Our ability to analyze surfaces is an important enabler in many fields of research and across many industries. It was not long ago that the field of surface analysis was its own niche, but today a literature search for any of the major surface analysis techniques will bring up mostly papers about other subjects, towards the understanding of which surface analysis methods have contributed.

• A wafer arrives in the lab from the manufacturing line with portions peeling off. The line needs to be told what interface is failing, and why.

• A series of treatments have been tried to make a material biocompatible. Some work better than others. What is it about those surfaces that make them work? What are the compounds that first adsorb onto the surface when the material is exposed to the biological system of interest?

• A new process is used to make a new polymer blend. The polymer that is present at the surface will determine the surface properties. Which polymer is it? Is it homogeneous, or can areas of the surface have different polymers at the surface? If there is more than one polymer at the surface, what portion of the surface is constituted by each? What are the sizes and distributions of the different domains? Can we determine the 3-dimensional (3D) structure of the material, at least near its surface? How best can the unusual and interesting phases in a cross section of a geological sample be found?

• A magnetic recording head has a smear of a material so thin it could only be imaged in the Secondary Electron Microscope (SEM) with a low 1 KeV electron beam. What is it?

• Trace contaminants can kill a semiconductor wafer, but you won’t know the wafer is defective electrically until it reaches the end of the manufacturing line, and your company has invested a large amount of money in it. How do you make sure that at key process steps the wafers have not become contaminated? If you can do that at large test sites or using homogeneous monitor wafers, how do you determine that nothing different is happening at the tiny actual devices on a real wafer?

• What can we learn about the chemicals present and their special locations in a thin tissue sample? What chemicals were used to paint that apparent masterpiece? Are they what you would expect for a painting from the time from which it was purported to come?

While many properties of materials and devices which are important to their use and function are bulk properties, a remarkable percentage of the properties that matter are functions of their surface. Those properties will be determined by the composition and bonding present at the surface. And when we talk about a surface, we are really talking about the top few monolayers of material at the surface. Perhaps only the top layer of atoms is important. And even when we are interested in the bulk (3D) structure of a material, often we need to get that by slowly eroding a sample, analyzing each new surface as it is exposed. In order to get information about the surfaces of samples, laboratories across the world spend large sums of money to purchase expensive and sophisticated equipment, and they pay good money to hire scientists and engineers to operate that equipment.

Surface properties that matter scientifically and technologically include surface energy (often determined by contact angle measurements), surface potential, conductivity and other electrical characteristics, biocompatibility (and this can amount to what type of cell, if any, likes to spread on that surface), topography (including roughness at a variety of length scales), hardness and surface mechanical properties, susceptibility to corrosion, and the coefficient of friction, among many others. The chemical state of a surface will determine whether the material can act as a catalyst or not, and how it changes when it gets fouled.

These important surface properties (topography excepted) are largely a function of the surface composition, the elements present, their bonding, and their structure. One can, of course, take a less sophisticated approach to the needs we have as researchers and engineers for surfaces with specific, well-understood, and reproducible surface properties by using trial and error to obtain them. It is certain though that in this, as in many human endeavors, actually knowing what you are doing is preferable, and often more efficient in the long run. Then, of course, there is the case where an unwanted interloper, a contaminant, is affecting the surface properties and in that case it can become critical to learn what the heck it is in order to get rid of it (or to intentionally apply it if it is helping). Much can be learned about the composition, bonding, and structure of a surface. One can try to determine the amount of each element present in a given portion of the sample. That may be expressed as the atomic percentage of all of the elements present in the analysis area to a depth of 3 nm, for example. Better still, one may try to determine the bonding present in the sample (what elements are bound to what other elements). Even more ideally, one would like to know what actual chemical compounds are present. And to take our insatiable desire for knowledge one step further, we would like to know how our chemical compounds are oriented, relative to the surface and to each other, and whether the surface has some kind of repeating order (as in a crystal) or is more random (amorphous). It is not possible to get all of this information using a single analytical method and a well-equipped laboratory filled with the latest equipment and the brightest minds will fail to learn all of this about most samples in a reasonable amount of time. Fortunately, to solve any given problem, a subset of this information will usually suffice. In most cases we will still need more than one technique at our disposal to learn all that needs to be learned.

The quintessential “black box” experiment is characterized by the inputs into the box and what one perceives coming out of the box. The surface as an unknown can be thought of in a similar way. With each analytical technique we interrogate the surface using some input and have our experiment set up to measure a specific output. These inputs and outputs are summarized in the following table.

In	Out	Technique(s)	Principles at work	Why surface sensitive
X-rays	Electrons	XPS/ESCA	Electrons of characteristic energy are emitted when they are ejected from atoms by absorption of X-rays.	Electrons lose their distinct energies when emitted at depth
Electrons	Electrons	Auger	Electrons of characteristic energy are emitted when excited atoms with missing core electrons relax	Electrons lose their distinct energies when emitted at depth
Primary X-rays	Secondary X-rays	TXRF	X-rays of characteristic energy are emitted when excited atoms with missing electrons relax	Below a critical angle, X-rays reflect from a surface without much penetration, so all the fluorescence occurs at the surface
Primary X-rays	Primary X-rays	XRR	The oscillations in X-ray intensity as a function of the reflected angle are modeled to reveal layer thickness and density	The method can measure the thickness of very thin layers assuming they differ in density, including the topmost layer
Force	Cantilever position	AFM	A tip is pressed against a surface with a set force/force profile, and the movement of the tip monitored	The topography is that of the surface assuming forces are kept beneath the point at which deformation occurs
Force / Light	Cantilever position	AFM-IR	The AFM measures photo induced forces allowing IR spectroscopy at nm scales	The degree of surface sensitivity depends on the specific method used
Primary Ions	Primary Ions	Ion Scattering (RBS, MEIS, LEIS)	The energy of scattered ions is a function of the mass of the atoms off which they are scattered and any energy loss that occurs within the solid	LEIS is quite surface-sensitive and because the ions start with such low energies, they won’t escape as ions if they penetrate very deeply
Primary Electrons	Primary and Secondary Electrons	SEM	Backscattered and secondary electrons that come from a focused beam give sharp images	Secondary electrons are representative of the top few nm of material, and have topographic contrast while backscattered electrons, which have Z contrast, may come from a much greater depth, since this value will increase with increasing primary beam energy
Primary Ions	Secondary Ions	SIMS	The primary ions force removal of atoms, fragments, and whole molecules from the surface, some of which ionize and can be detected	Experiments show that most of the ions detected originate within the top few nm of the surface

A number of these methods depend on the effects of ionizing radiation on samples. By definition, ionizing radiation makes portions of the sample ionic, generally by removing electrons. After that, the sample can lose electrons, photons, and ions, and we can learn about the sample by measuring all of these. In the Auger spectroscopy and XPS/ESCA (X-ray Photoelectron Spectroscopy/Electron Spectroscopy for Chemical Analysis) measurements, we look for the electrons. In TXRF (Total Reflection X-ray Fluorescence) and XRR (X-ray Reflectivity), we look for the photons. In SIMS, we measure the ions that are emitted.

The methods using the electrons and ions emitted by the sample require ultra-high vacuum systems to keep the surfaces clean and the space between the sample and the detector reasonably empty so that the characteristic species are not lost before detection. The more penetrating photons do not fix contamination to the surface as electron beams do, and air does not greatly impede X-rays, so techniques like TXRF do not require a vacuum system.

Ionizing radiation will often damage your sample. If your sample is nicely conductive and you hit it with an electron beam, the damage may be minimal. If you are analyzing a polymer, it may be chemically altered very quickly.

Other techniques in the table rely on the effects the sample surface has on the beam we use to interrogate the sample. In XRR (X-ray Reflectance), we look at the intensity of the reflected X-rays. In the various ion scattering techniques we measure the energies of reflected ions.

Finally, we have the probe techniques with which we poke directly at the sample surface (like putting our hands into the black box to feel what we can feel), represented here by the most commonly used set of techniques, Atomic Force Microscopy (AFM).

AFM-IR has been recently commercialized. This is a set of methods that detect the interaction of infrared radiation with the surface using a probe with the result that an infrared spectrum similar to standard infrared spectra is generated that provides chemical information about small areas of the sample.

What many physicists call ESCA, most chemists call XPS (Briggs 2003). This is just a matter of perspective. To physicists, this method provides a wealth of information about the chemistry of the surface. To chemists, this seems like pretty minimal chemical information. Whatever you call it, this method has proliferated recently far beyond the other methods in the table. Instrument sales and publications involving this technique far outstrip those of the other methods.

In these instruments, we hit our sample with a beam of mono-energetic X-rays. X-rays generate ionizing radiation; so as noted earlier, the sample ionizes, in this case, because atoms in the sample lose electrons when they absorb photons from the beam. If an electron that is removed is from the core of an atom, it will leave with an energy characteristic of the absorbed radiation minus the energy it takes to remove the electron from the atom. We measure the flux of electrons leaving the sample as a function of their energy. What we find is a continuous background of electrons at different energies, peppered with peaks that appear at specific energies. The presence of the background is the result of inelastic scattering of electrons as they bounce around inside the sample before finally emerging. The X-rays penetrate the sample deeply, and electrons are produced at all depths to which they reach. However, electrons produced deep within the sample scatter inelastically along with many electrons produced closer to the surface, losing their characteristic energy, and thus becoming part of the background. It is inelastic scattering that makes XPS surface sensitive. In order for an electron to have any chance of escaping the surface with all of the energy it had when it was removed from its atom of origin, it has to come from an atom close to the surface.

Subtracting the electron energies from the initial X-ray energy leaves the value of the binding energy for that electron. Electron binding energies reflect the electronic levels in the atoms from which they emerged and are therefore highly characteristic of the identity of those atoms. So the first bit of information one obtains is what elements are present near (typically within a few nm of) the sample surface. What puts the C in ESCA is the fact that these binding energies shift with the bonding in which the atoms of origin are participating. Bond an atom to a highly electronegative element and the binding energy increases. The electronegative atom has pulled charge away from the atom in question, leaving the core electron we are about to extract attached to an atom that is more positively charged than another atom of the same element not bonded to such an electronegative atom. It makes sense that it will require more energy to remove that electron.

The intensity of a given peak (after the background has been subtracted from it) is not only a function of the percentage of the surface made up of the element that produces electrons with that characteristic energy but also a function of the distribution of that element within the sample as a function of depth. The deeper the atom, the less it contributes to the peak. Because most of the physics surrounding the production of electrons from surfaces irradiated with X-rays is so well understood, one can get quantitative information about elemental composition from XPS, even quantification of the percentage of each element bonded within the surface one way or another. Add in variable angle measurements, and one can deconvolve intensity variations due to gradients with depth from intensity variations due to concentration changes alone. Given that instrument software takes care of much of the interpretation of the data, you have a powerful technique that may not require someone with a doctorate to operate.

However, as noted earlier, XPS cannot answer all the questions we have about our samples. The spot size of ESCA instruments has improved over the years, but to get data with a resolution of tens of microns requires long acquisition times, and that’s not going to work at various angles. X-rays can damage many samples. Finally, the chemical information one gets is quite limited. It is far from enough to identify unknown compounds on your surface.

Having an XPS next door, though, is marvelously helpful to a TOF-SIMS analyst. This slower, larger spot instrument is an important source for standards that will allow you to quantify similar samples at a smaller scale and much faster.

After the process that produces XPS spectra is complete, the ions that remain exist in highly energetic states. They can lose that energy (relax) only if they can lose this excess energy. There are two ways by which they do this; by losing another electron or by sending off a photon. The process of losing a second electron to produce a doubly charged ion is named after the physicist Pierre Victor Auger (Wolstenholm 2015).

The Auger electron has a characteristic energy that depends on the energy level of the core vacancy that was originally created, the energy of the level from which a higher energy electron within the atom drops to fill that core vacancy, and the binding energy of the electron that carries away the energy produced when that core vacancy is filled. All three of these energy levels are intrinsic to the element from which the Auger electron is produced, so the Auger electron’s characteristic energy is not affected by the energy of the initial ionizing radiation. This turns out to be very useful. It means that one can produce Auger electrons with an ionizing radiation that is not monoenergetic, and the easiest way to handle and to focus an ionizing beam is to use electrons. Though the electrons may lose energy as soon as they enter the sample, as long as they ionize, characteristic Auger electrons will be produced. As in XPS, electrons retaining their characteristic energies must come from close to the surface. In Auger instruments, because many Auger electrons have lower energies than the photoelectrons produced in most XPS instruments, the Auger tool is more surface sensitive than XPS. In fact Auger electrons are also observed and often prove useful in XPS instruments, but in an Auger instrument the electron beam gives you excellent lateral resolution, not remotely possible with an XPS tool. The Auger effect is now very well understood, so again, this technique is at least semiquantitative from first principles without standards.

The Auger signals are not strong and are found atop a significant background. To observe them from many elements in a reasonable amount of time requires a very high current of the primary electrons. The sample has to be conductive enough to receive electrons without charging so much that electron spectroscopy is no longer possible. Even if a sample is sufficiently conductive, if it is sensitive to the electron beam, it will be rapidly damaged.

Auger peak shapes have been shown to be repositories of chemical information just like XPS, but generally, Auger spectroscopy is used for elemental analysis only.

Like XPS, Auger can be a good source of standards for the SIMS analyst. Samples that may be well analyzed by Auger may also be analyzed by TOF-SIMS, but without the first principles quantification that Auger brings. So, given a series of similar samples, it is often worthwhile to do a few by Auger and TOF-SIMS, but then take over with TOF-SIMS in order to take advantage of its speed of analysis. There are actually situations in which the sensitivity of Auger is better than SIMS (sometimes this is chemical species specific). Finally, there are also times when the lateral resolution available in the TOF-SIMS is not sufficient, and Auger is the only recourse.

As noted above, an ion with a core level vacancy can “relax” when a higher-level electron drops into the core level that was vacant. The ion then needs to release the potential energy that has now been realized, and that can happen not only via ejection of another electron (the Auger effect described earlier) but also by emitting a photon. There are many techniques that involve detection of these X-rays. The degree to which they come from the surface rather than the bulk of the sample depends largely on the ionizing radiation used. X-rays are not well stopped by matter, so the emission of X-rays can be detected from within the bulk of a material. Most SEM instruments have energy dispersive X-ray detectors. Depending on the electron energy used, the X-rays detected may come from a fraction of a micron to a micron below the surface, basically limited by the interaction volume of the electron beam within the solid. This technique is thus far from surface sensitive. If one uses X-rays as one’s ionizing radiation, one can do what is essentially a bulk measurement by XRF spectroscopy. However, if the X-rays used come at the surface at a sufficiently grazing angle, the X-rays end up being reflected from the surface and do not penetrate into the material. The primary X-rays can still ionize the surface, but the emitted X-rays are now only coming from a very thin surface layer from which the reflection occurs. This TXRF technique is therefore quite surface sensitive (Hockett 1995).

TXRF has the advantage again that the physical principles behind it are well understood, and thus quantification is normally performed. It basically reveals the elemental composition of the surface, and it is supremely sensitive to some elements, but not all. A common use of the technique is in the analysis of wafer surface contamination where for some elements it can rival the surface sensitivity of SIMS. It has, however, very poor lateral resolution. X-rays are difficult to focus to start with, and the geometry of the measurement, requiring oblique incidence of the interrogating X-ray beam, widens the analysis area even further.

XRR measurements are made above the critical angle where X-rays penetrate the surface further (Spiga 2008). If you have a sample with discrete layers of differing densities, X-rays reflecting off the various interfaces interfere with one another. The reflected X-ray intensity oscillates with steadily changing angles of incidence. With some prior knowledge of the sample, even structures with multiple thin layers can be modeled and the thickness and roughness of the layers in the model iteratively adjusted until the simulation matches the acquired data. The results are, in effect, measurements against the wavelength of the X-rays, which makes the thicknesses that are determined among the most accurate that can be obtained.

XRR is more a tool for thin film analysis rather than for surface analysis. However, even very thin films of organic materials, oxides, and overcoats can be analyzed, and the quantitative results can be used to calibrate other techniques, including TOF-SIMS.

AFM and the other related probe techniques produce an entirely different and complementary set of information about surfaces (Yang 2014).

AFM itself produces highly resolved maps of the topography of a surface. The surface is probed with an extremely sharp tip. A piezo-electric crystal on the cantilever holding the tip can be used to apply a force, and a laser reflected off the cantilever will be deflected by small changes in the cantilever position. At each point on the sample you can measure how much force it takes to reach a certain distance relative to the sample surface, or you can measure the height at which you reach a constant force. The topography that is measured is more highly resolved than any of the other surface sensitive techniques. In conjunction with AFM, one can force the tip into the sample, see what happens, and thus get a sense of the physical properties of the surface (hardness, wearability, etc.). Because topography can affect ion yields in SIMS, and because SIMS performed beyond the static SIMS limit (the point beyond which subsequent analysis reveals aa surface altered by the irradiation of the sample by the primary ions) can produce a changed topography, AFM is a handy tool to have nearby. AFM can tell you if the feature imaged in the SIMS is a smear on top of the sample or if it has depth and is, instead, damage that has revealed layers beneath.

The variety of probe techniques that are used today is truly remarkable. These other methods are less commonly used in conjunction with SIMS, but they all have their places in the panoply of characterization methods. The original probe technique, scanning tunneling microscopy, also involves the presence of a sharp tip close to the sample surface, but in this case one measures the electron current between the tip and surface at a constant height or the height needed to maintain a constant current (Wiesendanger and Guentherodt 1996). This method has atomic resolution, but measures electronic states rather than the physical position of the atoms at the surface. Or one can use a tip with magnetic properties and measure magnetism of the surface using magnetic force microscopy (Sarid 1994). The shapes of the bits written into a magnetic recording disk can be directly measured. There are many other variations on this theme.

First published in 2005, the reports on the AFM-IR methods (Mathurin 2022) have led to the growth of a user base now fed by commercially available systems that are practical for widespread use. The use of a tunable Infrared laser in conjunction with the probe tip allows the system to obtain infrared spectra with a lateral resolution closer to that of AFM than to any other infrared spectrometer, and with high surface sensitivity. The infrared spectra obtained are exactly like those obtained by other infrared spectrometers, so that the large database and knowledge base of existing infrared spectra can be directly used to interpret the results.

AFM-IR thus has become something of a rival to TOF-SIMS, providing real chemical information and at lateral scales the SIMS cannot achieve. On the other hand, the method is not nearly as specific as the SIMS can be in identifying particular molecules and materials at the sample surface. Further, the need to scan the laser to get a spectrum at each pixel makes the method much slower than TOF-SIMS. AFM-IR provides important complimentary information to the SIMS analysis.

Shoot an ion at a solid surface and it may scatter back at you (Rabalais 2002). Classical physics determines the mechanics of this elastic collision. The scattered ion will lose some energy to the atom it strikes. If you hit an atom of the same mass, the ion will stop and the atom will take all its kinetic energy to fly forward into the solid. If the atom the ion hits is much heavier, the atom will budge only a little bit, and the ion will rebound with almost its original complement of energy. So by measuring the energy of the scattered ions, you know the mass of the atoms your ions collided with. This is simple billiard ball physics.

At low energies, (typically 500 eV to 20 KeV), ions tend to neutralize as they penetrate the solid, so if you are measuring ion scattering (and not neutrals), you are mostly limiting the information you get to scattering off the top layer of atoms. This, then, makes LEIS (Low energy ion scattering) the most surface sensitive of the techniques described in this chapter capable of compositional analysis. You can get some scattering of ions off atoms that lie deeper, but you will see both a drop in intensity due to ion neutralization and energy loss to the ion due to it’s transit through the solid. A shoulder on the low energy side of the surface peak for an element in the LEIS spectrum results from the scattering of ions off atoms beneath the surface. At these low energies, the physics is not so precisely known in all cases so as to say standards are not needed. If you are dealing with a crystalline surface, the orientation of the analysis beam with the crystal structure matters, and, in fact, you can learn something about the structure by plotting the spectrum versus angle and orientation.

In contrast, high energy ion scattering, more commonly known as RBS (Rutherford Backscattering Spectroscopy), is directly quantitative without standards. This is because at the MeV energies used, the cross sections and the energy loss is quite well understood. It is true that you need an accelerator to do RBS, but that is not as daunting as it may appear. Small accelerators are available for purchase that do not require massive projects and long tunnels. The method is not particularly surface sensitive, as one might expect. However, in situations where you are trying to characterize a thin surface layer of relatively high Z (heavy) elements on a sample constituted with lower Z (lighter) elements, the RBS can detect and quantify. Again, you can line up the ion beam with the crystal axes of a crystalline material and channel the beam into the sample, reducing scattering from all but the surface atoms and any impurities lying in the channels.

A technique related to RBS is elastic recoil detection where one mounts the sample at an oblique angle to the beam and looks for elements, lighter than the analysis beam used, that get knocked out of the sample. The method is most commonly used with alpha particles (He++) knocking H out of the film. The method is quantitative, like RBS, and can therefore be an excellent method for measuring H, the element that is very hard to detect but that SIMS can see and XPS and Auger cannot. Unfortunately, again, the method is not particularly surface sensitive, and is best used for surface analysis when it is known that only the surface layer contains H. There is an intermediate MEIS (Medium energy ion scattering) method that is not so quantitative as RBS and not so surface sensitive as LEIS and, therefore, tends to have less general application to materials characterization.

SIMS depends on the fact that shooting ions at surfaces will damage those surfaces (Benninghoven, Werner, and Rudenauer 1986). The primary ion may rebound, but the atom struck will be driven into the sample. The chances are that the collision will have been glancing, in which case both the primary ion and the atom it collided with will be off wreaking havoc within the solid. The typical primary ion energy used is many keV, so with bond energies in the few eV range, the bonding at the site of impact will be destroyed. The resulting chaos is known as the collision cascade. After a few collisions, some of the energy will be directed back up at the surface, and that is when pieces of the sample surface will be driven out into the vacuum.

This figure illustrates some of the processes at work at the site of an ion’s impact with the surface of the sample. In most instruments the primary ion’s trajectory is not normal but at an angle relative to the sample surface. The impact initiates the collision cascade here illustrated by the web of lines of trajectory of the resulting moving particles. As has been noted before, the ion beam constitutes an ionizing radiation, so there will be charge separation within the solid, with points within the surface becoming positively charged and electrons set in motion. At the site of impact where all the collisions that take place have relatively high energy, there is shortly be nothing left of the original sample structure, and the species that are ejected from the surface will be atoms. At some distance from the impact site, molecular fragments are ejected. Further from the impact site, whole molecules (if there are any discrete molecular species present) are ejected from the surface. Through all of this chaos, there is a chance that some of the species leaving the surface are ionized. It is that fraction of the sputtered material from the surface that is analyzed in the SIMS experiment. Fortunately that fraction is often high enough to make SIMS amongst the most sensitive analytical methods available. With sufficient mass resolution that many SIMS instruments can achieve, the formulae of the ion fragments detected can be directly determined.

Clearly, compared with the other surface analysis methods described above, the complicated process by which the species are formed and are detected in the SIMS experiment sets this tool apart. The sputter process is complex, making it a challenge to understand the chances of any molecular fragment or whole molecule leaving the surface intact, let alone to understand the chances that one or other ionization mechanism will make an ion out of that species. On the other hand, the potential for the highest sensitivity and molecular specificity of all the surface analytical methods is too great to ignore.

One way to simplify things, if one’s main interest is in atomic species, is simply to keep sputtering (Stevie 2015). While it may be difficult at first to figure out how much of each element will be sputtered, the problem, in fact, resolves itself. Those elements that sputter most readily will be depleted from the surface. Those elements that sputter more poorly will become concentrated at the surface. Eventually a dynamic equilibrium is reached where the flux of each element being supplied to the altered surface from the bulk has to equal the flux of each element being ejected from the surface. In this one sense, the depth profile using SIMS ends up being simpler than those that are performed using techniques that measure the remaining altered surface (Auger and XPS, for example). One is still left with that pesky problem of ionization probabilities, but then that’s what standards are for!

In fact, the probability that an atom of any given element will be an ion when it is sputtered is a strong function, not just of the properties of that element (it’s electronegativity in particular), but of the surface from which it is ejected. The term “matrix effect” comes from imagining the sample as a matrix of atoms, the relative numbers and the relationships (bonding, repeating structure) between them; all of these factors influence ion yields. It is for this reason that the use of standards for SIMS is restricted to standards that are very similar to the sample being analyzed in both structure and composition.

The rise of dynamic SIMS as a popular analytical technique is closely linked to the semiconductor industry. The electrical properties of semiconductors are not only a function of their structures but also of the dopants that are used to tailor their properties. Dopants can be used at very low concentrations, and measuring dopant amounts and distributions as a function of depth requires the dynamic SIMS’s sensitivity. Since the “matrix” that is a semiconductor such as Si or Gallium Arsenide is relatively invariant, the process of producing and maintaining a system of standards is manageable.

Dynamic SIMS analyses may be performed on a variety of types of SIMS instruments, but instruments designed with mass filter type spectrometers are pretty much used only for dynamic SIMS. A mass filter allows only ions of a given mass (in reality ions with a given mass to charge ratio, but the charge is usually 1) to reach the detector. The spectrometer can, of course, be scanned, but while it allows one mass to reach its detectors, the rest are being thrown away. If you only need to hop between a few masses while performing a depth profile, this is not so great a loss of information. In fact, a magnetic sector based instrument will give you the highest sensitivities for dopant profiles.

Of course, the challenges of ever advancing technology never ceases, and so the field of dynamic SIMS is beset by samples with ever thinner layer structures so that it is difficult to sputter enough to reach the dynamic equilibrium described before one is already at the point where one needs to measure the dopant concentration. At the same time, the range of materials of interest has widened. Another common use of a form of Dynamic SIMS is for geological (Wiedenbeck 2010), biological, and environmental samples (Hu, Zhang, and He 2013). In these studies, it is not composition as a function of depth that is required, but elemental composition with high special resolution coupled with isotopic analysis. Special instruments designed with magnetic sectors having multiple detectors have been used to measure the signals from a number of nearby masses, and this is of obvious use for isotopic analyses since an element’s different isotopes are always close in mass. Further, when measuring isotopic ratios, one does not need to worry about matrix effects, since all the isotopes of a given element will experience exactly the same matrix effects simultaneously.

At the other extreme, if your interest is in true surface analysis, the damage the ion beam does is a real concern. As a practical matter, unless you are using gas cluster ion beams, damage to the sample leads very quickly to the loss of virtually all information about organic compounds. In addition, if you are damaging the sample in the course of the analysis, the results you obtain are liable to be less reproducible.

Fortunately, the detection systems in SIMS instruments can be highly efficient, so that it is often practical to keep the dose of primary ions low. In the limit of very low doses, clearly, the damage done to the sample is minimized. If the probability of an ion impact is so low that an insignificant portion of the data obtained results from ions striking an already damaged section of the sample, the results obtained will be the same as that of the next analysis performed on the same area with a comparable dose. This is the definition of Static SIMS. As a rough rule, static SIMS ion doses need to be below ~10¹² ions/cm². Obviously, if the surface is particularly ion beam sensitive, this Static SIMS Limit may be lower, and similarly, the use of cluster beams such as Bi₃⁺, which can damage the sample over a wider area than a simple monoatomic primary ion, may also bring the limit lower.

If the purpose of the analysis is to identify the species at the sample surface, passing the static SIMS limit may be a reasonable approach. The damage that accumulates at the surface has the effect of reducing the molecular and molecular fragment signals, but to capture all the signals possible, you can acquire data until the sample no longer produces the signals of interest. This is known as near static SIMS analysis. Fortunately, it is unusual for new signals to appear that result from the damage to the sample surface (although there are some notable examples of this occurring in polymer analysis).

The static SIMS limit and the limitations of analysis in the near static SIMS regime place a physical limit on the ability to produce useful data at high lateral resolution. To gain higher lateral resolutions you must reduce the area of the analysis. You still need the same number of primary ions to get the signal of interest, but by putting them into a smaller area of the surface, you are increasing the dose. So if you had the perfect ion beam with as small a spot as you could wish, you are still limited to how small a feature you can analyze. There are only two ways to improve this situation. One is to increase the ratio of the ion yield to sample damage cross section. The other is to eliminate the static SIMS limit by removing all of the damage you create. It turns out that cluster ions seem to do both of these things to varying degrees, although in so doing you sacrifice surface sensitivity.

For many years, the primary ion gun of choice has been the LMIG (Liquid Metal Ion Gun) due to its brightness and the ability to get small spots with it for imaging applications. Ga⁺ was the original species used with this ion source, and, in fact, is still the beam of choice used in Focused Ion Beam tools used for TEM and other sample preparation applications. It was clear, early on though, that the relative molecular fragment and molecular ion yield for this ion was poor relative to ion beams with higher mass projectiles. While Ar⁺ ion sources have never had decent spot sizes, the spectra from these heavier ions were clearly superior for organic analyses. Thus there has been an ongoing effort to find higher mass projectiles for the LMIG. After Ga, In was tried for a time, followed by Au, and finally Bi. Au and Bi sources produce a variety of ions, not just the atomic species, but also cluster ions. It was an immediate empirical observation that the cluster ions promote the formation of many molecular fragment and molecular ions. Au and Bi are similar in mass, but it turns out the proportion of Bi clusters emitted from a Bi source is higher than the proportion of Au clusters emitted from a Au source. In TOF-SIMS instruments, the most popular ion source in use today is the Bi source. With its combination of having a relatively high percentage of emissions from that source and producing excellent secondary ion yields, the most popular primary ion species in use is likely the Bi₃⁺ ion. The use of this species has greatly reduced the size of features that can be successfully analyzed before the damage is large enough to eliminate molecular signals.

The larger the cluster size, the lower the energy of impact of each of the atoms within each cluster, the greater the tendency for the damage produced during each impact to be removed with each impact (Mahoney 2013). Thus ion clusters as small as SF₅⁺ have been shown to be used well beyond the static SIMS limit and still produce spectra with molecular fragments and ions. At the other extreme, massive large ion cluster beams (GCIB) can be used with most organic samples to etch those samples with hardly an effect on the SIMS spectra except those caused by the change in composition within those samples as a function of depth. The use of these cluster ions thus allows complete consumption of the surface for analysis and permits analysis in depth that was formerly only possible with elemental analysis methods. Unfortunately, with their very low energy per atom, large cluster ion beams are not a form of ionizing radiation, and so many mechanisms for ion formation are eliminated when using these beams (Mahoney book). Furthermore, these ion beams are hard to focus, so imaging becomes problematic. Thus a very common analysis mode uses a GCIB to etch the sample coupled with a Bi₃⁺ beam for the higher resolution higher sensitivity analysis. The trade off, of course, is a loss of sesitivity brought about by the fact that when the GCIB is etching the sample no secondary ions are collected.