turn causing the C (1s) BE to appear at higher than expected BE. The second factor that contributes to vertical differential charging is the capacitive nature of the non-conducting native oxide and the interface layer between the metal and the native oxide. These two factors are due to natural surface chemistry, surface physics, and electrical properties of the two interfaces and the resulting two dipole moments.
FIG. 24. Figure 24a is an overlay of Al (2p) spectra exposed to 0, 5, and 15 eV flood gun voltages. Figure 24b is an overlay of C (1s) spectra exposed to 0 eV (red color line), 5 eV (blue color line), and 15 eV (black color line) flood gun voltages. Example of vertical differential charging for a naturally formed, native oxide of Aluminum when the flood gun beam voltage is increased. This type of differential charging occurs when a non-conductive native oxide is irradiated with a low voltage beam of electrons varied from 0 to 5 to 15 eV. It is important to notice that the carbon peak shifted more than the aluminum oxide peak.
We can eliminate this vertical charging phenomenon by insulating (floating) the same sample to produce the spectra shown in Figure 25.
Figures 25a,b show that floating the native oxide of Aluminum produces an almost linear response to the flood gun voltage. After charge referencing the spectra in Figures 25 and 26 based on the Al (2p) pure metal BE, we see an almost perfect overlap of the peak-shapes and BE shifts. Figures 25 and 26 show us that differential shifting, and broadening seen in Figures 24a-b are eliminated by floating the sample.
By using the Al (2p) BE of pure aluminum (72.9 eV) to charge correct both Al (2p) and C (1s) spectra, we know that the correct Al (2p) BE for Al2O3 is 75.8 eV which matches the synchrotron work done in 1978.28 This Al (2p) BE is a particularly important BE for anyone working with Al2O3.22-23
FIG. 25. The native oxide of an aluminum sample was insulated, and the flood gun was turned on using 0, 5 and 15 V settings. These Al (2p) and C (1s) peak-shapes look reasonable, and here is no obvious distortion due to differential charging.
FIG. 26. Three Al (2p) spectra from the native oxide of aluminum (shown in Figure 25a) were deliberately insulated using double-sided tape and corrected by using the Al (2p) BE of pure Aluminum metal. The Al2O3 peak still shows almost no differential charging. Please note that the Al (2p) BE of the Al2O3 native oxide is 75.8 eV, not 74.4 eV as published in the NIST SRD20 database24 and PHI handbook.25
FIG. 27. Three C (1s) spectra (Figure 25b) from the native oxide of aluminum that were deliberately insulated using double-sided tape and corrected by using the Al (2p) BE of pure Aluminum metal. Because Carbon is on the very top it still suffers differential charging. Please note that the C (1s) BE is ~286.5 eV, not 285.0 eV which is due to a surface dipole moment effect that preferentially retards the KE of the C (1s) photoelectrons by ~1.5 eV.
True BE of Al2O3
Based on oxidation studies performed at a synchrotron facility in 1978, the true Al (2p) chemical shift for Al2O3 is 2.7, and the true Al (2p) BE for Al2O3 is ~75.9 eV (Figure 28). The BE difference between pure Aluminum (Al 2p = 72.9 eV) and Al2O3 is 2.7 eV which gives an Al (2p) BE at 75.9 eV for Al2O3.
This synchrotron based Al (2p) BE is 1.7 eV larger than the 74.2 eV BE published in the PHI Handbook of XPS in 199225 and the 74.3 eV BE in Surface Science Spectra in 1998.26
This difference is due to a surface dipole moment10,27, effect that preferentially retards the KE of the C (1s) photoelectrons in adventitious carbon on top of the aluminum native oxide by ~1.7 eV. This effect is common for elements in column 2 and 3 of the periodic table. The size of this effect depends on the dipole moment which is related to the bandgap of the oxides.
FIG. 28. Oxidation of a (111) surface of Aluminum monitored using h-bar omega = 130 eV synchrotron radiation published in 1978.28 The conversion of the chemisorbed state into the subsequent Al2O3 is shown. The BE difference between pure Aluminum (Al 2p = 72.9 eV) and Al2O3 is 2.7 eV which gives an Al (2p) BE at 75.9 eV for Al2O3. This Al (2p) is 1.5 eV larger than the 74.6 eV published in the PHI Handbook of XPS in 199225 and 74.5 eV in Surface Science Spectra in 1998.26 The adventitious hydrocarbon C (1s) BE is adjusted to give a C (1s) BE = 285.0 eV. This difference is due to a surface dipole moment effect that preferentially retards the KE of the C (1s) photoelectrons by ~1.7 eV.
Problem Type #6
Effect of sample size and distance from ground
When a large insulating sample is irradiated with an X-ray beam, the exposed area develops a positive surface charge. When the electron flood gun is then turned on, that positive charged area should develop a net negative charge if the voltage and emission current of the flood gun are high enough. Before we spend a lot of time collecting data from a large area with a ground point that is distant, we need to confirm that the area will produce high quality chemical state spectra.
An example of the effects of horizontal differential charging was found by analyzing a large piece of Aluminum Nitride ceramic (Figure 29). Chemical state spectra (Figure 30a-c) were obtained from the center of the Aluminum Nitride ceramic and from an area that was ~ 1 mm from the grounded copper clip holding the sample.
To minimize the effects of differential charging for the data obtained from the center area, the sample was moved in X and Y to allow electrons to buildup on neighboring areas.
After making overlays of the chemical state spectra from the two areas, the spectra from the center were found to be shifted by ~3 eV to higher BE, and to suffer peak broadening (Figure 30a-c).
The spectra from the area that was ~1 mm from the Copper clip produced symmetrical signals have Gaussian tails on the low BE side that were more intense with slightly smaller FWHM. This difference is attributed to the nearness of the Copper clip. We suggest that electrons from the grounded clip were able to migrate across the 1 mm distance of the surface to provide better charge control. The mechanism for this involves the monolayer level of water and various gases that are trapped on many surfaces.
FIG. 29. Aluminum Nitride sample. The initial 400 m analysis spot was changed from being ~1 mm away from copper ground clip to center of 50×50 mm sample far from ground, where ground was 20+ mm away.
FIG. 30. Overlays of (a) Al (2p), (b) O (1s), and (c) N (1s) chemical state spectra from two (2) different locations on a large (50×50 mm) insulating sample. One position was ~1 mm from a grounded clip. The other analysis position was in the center ~20 mm from the grounded clip.
EFFECTS THAT LOOK LIKE CHARGING
XPS work on insulators often involves the use of different pass energies, argon ion etching, and a flood gun, on materials having different degrees of surface roughness, electrical irregularities etc. In some cases, the industrial customer does not know how the sample was last treated, or the customer may have accidently contaminated the surface.
Industrial samples and samples from suppliers are sometimes extremely dirty having as much as 6-9 nm of organic contamination which usually require the unexpected use of the flood gun. For these reasons, the operator needs to be able to distinguish between differential charging, sample treatments, and analysis conditions that can produce spectra that appear to have differential charging. Figure 31 is an artistic depiction of the chemistry of a typical native oxide.
In this section we present examples of normal chemical state spectra that might appear to suffer the effects of differential charging. Because XPS typically measured ~100 angstroms (10 nm) depth and reports quantitation as a total of 100 atom%, we developed a Rule-of-Thumb where a 1 atom % of any element is approximately equal to 1 angstrom of that element. There are equations that can produce a more accurate value, but for practical work, this Rule-of-Thumb is useful.
- Surface Contamination on Native Oxides
FIG. 31. Artistic depiction of a typical native oxide having adventitious carbon, and monolayers of trapped gases, water, hydroxides or carbonates, and the native oxide(s) on a pure metal bulk.3
The drawing in Figure 31 indicates that the adventitious carbon layer is typically 3-4 nm thick on most native oxides. This is based on many years of experience. Using our rule-of-thumb the atom% amount of carbon is 30-40 atom%. This depiction includes a monolayer or two of water, which is enough to enable solutions and conductivity.
- Misleading effects due to different pass energy settings
Low levels of differential charging can produce peak broadening that makes the operator or analyst consider decreasing the pass energy to get better energy resolution. However, when your peak-fit has large FWHM (>1.5 eV) then you should consider that there can be 2 difference causes. One cause can be pass energy while the other can be differential charging. It is sometimes faster to check on flood gun settings instead of decreasing your pass energy setting and running a complete set of chemical state spectra. In either case, you should run just one of your chemical state spectra if you decide to decrease the pass energy. Be careful not to waste your time by jumping to the smallest possible pass energy. The smallest pass energy produces smaller FWHM and also a very large loss in signal intensity.
Figures 32a-c are example of differences in pass energy settings, that produce only modest improvement in FWHM for a nonconductive material. Due to the natural limit of FWHM of non-conductive chemical compounds, the smallest FWHM that can be measured ranges between 0.9 and 1.4 eV. Decreasing to a smaller pass energy will not produce a meaningful smaller FWHM, and to achieve the same S/N ratio, the number of scans and time used must increase by 50-100x.
FIG. 32. (a) Pure SiO2 (silica) – Freshly exposed edge of SiO2 disc. PE=50 and 200 eV. Not ion etched. (b) Si wafer with native oxide. Measured at PE=50 and 5 eV. (c) Poly-propylene – PE=200, 100 and 50 eV. FWHM did not change after PE was decreased from 100 to 50 eV because this is natural FWHM of the C (1s) in Poly-propylene.
Large pass energy settings provide the most intense XPS signals which allows the operator to quickly measure the presence or absence of any element, but large pass energy settings provide poor energy resolution that is not useful when we need chemical state information. Small pass energy settings are used when we want to measure the presence or absence of different chemical states. Small pass energies produce better energy resolution, but weaker signal causing us to spend more time to collect more spectra. It is a trade-off.
There is a natural limit to the FWHM of peaks obtained from pure elements, conductive chemical compounds, and non-conductive chemical compounds. See Figures 32a-c. These examples show that conductive materials can produce small FWHM, while non-conductive materials do not. This difference is important to remember because the time required to produce spectra having useful signal/noise (S/N) ratios grows exponentially as we decrease the pass energy to measure higher and higher energy resolution. There is a “middle-ground” for pass energies when measuring non-conductive chemical compounds. If your instrument provides pass energies ranging from 5 to 200 eV, then the most effective use of your time, to analyze chemical states from non-conductive compounds, is to use a pass energy setting 50-100 eV.
FIG. 33. Pure SiO2 (silica) – Freshly exposed edge of SiO2 disc. PE=50 eV. Before and After light ion etching. Only minor degradation.
- Effects of argon ion etching
Argon ion etching removes contamination from pure elements and chemical compounds. Ion etching is useful to remove adventitious carbon contamination and native oxides from pure metals to reveal a pure metal signal that has a narrow FWHM, but argon ion etching can cause peak broadening for XPS peaks from chemical compounds. This section shows the effects of using argon ion etching which can be confused with differential charging.
The analyst needs to be careful when considering using Argon ion etching to clean any material. Argon ion etching can and does degrade the chemistry in various materials which destroys the chemical state information that we are trying to measure.
Ion etching of various oxides, polymers and other chemical compounds can cause BE shifts, degradation of chemical states, preferential loss of one element, and the end effect is usually peak broadening. The following spectra show chemical state spectra before and after a few seconds (5-30) of Argon ion etching using only 500-1000 V Argon ions.
The metal signal of most metal oxides will have larger FWHM after being ion etched because the chemistry or physical structure has been degraded. There are only a few metal oxides that are only modestly affected by argon ion etching, such as Al2O3 or SiO2. Figure 33 shows the effect of lightly ion etching the freshly exposed bulk of pure SiO2 (silica). This broadening could be confused with differential charging. Pure SiO2, when ion etched, degrades to Si2O3 which has a lower BE.
FIG. 34. Si (2p) measured with ultra-high energy resolution. After ion etching the freshly exposed bulk of a n-type Silicon wafer, the BE of n-type silicon, decreased by ~0.3 eV as shown in this overlay. The BE shift is not due to differential charging, but is due to doping of the semiconductor with Argon ions. The peak broadening is due in part to the defects produced by the ion etching and is attributed to crystal structure degradation at the surface.
Crystalline Silicon <100> in wafer form (n-Si), suffers a significant change in peak-shape because Argon ion etching damages the crystalline structure of the top few nanometers, introduces defects, and changes conductivity as illustrated in Figure 34.
As-received, many chemical compounds, such as Cr2O3 powder and crystals, produce chemical state spectra that show fine structure due to Multiplet Splitting (M.S.). See Figure 35. After we lightly argon ion etch Cr2O3 powder, the fine structure is gone due to chemical degradation (e.g. lower oxidation states), and the fine structure due to Multiplet Splitting is gone (Figure 35a). These types of changes in peak-shape are not due to differential charging effects. These are chemical degradation due to ion induced reduction and the loss of one or more elements as ions or gases.
It is very often worthwhile to analyze an insulator after optimizing charge control conditions and before any argon ion etching, even light ion etching. New style, cluster argon ion gun systems, when used correctly, do not degrade chemistry of the many materials as they remove overlayers, but they are extremely slow to remove the damaged overlayers that were produced by very light mono-atomic argon ion etching (10s at 1kV).
FIG. 35. Cr (2p) spectra from Cr2O3. The smoothed Cr (2p) spectrum in Figure 35a is from the well known XPS PHI handbook.25 The other spectrum in Figure 35b, which shows a little noise, is from a single crystal of Cr2O3 measured using a 50 eV pass energy. The multiplet splitting shoulders and peaks are obvious when charging is well controlled and the sample has not been Argon ion etched. M.S. is abbreviation for Multiplet Splitting signal.
Chemical state spectra of the freshly exposed bulk of an amorphous silica (SiO2) sample were analyzed soon after fracturing in air, and then after argon ion etching. The C (1s), Si (2p) and O (1s) spectra (Figure 36) shown here reveal that the ion etching removes surface contamination which increases the intensity of the Si (2p) and O (1s) signals but has no obvious effect on the FWHM of these signals.
Due to the light argon ion etch one charge related change did occur. The Si (2p) and O (1s) BEs increased by ~ 1 eV, but the C (1s) BE of the remaining carbon did not shift. Why? This change can be attributed to a change in the static charge of the SiO2 bulk. This is a Bulk Charging effect, not a Surface Charging effect.
FIG. 36. C (1s), O (1s) and Si (2p) spectra from freshly cleaved natural crystal of SiO2 as cleaved and after light argon ion etching. After light ion etching the amount of carbon decreases, but the hydrocarbon C (1s) BE does not change (Figure 36a). After light ion etching the O (1s) and Si (2p) the intensity of these signals increase as expected, but the BEs of both signals increases by ~1 eV (Figures 36b-c). This situation reveals one reason why it is difficult to use adventitious hydrocarbon (AdC) to correct the charge shifted BEs.
- Degradation due to electrons from flood gun
In general, there are very few materials that are degraded by being irradiated with low voltage electrons from a flood gun. Two materials, that are exceptions to that statement, are Poly-acrylic acid (PAA) polymer film and sodium thiosulfite, Na2S2O3.
Damage (degradation) due to total X-ray flux is a concern that can distort peak-shapes, produce new chemical states, or cause preferential loss of one element. Even so, the rate of damage by X-rays is small for most materials. The same is true for damage due to flood gun electrons. The rate of damage by X-rays and flood gun electrons is low, but it does exist for various materials, such as: polymers with acid groups, chlorine, fluorine, and esters. Today’s X-ray beams have enough flux to degrade various polymers during the time spent to analyze them. For this reason, the operator should run a single scan survey at the start and the end of the analysis run. Other materials that can be degraded by high X-ray flux include high oxidation state metals (e.g., CrO3, KMnO4, K2Cr2O7, Au2O3) and carbonates, which can lose CO2. Again, these types of degradation are few, flux dependent, and usually slow, unless you are using a synchrotron beam line.
For both materials, the analysis area was changed after the first spectrum was recorded to avoid any damage due to the X-rays. Before turning on the X-rays, the flood gun was turned on for the specified time lengths shown on the figures. The X-rays were then turned on and the second spectrum was recorded. This method minimized any significant damage due to the X-ray beam itself.
The two overlaid C (1s) spectra for PAA (Figure 37) were from two fresh regions of the same film. The signal at ~289 is due to the acid group (RCOOH). The time difference between the two spectra is 15 minutes while using a 10 eV beam of electrons. The loss in intensity is due to the loss of CO2 as a gas.
The two overlaid S (2p) spectra for sodium thiosulfite (Figure 38) were obtained from the same region. The signal at 166 eV is attributed to the loss of oxygen from the SO3 group. The time difference between the two spectra is 60 minutes using a 2 eV beam of electrons.
FIG. 37. Overlay of the C (1s) chemical state spectra of Poly-acrylic acid. The decrease in the C (1s) acid chemical
FIG. 38. Overlay of S (2p) chemical state spectra of sodium thiosulfite (Na2S2O3). The production of a new SOx species at 166 eV is due to electron induced reduction of the SO3- chemical group at 168 eV.2
There are two catalyst materials (MoO3/Alumina and Pt/Alumina) that are known to be sensitive to electron flood gun electrons (personal observations, not published).
- Effect of surface roughness
A KBr pellet hand-press was used to produce a 3.0 mm diameter disc with ~0.5 mm thickness from pure Y2O3 powder. Chemical state spectra from the Y (3d) signal were collected from: (a) a powder firmly pressed onto double-sided tape (green line) and (b) powder hand-pressed into a 3 mm diameter disc (black line). To make sure the surface of the pellet disc was clean, a fresh piece of Aluminum foil was placed between the powder and the anvil to stop transfer of any contamination from the steel anvil.
Due to the added smoothness of the pressed pellet disc, the resulting Y (3d) spectrum has a slightly better peak-shape indicated by the arrow on Figure 39. This is attributed to a more uniform surface charge.
We conclude that when we need to analyze a powdered sample, we have 3 options for sample preparation, (1) drop powder into a cup to avoid any chance to contaminant the surface of the powder, (2) use a clean spatula to firmly press the loose powder onto double sided tape or indium metal foil, or (3) compress the powder into a thin pellet by using a handpress or a high pressure pellet press. Option 1 is acceptable if you do not need spectra with the best FWHM values. Option 2 gives smaller FWHM and makes charge control easier. Option 3 is best, but it requires more time and requires the use of clean aluminum foil or glycine paper as an interface to stop the transfer of rust from the anvil to the pellet.
FIG. 39. Overlay of chemical state spectra of Y (3d) from pellet made from powdered Y2O3. Pressed pellet and firmly pressed powder. The difference in the slopes at 155 eV is due to less charging that results when a surface is smoother. The difference was produced by using a simple hand-press. A layer of glycine weighing paper or Aluminum foil was used to stop the transfer of contaminants from the steel anvil.
MODERN FLOOD GUN SETTINGS
Modern electron flood guns have digital controls built into the instrument operating software. The spectra shown in this section were obtained by analyzing the center of a large piece (20×20 mm) of poly-propylene (Figure 40) in a Thermo K-alpha XPS. The grounded Copper clip was ~7 mm from the analysis area. The flood gun exists on the right side in the analysis chamber.
Using the flood gun menu in Thermo software (Figure 41) the operator can change beam voltages, extractor voltage, beam current, beam focus and XY positions of the flood gun shown here in this list and on Figure 41. Z sample height is adjusted on the normal user computer screen.
TABLE 9. Flood gun settings and ranges available to optimize quality and reliability of chemical state spectra.
- Beam voltage (0 to 10 V),
- Extractor voltage (0 to 25 V),
- Emission current (1 to 300 micro Amps),
- Beam focus voltage (0 to 25 V),
- Argon gas cell (0 to 25 V),
- Z sample height (1 to 50 micron steps per motion),
- X-position (0 to +/- 5 V),
- Y-position (0 to +/- 5 V).
The choice to use a pure hydrocarbon polymer (Poly-propylene) to test or choose a flood gun setting is due to the simplicity of chemistry and the knowledge that the C (1s) signal must be symmetrical because there is only one chemical state, and because the FWHM of the C (1s) peak is small, roughly 1.0 eV.
By changing the XY positions of the electron beam, the user can normally optimize the charge control conditions of each sample. In general, the user seldom needs to the other settings. However, when the operator must analyze materials that have different degrees of surface roughness or strong dielectric constants, then the operator may need to increase the Beam Voltage, and or the Emission Current. This author prefers to use a 1 volt beam voltage to make sure the surface is slightly negative in voltage. This author also prefers to use higher emission current which provides more than enough electrons to neutralize most samples, although a higher current shortens the lifetime of the filament, and risks electron induced reduction.
Based on the author’s personal experience to operate different XPS instruments and their flood guns, the best results for peak-shapes, FWHM, and charge control are produced when changing the XY positions of the flood gun causes the peak C (1s) BE to move to a smaller and smaller BE value. Depending on the instrument, XY adjustments can move the C (1s) BE for Poly-propylene by as much as 2-3 eV. As the C (1s) BE moves to smaller BE, you should find FWHM decreasing, and the sloping charge tail will disappear.
The following figures (Figures 42-45) show the effects of changing from a default flood gun value to other values, which might be necessary as your flood gun filament gets older or becomes dirty due to outgassing samples. (Depending on use, the flood gun filament can last 2-3 years.)
These figures are for teaching purposes. We strongly recommend that you first record your default settings (screen grab image), and then test the effect of changing each parameter to improve your ability to control charging. Record each image. Please note that when you change any setting the charge conditions on the sample may or may not change if you make only a small change. Larger steps are more revealing and useful.
Based on the author’s experience, the major objective of changing any of these settings is to move the C (1s) or O (1s) peak to as small a BE value as possible, which normally gives the smallest FWHM, eliminates sloping tails or shoulders, gives a 20-40% stronger signal, and produces optimum charge control.
FIG. 40. This photo displays 3 different materials used to study the effects of changing the electron flood gun settings. The materials include: poly-propylene, poly-ethylene, and Teflon tape. The narrow wedge sample (top right) was used to study the effect of sample size. The electron flood gun is located at the far right.
FIG. 41. Pop-up screen showing the Flood Gun menu for the Thermo K-alpha XPS. This menu displays the dual-beam flood gun parameters that can be varied to improve the charge control conditions of the active sample. The monitored values are the actual (readback) values that result when the Target values are changed.
- Beam voltage settings
The beam voltage control (Figure 42) provides a linear response to increases in beam voltage. Increasing to 1 eV causes BE to decrease by 1 eV as shown in Figure 42. In new instruments, the upper limit for beam voltage can be 5 volts. In older instruments the upper beam voltage can be as high as 20 volts.
The flood gun in the Thermo K-alpha XPS instrument can be set to -0.1 eV which provides good charge control for smooth surfaces. Based on 10+ years of use, it is better to use a -1.0 eV setting to produce a truly negative surface.
FIG. 42. Peak shifts when beam voltage is varied from -1, -2, -3, to -4 volts. A linear response. As the negative voltage is increased, the measured BE moves to smaller BE. (Lowest possible voltage is 0.1 eV which is the default setting. Maximum voltage is 5.0 eV).
- Extractor voltage settings
The extractor voltage control for the Thermo K-alpha is normally set to 40 volts. This voltage of the extractor pulls and accelerates electrons out of the heated W (ThO2) filament and downstream. If the extractor voltage is decreased from its optimum value, then the measured C (1s) BE increases, and your spectra will have a sloping charge tail. The optimum value is between 30-40 volts. See Figure 43.
FIG. 43. Extractor voltage. Normal setting is 40 volts. Using 30 volts causes a small shift to lower BE. Using 10 volts causes a large 6 eV shift to higher BE and produces a sloping tail. Lower BE results are better.
- Emission current settings
The emission current for different flood guns ranges from a low of 1 mA to maximum of 350 mA. If your sample has a rough surface, then you might want to increase the current to increase the number of electrons deposited on the surface, but you increase the chance of chemical damage by reduction. High oxidations states and polymers might degrade under higher emission current. This author used 200-300 mA because the samples had many different dielectric constants. When you use a higher emission current, then you decrease the lifetime of the flood gun filament. Normal lifetime is 1-3 years. An example of the change produced is shown in Figure 44.
FIG. 44. Emission current. Normal setting is 150 micro-Amps. Using 20 micro-Amps causes BE to increase and decreases the signal intensity and causes peak broadening. Lower BE results are better.
- Beam focus settings
Beam focus controls the size of the electron beam. The electron beam size is <2 mm diameter. The normal beam voltage is 25 volts. A smaller beam voltage, ~5 volts, produces a smaller BE with a smaller FWHM which indicates better charge control. The smaller beam voltage is probably producing an electron beam area larger than 1 mm and smaller than 5 mm. An example of the change produced is shown in Figure 45.
FIG. 45. Beam focus voltage. Normal setting is 25 volts. Using a 5 volt setting allows the electron beam size to grow which is beneficial in this example. Lower BE results are better.
- Argon gas setting
In the argon gas ionization chamber the pressure is on the order of 10-1 Pa. The pressure downstream is 10-4 to 10-5 Pa which results in a 2×10-7 torr pressure in the main analysis chamber.
The default setting for the gas cell is 25 volts. Based on the patent the voltage can be dropped to 1 volt with the resulting ion beam current being 10 to 100 nA.
The Argon ionization voltage can range from 1 to 25 volts.
FIG. 46. Argon gas ionization setting. Normal setting is 25 volts.
The approximate size of the Argon ion beam used in the Thermo Dual Beam flood gun is <3 mm).
- Z-position of sample.
The Z position of a sample naturally affects the XPS signal intensity for all focused X-ray systems. But the maximum counts and best Z position for charge control and data collection are often not the same as the optical focus of the sample surface as provided by optical cameras on XPS instruments. If you have Snap-shotTM or Un-scanned mode capability, then you can use that to determine the optimum Z position for collecting XPS data. Optimum Z position for electron counts is not necessarily the same as optimum optical focus of the sample surface. Figure 47 shows the variation in electron count-rate for the Thermo K-alpha when the Z position is dropped vertically. As the Z position drops the count-rate drops. Optical focus may or may not change until the Z position is changed by >200 microns. This behavior probably occurs on other systems equipped with optical cameras.
FIG. 47. Z position of sample height. Sample height was decreased in 100 micron steps to 600 microns from max counts position. This produced a 75% decrease in signal intensity which is due to X-ray focus. Focus based on high electron count-rate is often different from focus based on the optical microscope due to mis-alignment of the camera focus.
- X-Y position settings of “focused” flood gun beam
Based on the author’s efforts to produce high quality, high energy resolution, chemical state spectra, the first objective is to change the XY positions of the flood gun so that the C (1s) BE or O (1s) BE are as small as possible, which normally gives the smallest FWHM, eliminates sloping tails or shoulders, and produces best charge control.
For that reason, section VIII is dedicated to “Optimizing XY positions of focused low voltage beam of electrons”. There are 3 sub-sections dedicated to this topic because optimizing XY has been a re-occurring necessity when optimizing any flood gun system on different XPS instruments,
Figure 48 gives a quick overview of what happens when the XY positions are not optimized for the sample currently being prepared for XPS analysis.
Brand new XPS instruments and their electron flood gun systems are normally installed by engineers from the maker. Nowadays, those engineers initially adjust the flood gun settings to a default set of values. The engineers finalize the electron flood gun settings based on the C (1s) FWHM of the ester signal (at ~289 eV) from a piece of freshly cleaned Poly-ethylene terephthalate (PET, Mylar) which has dielectric properties that are unique to PET. Their PET sample is usually a smooth film or sheet that was cleaned with alcohol, acetone, or scraped to expose fresh bulk. A smooth clean surface is much easier to use to find optimum flood gun settings. Making the sample small also helps. Ask them if you can keep that PET sample.
FIG. 48. For this test, the X-position was varied from -5 V, to 0 V, to +5 V. The Y-position was fixed at -2.3 V because it had already been optimized. Notice that the optimum position has the lowest BE, a Gaussian shape, the highest count-rate, smaller FWHM and no sloping tail.
OPTIMIZING XY POSITIONS OF FOCUSED LOW VOLTAGE BEAM OF ELECTRONS
Test method useful to optimize XY positions
Because different insulating samples have different shapes, dielectric properties, roughness, contamination, etc., the operator may need to optimize one or more of the electron flood gun settings. The questions are: Which ones to adjust, why, and how much? Real-world experience indicates that optimizing the XY position of the flood gun should be your first effort.
Before spending an hour or more to collect chemical state spectra that might have differential charging features, the instrument operator should test if the current flood gun settings are optimized for the sample currently being analyzed instead of the previous sample. This objective is one of the main reason for this publication.
Because each material can have different charging effects due to differences in chemistry, dielectric constant, roughness etc., the operator may need to change the flood gun settings for each individual sample if the operator or the customer needs or wants the best peak-shapes for each sample. If the operator or customer is not looking for the best FWHM to help resolve the presence of absence of various chemical state components, then there may be no need to adjust the flood gun settings for each individual sample on the sample stage. The needs of the customer determine the design of the experiment and the time spent to optimize the flood gun settings.
Because Oxygen is present in most materials it is easy to use the O (1s) signal to optimize the XY positions of your flood gun settings for your real-world samples.
However, for learning purposes, the C (1s) peak from freshly scraped Poly-propylene is used. The following Figures 49 and 50 show the effects of changing XY flood gun positions after collecting data using the normal “Scanning mode” of data collection. For these learning measurements, the flood gun beam voltage was set to -2 eV.
FIG. 49. A transparent ruler, shown here, should be placed on the ledge of computer monitor to serve as a reference point when testing different X and Y flood gun positions.
To observe the BE shifts in the C (1s) spectrum as we test different XY positions, we place a transparent ruler on your computer monitor which can sit on the edge of the monitor. As an example, look for the refer to Figure 49. Watch the relative position of the peak max BE.
The normal scanning mode of data collection was used to produce Figures 49 and 50 because the operator needs to know what happens to the BE and intensity of the chemical state spectra when XY flood gun positions are varied.
(Later in this section, the Snap-shot (un-scanned) mode results are presented. Please remember that energy resolution is low for Snap-shot mode, but data collection is fast, so Snap-shot mode is used to watch the BE and the peak-shape change as the XY positions are changed. The effects of using the Snap-shot or Un-scanned mode are shown in Figure 51-52.)
As the C (1s) peak BE moves to smaller and smaller BE, then you have made an improvement in charge control which will provide an improvement in FWHM and peak-shape and minimize any differential charging.
Align the edge of a transparent ruler with the current position of the peak maximum from C (1s). While observing the peak max, change either the X or Y settings by 25-50% and start the next recording. If a large movement in C (1s) BE is observed in the new spectrum, then decrease the size of the changes by 2x. The size of these steps depends on the chemistry, the analysis position, roughness of the sample, and the settings used by the previous operator.
As the C (1s) BE becomes smaller and smaller, the operator should notice an increase in counts or count-rate. The chart shown in Figure 50 is the result of changing XY positions (10 positions) until the BE was as small as possible. If adjusting XY positions does not produce a measurable change, then increase the beam voltage by 1 eV and start over. By increasing the beam voltage, the C (1s) will have moved to lower BE by 1 eV. Once more, adjust XY positions to try to move the C (1s) peak BE even lower. If there is no change, then the next option to improve charge control is emission current. Increase the current by 10-50% and again test BE or FWHM.
In the past, analysts were trained to increase the beam voltage as the first step to getting better results. You should test both methods.
Based on the author’s experience, it is best to document what happens when any of the flood gun settings are changed by 10-20% from the default setting by recording the C (1s) spectra of Poly-propylene exactly as presented in sections VII and VIII in this report. It is very convenient to use a Screen-Capture software to record the changes made as we change XY positions, current or voltage. The operator should save those spectra and write up a small report to share with all users and customers, so they know the extra work that is needed to get the best result for them. Figures 51 and 52 were produced from Poly-propylene by changing the XY positions of the flood gun beam using 10 different settings and recording the C (1s) spectra that result. Figures 51 and 52 summarize the results of those peak-fits.
After recording the effects of choosing different XY positions of our dual-mode flood gun, we discovered from scanned mode spectra that the raw as-measured C (1s) BE correlates inversely with the electron count-rate. In other words, smaller C (1s) BEs have higher count-rates (Figure 50). The graph in Figure 59 clearly shows the direct correlation between electron count-rate and peak BE when XY positions are varied. This indicates that every effort should be made to adjust the flood gun settings to produce lower BEs which correlate with higher electron count-rates.
This observation is based on the scanned mode results shown in the graph shown in these Figure 50 and 51.
When the flood gun is properly aligned in X and Y the smallest C (1s) BE, correlates with an ~40% increase in count-rate (Figure 50). Scanned mode C (1s) spectra in Figure 51 reveal that the FWHM is smaller by ~20% at the XY position that produces the smallest C (1s) BE.
The scanned C (1s) spectra from a piece of poly-propylene (Figure 51) demonstrates what happens when the XY positions of a modern flood gun are changed from a position that produces good peak-shapes to other potentially better or less useful XY positions. The electron beam voltage for this set of spectra was set to -2 eV. Only the XY positions were varied. All BEs are “as-measured” and were not corrected. The middle two spectra (Figures 51b and 51e) show symmetrical peak-shapes, have the smallest FWHM values, and have the smallest BE values.
As a rule of thumb for charge control: adjust XY positions to produce the lowest BE for the test peak.
This series of spectra shows that the lowest BE spectra do not have sloping tails and have the smaller FWHM values. For most real world samples, O (1s) is the most intense signal. For that reason, it might be useful to repeat this series of XY position changes of the flood gun beam on your real world sample using the O (1s) signal.
If XPS makers would automate this charge control test routine into their software, then operators and users would be able to quickly optimize the flood gun settings that would help operators to collect spectra with less distortion, higher count-rate, and smaller FWHM. These measurements were performed by using the scanned mode for data collection using a 20 eV window, pass energy 50 eV, and the C (1s) spectra for Poly-propylene. Other high purity materials that produce single symmetrical peaks, such as SiO2, Al2O3, Poly-styrene, or HDPE are readily available in many labs, and can be used in place of Poly-propylene.
FIG. 50. After recording the effects of choosing different XY positions of the flood gun, we discovered that the raw as-measured C (1s) BE correlates linearly with the electron count-rate. Smaller C (1s) BEs give higher count-rates. When the flood gun is properly aligned in X and Y the smallest C (1s) BE, correlates with a 40% increase in count-rate. Scanned mode C (1s) spectra revealed that the FWHM is smaller by ~20% at the XY position that produces the smallest C (1s) BE. If XPS makers, automate this routine into their software, then users would be able to measure spectra with less distortion, higher count-rate and smaller FWHM. These measurements were performed by using the fast snapshot (unscanned) mode while measuring ~20 eV wide C (1s) spectra from Poly-propylene. Dashed trendline shows linearity.
Figure 51 shows a set of screen-shots captured from the software pop-up windows that were used to determine the effects of changing the XY positions while using the SnapshotTM (Un-scanned) mode to measure the C (1s) spectra from Poly-propylene. The objective is to locate the optimum XY positions that produce the best peak-shapes for measuring chemical state spectra from this sample. The ~20 eV size Snap-shot (Unscanned) mode window shown here is used for determining which XY positions are best for collecting normal Scanned mode spectra with optimized peak-shapes. The pop-up panel on the top left (Efficiency Data) gives the electron count-rate obtained from the spectrum shown in the pop-up screen on the top-right which is the Snap-shot View (parallel mode) spectrum. To simplify the work, a transparent plastic ruler is placed on the ledge of the computer monitor. After the first spectrum is measured, the left edge of the ruler is aligned with the peak position. The operator then changes either X or Y positions and re-measures the spectrum, noting if the peak maximum moved to the left or the right. Best results are obtained when the peak maximum moves as far to the right as possible producing the lowest BE. In this figure we only show 4 of the 10 spectra used to determine the optimum XY position for this sample.
FIG. 51. This set of scanned C (1s) spectra from a piece of Poly-propylene serves as an example guide showing what happens when the XY positions of a modern flood gun are changed from a position that produces good peak-shapes to other potentially better or less useful XY positions. The electron beam voltage for this set of spectra was set to -2 eV. Only the XY positions were varied. All BEs are “as-measured” and were not corrected. The middle two spectra (Figures 51b and 51e) show symmetrical peak-shapes, have the smallest FWHM values, and have the smallest BE values. As a rule of thumb for charge control: adjust XY positions to produce the lowest BE for the test peak. This series of spectra shows that the lowest BE spectra do not have sloping tails and have the smaller FWHM values. For most real world samples, O (1s) is the most intense signal. Repeat this series of XY position changes of the flood gun beam on your real world sample using the O (1s) signal because it is usually the most intense and it usually appears as a single peak.
FIG. 52. This set of screen-shots show the pop-up windows that were used to determine the effects of changing the XY positions of the flood gun while using the Snapshot (Un-scanned) mode to measure the C (1s) spectra from Poly-propylene. The ~20 eV size window (from a 150 eV pass energy) shown here is useful for this test of finding which XY positions are best for collecting normal Scanned mode spectra with optimized peak-shapes. The pop-up panel on the top left gives the count-rate obtained from the pop-up screen on the top-right which shows the Snap-shot (parallel mode) spectrum. To simplify the work, a transparent plastic ruler is placed on the ledge of the computer screen (see figure 61a). After the first spectrum is measured, the left edge of the ruler is aligned with the peak max position. The operator then changes either X or Y positions and re-measures the spectrum, noting if the peak maximum moved to the left or the right (as indicated in Figures 52b-d). Best spectra peak-shapes are obtained when the peak maximum BE moves as far to the right as possible to lowest BE. In this figure we only show 4 of the 10 spectra used to decide that the best XY position is Y= -2.3 and X= 0.0. In Figure 50 we present a correlation between the count-rate and the C (1s) BE which shows that the lowest BE value has the highest count-rate.
Commonplace materials used to optimize XY positions and charge control settings
Non-conductive commonplace materials are extremely useful to practice how to optimize the peak-shape and FWHM of your high energy resolution, chemical state spectra. The following materials – Poly-Propylene, HDPE, Teflon, PET, SiO2, and Al2O3 are useful to practice optimization. Figures 53, 54 and 55 are chemical state spectra from these commonplace materials. The simple symmetry of the XPS signals in the chemical state spectra from Teflon, SiO2 and Al2O3 make them easy to use when testing flood gun settings.
Three (3) hydrocarbon polymers (e.g., Poly-propylene, HDPE, LDPE) that are useful for these tests are commonly used to make plastic boxes, trays, and wrapping films. These materials are translucent and are easily cut. Each of these hydrocarbon polymers produces a single symmetrical C (1s) peak. If you use one of these products, you should wipe the surface clean using 90% IPA or acetone. Or, if you have a polymer sheet, you can use a clean single-edged razor blade to scrape or expose the bulk, which is free of surface contaminants. Exposing the bulk of any sample helps you to learn what is the true chemistry of the bulk of that material.
TABLE 10. Commonplace materials, available in labs, that are useful to practice charge control by adjusting XY positions of the flood gun and to learn how to adjust other settings.
- Poly-propylene (translucent) wiping with 90% IPA or after scraping with a clean single edged razor blade,
- Poly-ethylene (HDPE/LDPE) after wiping with 90% IPA, or after scraping with a clean single edged razor blade,
- Poly-ethylene terephthalate, PET (transparent soft-drink bottles). After wiping with 90% IPA, or scraping with a clean razor blade,
- Teflon tape (Plumber’s tape). But has static charge,
- Teflon sheet (1-2 mm thick). Cut to expose bulk, or scrape surface with a clean razor blade,
- Silica (quartz) glass (amorphous SiO2 disc, not soda-lime microscope slide) Expose fresh bulk by fracturing,
- Alumina glass (amorphous Al2O3). Wipe with 90% IPA or expose fresh bulk by fracturing.
By adjusting XY positions of the flood gun, and sometimes Z focus, we can reproduce the symmetrical peaks shown below in Figures 62-65. If adjusting XY of the flood gun does not produce a symmetrical C (1s) signal, then adjust other flood gun settings, such as Beam Voltage, or Focus, or Extractor, or Emission Current. (Maybe the flood gun is not fully inserted to the proper translation position.) If you can bias your stage, then try a small positive bias to attract more flood gun electrons. Maybe you need a smaller sample. Maybe you should analyze near one of the corners of your sample. By testing the effects of changing these different variables on a simple symmetrical peak, we can more easily learn more about our flood gun system and how to optimize it. A list of variables that might help improve the peak-shape and FWHM is shown later.
Please keep in mind that different materials have different dielectric constants, surface roughness, and mixed chemistry which requires you to adjust one or more of the flood gun settings when you need optimum peak-shapes that can yield a bit more chemical state information. Very often, we only need to adjust the XY positions of the flood gun or the beam voltage, but not always. Clean Teflon produces a single symmetrical C (1s) peak as well as a single symmetrical F (1s) peak; both can be used to develop your expertise to producing charge controlled spectra. Teflon with its higher dielectric constant will provide a bit more challenge than the pure hydrocarbon polymers. See Figure 53.
There are two commonplace amorphous glasses that provide single symmetrical peaks from O (1s), Si (2p), and Al (2p). The Si (2s) and Al (2s) peaks are also symmetrical peaks, but they have FHWM that are broader due to differences in core hole lifetimes. See Figures 63 and 64.
If, by chance, you need to also check your sensitivity factors or atom% quantitation, then you can use the pure freshly exposed bulk of SiO2, Al2O3, PET and Teflon.
The following spectra (Figures 53-65) show the true peak-shapes from a few commonplace (lab) materials which serve as a tool to help you to practice optimizing your flood gun system.
FIG. 53. C (1s) spectrum from Teflon tape (Plumber’s tape). The -CF2- signal is symmetrical without a tail. Adventitious hydrocarbon is highlighted with an arrow because the -CH2- signal is small. This example spectrum the C (1s) peak-shape and BE expected, when you use Teflon tape to test flood gun settings.
FIG. 54. These O (1s) and Si (2p) spectra (Figures 54a-b) are from freshly fractured SiO2 (silica or quartz). Both signals are symmetrical which means that the surface has Controlled Charging. No tails. FWHM <1.4 eV. The peak-shapes shown here show what is typical for good charge control. This material, freshly fractured, is useful to learn how to operate the flood gun settings and learn how to locate optimum XY positions.
FIG. 55. Al (2p) and O (1s) spectra (Figures 55a-b) from freshly fractured Al2O3 sheet (alumina). The Al (2p) signal is symmetrical, while the O (1s) signal has a tail on the high BE side. The question you may ask: Is that tail due to charging? The answer is: No. Charging tails only appear on the low BE side of the complete set of chemical state spectra for a material. Neither of these two peaks have tails on the low BE side. The O (1s) peak at ~534 eV is most likely due to gamma-type Al2O3, but this is speculation.
Using C (1s) or O (1s) peak FWHM and BE to optimize and check quality of charge control
The FWHM of the C (1s) of adventitious hydrocarbons on the active sample from scanned mode spectra can be used as an indicator of the quality or goodness of charge compensation. Smaller FWHM indicate better charge control. Therefore, if the FWHM of the adventitious hydrocarbon peak in your C (1s) spectrum is >1.5 eV, you should consider that the sample might suffer from differential charging. If the FWHM of the O (1s) and main metal signals are also >1.5 eV, then you have more evidence that the sample might suffer from differential charging.
Based on this understanding, a rule-of-thumb was developed to guide charge control quality by measuring FWHM of C (1s) and O (1s).
The Rule of thumb for best peak-shapes and charge control is:
Adjust XY positions to produce lowest O (1s) BE (or C 1s) with highest count-rate. This corresponds with smaller FWHM.
Our experimental results in Table 10 show that when the FWHM of the adventitious hydrocarbon C (1s) signal is small, then the FWHM of the main metal peak and the O (1s) peak are also small. In like manner, when the FWHM of the C (1s) is large, then the metal and O (1s) peaks are large (Table 10). This means that the C (1s) FWHM and BE can be used to check on the quality of the charge control of your current sample.
TABLE 11. FWHM of C (1s) and O (1s) indicate quality of charge control3 for bulk metal oxides. These FWHM show a rough correlation between all three (3) FWHM. Carbon is a common contaminant. When C (1s) FWHM is small (<1.4 eV) then charge control is good.
|C (1s) FWHM (eV)||Metal FWHM (eV)||O (1s) FWHM (eV)|
One of the goals of this publication is to guide the instrument operator and the data analyst on methods that can be used to obtain high energy resolution spectra with excellent peak-shapes that have little or no differential charging that might broaden FWHM or produce unexpected shoulders. The next goal is to provide the operator and data analyst with a useful set of FWHM values to be used to peak-fit raw spectra.
To derive as much useful information as possible from peak-fitting, data analysts need not only true peak-shapes free from distortion and reliable5 BEs, but also useful or reliable FWHM to maximize the information derived from peak-fitting. Differential charging can produce peaks that are falsely broad, leading the data analyst to use a FWHM that is not useful or meaningless. By providing data analysts with a useful set of FWHM, they can improve their efforts to properly peak-fit complex chemical state spectra by using FWHM derived from high quality XPS reference spectra. With this goal in mind, Table 11 is a list of FWHM from the main metal XPS peaks and O (1s) peaks for a list of 48 commercially pure binary metal oxides and 2 metal carbonates. These FWHM were obtained from peak-fitting spectra from commercially pure binary oxides published in the XPS International SpecMaster database2 and The XPS Library.3
In Table 11, the metal oxides that have multiplet splittings for the metal signal are highlighted with the abbreviation M.S. The FWHM shown for these metal oxides are due to the initial peak belonging to the splitting. The remaining peaks in that peak-fit use the same FWHM. This style produced a reasonable fit between the first peak and the peak envelop.
TABLE 12. FWHM of O man metal peak and the O (1s) peak from bulk metal oxides using charge-control mesh-screen.2 These FWHM are provided to help data-analysts to choose a useful FWHM for peak-fitting pure oxides and other materials.
|Bulk Metal Oxides
Charge Referenced to
of Metal Signal
|Metal (main peak) FWHM using
Charge Control Mesh-Screen(eV)
|O (1s) FWHM using
Charge Control Mesh-Screen(eV)
|CeO2*||3d5*||3.0* (1st MS peak at 877.0 eV)||1.5|
|Co3O4*||2p3*||1.51* (1st MS peak at 779.5 eV)||1.03|
|Cr2O3*||2p3*||1.10* (1st MS peak at 575.6 eV)||1.27|
|Fe2O3*||2p3*||1.31* (1st MS peak at 709.8 eV||1.04|
|La2O3*||3d5*||2.50* (1st MS peak at 834.4 eV)||1.66|
|MnO2*||2p3*||1.12* (2nd MS peak at 641.4 eV)||1.02|
|NiO*||2p3*||1.19* (1st MS peak at 853.7 eV)||1.06|
|Rh2O3*||3d5*||0.90* (1st MS peak at 308.6 eV)||1.09|
|RuO2*||3d5*||0.77* (1st MS peak at 280.7 eV)||0.95|
|V2O5*||2p3*||0.99* (3rd MS peak at 517.0 eV)||1.33|
SOURCES OF XPS REFERENCE GRADE SPECTRA
XPS is rapidly growing; being used in hundreds of different research fields because it is extremely useful to identify different chemical states at the surface of any solid material. Researchers that need XPS results use central research centers that provide raw XPS spectra without chemical state assignments. To help them to process and peak-fit their raw spectra these researchers use the NIST XPS database and a free standalone peak-fitting software, CasaXPS.
The NIST database supplies XPS BEs but no actual spectra, only numbers. These simple numbers2 have been found to have various problems with reliability. Calibration in the NIST database is usually limited to BEs from Silver or Gold, but never Copper. The scientific literature that publishes XPS data normally only publish 2-3 raw or processed chemical state spectra even though the research analyzed various chemicals. Data-analysts need actual reference XPS spectra to cross-check spectra obtained from central labs. These reference XPS spectra allow the data-analyst to more accurately and more reliably peak-fit their raw spectra.
To help data analysts and instrument operators, two groups have been publishing complete sets of actual monochromatic XPS spectra. Actual spectra are extremely helpful and valuable guides to processing raw spectra and checking on the presence or absence of differential charging. The products offered by these groups are:
- The XPS Library, Spectra-Base2,3 with >70,000 Mono-chromatic XPS Spectra from pure materials, natural crystals, single crystals, polymer films, polymer beads, etc. (https://xpslibrary.com)
- Vincent Crist, Handbook of Monochromatic XPS Spectra, The Elements and Native Oxides, Wiley, (2000) ISBN: 978-0-471-49265-829
- Vincent Crist, Handbook of Monochromatic XPS Spectra, Semiconductors, Wiley, (2000) ISBN: 978-0-471-49266-530
- Vincent Crist, Handbook of Monochromatic XPS Spectra, Polymers and Polymers Damaged by X-rays, Wiley (2000)31
- Vincent Crist, PDF Handbooks of Monochromatic XPS Spectra, Five (5) Volume series – 5 PDFs, XPS International LLC, (https://xpsdata.com) (2018)2,3
- Surface Science Spectra (Am. Vac. Soc.) with >1,500 monochromatic XPS spectra, Editors: J.E. Castle and R. Haasch.
SUMMARY AND CONCLUSIONS
Current day XPS instrument makers have made significant advances in charge control systems over the last 20 years that make it easier to analyze insulators, but samples still have many differences in chemistry, dielectric properties, sizes, surface roughness, etc. that force instrument operators to tweak flood gun settings if they want or need to obtain high quality chemical state spectra that provide the most information.
Current day data-analysts and researchers, who use XPS spectra, have little if any experience with the difficulties of preparing samples for XPS analysis and various problems that can and do occur when running an XPS instrument, maintaining that instrument, and calibrating that instrument. As a direct result these researchers do not recognize differential charging problems because they have no idea what differential charging looks like. If the data analysts would run the instrument, then they would learn what good spectra look like and what bad spectra look like.
This guide teaches what differential charging looks like by showing a significant number of example spectra together with spectra from the same sample that are free from differential charging.
This paper discusses the origins and types of charging, charge neutralization within the atom, and how the atom deals with the loss of a photoelectron.
To help learners and mid-level users to learn which flood gun variables to check, we present tables show the effect of changing those variables. Learning how to optimize electron flood gun settings by presenting high energy resolution, chemical state spectra makes learning easier and faster.
By focusing on the XPS measurement of insulators – non-conductive metal oxides and polymers this paper reveals what happens to >70% of all materials analyzed by XPS. This guide shows that by measuring commonly available polymers (Poly-propylene and PET) or ceramic materials (SiO2 and Al2O3), that the operator can easily characterize the good and bad effects of XY position settings, and other settings provided by modern electron flood gun systems.
Many original, never-before published peak FWHM are provided that will greatly assist peak-fitting efforts. This guide reveals a useful direct correlation between electron count-rate, measured BEs, and best charge control settings that should be a common phenomenon.
This guide discusses FWHM and BE of C (1s) or O (1s) spectra produced from the sample currently being analyzed that can be directly used to decide if charging has or has not been minimized. A list of other charge control methods is provided; along with advice and a author’s recommended method (ARM).
A short list of large extensive databases of actual XPS spectra is extremely beneficial to users who need real world examples of high quality chemical state spectra to guide their in-house efforts to collect high quality spectra and to interpret valuable information from the peak-fits of those spectra.
This study was supported by The XPS Library, a Not-for-Profit foundation. We acknowledge the assistance of and discussions with Dick Brundle, Mark Engelhard, Paul Bagus, and Noel Casey. We are grateful to Hakuto Co. Ltd in Japan, Nanolab Technologies in USA, and IPG Photonics in USA for the use of the XPS instruments to produce the spectra used in this work. Useful institutional guides and publications involving charge control, differential charging, and patents on new type of electron flood guns are listed at the end of the reference section, as references 33-38.
The data that support the findings of this study are available from the corresponding author upon reasonable request. All spectra and images are copyrighted by the author and available free-of-charge. https://xpsdata.com/charge-control-guide-JVST-2020
All XPS spectra, data, data tables, and images were produced by B. Vincent Crist, the author. There are no funding agents as the author paid for all materials and analyzed all materials.
All spectra are part of The XPS Library and The XPS Spectra-Base with 70,000+ XPS Spectra, which are commercially available from XPS International LLC. The software “SDP v8” was used to process all spectra. SDP v8 is produced by XPS International LLC (https://xpsdata.com).
1B. Vincent Crist, J. Electron Spectrosc. Relat. Phenom. 231, p76-87 (2019).
2XPS International SpecMaster Database of 70,000 Monochromatic XPS Spectra (https://xpsdata.com)
3The XPS Library (https://xpslibrary.com).
4C. E. Bryson III, Surf. Sci. 189/190, p50 (1987).
5Reproducibility and Replicability in Science. Washington, DC: The National Academies Press. https://doi.org/10.17226/25303.
6D. R. Baer, K. Artyushkova, H. Cohen, C. D. Easton, M. Engelhard, T. R, Gengenbach, G. Grecynski, P. Mack, D. J. Morgan, and A. Roberts, XPS guide: Charge neutralization and binding energy referencing for insulating samples, J. Vac. Sci. Technol. A 38(3) May 38 (2020).
7V. E. Henrich and P.A. Cox, The Surface Science of Metal Oxides, Cambridge University Press p102-156. (1996)
8S.G. Davison and M. Steslicka, Basic Theory of Surface States, Oxford Science, Monograph p46 (1992).
9G. A. Somorjai, Introduction to Surface Chemistry and Catalysis, Wiley, p1-16 (1994)
10A. Zangwill, Physics at Surfaces, Cambridge University Pres, p104-109 (1990).
11M. Prutton, Introduction to Surface Physics, Oxford Science Publications, p109-138 (1994).
12M. A. Kelly and C. E. Tyler, HP Journal, 24, p2 (1973).
13M. A. Kelly, J. Elec. Spectrosc. Relat. Phenom. 176, p5 (2010).
14G. Barth, R. Linden, and C. Bryson, Surf. Interface Anal. 11, p307 (1988).
15S. A. Flodstom and C.W.B. Martinsson, R.Z, Bachrach, S.B.M. Hagstrom, and R. S. Bauer Phys Rev Letters, 40, p907 (1978).
16J. Cazaux, J. Electron Spectrosc. 105 (2), p155-185 (1999).
17J. Cazaux, J. Electron Spectrosc. 113 (1), p15-33 (2000).
18G.L. Nyberg and S.E. Anderson, J. Electron Spectrosc. Relat Phenom. 52, p293-302 (1990).
19D.L. Perry and J.Q. Broughton, Surface Science 74, p307-317 (1978).
20Kratos – Charge Compensation Presentation 2004. DIFF CHARGING S.L. Bernasek, Chem Materials, (2017) DOI: 10.1021/acs.chemmater.7b00621.
21P. Nachimuthu, (PNNL) Working with difficult samples: Preparation, Damage, Charging and Data Analysis, presented at AVS 58th International Symposium Nashville, Tennessee – Oct 30-Nov 4, 2011.
22D. R. Baer, C. F. Windisch Jr., M. H. Engelhard and K. R. Zavadil, J. Surf. Anal. 9 (3), p396-403 (2002).
23D. R. Baer, M.H. Engelhard, D. J. Gaspar, A. S. Lea and C. F. Windisch, Surf. Interface Anal. 2002; 33: p781–790 DOI: 10.1002/sia.1454.
24NIST X-ray Photoelectron Spectroscopy Database, NIST Standard Reference Database Number 20, (National Institute of Standards and Technology, Gaithersburg MD, 20899) available online https://srdata.nist.gov/xps/Default.aspx.
25J. F. Moulder, W. F. Stickle, P. E. Sobol, and K. D. Bomben, Handbook of X-ray Photoelectron Spectroscopy, publ. Perkin Elmer Corp. (1992)
26P. M. A. Sherwood and J. A. Rotole, SSS, 5, p11 (1998).
27C.Noguera, Physics and Chemistry of Oxide Surfaces, Cambridge University Press, p43 (1996).
28S. A. Flodstom and C.W.B. Martinsson, R.Z, Bachrach, S.B.M. Hagstrom, and R.S. Bauer Phys Rev Letters, 40, p907 (1978)
29B. Vincent Crist, Handbook of Monochromatic XPS Spectra, The Elements and Native Oxides, Wiley, (2000) ISBN: 978-0-471-49265-8.
30B. Vincent Crist, Handbook of Monochromatic XPS Spectra, Semiconductors, Wiley, (2000) ISBN: 978-0-471-49266-5.
31B. Vincent Crist, Handbook of Monochromatic XPS Spectra, Polymers and Polymers Damaged by X-rays, Wiley, (2000) ISBN: 978-0-471-49267-2.
32B. Vincent Crist, PDF Handbooks of Monochromatic XPS Spectra, Five (5) Volume series – 5 PDFs, XPS International LLC, (2018).
33ASTM E1523-15, Standard Guide to Charge Control and Charge Referencing Techniques in X-Ray Photoelectron Spectroscopy, ASTM International, West Coshocton, PA, (2015).
34ISO 19318:2004 “Surface chemical analysis – Reporting of methods used for charge control and charge correction” International Organization for Standardization, Geneva, (2004).
35P. E. Larson and M. A. Kelly, Control of surface potential of insulating specimens in surface analysis US Patent 5,990,476, JP Patent P3616714, and EP Patent 0848247B1. (1999).
36Journal of Surface Science Spectra, AVS Publication, J. Castle and R. Haasch, editors.
37B. R. Barnard, M. H. Humpherson, and A. R. Bayly, Flood gun for charge neutralization patents US20050205800A1(2011) GB2442485B (2006) (2005).
38Thermo Dual-Beam Argon Electron Flood Gun D16107, App Note 31071 FG01 Combination Low Energy Electron/Ion Gun (2008).
FIG. 56. Artistic schematic of core hole – electron cascade for an insulator (e.g. GeO2). The sequence of actions are 1-5. Internal electrons are the source of the neutralization. The time scale for the electron-cascade to produce a valence-hole is <10-17 sec. This depiction uses a simple Bohr orbital model to simplify the display.
FIG. 58. C (1s) spectrum of PET (Mylar) (a), O (1s) of PET (Mylar) (b), and C (1s) of Poly-vinylidenedifluoride (PVdF) (c). The hydrocarbon C (1s) peak from PVdF has a small shoulder at low BE which is due to adventitious hydrocarbons. If that was due to charging, then the CF2 peak would have the same small shoulder. There is no tail, which means that charge is well controlled.
FIG. 59. C (1s) and O (1s) spectra (Figures 59a-b) from freshly scraped PET. PET (MylarTM). produces more complicated peak-shapes that will test not only your charge control skills but also your peak-fitting skills. PET has become the reference material used to check on the performance of modern flood gun systems in an XPS instrument. Peak area ratios are shown as numbers inside each peak shape. The 3 peaks for C (1s) should have a 3:1:1 ratio for the 3 species shown. If your peak-fit produces a ratio = 4:1:1 or 4:2:1, then the surface is contaminated and needs to be cleaned. You can add an extra constrained peak to learn how much contamination is present. The most likely contaminants are hydrocarbons and alcohols.
FIG. 60. Photos of S-Probe flood gun system (rf Service-Physics LLC).
FIG. 62. Example of “Un-controlled (Differential) Charging” (a) versus “Controlled Charging” (b) from the same sample of poly-acrylate adhesive on the backside of a paper label. The spectrum on the top (a) shows differential charging effects on the low BE side of the O (1s). A charge-control mesh-screen was then placed ~1 mm above the same sample to produce the O (1s) spectra on the right. Note the symmetry and valley in the O (1s) spectrum at the bottom (b).