As was noted in Section 1, the sputter event is complicated, with both the sputter process and ionization important to whether and how secondary ions will form and survive to be detected. While analysts are unable to quantitatively interpret a time of flight secondary ion mass spectrometer (TOF-SIMS) spectrum from first principles, a qualitative understanding of the process by which molecular and fragment ions are formed and survive to be detected has been briefly described (Pachuta and Cooks 1987).

This section enumerates in more detail the various ways that secondary ions can form from a primary ion impact. From the most relevant models for the sputter process that have been proposed, the section attempts to lay out the continuum of processes that lead to material removal from surfaces relevant to TOF-SIMS. Following this, an overarching view of the mechanisms for ion formation during the sputter event are described, along with the supporting evidence and implications this model of the process has for how positive ion spectra are formed, and what one might expect under different experimental conditions and from different samples. Detailed analyses of spectra from simple compounds are discussed, organized by their differing functional groups. Particular attention is paid to similarities and differences between TOF-SIMS spectra and electron impact mass spectrometry (EI-MS) spectra. The section ends with a discussion of the less well-studied negative ion formation and negative ion spectra.

As noted in the preceding sections, the primary ions used in TOF-SIMS analyses vary from relatively low mass atomic species (Mn⁺ and Ga⁺), to higher mass atomics (Au⁺ and Bi⁺), to small cluster ions (Au₂⁺, Bi₃⁺, and Bi₃⁺⁺), to molecular species (C₆₀⁺), and finally to large clusters (Ar_n⁺ and [H2O]_n⁺). Over this wide range of projectiles, the mechanisms by which sputtering occurs vary substantially.

The study of sputtering mechanisms aims to reach an understanding of the yield of sputtered material, the nature of that sputtered material (atoms, clusters, fragments, molecular species), the energy states of the emitted ions and consequent unimolecular fragmentation, the individual yields of each species, the damage induced in the sample by the impact, and the depth of that damage. For the purposes of this wiki, the alteration of the surface is of less interest, since the damaged incurred via depth profiling has been well covered elsewhere (Stevie 2015).

At the energies used in commercial Secondary Ion Mass Spectrometer (SIMS) instruments, the collision of the primary atomic ion with the surface is mostly elastic, with the atoms acting as spheres having radii determined by their clouds of electrons. At the KeV energies used in commercial instruments, the primary ion is neutralized as it approaches the surface and comes close enough for an electron in the sample to transfer to it. We shall therefore refer to it as the primary projectile. It may collide with an atom at the surface. Alternatively, there are spaces between the atoms through which the primary projectile may pass, so that it may penetrate into the sample before hitting an atom in the sample. A collision is inevitable, however, and once one occurs, there will now be two atoms moving within the solid. The collision is mostly elastic because in order for the second atom to move, bond breaking must occur. However, the energy of the bonds (only a few eV) will be trivial compared to the KeV energy of the primary projectile. Now the fast moving two atoms will have their collisions and the process is repeated. This ever increasing series of events is known as the collision cascade, and it is now a well-understood phenomenon (Rabalais 1994, 2002; Garrison and Postawa 2008). When the projectiles are atoms, the collisions that take place within the sample are pretty much independent of each other, separated in space and time, at least for most of the time the cascade progresses.

As the cascade develops, some of the spreading chaos will rebound up toward the surface. Near the impact site where the energy of the moving atoms is highest, any atoms directed back up toward the surface will likely do so with so much energy that no bonding will survive the process, and atoms will be the primary species sputtered. Further away from the impact site, the moving atoms reaching the surface will have much lower energy, and some bonding may survive the impact that drives material from the surface producing clusters and molecular fragments. At some distance from the impact site, only atoms with low energy may reach the surface, and it may be possible for whole molecules to be detached from the surface. At this point in the cascade, the exponential increase in the number of moving atoms may result in multiple simultaneous movements toward the surface that are believed to be necessary to desorb large molecules. The difference between this portion of the cascade and what some authors have referred to as a “thermal spike” is mainly a matter of semantics. This figure is a cartoon of a sputter event, showing an artist’s imagining of the different areas of the surface from which material may be sputtered as atoms, fragments, or whole molecules. The collision cascade has a randomizing effect, though, that should be understood. When the projectile is atomic, the sputter event produces secondary ions whose emission direction is mostly not determined by the angle or direction of the initial projectile’s impact. Secondary ions are most likely to fly directly away from the sample along a line perpendicular to the sample surface; otherwise their emission has a cosine distribution with respect to the sample normal. As might be expected, atomic secondary ions have a larger range of energies than molecular fragments or molecular ions.

Given a layer of atoms of a given element and a specified primary projectile, the chances of a collision with that layer is a function of the ratio of the surface area taken up by the atoms to the total area of the surface. If the atoms at the surface are big and the projectile is big, the chances of an impact is higher than if the atoms at the surface are small and the projectile is also small. The area of the surface taken up by the atoms (and not the empty space) and the area of the impacting particle combined lead to the parameter known as the scattering cross section. Much work, both theoretical and experimental, has gone into determining these scattering cross sections. The result is that programs have been developed (ex. SRIM) that do an excellent job of predicting how, given any specified projectile and surface, the collision cascade will drive the sputter process.

It is not surprising that heavier projectiles tend to have higher sputter yields. That is, they tend to remove more material per impact. It may be less obvious that the heavier projectiles also produce a higher ratio of larger fragments and molecular species than do smaller projectiles. The bigger atomic projectile will certainly have a much larger collision cross section. Where the lighter atoms, such as Mn or Ga, may pass many atomic layers before finally hitting something, the much larger Au and Bi atoms are much more likely to hit a sample atom close to the surface. This has the effect of moving the collision cascade closer to the surface and getting more energy reflected back toward the surface, producing more sputtering. Equally important, the lower energy collisions happening further down the cascade are now many times more likely to happen within reach of the surface. Whenever atomic primary ions are used, the heavier ones turn out to provide a much better result than the lighter ones.

Of course, the sample plays a large role in how the collision cascade progresses. The cascade of collisions that forms upon the impact of a primary ion with a sample, consisting of atoms with an overall low atomic number, will quickly progress deep into the sample, with the result of more and deeper damage but with less sputtering and ion yield. Conversely, the impacts of primary ions on a sample surface with a higher average atomic number will produce cascades that will stay closer to the surface, yielding more sputtering and more secondary ions. It turns out that many real life surfaces coincidentally are better for TOF-SIMS analysis than our imaginary idealized surfaces used in calculations. This is because they typically have an over layer with lower atomic number than the near-surface region. Adventitious hydrocarbons, various adsorbed species, and oxide layers for the most part are less dense and have a lower atomic number than the subsurface region of the sample. Upon hitting the denser subsurface, the cascade will tend to rebound, driving the weakly bound surface layer off the sample.

It is less common for collisions within the cascade to be inelastic, with some of the energy of the collision exciting and, therefore, removing electrons or creating molecules in an excited electronic state. While these sorts of collisions are rare enough that their existence produces little error in the collision cascade calculations, they are a source of ions within the solid that likely have importance for secondary ion formation. More is discussed about this below.

As was noted above, the number of atoms in motion within the solid increases geometrically as first one, then two, then four, then eight atoms and so on join the cascade. The end of the cascade then takes place when the atoms in motion have less and less energy, the last collisions becoming so weak that no further bond is broken. At this point, there are a lot of atoms in motion, which means the movements of these atoms may cease to be completely independent. It may be at this point, where motions that happen to take place at the surface may successfully dislodge weakly attached adsorbed molecules intact, even when atomic projectiles have been used.

Bi_{3</sub⁺ and other small cluster ions produce enhancements to sputter and ion yields well beyond what can be achieved with any atomic species. Certainly, if a heavy atomic ion is more likely to have a collision with a sample atom before traveling very far from the surface than a lighter atomic ion, a cluster will be even less likely to penetrate the surface before a collision occurs. The higher effective cross section for collisions alone would be expected to produce significant enhancements to sputter yields. In other words, a Bi3</sub⁺ ion should produce at least three times the cascade that a Bi1</sub⁺ ion would (somewhat attenuated by the lower energies of each atom in the cluster).}

In fact, the enhancements seen go well beyond those imagined from these simple reasons. The game changes because the individual collisions in the cascade can no longer take place independently. The cluster impact sets up such a high density of events that multiple collisions taking part in close proximity affect one another. The molecular dynamics simulations (Garrison and Postawa 2008) show something very different taking place amidst the collision cascade. There is cooperative motion. There is a hint of flow. This is particularly important in understanding how these small clusters can produce such remarkable increases in the yield of large fragments and whole molecules. Multiple impacts and the movement of larger sections of the surface in coordinated motion are more likely to displace larger structures at the sample surface.

When you start hitting samples with C₆₀⁺, Ar_n⁺ or [H2O]_n⁺ projectiles, you have entered quite a different sputter regime. Gone is the individual atomic collision cascade. Molecular dynamics simulations indicate that cooperative motion is now a very important cause of desorption from the surface.

C₆₀⁺ is the lightest of these clusters, and experimentally, it shows somewhat intermediate behavior. C₆₀⁺ depth profiles of organic samples tend to show more surface damage that is not completely removed by the ion beam., C₆₀⁺ ion beams are most effective in the analysis of materials polymers prone to chemical degradation of the polymer backbone, and less successful with materials that tend to cross link upon irradiation. All of this suggests that energy from this projectile can affect layers beneath those removed in the ion impact. The heavier clusters, in contrast, seem to remove almost all of the material affected by the impact.

The effect of impact of a large cluster on an organic material has been described as being more akin to fluid flow than to any kind of collision cascade. One imagines a sort of liquefaction at the impact site. The whole cluster energy is in the KeV range still, but there are so many atoms in the cluster that each of the individual atom constituents of the cluster has quite a low energy. The collective motion of many atoms does create bond breakage and plenty of damage, but there is less of a tendency toward complete atomization. Large motions of many atoms create effects that look like springboard motions, where lower layers are temporarily compressed but then rebound, casting large sections of the surface into the vacuum. Other related motions can be seen to catapult larger molecules off the surface. The simulations suggest that much more of what is sputtered leaves the surface as part of a larger group of atoms, as intact molecules or as larger sections of any polymer present. A simple way of thinking of this is to imagine using a hammer (atomic and small cluster ions) to using a scoop to remove material from a surface.

While the sputter process determines what and how much material leaves the surface, the ionization process determines what can be detected by the mass spectrometer. While the differences in sputter yields from sample to sample have some effect on the sample-to-sample ion yield differences, the many orders of magnitude variations that have been seen are much more a function of ion formation and ion survival. In fact, for many samples, even with Bi₃⁺ irradiation, an ion is detected only from a fraction of the primary ion impacts, since many impacts only produce neutral species. Most of the material sputtered from the sample leaves the surface with no charge. Because of a longer history of fundamental study in the pursuit of quantitative SIMS depth profiles, the formation of atomic secondary ions is a better understood process than the formation of molecular ions. Nonetheless, it is possible to enumerate possible ionization mechanisms and there implications for how SIMS spectra will look.

Possible ionization mechanisms come in two categories, those that require that the primary ion beam be a form of ionizing radiation, and those that do not.

Atomic and small cluster primary ions represent a form of ionizing radiation. The ionization of the sample that occurs upon irradiation with these primary ions is not so different from the ionization that will occur when the sample is struck by other forms of ionizing radiation. Bombardment by electrons, X-rays, and lasers at a variety of wavelengths will also ionize the sample. The damage they do to polymers, for example, is similar, and a polymer’s relative susceptibility to damage by ionizing radiation is not dependent on the nature of that radiation. The differences are mostly due to the fact that the action with ions takes place so much closer to the surface. The exact nature of the sputter process itself is also quite different. However, remarkably similar mass spectra can result from electron radiation (electron stimulated desorption) and from laser ablation mass spectrometry. The concept of “energy isomerization” has previously been invoked to explain these similarities (Pachuta and Cooks 1987). If the energy deposited in the sample ends up producing motion in the solid that can push molecules at the surface into the vacuum, and this is accompanied by ionization of the sample, the ions produced should be substantially similar. Such a concept suggests that the direct effect of the collision cascade on the ultimate formation of molecular ions and fragments is minimal.

Large cluster primary ions such as Ar_n⁺ where n is generally > 500 are not a form of ionizing radiation, or at the least are much less so, because their KeV energies are spread out amongst so many atoms within the cluster. Bonds can be broken and material removed from the surface, there can be vibrational excitations induced, but electrons will not be promoted from surface species to induce direct ionization.

This first set of possible ionization mechanisms requires ionizing radiation, that is, atomic or small cluster primary ions. Note that all of these mechanisms initially lead to the formation of odd electron ions, with subsequent collisions producing the more commonly observed even electron ions.

1. Direct ionization of a species leaving the surface can occur via collision with material moving up toward the surface. The direct results of this process are positively charged odd electron ions, as the direct impact would knock an electron out of the impacted species.

2. A neutral species sputtered via an inelastic collision can leave the surface electronically excited, and then lose an electron later. The direct results of this process are positively charged odd electron ions.

3. Charges form within the sample as a result of inelastic collisions during the sputter event. Electrons can transfer between these points of charge near the sample surface and species leaving the surface. The direct results of this process are odd electron ions of both polarities.

4. A leaving species can be ionized via impact with or capture of a secondary electron released due to another one of those inelastic collisions. The direct results of this process are odd electron ions of both polarities.

The second set of possible ionization mechanisms only require bonds to break and new associations to form. These mechanisms never produce odd electron ions, although additional bond scission (for example, a second break in a polymer chain) that occurs homogeneously could produce an odd electron in the fragment.

5. Heterogeneous bond scission will lead to charge separation and the formation of ions of both polarities. In the case of highly asymmetric bonds, this could be a source for species such as protons needed for process four here. The direct results of this process are even electron ions.

6. Cationization of a species leaving the surface can occur when it associates with a proton (protonation). The direct results of this process are even electron positively charged ions.

7. Ionization of a species leaving the surface can occur via attachment to either a cation or an anion. The direct results of this process are even electron ions of either polarity.

8. Cationization of a species leaving the surface can occur when it associates with a metal. The direct results of this process are positively charged even electron ions.

As the nascent ions leave the surface, they can be neutralized. Electron transfer from the sample surface (more likely when the surface is completely conductive) neutralizes many atomic ions. Organic molecular ions and fragments, especially if they are in an excited state, may lose an electron to the surface. To the extent that the nascent ion can stabilize itself (for example, by H attachment to eliminate an odd electron) by a rearrangement or by loss of excess energy via fragmentation or photon loss, the ion will be more likely to survive. Most of these, though, require time. Electronic transitions are orders of magnitude faster than the motion of atoms and molecules. Also, the heavier the molecule, the slower it moves as it leaves the surface and the longer it is within electron transfer range. Ion survival probabilities may be nearly as important to the shaping of the TOF-SIMS spectrum as ion formation.

All of the ion formation processes can leave a vibrationally excited molecular ion, and that, along with the ionic and possibly radical nature of the ion can lead to further fragmentation. When the fragmentation is delayed, the process is known as metastable decay and leads to anomalous peaks or background in the spectrum, especially in reflectron based instruments.

Some matrix effects enhance molecular ion formation via a solvation of the molecular ion followed by loss of the solvating matrix. The result cools the ion, lowering its internal temperature below the point at which it will fragment. The same molecule emitted from a solid surface without this solvation does not fare as well.

In the following eight sections, each of the ionization processes mentioned earlier are explored in more detail.

A direct ionization event involves the loss of a single electron, leaving a likely excited ion. Vibrational excitation in one bond can quickly spread over an entire molecule. This is the one advantage a larger molecule or molecular fragment has; it can accommodate a larger amount of excess energy than a smaller molecule or fragment without fragmenting. Direct ionization in particular has been suggested as a reasonable mechanism for the formation of polymer spectra (Leggett et al. 1992). The similarities between low energy collisionally activated desorption (CAD) and direct static SIMS spectra led to the hypothesis that low energy collisions at the end of the collision cascade could directly account for both the sputter event and ionization. High molecular weight polymers (which have few end groups and even less of these at the polymer’s surface) must have bonds broken in at least two places to produce polymer fragment ions. It makes sense that these bonds may be broken in the sputter event itself. It only remains for the collision that breaks the last bond to also be inelastic, driving an electron from the resulting fragment and, thereby, creating a detectable secondary ion. There is no obvious way to produce negative ions from this process.

For each process we consider, it is worth understanding how the matrix influences ion yields. It is largely because of the extreme matrix effects that are encountered by analysts that we are so interested in how the ions form and survive. In this case, the sputter process itself would largely drive ion formation by this mechanism. Ion survival would be dependent on the ability of the nascent ions to stabilize and on the localized work function of the sample. Since many of these factors can be modeled or calculated, it may be possible in the future to further test the relative likelihood of this process.

Just as metal cationization may be the result of autoionization resulting during the association of an excited metal atom and a molecule, a subset of excited molecules may also autoionize if they leave the surface in an electronically excited state. The distinction between this mechanism and the direct ionization process is that ionization does not occur immediately upon excitation. Not until the excited molecule has moved away from the sample does it auto-ionize, at which point electron transfer from the surface will not as readily neutralize the ion formed. The excitation would be the result of an inelastic collision or of energy transfer from another excited molecule nearby. Autoionization would lead to the formation of positive ions exclusively. At this time, there is no experimental evidence for this process.

In order for the excited molecule to have a significant lifetime, it needs to be in a “triplet” excited state so that relaxation is slow. Since this is only likely for a subset of molecules found in SIMS spectra, this process likely has applicability only for specific cases. In this it is akin to the process that produces phosphorescence.

As always, the matrix affects the efficiency of intact molecular desorption in an ion impact. A matrix that can transfer energy from the site of an inelastic collision to the molecule leaving the surface to be excited might be expected to provide significant enhancement to this process.

Atomic secondary ions are known to form after an initial release from the surface via the sputter process by electron transfer between the leaving atom and the sample. This process is as likely for organic molecules as it is for atoms. Ionizing radiation, including atomic and small cluster bombardment, will ionize the sample. Electron transfer is then all it takes to create ions out of molecules leaving the sample. The initial result will be the creation of a singly charged radical ion, but given the collisions that are likely to take place as the molecule leaves the surface, the radical will often be eliminated via hydrogen extraction (either from or to the nascent ion) or fragmentation.

This electron transfer process with subsequent collisions and protonation at first glance will produce similar results for some compounds. The main molecular ion is the M+1 ion. Fragment ions, largely the same as will be seen from collision induced dissociation (CID) used in MS/MS experiments, result from chemical processes driven by the excess energy the molecule contains as it leaves the sample (Pachuta and Cooks 1987). In the case of electron transfer, some of that energy can come from the ionization process itself. However, there are plenty of odd electron ions in the TOF-SIMS spectra produced by atomic and small cluster primary ions. These are not products of the protonation process.

Molecules that are poor candidates for protonation due to the absence of lone electron pairs produce more pronounced odd electron molecular ions, and the intensity of these ions will be greater for those that have π systems that will stabilize the radical (Spool 2004). Direct ionization or electron transfer are the most obvious explanation for the presence of these ions in SIMS spectra.

Electron transfer need not be limited to a single hop from, say, a leaving molecule to a nearby bit of the sample with a vacancy. It is interesting to consider, as one contemplates the possibility of multiple electron transfers, the example of the respiratory electron transport chain (ETC) in mitochondria (Rich 2003). This efficient pathway for electrons begins with the transfer of an electron from a coenzyme nicotinamide adenine dinucleotide hydride, a reasonably powerful reducing agent. The electron moves down the chain, which uses each drop in the potential of the electron to produce a proton gradient from one side of the membrane to the other. The enzyme Cytochrome C Oxidase eventually collects 4 electrons and reduces an O₂ molecule to water. The electrons moving through the ETC do so across remarkable distances, each step involving a relatively small drop in potential. The trial and error of evolution has perfected the ETC, but the physical principles behind the electron transfer can also apply to electron transfers in irradiated samples.

It is worth noting in this context that the organic chemicals that have been found to be the best matrices for Matrix Assisted Laser Desorption and Ionization (MALDI) are pretty much always unsaturated and often aromatic (Karas and Krueger 2003). MALDI matrices have been shown to also work well in the TOF-SIMS to enhance molecular ion formation. In MALDI the unsaturation is believed to be necessary in order for the matrix to absorb laser light (which results in the heating and the plumes of gas that force desorption), but the light absorption also leads to ionization, and these matrices will be good mediums for electron transfer across longer distances. Certainly some of the enhancements seen in MALDI and matrix assisted SIMS are due to protonation by acidic matrix compounds, and some of the increase in molecular ion formation over fragmentation is due to the emission of molecule or matrix clusters for which the loss of matrix molecules from the larger molecule of interest cools that molecule and reduces the chance that it will fragment. However, another possible source of matrix enhancement is the ionizability and electron transfer readiness of these matrixes.

One added consideration for this process is the relative time scale of atomic and molecular motion versus that of electronic transitions and electron transport. A helpful analogy is to thunder and lightning in a storm. The thunder is like the motions of atoms and molecules. The much faster lightning is electronic excitation and electron transfer. Ionization occurs all along the collision cascade (for atomic projectiles) and also at all points, even for small cluster projectiles and the cooperative motions that they produce. However, ionization taking place at the beginning of the impact event occurs while most species are still moving within the solid and before atoms, fragments, and molecules have begun to leave the sample in earnest. Locked within the sample, even if such atoms and fragments destined to leave the surface become ions, there is a high probability they will be neutralized before they actually make their exit. Toward the end of the impact process, when the atomic motions have moved some distance from the initial impact point and when disruption to the surface allows desorption of whole molecules, ionization events are more likely to lead to the formation of ions that will leave before they become neutralized. This may be why the fragments in SIMS spectra look like they come from whole molecules and have chemical specificity, rather than looking like the random breakage of bonds that molecular dynamics shows occurs right at the site of impact.

Matching of energy levels between the source and the target in an electron transfer process will enhance the probability of such an event, and thus the electronic configuration of the sample and the leaving molecule can play a significant role in ion yields for this process. This may explain some interesting observations of matrix effects (Shard et al. 2015). These may be similar to what is seen for atomic ion emission, and indeed, surfaces that produce more intense atomic ions tend also to produce more intense organic ions.

Ionizing radiation generally acts upon irradiated samples by driving electrons out of atoms and molecules. These secondary electrons are readily detectable. Indeed, the ion beams in TOF-SIMS instruments are typically focused using images derived from detected secondary electrons. The impact of a secondary electron with a molecule leaving the surface can remove an electron from the molecule, leading to the formation of a radical cation. This is the process used in the typical electron impact gas chromatograph mass spectrometer (McLafferty 1993). The capture of a low energy secondary electron by a molecule can result in the formation of a radical anion. Such a process has been specifically proposed to explain the spectra of perfluoropolyethers (Spool and Kasai 1996).

EI-MS and SIMS do produce quite different spectra, although there are similarities. Interestingly, degradation of the vacuum in an EI-MS instrument can go a long way toward replicating a SIMS spectrum (McLafferty 1993). Collisions that follow the ionization event reduce the number of odd electron ions detected, and increase the number of even electron ions.

The main difference between this process of direct interaction with secondary electrons and the electron transfer mechanism is the effect to be expected from the matrix. The yield of secondary electron produced ions will be primarily a function of the density of secondary electrons that reach the surface during ion impact and their energy spectrum. Electron transfer will be more directly affected by the details of the energy levels available at the sample surface.

While the literature does not explicitly describe the creation of charge through heterogeneous bond breakage, it is implicit in many discussions of the sputter process. The “precursor model” describes ions as being preformed on the surface (Benninghoven 1983), but what it really suggests is that in molecules or associations between molecules where there is significant polarization, bonds may break heterogeneously. The effects of shock waves have been invoked to explain ionization with high energy massive cluster impacts (Mahoney et al. 1992; Zhang, Williams, and Lee 2015). Such processes in miniature could take place during the sputter events more routinely used for TOF-SIMS analysis.

One of the attractions of the idea of a heterogeneous bond cleavage process is the fact that it is one that results directly in an even electron ion without the molecule having passed through an odd electron state.

Heterogeneous bond cleavage would obviously require that the molecules so cleaved have initially asymmetric bonds, with elements having quite different electron affinities. C-H bonds certainly cannot be expected to break heterogeneously very often. O-H bonds might. C-F bonds also may be highly susceptible to heterogeneous cleavage. Interestingly, the spectra of fluorocarbons and perfluoropolyethers (PFPEs) have some of the strongest odd electron peaks one is likely to find in a SIMS spectrum, not necessarily what one would expect as a direct result of heterogeneous bond cleavage. This could be the result of two processes, heterogeneous bond scission to produce a charge on one side of the fragment, and a homogeneous bond break on the other to produce the odd electron. This is like why these intense odd electron ions can be found in the spectra of per fluorinated materials produced by large cluster ions. The strong ion yield enhancements due to [H2O]Z_n⁺ bombardment (Sheraz et al. 2015) may be due to heterogeneous bond cleavage within the impacting cluster ion, followed by protonation.

For this process, differences in ion yield with matrix differences will be largely a factor of differences in the collision cascade or motion within the sample created by ion impact. Because the ions formed are from the start even electron ions, electron transfer is a less likely mode for the quenching of these secondary ions, so this process should be less sensitive to the localized sample work function.

Protonation differs significantly from both metal cationization and ion attachment. It differs from metal cationization in that the simple H atom will be unlikely to come to the molecule to which it attaches in an excited state ready to ionize. It would also be less energetically favorable for it to associate with a molecule as a radical (it would not have the tendency to associate with a nonbonded electron pair as a metal might). Protonation differs from ion attachment in that protons are not present in most solids as pre-formed ions. H atoms generally form covalent bonds and only hold a partial charge, even in acid molecules, and thus will need to be heterogeneously parted. Protonation, for obvious reasons, can be expected to produce only positive ions.

Matrix effects that would alter the probability of protonation can be both chemical and physical. Hydrogen bonding is the most obvious precursor to protonation as mentioned above. Clearly any surface that will likely produce a higher density of protons upon ion impact will be more likely to produce protonated ions. One clear example where this takes place is in the spectra of poly-methyl methacrylate which are a strong function of the hydrogen bonding in the polymer's super-structure Nowak et. al. 2000.

Protonation is probably not as common a process as is generally believed, at least when small primary ions are used. The enhancement seen in small cluster impacts of even electron M+1 molecular ions is invariably accompanied by an equal enhancement in the presence of odd electron ions in the spectra. The M+1 ions in many of those cases are likely to be the result of hydrogen atom extraction or attachment to radical cations rather than protonation. Protonation may be responsible for the production of a larger percentage of secondary ions by large cluster impacts.

The alkali metals Li, Na, and K are so electropositive that they typically only form ionic bonds and are rarely present in neutral form. The alkalis form stable complexes with compounds having multiple O atoms with appropriate distances between them. Crown ethers in particular come to mind, but polymers such as polyethylene glycol will also form stable complexes with alkali metals. In such cases, the alkali ion becomes separated in space from its anionic partner, and thus charge separation is less difficult. It is also interesting to note that ionic compounds such as NaCl are much less capable of cationizing molecules than more easily separable ionic compounds such as NaBr or NaI (Gusev, Choi, and Hercules 1998). Thus it appears that there are two factors that allow successful cationization by alkali ions; complex formation and charge separation. Alkali cationization, for obvious reasons, produces only positive ions, but there are also reports that attachment of anions (such as from halides) can produce negative ions.

Cationization by alkalis can be the result of intentional sample preparation, but can also happen with as received samples. In the latter case, it is important to recognize the ions in the spectra. Na in particular is monoisotopic, so the presence of a Na adduct in the spectrum can sometimes elude the analyst. Obviously the presence of strong alkali signals in the spectrum is the first clue to the possibility of alkali cationization. In most cases, an adventitious alkali is not pure. Especially with biological samples, K usually accompanies Na. It is worth noting that the mass difference between the alkalis (Li, Na, K) is 16 amu. Therefore peak series with 16 amu mass differences in a sample with strong alkali signals is an indicator of alkali cationization. The glass substrates used in magnetic disk manufacturing have Li, Na, and K in the glass, with strong signals for all three in the TOF-SIMS spectra. Adsorbates to the surface will often appear in the spectra as alkali adducts with three peaks 16 amu apart.

Obviously, the matrix effects that drive this mechanism are chemical in nature. The presence and relative affinities of the attaching ions will determine what species form and in what abundance.

When a monolayer or less of an organic material is present at the surface of many metal substrates, metal cationization is often observed, producing positively charged {M+Metal}⁺ ions. The situation is different than what is found for alkali ions. First, the metals are not already ions before primary ion impact. It does not work to mix the metals into an organic material. Instead, the presence of the metal surface is important to the process. Further, the organic being cationized needs to sit atop the metal, and not the reverse. In all cases in which deposition of metal on top of a sample of interest has produced metal cationization, this has occurred because the organic had some mobility and had diffused out onto the surface of the metal. Finally, the process does not appear to be due to the presence of a high concentration of ionized metal atoms combining with molecules leaving the sample surface. Ag, which by far produces the best metal cationization results, does not do so because it has the most impressive ion yield.

The most likely hypothesis explaining the metal cationization process involves the loss of an electron when a neutral molecule leaving the surface associates with an excited metal atom leaving on a similar trajectory (Wojciechowski 2003). Calculations show the likelihood of the event. The mechanism has the advantage that the proposed electron loss occurs while the molecule is moving away from the surface, thus reducing the chance of neutralization via electron transfer from the sample. The main metals that will cationize in the SIMS are Ag, Au, and Cu, all odd electron atoms, that when ionized join with a neutral molecule to make stable even electron ions.

It is likely that the molecule being cationized and the metal are already bonded (if weakly) or at the least are very close together at the surface before the primary ion arrives. Certainly, an increase in the thickness of the organic over layer at the metal surface rapidly reduces the cationization efficiency. Metal atoms leaving the sample from some depth are clearly not capable of cationizing.

Metal cationization is a special case in that it occurs in a select subset of samples one might analyze. The matrix involving a metal substrate and a thin organic layer is specific. Variations from this formula disallow this process.

Ionization can directly result in excitation of the affected molecule. It is not necessary for the molecule to obtain energy from the sputter process for it to leave the sample in an excited state. This is especially true in cases where the ionized molecule has bond lengths that are appreciably different from that of the neutral molecule. The reason for this is that the ionization process is nearly instantaneous on the time scale of molecular motion. If the bond lengths are very different, the molecule cannot transition from the neutral state to the ground state of the ion directly. The nuclei are at a different distance in the neutral molecule than in the ion, so the ion must start out with the bond compressed or extended relative to where it would end up in the ground state of the ion. One way to look at this is shown in below. The transition has to be directly vertical, but the lowest vibrational states of the ion do not lie directly above the ground state of the neutral molecule. The ion thus begins life vibrationally excited. For per-fluorinated materials, the vibrational excitation that results from ionization can be extreme, likely because the ionized molecule has such different bond lengths. Thus it is common to see, for example, the C⁺ ion resulting from the ion fragment vibrating completely apart.

The positive ion spectra one obtains with a TOF-SIMS instrument are similar in many ways to CID spectra (Pachuta and Cooks 1987), and have similarities to and differences from EI-MS (Spool 2004). This section describes a selection of static TOF-SIMS spectra in order to further explore the mechanisms behind positive secondary ion formation. Further, this survey seeks to provide a more advanced understanding of the static SIMS spectra that can further aid interpretation. This review of spectra begins with the analysis of spectra from simple compounds and progresses toward those of somewhat more complicated molecules. The spectra described were produced with atomic ions and small clusters.

The spectra sport a variety of molecular and fragment ions. The molecular ions include the ions most often produced by EI, M^+., and in addition to this, the [M+1]⁺ protonated ion, the [M−1]⁺ de-protonated ion and sometimes even the [M−2]^+.. radical cation. There are many fragments detected with masses below that of the molecular ion, but the dominant ones have chemical significance. That is, they tend to be products of fragmentation that make chemical sense. They make sense from the point of view of kinetics, the ease with which the transformation that leads to the fragment occurs. They also make sense from the point of view of thermodynamics, the stability of the resulting fragments (both neutral and ionic). Many of the mechanisms established in the study of EI-MS (McLafferty 1993) explain the fragments found in TOF-SIMS spectra.

Sometimes, there are also ions present in the spectra at masses significantly greater than a compound’s molecular weight. Many of these are combinations of molecular ions (dimers, trimers, etc.), but there are also interesting additional products, the result of the combination of several molecules and the subsequent fragmentation of that larger molecule. The possibility of such reactions obviously depends on the amount of a compound on the sample. Bulk material will produce these larger fragments, where submonolayer levels clearly cannot. Thus chemical concentrations produce another matrix effect.

The figure below shows the TOF-SIMS spectrum for bulk stearic acid taken with an IonTof instrument using a 30 KeV Bi₃⁺ primary ion source. The molecular ions around the mass of stearic acid at the nominal mass of 284 amu are prominent. While the [M+1]⁺ ion is the largest of these at nominal mass 285 amu, the other possible molecular ions that are present in the spectrum include the [M−1]⁺ ion and the M^+. radical cation. The peaks at masses higher than the [M+1]⁺ ions are the [M+1]⁺ ions that contain ¹³C isotopes. The prominent [M+1]⁺ ion may be the result of heterogeneous bond cleavage in dimers and trimers or due to protonation, but the prominence of other odd electron species in the spectra also suggests the H atom extraction reaction shown in Equation [1]. Note that the highest occupied molecular orbital (HOMO) in a simple carboxylic acid containing molecule is the nonbonded electron pair on the carbonyl O.

Because this is a bulk sample, interaction between stearic acid molecules at the surface is probable. In fact, most of the molecules in the sample are present as multimers, primarily dimers, bonded together via hydrogen bonds. This then is an ideal sample for which protonation could occur. It is thus a bit surprising that the [M+1]⁺ molecular ion is not more dominant within the molecular ion cluster of peaks.

The presence of the lower mass molecular ions M^+. and [M−1]⁺ could be due to an excess of vibrational energy in the sputtered protonated molecular ion leading to fragmentation and the loss of one or two H atoms. This hypothesis, however, is not supported by observations of relative ion intensities using different primary ions. The preservation of more molecular ions and those of larger molecules in the cluster induced spectra are believed to be partly due to the ability of these primary ions to produce cooler ions less likely to fragment due to excess internal energy. The energy of collision of cluster ions with the surface produces more cooperative motion, which gives leaving molecular ions more translational energy and presumably less internal or vibrational energy when compared with what is seen from atomic ion impacts. However, a comparison of the spectra from monoatomic and cluster ion-induced spectra show the M^+. and [M−1]⁺ keeping pace with the [M+1]⁺ ion intensity. In fact, the M^+. ion is relatively more intense in the cluster-induced spectra, as shown below. This observation supports the idea that these ions are the result of other ionization processes such as electron transfer and electron impact. The mechanism for the formation of the [M+1]⁺ is likely to be due to hydrogen extraction during collision of a M^+. with a neighboring molecule (Equation [1].

While it is true that the EI-MS stearic acid spectrum below has a much larger M^+. peak than its [M+1]⁺ peak, a result that is not surprising given the relative infrequency of molecule–molecule collisions in the gas phase, it is actually more striking that the [M−1]⁺ peak is completely absent from the EI spectrum. The much higher probability of collisions must be partly responsible for the presence of this species in the SIMS spectrum. This ion is in some sense the opposite of the M+1 ion. Where the M+1 ion may be the result of a radical cation snatching a H atom from a neighboring molecule, the M−1 ion is the result of a radical species snatching a H atom from the leaving radical cation.

Another peak that is intense in the SIMS spectrum but weak in the EI-MS spectrum is the [M−OH]⁺ peak at 267 amu (Equation [2]). The most propitious mechanism for the formation of this ion requires starting with the dimer. The 1,3 H transfer that would be needed in an isolated molecule is forbidden. Interestingly, in di-carboxylic acid molecules where the acids can hydrogen bond, the [M−H₂O]⁺ ion is intensely observed in EI-MS spectra. Thus the mechanism may be common with that seen in EI-MS, except that as noted above, bulk samples in the SIMS consist of multimers of the acid.

Equation [3] illustrates the McLafferty rearrangement (McLafferty 1993). Starting with the radical cation at the carbonyl, a H atom transfers from the chain to the carbonyl O via a six-membered ring transition state (the lowest energy configuration possible). The scission of the C-C bond closest to the carbonyl results in the loss of the fatty acid chain and the formation of a radical cation that is stabilized by conjugation with the carboxylic acid’s π system. The peak at 60 amu (the peak labeled (3)) is prominent in both EI mass spectra and in TOF-SIMS spectra taken with Bi₃⁺. The peak is present, but far less prominent in TOF-SIMS spectra taken with monoatomic projectiles (Spool 2004). In fact, the 60 amu fragment nearly keeps pace with the enhancement of the molecular ion in the small cluster-induced spectra. This is another indication, if needed, that this fragment is a direct descendant of the whole molecular ion.

Once the radical transfers to the chain, it can break a bond in the other direction, leaving a double bond on the ion while losing a neutral radical, as shown in Equation [4]. The resulting series of fragments start with the peak at mass 73 amu and proceed up in mass in 14 amu (CH₂) increments as shown in the figure. In the EI spectrum there are four major peaks in this series at 73, 129, 185, and 241 amu. This is a result of the McLafferty rearrangement moving the radical down the chain in even six-atom ring transition states, moving the radical four carbons down the chain each time. In the SIMS, the leaving ion has a lot more internal energy, and transition states with five or seven atoms are easily achieved, leading to more randomization in this fragmentation pattern as the radical is passed up and down the chain.

Just as the [M+1]⁺ molecular ion can lose H₂O (18 amu) to form the ion shown in Equation [2], the fragments that result from Equation [4] can lose H₂O to form the fragment series shown in Equation [5]. Again, this will be a unique series possible due to the hydrogen bonding taking place in the bulk acid surface. This series begins at 55 amu and proceeds at 14 amu intervals to shadow series [4].

Equation [6] shows one possible route to a curious set of three peaks found in the stearic acid spectrum and those of other fatty acid derivatives. The most intense of the three at 98 amu is 14 amu higher than the lowest mass peak in the series at 84 amu and 14 amu lower than the higher mass peak in the spectrum at 112 amu. The peaks are also found in the EI mass spectra. They are remarkable in the SIMS because they are odd electron ions. This in itself implies that the radical is stabilized in some fashion, and that means it must be conjugated with the carbonyl π system. The trio of peaks suggests ring structures, where the most intense ion has the most stable six-membered ring, and the other ions the less stable five and seven membered rings. The series is particularly prominent in the EI-MS spectrum of the dimethyl dicarboxylic ester shown below (“Mass Spectra of Derivatives of Dicarboxylic Acids” 2015).

The acylium ions (with an R-CO⁺ cationic end group), shown in Equations [2] and [5], can readily lose CO to form carbenium ions (R⁺), a reaction that is prominent in EI-MS (Gross 2004). From there both the shrinking of the resulting chain via loss of alkenes or dehydrogenation are both common fragmentation pathways, again common with EI-MS. The result is the series of peaks, mostly at low mass, previously dubbed the “hydrocarbon envelope”.

The esters of simple carboxylic acids produce in the TOF-SIMS ions that are both analogous to those found in the carboxylic acid spectra and the new one. The figure below shows TOF-SIMS spectra for four different straight chain carboxylic acid esters (the high mass region has been aligned to remove differences in position induced by differences in the long fatty acid chain length). The similarities and differences between these spectra and the spectra of the carboxylic acids on which these compounds are based are pretty much what might be expected based on what is seen in the EI-MS.

Source: Adapted from Spool (2004).

For all of these compounds, the molecular ion clusters are remarkably similar to each other and to those of the parent carboxylic acid. Again, the [M+1]⁺ ion is the most intense in each cluster, but the [M−1]⁺ and M⁺. are present in relative amounts similar to what was seen for the carboxylic acid. Upon reflection, this is a remarkable observation. Heterolytic cleavage leading to protonation should surely be more likely in a bulk carboxylic acid’s spectrum than in an ester spectrum. This is further evidence that at least in the TOF-SIMS spectra produced from monoatomic and small cluster primary ions, the protonation process does not dominate ion formation, even for molecular ions. The intensity of the molecular ion cluster relative to the rest of the spectrum decreases with increasing ester chain length, consistent with what is seen in EI-MS.

The odd electron ions resulting from the McLafferty rearrangement with the main chain and leading directly to olefin formation as shown in Equation [3] above are similarly present in the ester spectra, bearing the added weight of the ester chains. Similarly, the ion series, originating in the processes shown in Equations [4] and [5] in the same figure but bearing the added weight of intact ester chains, are also present in the ester spectra.

Molecules with ester chains with two or more carbons can undergo a McLafferty rearrangement that places the radical onto the ester chain. Equation [7] in Figure 4.8 shows how this can lead to ions in the molecular ion cluster for the parent carboxylic acid. The rearrangement is much less likely and cannot lead to the same result for the methyl ester. Equation [8] in Figure 4.8 shows a mechanism for directly producing the even electron ion without the need for further H atom extraction. For molecules with ester chains three or more carbons long, Equations [7] and [8] become major pathways for ester ion fragmentation. Esters with very long chains attached to the carbonyl and the O atom of the ester group may produce characteristic peaks for the parent carboxylic acid but no molecular ion. Again, this behavior is similar to what is seen in EI-MS.

In the absence of hydrogen bonding to facilitate fragmentation leading to formation of the acylium ion, the ion is much less likely to be produced by the mechanism shown above in Equation [2]. For the methyl ester, the process is almost completely inhibited. The mechanism shown in Equation [9] provides a path to the acylium ion. The absence of hydrogen bonding, though, leaves the acylium ion a less common fragment in the ester spectra.

The carbenium ion representing the chain attached to O is prominent in the ester spectra shown below. The length of the chain is easily seen upon examination of the lower mass regions of these spectra. Within the series of esters, a comparison of the spectra shows that this peak is more intense for longer chain lengths. Equation [10] shows how this ion can form. Unexpectedly, the use of Bi₃⁺ as the primary ion instead of Ga⁺ decreases the relative intensities of these ions relative to the hydrocarbon envelope (see figure below)). There is still enhancement of the intensity of the carbenium ion associated with the ester chain, but the peak is no longer quite so obvious.

With the addition of N to a molecule, the mix of peaks in the TOF-SIMS spectrum with odd and even nominal mass no longer have the same meaning to the analyst, as discussed previously. Ions with an odd number of N atoms that have even mass are even electron ions. Ions with an odd number of N atoms, but an odd nominal mass, are actually odd electron ions. Simple amides have a single N atom, so any peaks with the amide piece of the molecule now obey these reversed rules. On the other hand, the hydrocarbon envelope, sourced from the portion of the molecule from which the amide portion of the molecule has been removed by fragmentation, looks much the same as it always has.

An amide is, of course, much like a carboxylic acid except with an NH₂ where the OH used to be. In contrast to the carboxylic acid, however, the amide does not readily lose its NH₂ to form the acylium ion, even when the molecule is hydrogen bonded to its neighbors. In other ways, however, there is a good deal of similarity to the spectra. Again, most of the ionization is expected to begin at the amide functional group where the highest occupied molecular orbital (HOMO) can be found (the molecular orbital from which it will be easiest to lose an electron).

The main molecular ion is the [M+1]⁺, now an even electron ion with an even nominal mass at 284 amu (above), but there are others present in the cluster as well. The odd electron ion M^+. of stearamide is relatively weak.

The McLafferty rearrangement is just as important to the formation of the stearamide spectrum. The parallel fragmentation process to that, shown in Equation [3] above for stearic acid, produces the prominent odd electron ion at 59 amu. Here in particular, the ion could be mistaken for an even electron ion fragment, but exact mass determination shows it to contain a single N atom and to have the formula C₂H₅NO. The series that is comparable to that produced by the process shown in Equation [4] above dominates the middle portion of the stearamide spectrum. The series is also more closely akin to that found in EI-MS with the most intense peaks in the spectrum progressing down the chain in the six-membered ring transition states producing the most intense ions in the series at 72, 128, 184, and 240 amu. Again, because formation of the acylium ion is not favorable for the amide group, the series of ions comparable to that produced in Equation [5] for stearic acid is absent from the stearamide spectrum.

There is one other ion in the spectrum that requires explanation, and its presence highlights one of the pitfalls in the interpretation of the TOF-SIMS spectra. The peak labeled [9] in the stearamide spectrum at 256 amu has an exact mass that indicates that it has two carbon atoms and four hydrogen atoms (C₂H₄) less than the molecular ion [M+1]⁺ at 284 amu. It is not a fragment, but rather a homologue. It is the molecular ion for palmitamide, the fatty amide with a two carbon shorter chain. One would not find such a feature in an EI-MS spectrum, because the GC column would separate it from stearamide, and the spectrum obtained is pretty well guaranteed to be that of a pure compound. In other samples, impurities that tend to surface segregate can come to dominate spectra even at very low concentrations. The SIMS analyst must proceed with caution.

Ketones are at once a simpler carbonyl containing functionality, but with their own individual quirks in a mass spectrometer. The carbonyl group itself has no significant acidity or basicity, but it is reactive under the right circumstances, and can be readily ionized by electron loss. With two pendant alkyl groups, the ketone radical cation can react with either side, and then the radical can cross from one alkyl group to the other.

The TOF-SIMS spectrum of bulk stearone shown above is that of a symmetric long chain ketone with 17 carbons on each side of the carbonyl. The molecular ion cluster ratios for stearone show more relative intensity for the M^+. radical cation than in stearamide, but otherwise at the low end of the range shown in that figure. It is worth noting that the H extraction available to stearone in the TOF-SIMS likely stabilizes the molecular ion, leaving an intense peak. Here the TOF-SIMS has the advantage over EI-MS in which the molecular ion for stearone is quite weak. Comparable to what is found in the EI-MS (not shown), there is a strong peak for the acylium ion that results from loss of one of the pendant alkyl groups. A possible mechanism for such an ion’s formation is shown below. The neutral alkyl radical that forms is more stable and, therefore, more likely to form than a hydroxyl, oxy, or an amine radical, so this process can take place in a unimolecular fragmentation pathway that leaves an ion containing an alkyl group, a process that in acids, esters, or amides can only result in loss of the alkyl group.

Other reactions can proceed much as has been described previously. The even electron ion that is formed in a process comparable to that shown for a carboxylic acid in Equation [3] occurs for the ketone as well, but now that ion contains one of the long alkyl chains and, therefore, still remains with a fairly high mass (species 3 in the stearone spectrum). Olefin formation after the McLafferty rearrangement and randomization of the radical up and down the chain leads to the series of ions labeled (4). Note that the radical can potentially hop from one alkyl chain across the carbonyl to the other via a McLafferty rearrangement, but olefin loss will still produce ions with the same mass as in series (4).

If a carboxylic acid should be remarkable for its ability to donate a proton relative to many other molecules, an amine’s basicity makes it an optimum acceptor. If protonation is a major path for ionization of anything, it should be so for simple amines. Indeed the [M+1]⁺ ion does very much dominate the high mass range of the TOF-SIMS spectrum of the primary n-tetradecyl amine as shown below. The secondary amine spectrum shown in below that has a similarly intense [M+1]⁺ molecular ion.

The HOMO of a simple amine is the nonbonding electron pair on N. If any of the ionization occurring for amines, it is via electron loss, and the nonbonding orbital is the one most likely to lose an electron. It turns out that it is easy to rationalize the major features of the SIMS spectrum by proposing mechanisms beginning with the odd electron radical cation molecular ion. Without a π system, however, there is no way to stabilize a radical, and so no prominent odd electron ions are present in amine molecule spectra. Nonetheless, the chemical specificity of the fragmentation found in the simple amines’ spectra is again compelling evidence for a significant role being played by electron loss in the ionization of these molecules.

The McLafferty rearrangement followed by radical movement up and down the chains ending in radical loss and olefin formation comparable to the process shown in Equation [2] below likely produces many of the fragment ions that dominate the amine spectra. The simple elimination reaction shown in Equation [1] leads to the ions labeled (10) in the amine spectra shown in Figure 4.14. For the primary amine (Figure 4.13), the mass and the intensity of the ion are quite low, but for the secondary amine, the higher mass and intensity make it a prominent ion in the spectrum. The McLafferty rearrangement shown in Equation [2] below can lead to the olefins already discussed earlier, but can also lead to a second McLafferty across the alkyl chains in the secondary amine, leading to ions (11) and the shortening of both alkyl chains. Presumably if the elimination occurs in the other direction leading again to a radical cation fragment, the unstabilized radical either further rearranges to fragment further, or manages, by a collision at the surface, to extract an H atom and become an even electron cation.

The figure below shows the TOF-SIMS spectrum of bulk tri-benzyl amine. This tertiary amine has its N separated from the aromatic rings by saturated carbons, so that the compound is still a strong base. Nonetheless, the [M+1]⁺ ion is not the most intense peak within the molecular ion cluster. Instead, that distinction goes to the even electron [M−1]⁺ ion, which is particularly stable because in that ion the N lone pair ends up being conjugated with one of the rings’ π systems. Whether a H radical can dissociate from the molecule in a unimolecular reaction, or if a collision is required to extract the H atom is unclear. Within the molecular ion cluster, the M^+. radical cation is more intense than the [M+1]⁺ ion, likely because ionization can also occur via loss of an electron from one of the rings. The peak at 210 amu (M-77) likely forms by a mechanism akin to that shown in Equation [1] above, which in this case will eject a neutral C₆H₅^. radical.

A saturated hydrocarbon is a hard molecule to ionize. There is no place to park an added proton, so the M^+. molecular ion is the easiest to produce with loss of an electron from a C-H or C-C bond. Whichever one it is, that bond will now be substantially weakened. This is unlikely to be a stable ion, able to survive collisions with the surface or with fellow travelers to the vacuum.

Instead of M^+. or even [M−1]⁺, the figure below shows that in the TOF-SIMS spectrum of n-nonadecane, taken with a cold stage, it is the [M−2]^+. radical cation that is the most intense of the molecular ions. This is a more stable ion. As noted above, with only one electron gone, there will be a weak one electron bond in the ion. With an electron and a hydrogen atom gone, one carbon can reorganize itself into an sp² configuration, placing the lone electron into a p orbital. With two hydrogen atoms gone along with the single electron, the lone electron is somewhat stabilized by living in a π bond. The intensity of the peak is weak. The mechanism for the formation of the ion is unclear (although a unimolecular dissociation from an initially strongly vibrationally excited molecular ion is a strong possibility), and whether a collision is required is also uncertain.

The rest of the n-nonadecane spectrum is unremarkable, with the hydrocarbon envelope in evidence but little else. As has been noted previously, branching and molecular weight will affect the relative intensities of the peaks in the hydrocarbon envelope in ways that can be rationalized (Galuska 1997). In these respects, the TOF-SIMS spectra are similar to what is found in EI-MS spectra of saturated hydrocarbons.

The introduction of unsaturation into the hydrocarbon makes ionization easier and stabilization of the odd electron possible. The figure below shows a TOF-SIMS spectrum of n-decyl benzene taken using a cold stage, which has, as its primary molecular ion, the M^+. radical cation. Also present at lower intensities is the even electron [M+1]⁺ molecular ion and the ion seen in the saturated hydrocarbon spectrum, the [M−2]^+. radical cation.

Peaks at 91 and 105 amu are the result of scission of the chain close to the phenyl group. The figure below shows a possible mechanism for the formation of the 91 amu peak. The resulting ion can rearrange itself into the aromatic tropylium ion. Studies of alkyl benzenes in the EI-MS with labeled samples suggest, however, that the mechanism for tropylium ion formation, and also for formation of the 105 amu ion with one more carbon are more complicated than the equation below implies. In any case, the processes at work in the bombardment of this sample are similarly at work in other cases of molecules with phenyl groups, for example, the tribenzyl amine whose spectrum was discussed previously.

One other odd feature in the spectrum shown above is the presence of the [M+CH]⁺ ion. This observation is not an isolated one. Spectra of many bulk aromatic materials produce this peak. The ion is likely a tropylium ion with an attached decyl chain. The ion must be the result of the reaction of a molecular radical cation with neighboring phenyl group molecules in which first the dimer is formed, and then splits unevenly.

At the other extreme of the range of hydrocarbons are completely unsaturated molecules. An example of spectra of such a molecule is shown below. The tetra-phenyl butadiene molecule naturally produces a relatively stable M^+. radical cation. In fact, the entire TOF-SIMS spectrum quite closely resembles the EI-MS spectrum, an indication that most of the processes involved in fragmentation of this molecule are primarily unimolecular.

The TOF-SIMS positive ion spectrum for glyceryl mono-stearate, chosen as an example of the multifunctional molecule, is shown below, as taken at −150°C using a cold stage. This compound is an example of an important series of compounds studied by TOF-SIMS that are lipid-based. It turns out that the molecular ion cluster dominated by the [M+1]⁺ is very weak at room temperature and higher temperatures, but cooled, the bulk material begins to produce a substantial molecular ion signal. The fragment signals are not suppressed at low temperatures, so it does not appear that the molecule is less likely to fragment when cold. Instead, it appears that neutralization of the molecular ion is reduced at lower temperatures.

The more dominant high mass ion is a result of a McLafferty rearrangement followed by loss of OH (possibly assisted by H bonding in the bulk solid). The process is shown below. Triglycerides, molecules with fatty acids attached to all three of the glycerol’s hydroxyl groups, similarly produce TOF-SIMS spectra that lack the molecular ion at room temperature, and are dominated by the olefin with one ester removed.

Without a library spectrum, running the analysis at room temperature, and not knowing that this compound was present at the sample surface, would an analyst be likely to definitively identify it? Neither the extensive library nor the interpretative tools available to other mass spectroscopists are at this time available to the TOF-SIMS analyst. Without a molecular ion to interrogate, even MS/MS would fail the analyst. The state of the art is still developing.

The understanding of the ionization mechanisms and fragmentation pathways for positive ions in the TOF-SIMS is still under study. The comparable understanding of TOF-SIMS negative ion spectra is even more in its infancy. For one thing, there is not the wealth of knowledge for negative ion formation mechanisms that is available for the researcher studying positive ions. EI-MS produces only positive ions. The wealth of research into mechanisms of fragmentation for EI-MS and, for that matter, CID is almost exclusively limited to positive ions. Negative ions are produced in electrospray MS analyses, but here we can expect a somewhat different process for ion formation, and the solvation of the nascent ion in electrospray analysis reduces fragmentation.

In general, materials that produce strong positive ion spectra often produce strong negative ion spectra. Obviously this is not always true, and especially for materials that are ionic in nature, this is decidedly untrue. But, in particular, for positive ions that lose electrons from weaker bonding orbitals, HOMOs, that are reasonably high in energy, those molecules are amenable to receiving electrons into their lowest unoccupied molecular orbitals (LUMOs), which will tend to be relatively low in energy. One example of this behavior is in silicone compounds. These materials have a conjugated system in their backbones with the nonbonding electrons on the O atoms conjugated with the empty d orbitals on the Si atoms. The bonding is weak, so the HOMO is high and the LUMO low. The molecules therefore easily ionize both by losing electrons and by accepting them, and this leads to strong spectra in both positive and negative ion modes.

The analysis of perfluoropolyether negative ion spectra, in particular, strongly indicates that electron capture by the molecule initiates ionization and fragmentation (Spool and Kasai 1996). For these compounds, there is a wealth of information in the negative ion spectra that is not found in positive ion mode that impels analysts working with these compounds to further study these spectra.

While “anionization” occurs in specific circumstances, there is no equivalent to protonation possible in the formation of negative ions. The capture of an electron, either a low energy secondary electron or via electron transfer from the ionized sample, pretty much has to occur to create molecular and molecular fragment negative ions. The result, initially, is a negative ion radical, a species that is known to be particularly unstable. There is little likelihood that odd electron ions will be observed in negative ion spectra. Some kind of fragmentation or rearrangement separating the odd electron from the charge must occur in order for an odd electron negative ion to survive into the vacuum.

The figure below shows the negative ion spectrum of stearic acid. Instead of a prominent [M+1]+ ion, the molecular ion is instead at [M−1]−. By losing a hydrogen atom, likely at the C adjacent to the carbonyl producing a conjugated anion, the nascent ion gains some stability. The process may be as simple as the result of homolytic cleavage of the C-H bond in the newly ionized and, therefore, vibrationally excited molecule.

The even electron olefin containing fragments labeled as (4) in the positive ion spectrum shown forming in Equation [4] have their negative ion relations evident in above. However the negative ion fragments are 2 amu lighter than their positive ion versions. It is as if the anionic end of the molecule attained its even electron [M−1] nature even while a radical moved down the chain, ending up fragmenting in the usual way, forming the olefin. How that might occur is unclear.

Similar observations linking the negative ion spectrum for a compound with its positive ion spectrum are common, but not well explored. For the analyst operating with the knowledge we have of the negative ion spectra formation processes today, it is fair to say the approach to spectral interpretation will be fairly empirical. As always, if one has an idea of the compounds one is likely to encounter, the acquisition of standard spectra is essential, especially if negative ion spectra are to be employed.