A first glance at a mass spectrum from a TOF-SIMS instrument can be bewildering. The Y-axis of the raw data is simply secondary ion counts. The X-axis is, in its original form, the time to digital converter (TDC) channel number, each channel representing a quantum of time set by the TDC bin size (e.g., 156 ps). The X axis is rarely plotted in units of time, however. Instead the channels are converted to m/e (mass to charge ratio) with a proper calibration. Secondary ions with the same charge (usually 1) have been given essentially the same energy; so given the fact that E = (1/2)m(x/t)², where E is the energy, m is the mass, x is the flight distance, and t is the time, secondary ions with different masses will have different arrival times and, therefore, will fall into different TDC channels. Solving for t one finds t = x(m/2E)^0.5. E and x are constant. The flight time t is therefore proportional to the square root of the mass. The higher the mass, the longer the flight times, but the relationship is not linear. The larger the mass, the fewer channels there are in each mass interval. Thus a restriction of the modern TDC with a set channel size is decreasing mass resolution with mass. Fortunately, in most situations the channel duration is small enough so that this has little practical impact on the results.

Unlike many other mass spectrometric methods, TOF-SIMS has as much as six orders of magnitude of dynamic range, so to see all the peaks in the spectrum, it is convenient to set the Y-axis to be the log of the secondary ion counts.

This plot reveals a peak at almost every integral mass. Blown up, a high-resolution spectrum will have many peaks at each integral mass. The most significant peaks are often not the most intense. Clearly there is a wealth of data here, but the task of interpreting it may seem daunting.

The reason there can be many peaks at each integral mass is that the elements of the periodic table have masses that are nearly but never quite integral (except for carbon, C), and the mass differences from the integral “nominal” mass varies from element to element. Therefore, species with different combinations of elements or different numbers of the same elements that turn out to have the same nominal mass are nonetheless distinguishable from each other when the analysis is performed with sufficient mass resolution. A high mass resolution measurement allows the determination of the empirical formulae for many of the peaks in the spectrum. This next figure shows an example of how a high mass resolution TOF-SIMS spectrum can reveal many peaks at a single nominal mass.

The sputter event can produce ions of either polarity. That is, ions will be produced with (usually a single) positive charge or with (again, usually a single) negative charge. In the commercial TOF-SIMS instruments currently available, you cannot detect ions of both polarities at once. The extraction field that gives the secondary ions most of their energy must be set at the start of each primary ion pulse, and depending on its polarity, will either accelerate positive ions away from the sample into the spectrometer while simultaneously driving any negative ions formed back into the sample or it will accelerate negative ions into the spectrometer and drive the positive ions back into the sample. In fact, the whole instrument is placed into either positive or negative ion mode, with most of the voltages on the ion optics of the spectrometer almost exactly reversed (switching from pulse to pulse puts an enormous strain off . system's electronics, and so such switching is rarely done within a given acquisition).

Neutral species, materials sputtered from the surface that are not ionized, comprise the majority of the sputtered material but they are not detected in TOF-SIMS. Many methods have been tried for post-ionization. For various reasons, none of these methods have achieved the success of the TOF-SIMS method itself.

The surfaces that have been and will be analyzed by TOF-SIMS are diverse, and no single work can capture all the spectral features one may find. Much interpretation is made possible by the fact that much is known about these surfaces before the analysis. Labs that specialize in the analysis of one type of sample or another will have expertise suited to the interpretation of the spectra from those types of samples. Industrial labs maintain private collections of standard spectra suited to their work. For any given sample set, the general guidelines given that follow will have to be supplemented with specialized knowledge.

In practice, in many (but not all) situations, much of the information of interest is obtained in positive ion mode. Positive ion spectra are more likely to contain molecular ions and large organic fragments. Positive ion spectra also have more parallels to results that are obtained by older, better-understood mass spectrometric methods.

The most common mass spectrum taken today still uses electrons to ionize neutral molecules that flow into the spectrometer in the gas phase (Gross 2004). The gas chromatograph mass spectrometer (GC-MS) separates out the different species previously injected in the chromatograph (differential affinities of different chemicals to the column slows their passage differentially) and then ionizes them with a beam of electrons. The impact of the electron knocks an electron out of the molecule (again, this is an ionizing radiation) sometimes fragmenting it, and then the resulting ions are subject to electric fields, which pull them into the spectrometer and ultimately determine their mass. Electron impact mass spectrometry (EI-MS) spectra are, therefore, positive ion spectra.

EI-MS is an old technique and it has been well studied. The ion formation mechanisms are reasonably well understood (McLafferty 1993) and there are enormous databases of spectra against which to match the unknown spectrum you might generate. It would be so much easier for the TOF-SIMS analyst if the study of EI-MS would reveal most of what you need to know to understand TOF-SIMS spectra. As you have already guessed, it does not.

One of the differences between EI-MS and TOF-SIMS are the different probabilities of post-ionization collisions. In the EI-MS, collisions are rare (unless intentionally produced) and so after ionization, the mechanisms that lead to the formation of any given ion are unimolecular. In the TOF-SIMS, the nascent ion begins life in contact with a surface. Collisions are almost inevitable.

Collisions tend to eliminate odd electron ions. While the EI-MS will produce the original molecule minus 1 electron, the radical cation M^+., after a collision in the TOF-SIMS, will snatch a H atom from the neighbor it collided with, producing the MH⁺ ion, which contains an even number of electrons. Interestingly, this can have the effect of stabilizing the resulting ion against fragmentation (that and the collision, which can have a cooling effect), so that in some cases the M+1 MH⁺ ion may be relatively more intense than the M^+. ion in the EI spectrum. On the other hand, the sputter event with typically a high energy 20 to 30 keV beam initiating the collision cascade will tend to produce ions with lots of internal energy. so fragmentation is often more pronounced and more random in the TOF-SIMS spectrum than in the EI spectrum (Spool 2004).

Because the EI-MS almost always has a chromatograph in front of it, the study of EI-MS has been aided by the fact that it is easy to obtain spectra of pure compounds. This is, unfortunately, not the case for TOF-SIMS. Most TOF-SIMS spectra are to some degree spectra of mixtures. Different molecules can have orders of magnitude differences in ion yield in the TOF-SIMS, so it is possible for trace contaminants to dominate the TOF-SIMS spectrum.

On the other hand, in order to do GC-MS, the sample has to be volatile enough to enter the gas phase. Many materials are nonvolatile. Polymers in particular are difficult to characterize by other mass spectrometric techniques, but can produce rich spectra in the TOF-SIMS (Briggs 2005). In fact this application, the analysis of polymers, was certainly a major impetus for the development of TOF-SIMS.

All these things make TOF-SIMS spectra decidedly different from EI spectra. This is not to say that there are not lessons that can be learned from the understanding of EI spectra or that having access to an EI-MS database is entirely useless to the TOF-SIMS analyst. It’s just that the use of EI-MS theory, practice, and data is not directly applicable to the interpretation of TOF-SIMS results.

The X-axis of a TOF-SIMS spectrum, as noted earlier, is initially in units of channel numbers, each channel recording a discrete window of time at the detector relative to the initial pulse of the primary ion beam. The Y-axis represents the number of secondary ions that reached the detector and were detected during that channel’s time window. In practice, the data is never viewed in this way. Instead, using a mass calibration, the channels are converted to m/e, given in units of amu (atomic mass units). On this scale, the ¹²C isotope has the exact mass 12 amu. H has a nominal mass (mass equal to the sum of the number of protons and neutrons in the ions atomic nuclei) of 1 amu, O may be found at 16 amu and so forth.

Except for ¹²C, though, none of the major isotopes of the elements have an exact mass that is the same as the nominal mass. F comes close with a single isotope at 18.998 amu. The minor isotope ¹⁵N also has an integral exact mass (15.000 amu). Only the stable isotopes of a few elements have mass excess, that is, an exact mass greater than the nominal mass. These are ¹H (99.989 percent abundant) at 1.008 amu, ²H (0.0115 percent) at 2.014 amu, ⁶Li (7.59 percent) at 6.015 amu, ⁷Li (92.41 percent) at 7.016 amu, ¹⁰B (19.9 percent) at 10.012 amu, ¹¹B (80.1 percent) at 11.009 amu, ¹³C (1.07 percent) at 13.003 amu, and ¹⁴N (99.632 percent) at 14.003 amu (“WebElements Periodic Table of the Elements” 2015). All the rest of the elements in the periodic table have some mass defect, that is, an exact mass less than the nominal mass. These differences from nominal mass may seem small, but they are readily resolvable with decent mass resolution.

Further, these differences from nominal mass add up. Fragments of organic species will often have many H atoms, so the total mass excess for the ions tends to grow with the size of the fragments. Similarly for most inorganic clusters, the mass defect for the ions will get larger as the size of the cluster increases. The exception for this last statement would be inorganic clusters with many B or Li atoms, but these will be unusual materials for most analysts, and it is unlikely that they will arrive inside the spectrometer as an unexpected guest. In this figure, an inorganic cluster (CCo⁺) is well separated from the other peaks due to its mass defect.

Many inorganic atomic species can be recognized from their isotope patterns.

At first glance, the TOF-SIMS has perfectly reproduced the expected isotope pattern, but close inspection shows anomalies. Note in particular, the peak at 103 amu with some mass defect. There is no ¹⁰³Ru isotope. In fact this peak is due to a signal for ¹⁰²RuH⁺, the hydride ion. There are equivalent levels of RuH⁺ ions for all of the Ru isotopes, distorting the isotope pattern for mass spectra taken with mass resolutions insufficient to separate the hydride ions from the Ru signals. Some level of hydride is found in many inorganic isotope patterns.

If you know the empirical formula of an ion, you can tell if it is an even or an odd electron ion. For example, C has atomic number six, so as a neutral species, it has six electrons. When this atom is transformed into a singly charged positive ion, it loses one electron and is then left with five, an odd number. If you add a hydrogen atom to make the CH+ ion, the hydrogen atom has brought one electron with it, and there are now six electrons in the ion, making it an even electron species.

Nature has made the determination of whether an organic ion has an even or odd number of electrons easy. This is due to the following circumstances:

• Almost all atoms in organic molecules that form chains have an even nominal mass and form an even number of bonds. Atoms that fall into this category are C, O, Si and S. O and S can be attached atoms but when they are, they form double bonds.
• Almost all atoms in organic molecules attach to chains because all they can do is form single bonds, have an odd nominal mass, and form an odd (single) number of bonds. Atoms that fall into this category are H, F, Cl, Br and I.

When an organic ion formed from the elements listed in the previous list has an even nominal mass, the ion has an odd number of electrons. An example of this is the CH₂⁺ ion. It has an even nominal mass (12 from the C plus one each from each H equals 14). It also has six electrons from the C atom, one each from the H atoms, but has lost one electron when it became a singly charged ion, leaving it with an odd seven electrons.

Conversely, when an organic ion formed from these elements has an odd nominal mass, the ion has an even number of electrons. The CH₃⁺ ion at an odd nominal mass of 15 amu is an example. It starts with C and its six electrons plus one from each of the three attached H atoms, minus the one that was lost when it became a singly charged ion, for a total of an even eight electrons.

The major exception to all this is N. It is an atom that can be present within a chain, has an even mass, but it forms an odd number of bonds (three). When it attaches to the end of a chain, it still forms an odd number of bonds. An even number of N atoms contributes an even number of bonds, so as long as there is an even number of N atoms in the ion, the rule described here stays the same. An odd mass ion has an even number of electrons, and an even mass ion has an odd number of electrons. But when you have an odd number of N atoms in the ion, the rule reverses. Odd mass ions will have odd numbers of electrons, and even mass ions will have even numbers of electrons. It is less usual to find P in organic molecules and fragments. P has the same bonding characteristics but not the same effect, since its primary isotope has an odd mass (31 amu).

When there is little to no N on the surface, the even or odd electron nature of the ions in the spectrum is easy to tell. If your spectrum has sufficient mass resolution, it is possible to determine the empirical formulae of many of the low mass fragments from their exact masses, and from this you can determine if there is a significant presence of N at the sample surface. Because odd electron ions signals tend to be less intense, the prominence of peaks with even masses is an immediate sign that N is likely to be present. The following two figures demonstrate the effect the absence or presence of N has on TOF-SIMS spectra.

In the stearic acid spectrum, the main fragments have odd masses. In the spectrum of stearamide, the major fragment ions are even in mass.

The significance of prominent odd electron ions in TOF-SIMS spectra, which are otherwise dominated by even electron ions, is discussed as it pertains to ion formation mechanisms.

Atomic ions are limited to the first 225 amu of the spectrum, simply because pretty much all the stable isotopes that exist have mass less than 225. Clearly, if you want to determine the empirical formulae of larger fragments in the spectrum where mass resolution alone is not sufficient, you want to know what elements are present at the sample surface. Finding atomic ions at low masses, where there is less interference from cluster ions, is easy but above 60 amu or so it is more difficult. One is aided by the fact that many of the elements have significant abundances of multiple stable isotopes, so the isotope patterns that will be found in the spectrum definitively identify those elements. The patterns can also aid in distinguishing the presence of inorganic clusters from atomic species in the spectra, since those clusters will often have combined isotopic patterns of their own. The intensity of an atomic ion, relative to the surface concentration of its elemental source, is a strong function of the element’s electronegativity. Thus, it is no surprise that trace levels of alkali elements can produce quite strong signals, whereas high concentrations of halides will produce only weak signals in the TOF-SIMS positive ion spectra. Fortunately, those elements most difficult to find in the positive ion mode can be found in the negative ion mode. The most problematic elements for TOF-SIMS are the ones with intermediate electronegativities.

With the rare exception of species containing multiple B or Li atoms, organic and inorganic ions are well separated in the first 250 amu of the spectrum. Organic fragments (containing C, H, O, and N, but not enough O to take the fragment in mass defect territory) have mass excess. Fluorocarbon fragments will have almost nominal mass. Inorganic materials will produce ions with mass defects.

Organic ions in the first 225 amu are generally fragments of larger molecules. There are exceptions, but to a large degree, molecular species with masses less than 200 amu are almost always too volatile to stay present at the surface within the TOF-SIMS vacuum. There are more exceptions to this rule between 200 and 250 amu (Myristic acid, the 14 carbon long saturated fatty acid with a major peak M+1 at 229 amu comes to mind), but still, most ions with mass excess that will be found in this range will be fragments and not molecular ions. This will obviously be less true if a cold stage is used to reduce the loss of volatile species.

Salts with organic cations will generally produce the expected ion in the spectrum. Strictly speaking, these are fragments of the whole molecule, so the rule that the spectrum below 225 amu contains mostly fragments is not violated, but it is certainly true that much smaller cations that make up the bulk of the mass of such salts can have masses well below 225 amu. The signals from these salts can be surprisingly weak. Even though the cation is preformed at the surface, the sputtering of the intact ion remains a problem. Ionic attraction in salts is stronger than van der Waals forces or even hydrogen bonding, so it may take a fair amount of energy to send a portion of a salt into the vacuum. More energy is present closer to the impact site, but close to the impact, damage to the molecule is also much more likely.

Within this low mass range, and indeed up to much higher masses, for most samples there will typically be a peak at almost every nominal mass. The intensities of many of these peaks will be much smaller than a few prominent nearby peaks, so at least in the initial review of a spectrum, these signals will be difficult to assess. Some of these may come from species that are minority contaminants of the surface. Some may be unusual ions (odd electron, or due to rare collision events or seldom seen ion–molecule interactions). A few may be of significance to the analyst, but determining which of these are may require multivariate statistical analysis.

For most polymers, characteristic ions for the polymer will be found in this, lower, mass range (Briggs 2005). The monomer fragment is usually prominent.

Surfaces of low surface energy polymers such as PTFE (Polytetrafluoroethylene, commonly known as Teflon) or silicone rubbers will not adsorb adventitious hydrocarbons and will show clean spectra with peaks characteristic of each polymer, but most surfaces will have some hydrocarbon character. Some materials will simply pick up hydrocarbons and species with hydrocarbon portions. Others will already contain hydrocarbon chains or be hydrocarbon-based polymers. In all of these cases there will be what may be referred to as the “hydrocarbon envelope” in the first portion of the spectrum.

Surfaces with saturated or mostly saturated hydrocarbon portions will have a series of peak clusters separated by approximately 14 amu (the mass of the -CH₂- unit) starting with one carbon (peaks at 12, 13, 14, and 15 amu). Within each cluster, the even electron ions will be more intense than their odd electron neighbors. For the one C clusters, this means that the 13 amu (CH⁺) and 15 amu (CH₃⁺) ions will be more intense than their odd electron neighbor at 14 amu (CH₂⁺).

The shape of the hydrocarbon envelope will be affected by both the primary ion beam and by the nature of the sample. As the primary ion becomes heavier, and especially when small clusters such as Bi₃⁺ are used instead of monoatomic ions, the average mass of the peaks in the hydrocarbon envelope shifts toward higher mass. Within each cluster of ions there will also be some shift toward the more saturated ions with heavier primary ions. Nonetheless, the shape of the envelope for a given material will tend to be recognizable, and thus spectral library entries obtained using different primary ions will still be of some use in identifying the material. The figure below shows an example of a hydrocarbon envelope as produced by three different primary ions, showing the trends described above.

The simplest hydrocarbon envelope, and the one most similar to what one generally sees from adventitious hydrocarbons, is that one sees in a spectrum of Polyethylene.

If the materials at the surface contain some prominent side groups, end groups or is composed of a polymer with a unique repeat unit, the hydrocarbon envelope will show that feature prominently.

The following spectrum of polyisobutylene shows not only the prominent butyl signal but other structure specific ions. Note the presence of the unusually intense odd electron radical cation C₄H₈^+. at 56 amu.

As you start adding unassociated double bonds into the structure, the hydrocarbon envelope shifts such that less saturated ions become relatively more intense.

Once the double bonds in a structure become conjugated, and especially if the material has aromatic rings, ions with ring structures come to dominate the spectra. In the spectrum of polystyrene, C₇H₇⁺ produces the most intense signal despite the fact that this is a arrangement fragment from the initial structure.

Additional detailed information about the sample can be deduced from the hydrocarbon envelope, but only with the use of standard spectra taken under substantially similar experimental conditions. Although hydrocarbon envelope patterns may look similar from sample to sample, the relative ratios of the different peaks can be used to determine information about polymer blends (if the surface is polymeric), the molecular weight (MW), sequence distributions, and unsaturation (Galuska 1997) given a sufficient supply of the appropriate standards.

Of course, most samples contain more than hydrocarbon-based materials, and the low mass region of the spectrum will reflect this. Often, though, extra peaks indicative of these interesting species will be nestled among peaks recognizable as part of a typical hydrocarbon envelope.

As was noted above, inorganic species will have mass defects, organic species mass excess, and fluorocarbons will have nearly nominal mass.

Above 225 amu, there are no atomic species to be found amongst the peaks with mass defects. All peaks with mass defect (which have less than their nominal mass) will be ion clusters, usually entirely inorganic in nature. The exception to this last is in the presence of organometallic cluster ions, which may have more mass defect provided by the inorganic species than mass excess provided by the H present in the organic portion of the ion.

It is a handy rule to keep in mind that a fully saturated hydrocarbon-based molecule, which will have the maximum amount of hydrogen relative to its mass that is possible for any species, will have at most ~0.1 percent of its nominal mass as mass excess. That means a hydrocarbon dominated species with a nominal mass around 250 amu will have an exact mass of no more than 250.25 amu. Add in O atoms or unsaturation and the mass excess of the ion will decrease accordingly.

Above 225 amu, the slight mass defect of the F atom begins to become noticeable in fluorocarbon-based ions. These larger fragments will now have a slight but quite measureable mass defect. The addition of O into the fragments (e.g., in the case of the perfluoropolyethers used as synthetic oils and lubricants) will increase the mass defect of these fragments, but they will still tend to be much closer to the nominal mass than most inorganic clusters.

In this mass range many discrete compounds will have their molecular ions. Most often these ions will be found at mass M+1 as the odd electron molecular ion will most often snag a hydrogen atom or will have been protonated in the original ionization event. The exception will be ions that have a lot of unsaturation in which the delocalized lone electron is less reactive and hydrogen extraction is less probable. The sputter event and the ensuing collision cascade give molecular ions enough energy to leave the surface, but if too much energy is imparted, the ion will fragment. It makes sense that the more energy it takes to remove a molecule from the surface, the less likely it is that the molecule will remain intact. Also, the larger the molecule, the more the attachment points (Van der Waals attraction or hydrogen bonding) that will be present to increase the energy needed for desorption. This is why molecular ion intensities tend to decrease with increasing mass. The most prominent molecular ions will often be found in this mass range.

Fatty acids and their derivatives are particularly visible in TOF-SIMS spectra when they are present. The majority of these molecules have long saturated hydrocarbon chains which cannot hydrogen bond, and therefore can only weakly attach to the surface. Fatty acids that commonly appear in this region of the spectrum include palmitic acid with 16 carbons, stearic acid with 18, and arachadic acid with 20. The acid, ester or amide functionality in the molecules is much more readily ionizable than a hydrocarbon, greatly enhancing ion yield. Unfortunately, large molecules with fatty acid appendages will often produce fatty acid ion fragments that are identical to fatty acid molecular ions. Most, but not all of these derivatives, will have distinct fragments of their own at a higher mass.

Fatty acid derivatives, like many discrete molecular species, commonly produce a series of peaks in the TOF-SIMS spectrum with a consistent mass difference. For fatty acid derivatives that difference is 28 amu, the mass of the C₂H₄ unit. This is not the result of an ion fragmentation pattern. Instead, this is due to the fact that these compounds are rarely pure, and they contain hydrocarbon chains that vary in length. Since the origin of fatty acids is typically biological, they tend to have an even number of carbons, and so the common difference between the different species in the mixture is C₂H₄. Such related species are known as homologues. Note that the appearance of homologues in TOF-SIMS spectra is another example of a difference between these spectra and those that can be obtained from GC-MS. The GC separates homologues, so they will not appear together in the same GC-MS spectrum as they do in the SIMS.

Of course, you can have a series of ions formed via fragmentation with a consistent mass difference. Such fragmentation typically produces a series of peaks at low mass that steadily declines in intensity as the series progresses to a higher mass. This is also true for many inorganic cluster series as well. In contrast, homologues can be found as a series of peaks first noticeable at a higher mass that rises with repetition to even higher masses and then falls again. The series of peaks is representative of the original distribution of homologues in the mixture at the surface. However, as noted above, the larger the molecule, the lower the probability that a parent ion of that molecule will be detected. The distribution as measured in the TOF-SIMS will tend to be somewhat skewed, favoring lower mass homologues. If quantification of such distributions is needed, it will be necessary to calibrate the measurement with known distributions.

Surfactants tend to produce beautiful homologue series in TOF-SIMS spectra. Many of these are based on polyethylene oxide, and so their homologues differ in mass by C2H4O, 44 amu. This figure shows a spectrum of polypropylene glycol with homologues that differ by C₃H₆O, 58 amu. Ionic surfactants in particular can be hard to completely remove from surfaces, so they may often be found as contaminants on TOF-SIMS analyzed samples.

Many of the common polymer additives will have either molecular ions or major fragments in this part of the spectrum. The fragments in this mass range will often be more intense than the more definitive molecular ions that may be found at higher mass.

Looking only at the spectrum above 500 amu can become confusing. Highly saturated organic ions will have a mass excess of greater than 0.5 amu, and inorganic cluster ions can have a mass defect greater than 0.5 amu. From 500 amu onward, it is no longer obvious what is organic and what is inorganic. Usually, though, inorganic cluster series begin at lower masses. The elements present can generally be established through examination of the low mass portion of the spectrum. Consequently, the other portions of the spectrum can inform the analyst sufficiently for inorganic clusters to be distinguishable from organic ions.

Above 500 amu, spectra tend to get sparser. Lower MW hydrocarbons can produce a series of peaks between 400 and 600 amu of low intensity in which it may be difficult to establish a pattern. There may also be low intensity peaks at almost every mass trailing off into the higher mass regions. As a consequence of the relatively sparser appearance of the higher mass regions, even relatively low intensity peaks may appear prominent at higher masses. In fact, these peaks may be more important to an understanding of the sample than far more intense peaks that are nonetheless dwarfed by their neighbors at lower masses. This is because they are either large fragments of even larger molecules, or they themselves are molecular ions, directly indicating the MW of a species on the sample surface.

There are still many polymer additives with molecular ions above 500 amu. Some of these may appear in the spectrum just like they appear in their standard spectra. Some, however, are antioxidants, and they may act in this capacity by oxidizing selectively. Thus instead of the pure compound to be found in the standard spectrum, the spectrum may reveal a mixture of the antioxidant and its O reacted product. For example, a peak at M+17 (M+OH) often accompanies the M+1 molecular ion for Irgafos 168.

At a higher mass, organic species accumulate a higher percentage of molecules with relatively rare ¹³C and D (²H) isotopes. The isotope pattern for organic species will become more pronounced. While it is not very common to get prominent peaks in the TOF-SIMS spectrum above 1,500 amu or so, select samples may have many peaks at higher masses. At these masses, the isotope pattern will become pronounced.

Whatever the mechanism that drives ion formation during the sputter event, it is a fact that in the TOF-SIMS, many ions consist of a molecule attached to a cation. Indeed, the M+1 ions discussed earlier can be thought of in that way, with the cation being a proton. Primary ions using clusters of water molecules have been used to enhance this form of ionization (Sheraz et al. 2015). The intention is to enhance the yield of ions that are recognizable as containing the entirety of the molecules of interest at the surface, and thus to know their MW (Hagenhoff 2013).

Ag surfaces are an excellent source of cationized molecular species (Delcorte and Bertrand 1998). To work well, they are etched (which roughens the surface and removes the oxide) before the sample is prepared. The distribution of molecular species (M+Ag) that are captured in the spectra suggests direct measurement of polymer MW distributions, although for the reasons mentioned earlier, these will tend to be skewed to lower MWs and will need to be calibrated for accuracy.

However, the fact that for this work, samples require preparation defeats one of the main strengths of the TOF-SIMS, that it is a method for the direct observation of surfaces as they are, rather than as prepared for analysis. If you have enough material to manipulate, there are often better analysis options.

Alkali elements will also cationize organic molecules. There are several situations where this may be a natural product of an as-received surface. Biological samples often have high concentrations of Na and K, and thus can produce cationized species in the TOF-SIMS. The surfaces of some glasses also have high concentrations of alkali elements, and adsorbed organics are often cationized in the sputter event. In both cases it is worth remembering that the mass difference between the alkali elements Li, Na, and K is 16 amu. Finding peaks with consistent 16 amu differences in the middle and high mass regions of the TOF-SIMS spectrum in the presence of multiple alkali elements is a strong clue that should prompt you to suspect cationization.

As of this writing there are three TOF-SIMS spectral libraries available to the analyst. Two of these are provided by each of the instrument manufacturers. The last, the Static SIMS Library is independently available for purchase (“Static SIMS Library” 2015). All of these databases are remarkably small (hundreds of samples total). The publication “Surface Science Spectra” is now accepting TOF-SIMS spectra, but the numbers of entries is also small, especially relative to other mass spectral databases. The entries include many of the most pervasive materials, so when attempting to identify an unknown it is worth searching these databases. Unfortunately, finding a match for the spectrum of an unknown in this small set of spectra will be the exception rather than the rule.

In contrast, there are more than 240,000 spectra in the NIST/EPA/NIH Mass Spectral Library (EI), and it is not the only EI-MS database available. The search capabilities, (e.g., pattern matching) are also impressive. Unfortunately, EI-MS is sufficiently different from TOF-SIMS so that the EI-MS databases are not directly useful. However, given a high mass ion in the TOF-SIMS spectrum that you suspect to be a molecular ion, you can attempt to find some possible matches in an EI-MS database. If the TOF-SIMS ion is an odd electron ion (seen for highly unsaturated molecules) you can use its mass directly in a search. If the ion is an even electron ion, be sure to subtract (or in rare cases, add) the mass of a H atom before doing a search. If you have some decent mass resolution, you might narrow your search sufficiently to reduce the possible matches to a reasonably number. With some molecular structures in mind, you can then proceed to decide if the ion fragments seen in the spectrum make more sense with one or another of them.

It is surprisingly useful to simply make sure that the distinct peaks that you find in your spectra get named in reports that you can later text search. The reports of most modern laboratories are readily accessible using a simple Boolean logic search. You will find that your own experiences with both knowns and unknowns form a remarkably rich data set. Even if you can’t identify peaks in a spectrum, keeping track of when and where you have seen them before can be remarkably helpful.

Ultimately, the cure for TOF-SIMS signals from unknown sources is the MS/MS method now available as an add on in two of the commercially available TOF-SIMS instruments.

With a simple setting change, a TOF-SIMS instrument can measure negatively charged ions instead of positive. It’s so easy that you would think it would be done more often. It probably should. As a practical matter, there are reasons why negative ion spectra are often omitted from an analysis.

• Much of the work needed in one mode needs to be repeated for the second. Obtaining both positive and negative ion data takes just a little short of twice the time.
• If one has a limited sample (usually because the size of the feature of interest is small), then analyzing both positive ions and negative ions means splitting the possible dose before the static SIMS limit (or if you are not so picky, before the sample is damaged beyond recognition). You may want to save your limited signal-to-noise for one data set.
• For many types of samples, (e.g., hydrocarbon-based polymers) the negative ion spectra can be quite uninformative. Instead of the hydrocarbon envelope, one may see a much weaker set of signals including atomic species and small fragments.

However, there are many electronegative elements and anionic species that can only be detected in the negative ion mode. Ions that are mysterious in positive ion mode may contain heretofore unsuspected elements that become apparent when the negative ion spectra are obtained. There are also certain materials for which the negative ion spectra are richer in information than positive ion spectra.

The information to be found in the mass excess or defect an ion has is the same for negative ions as for positive. Because a negative ion is more likely to lose a H atom than to pick one up and also because it will be more likely to produce an intense signal in negative ion mode if it has heteroatoms such as O or lots of unsaturation, organic fragments in the negative ion mode tend to have less mass excess than those generally found in the positive ion mode.

As in positive ion spectra, atomic species are, for the same reasons, only present at lower masses. In the reverse of the positive ion mode, halides will produce the strongest signals relative to their concentrations, while alkalis will not be detectable. Because of oxygen’s electronegativity, clusters of oxides are generally more pronounced in the negative ion mode.

Odd and even electron species follow the same rules in the negative ion mode as in the positive. As before, odd electron species tend to be rare (although this is less true for per fluorinated species). To the extent that molecular ion species are present, they tend to be present as M−1 even electron ions rather than as the odd electron ion with the same mass as the molecule. As in the positive ion mode, the exception is in cases where the odd electron molecular ion will be well stabilized, usually in highly unsaturated molecules.

One major reason analysts turn to negative ion spectra is to search for counter ions. Positive ion spectra may have revealed an alkali element or a tetra-alkyl ammonium cation, and for a complete picture, they take the negative ion data to determine the nature of the inevitable anion. Halide, sulfide, and oxide anions of all types (carbonate, nitrogen oxide anions, oxidized S, oxidized P, etc.) are readily detected. Unfortunately, the interpretation of the oxide anion spectra is not as straightforward as one might hope. This is because in addition to the anion actually present, fragments and collision products of the anion will also be present. So, for example, surfaces with sulfates and sulfites will display the same peaks, with the major difference being the relative ratios of those peaks.

Standard spectra may be needed to identify which is which. If the sample is amenable to X-ray photoelectron spectroscopy (XPS) analysis, this is a good situation in which a combination of both the methods of analysis may be particularly helpful. The XPS is usually quite capable of determining the compound in which the oxidized element is to be found. Another problem with the various oxide anions is that their intensities can be wildly different in the TOF-SIMS. For example, the TOF-SIMS is quite sensitive to sulfide and sulfate anions, but some of the intermediate S oxide anions produce very weak spectra. Sulfide and sulfate can be detected in the TOF-SIMS at levels beneath the detection limits of Auger and XPS, but some of the intermediate oxides have been readily detected by Auger and XPS, whereas they were not seen in the TOF-SIMS spectra.

Often the search for a counter ion is fruitful, other times not. Surfaces can themselves hold a charge, and so cations may be present without there being an obvious anion. For example, surfaces that are washed with an ionic surfactant will often retain a readily detectable trace of the surfactant that is visible in positive ion mode, but analysis does not show the anionic portion of the surfactant. It is also true that the TOF-SIMS sensitivity for alkali elements is so extreme that these may be easily detected when the anion cannot be found. Finally, OH⁻ is a common fragment in the negative ion spectra of almost any surface containing O. It would be impossible to tell if this were the anionic partner of a cation detected in the positive ion mode.

The CN⁻ ion is another example of a negative ion whose interpretation is not as direct as one might hope. While certainly one will find CN⁻ ions in the negative ion spectra of samples containing cyanide or samples with organic species containing nitrile functionality, it is also detected from a wide variety of other materials. In short, it will be found, and usually will be found to be quite intense, in the spectra of any sample of a species having N bound to C in any kind of structure. Substances such as sputtered carbon with added N and discrete organic molecules containing N will produce spectra with the CN⁻ ion. This can be useful if, for example, there is doubt as to how to interpret a positive ion spectrum with a few even mass ions. To prove that N is present in the organic species on the surface of this sample, simply look for CN⁻ in the negative ion spectrum.

Inorganic oxides can produce quite remarkable negative ion spectra in the TOF-SIMS, with the surface fragment series extending to quite a high mass. The exact masses of the lower mass fragments in these series coupled with the mass differences between the clusters allow ready identification of the formulae for the ions in these series. The example shown below has characteristic 43 amu (AlO) mass differences between the peaks in the series.

As for polymers, the ratios of peaks in these series tell about the materials from which they originate. However, the meaning of the peak ratios of inorganic oxide clusters is less clear in many cases. Polymers often consist of chains, but inorganic oxides are 3D structures. The absence of defects in these structures tends to suppress the formation of high mass inorganic oxide cluster ions. Passing the static SIMS limit and thereby damaging the structure can actually enhance inorganic cluster ion yields.

For every ion found in a TOF-SIMS spectrum, in addition to the time of arrival at the detector, the instrument will keep track of the X and Y coordinates within the raster of the ion beam. The instrument will also keep track of the sequence of the ions detected though this information is not important when keeping within the static SIMS limit. For ion beams with large spots, the beam will be moved across the surface during analysis to ensure even exposure of the surface being analyzed to the beam, but for smaller spots, the instrument will generate meaningful images. In principle, one could generate an ion image for every channel in the spectrum, but generally it is more meaningful to bin channels that correspond to a single ion species, or where necessary, multiple ion species whose signals cannot be resolved.

In most cases, an ion image is easy to understand as pixels with higher ion counts correspond to locations on the surface with higher concentrations of the species from which the ion originates. An ion image is often depicted in a thermal scale, with black pixels having no ions, white representing the maximum count of ions found in a single pixel within the image, and a gradient from black to dark red to light red to dark yellow to light yellow to white displaying visually the continuum of ion counts in the pixels. The following figure shows examples of ion images displayed using the thermal scale described above.

The ion images are less commonly represented as grey scale images (possibly because they are then likely to be confused with secondary electron microscope (SEM) images that are usually displayed in grey scale). So, then, an area of the image with a cluster of white pixels represents an area with a higher concentration of the species represented by the imaged ion rather than a surrounding darker region. It is natural to interpret the visual scale as representative of concentration.

There are numerous exceptions to the rule that ion count varies proportionately or even in the same direction as concentration. This is because of the matrix effects previously mentioned. If the surface is sufficiently inhomogeneous with different regions having drastically different chemistries, ion intensities can change across the surface with no change in the originating species’ concentration. In some cases, ion intensities can be higher in locations where the actual concentration of the material on the surface is lower.

A common situation in which the interpretation of ion images is counterintuitive is the case in which there are trace contaminants on a surface with varying base substrate materials. A wafer with gold pads is a good example. Ion yields on the Au surface will be strongly enhanced relative to the yields obtained from the surrounding wafer when that wafer is coated with a much lower atomic number species such as Si, silica, or alumina. Traces of the surrounding wafer material (Si or Al in this example) present on the Au pad may be more intense than the same atomic signals obtained from the surrounding area where the element is a major component. Trace organic contaminants present everywhere on the wafer will produce much more intense ion signals on the Au pad. The higher atomic number of Au keeps the collision cascade close to the surface, enhancing ion yields. The Au surface is relatively noble, so bonding to the surface is typically weak, making desorption of nascent ions easier. Finally, Au does adsorb a thin oxygen layer from the air, which acts as a decent barrier to electron tunneling, a mechanism that destroys more secondary ions than any other. As a consequence, Au is an ideal substrate for ion for producing high ion yields; often orders of magnitude higher ion yields than will be produced from surrounding materials. For these types of samples, ion intensity definitely cannot be equated with concentration.

One of the principal responsibilities of the SIMS analyst is to have a healthy skepticism of simple interpretations of ion images. In most laboratories there will be types of samples that are generally introduced into the instrument. Sometimes divergence from the maxim “ion intensity is proportional to concentration” will be expected from first principles (as in the case with Au pads). In other cases, the analyst knows that the regions of the sample have completely different chemistries, but the effect this will have on ion yields is not obvious. In many of these situations, it may be worth checking the SIMS result against those from another technique, even if unusual samples with larger areas must be produced so that methods having poorer lateral resolution can be employed.

TOF-SIMS data consists fundamentally of the following information: x, y, t (time of arrival at the detector after the primary ion pulse), and the point during the analysis at which the secondary ion was detected (the raster or analysis cycle). All of this is stored in what is known as the raw data file. Sort all of the data into time channels, mass calibrate the data, and you have the total spectrum for the analysis. Sort all the ions into their respective pixels, and you have the total ion image. Choose a subset of pixels, and sort only the ions that were detected when the ion beam was pointed at those locations in their respective time channels, and after calibration you have a spectrum that corresponds only to that region of interest (ROI). Choose a subset of time channels that correspond to a peak in the spectrum, and you may produce a single ion image.

The use of the raw data to explore regions of the image or to produce ion images from any ion detected during the analysis is a powerful data analysis technique. The following sequence, utilizing the raw data, is a common one employed by analysts to answer that age-old question ”what is that?”

1. Acquire data from an area that includes the feature of interest.

2. Examine the total area spectrum, looking for peaks of interest. Define these as peaks in the instrument software. Alternatively, you can have the instrument software automatically identify a peak list. Another approach is to use a master peak list that you typically see for this type of sample (if this sample is a common variant on samples typically analyzed in you laboratory).

3. Sort the defined peaks into separate images using the raw data.

4. Examine the ion images to determine, where in the rastered area, is an area of particular interest. It may be a physical feature discernable in the total ion image. Often, it is an area with specific chemistry of interest defined by one or more ion images.

5. Define an ROI. You can do this by literally drawing a shape on an ion image. Or, you can take an “intensity slice” of an image. You may be interested only in pixels that have a minimum intensity of a peak of interest. Or you may be more interested in an area from which a species is excluded. The instrument software will allow you to define your ROI as needed.

6. Replay the raw data again if needed, this time looking only to acquire the spectrum from your ROI (or in some cases, multiple ROIs).

7. Examine the spectrum (or spectra) obtained. Now you can have a better idea of peaks that are only associated with your ROI(s). The relative intensities of peaks will be clearer.

8. Optionally, you can replay the raw data, getting ion images for those peaks that you now find significant, but had not before, based on the ROI spectra.

It is a drawback for the TOF-SIMS that there is no separation before analysis, so unlike in the GC-MS tool, mixtures of materials are the norm. The image data makes up for some of that lack of chromatography. Lateral inhomogeneity in mixtures of materials at surfaces is also the norm. The use of raw data allows the analyst to make full use of the variability that will be found across the sample surface.

With the use of atomic and small cluster primary ions, beyond the static SIMS limit one generally found damaged samples from which only ion beam altered surfaces could be measured. The advent of cluster ion beam sources (Mahoney 2013), and in particular massive Ar cluster ion beams, has made possible in depth analyses where the nature of the organic material is largely preserved for further analysis. These ion beams, with low amounts of energy per incident atom and lower velocities, are known to sputter the surface in a completely different way from atomic ions. Instead of the collision cascade, a very different process takes place, one in which the incident atoms do not penetrate the sample very far, in which there are cooperative movements that can propel surface species off the surface intact, and in which very little subsurface damage is produced. The high sputter yields of these clusters efficiently remove the damaged surface before it can build up. It is also quite possible that some of the damaged material is volatile and simply pumps away. In any event, XPS analyses of Ar cluster sputtered organic materials show very little change from that analyzed before sputtering begins.

The large cluster ion sources (C₆₀, massive Ar) produce very different mass spectra from those produced by smaller primary ions. There is generally less fragmentation, and a sometimes very different pattern of peaks is found in the spectra. Unfortunately, the ion beams are generally much less well focused (especially for the massive Ar clusters, which are also difficult to focus in time, making for poor mass resolution). Using these beams alone, one is generally limited to obtaining molecular depth profiles, and at best limited 3D imaging. For this reason, these beams are often used in conjunction with the liquid metal ion source (LMIS) beams that are commonly used in static SIMS 2D analyses as the sputter gun.

There are multiple ways in which a given cluster ion source, used in conjunction with a liquid metal ion gun (LMIG) primary ion source, can produce 3D images. The first is simply to analyze the surface while etching the sample, getting TOF-SIMS data for each successive layer. Generally, one will be analyzing, in these cases, an area of the surface much wider than the depth of the sputter profile. A second method is to use a focused ion beam (FIB) to etch the sample, producing a cross section. You then etch further (“polishing” the surface) using your cluster ion source in order to remove the damage created on the surface of the exposed cross section by the FIB. This is followed by TOF-SIMS analysis of the exposed surface, usually with an LMIG ion source. These steps are repeated until a full 3D data set is obtained, a kind of tomography. This set will have a much lower aspect ratio than that obtained using direct profiling. In either case, you are doing what amounts to a normal TOF-SIMS analysis on surfaces exposed by the cluster source.

Exciting applications of 3D imaging of organic materials, in particular, include the analysis of biological systems from tissue samples to cells. For this work, analysts are constantly seeking high sensitivities, since the number of actual analyte molecules in the small volume of a voxel at the lateral and depth resolutions being attempted is small. TOF-SIMS has the advantage over most competing techniques in that the sample preparation involves getting the sample into a shape and size and condition (ex. frozen) that is vacuum ready, but that chemical treatments are not necessary. It has the disadvantage, mentioned earlier, of orders of magnitude variation in sensitivities for different organic species. The TOF-SIMS has proven its worth in the analysis of lipids and a variety of small metabolites. It has not done well in the analysis of proteins. Experiments with novel ion beams intended to maximize ion yields, especially in biological systems, show that protonation enhanced ion formation is possible (Sheraz et al. 2015).

3D imaging can compound sensitivity problems. Many 3D analyses involve analysis with an LMIG for the lateral and mass resolution it enables, mixed with sputtering using an Ar cluster source. However, the material sputtered with the Ar cluster source is not analyzed. Unfortunately, the LMIG will quickly damage the sample. In organic depth profiling most of the sputtering must be done with the Ar clusters. In a biological system in particular, this can be problematic.

To solve issues of loss of materials during profiling, researchers have developed new TOF-SIMS instruments designed for the analysis of DC sputtered materials.

3D imaging is, of course, subject to the same issues as 2D imaging in the TOF-SIMS. It is no surprise that matrix effects can cause molecular depth profiles and 3D images to fail to produce direct information on species concentrations. The attempts to do 3D imaging have brought some of these issues to the fore. In particular, it has been noted that mixing compounds of different acidities or basicities leads to very nonlinear responses versus concentration. Aside from the possible issues with protonation leading to M+1 ion formation, salt formation changes the ease with which species will desorb intact with sputtering.

Instrument software will generally show 3D images assuming that the surface is flat. If the sample has significant topography, this will obviously lead to distortions in the 3D images. One instrument manufacturer has a built in AFM option and accompanying software precisely to meet this challenge. Analysis of the topography before analysis can be used to adjust the results.

Another problem, one that is well understood for depth profiling inorganic materials, is that profiling a mixed material surface where the different materials have different sputter rates will lead to distorted 3D images. The problem can be particularly difficult for some polymers for which sputtering tends to break up the backbone of the polymer, leading to very high sputter rates (in some cases, simple evaporation of the small molecules produced adds to the material loss). If such a polymer is mixed with less readily sputtered materials, the distortion of the 3D images can be extreme. Topographic analysis before and after the profile can be used to help correct the data, but obviously if the sample is not columnar but instead has mixed phases, this can become quite a complex process, possibly involving topography checks at multiple points in the profile. Sputter rates derived from the topographic analyses need to be assigned to the different regions to allow expansion of the apparently thinner regions with the higher sputter rates to have their depth better match reality. Composite materials with inorganic fillers in an organic matrix represent the ultimate challenge, since inorganic species are barely sputtered by cluster ion beams. Reducing the size of the cluster (with Ar to Ar₅₀₀⁺ or so) allows for etching of inorganic species, but then damage to the organic signals turns up in the profiles. This is a challenge that still remains for the method.