Ion guns are used in many types of apparatus including sputter systems, focused ion beam (FIB) tools, and in surface analysis equipment. The use in TOF-SIMS as the source of primary ions is the most demanding application of an ion gun. The ideal primary ion source for TOF-SIMS produces quick repeatable pulses of ions that arrive at the sample focused both in time (all ions in the pulse striking the sample at nearly the same moment, at a repeatable time relative to the pulse captured by the electronics) and in space (with a minimum spot size). There is an exception to this, in the case of an instrument that pulses the secondary ions rather than pulsing the primary ions. That exception is dealt with separately.

Additionally, the primary ion should be a heavy atomic ion, or better still, a cluster ion, or even better, a large cluster ion (Kersting et al. 2004). Recent work suggests that the chemical nature of the large cluster ion can be chosen to give large secondary ion enhancement (Sheraz et al. 2015). Finally, the source should be bright, producing pulses with plenty of ion current. It should come as no surprise that the ideal primary ion source is yet to be built. As a result, multiple ion sources may often be present on a single TOF-SIMS instrument.

The liquid metal ion gun (LMIG) is the ion gun of choice for FIB tools, and it is also the most important of the primary ion guns used in most of the TOF-SIMS instruments shipped today.

• The gun can produce ion pulses that arrive at the sample in a very narrow time window, allowing for high mass resolution.

• The gun can alternatively produce very small spots on the sample, the smallest of any ion gun available, spots < 50 nm in width.

• The gun can simultaneously produce submicron spots with short pulses that allow for high mass resolution (<600 nm, the so called “compromise mode”).

• High mass resolution, high lateral resolution, and compromise mode operation can be achieved with decent primary ion flux and therefore reasonable data rates for the instrument.

• With a Bi alloy as the metal, LMIGs can produce cluster ions with reasonable brightness. The Bi₃⁺ cluster in particular is a very practical ion to use in TOF-SIMS analyses.

The chief disadvantage of the LMIG is that the ions it can produce are limited to metals, and it cannot produce the sort of large clusters that have enabled exceeding the static SIMS limit without degrading the sample allowing an analyst to perform organic depth profiling.

The liquid metal ion source (LMIS) consists of a reservoir of metal with an attendant heater attached to a needle, which is coated with the same metal that forms the actual ion source (Komuro 1982; Komuro and Kawakatsu 1981).

The heater keeps the metal liquid. A metal extraction plate with an aperture is placed near the tip of the needle. When a high potential difference is placed between the needle and the extractor, a very high electric field is formed at the tip of the needle. The action of the field is to charge the metal positively near the tip. When the tip “lights,” the metal surrounding the tip elongates, is driven into a cone shape (known as the Taylor cone) by the intense field, and ions begin to be extracted from the tip. One of the most ingenious features of this device is the fact that ultimately the needle itself is buried under the distorted liquid metal and the shape of the cone becomes almost entirely a function of the electric field. The exact shape of the needle itself is not so important and, therefore, the engineering effort needed to consistently create these sources is not so demanding. The source of the ions is very sharp, an emission point of 10 nm or so in diameter. A dynamic equilibrium is set up, driving liquid from the reservoir up the needle where the electrons are stripped from the metal and driven back down the needle, while ions fly out into the ion gun column.

LMIS are commonly manufactured with one of three metals (Swanson 1994). Gallium is the metal of choice for FIB tools. FIB instruments use LMIG sources in DC mode to slice and dice samples, to open up cross sections for analysis, and to produce thin sections for transmission electron microscopy (TEM) analysis. The species formed at the tip are mostly atomic ions. The FIB tools are used to cut, so the direction of the ion beam is parallel to the FIB exposed surfaces. This means that damage caused by penetration of Ga into the material being cut is minimized naturally by the geometry of the cut. In contrast, in the TOF-SIMS tool the ion beam comes in at a much higher angle, and the penetration into the sample along with the subsurface damage that it causes, along with the reduced sputter rates and ion yields that result from much of the energy of the ion beam being deposited well below the surface, makes Ga⁺ a poor primary ion for TOF-SIMS.

Nonetheless there are labs that still use Ga⁺, principally for inorganic analysis. Ga requires little heating, and therefore is more monoenergetic. Until recently, Ga⁺ beams had a superior spot size. Ga is also available only as the ⁶⁹Ga isotope, and a Ga source produces atomic ions only with no clusters. As a result, no separation in the column is required when using a Ga source (see below). Finally, when a set of analytical protocols have been developed in a lab using a specific source, it is difficult to make the transition to a new primary ion.

Au sources still see some use (Hill and Blenkinsopp 2004). To bring the temperature of these sources down, a AuGe eutectic is used, and the Au and Au cluster ions are separated from ions containing Ge. However, only a small percentage of the ions produced from the Au source are Au clusters, which represents a significant disadvantage for this source in comparison to Bi.

Finally, there is the Bi source, which is the preferred source in use today for TOF-SIMS. The percentage of cluster ions, in particular Bi₃⁺, is so high that with the enhanced secondary ion yield from the cluster, the ion produces both faster and richer results.

The small source of the LMIS is what makes it possible to focus a small spot onto the sample. Large sources can be focused into small spots (think of a magnifying glass in the sun being used to burn paper), but those sources need to be collimated, not an easy thing to achieve in an ion source. The ions flying out from the LMIS point source can be manipulated and focused back into the image of its source, a point.

The art of manipulating ions is analogous to the manipulation of light. Instead of clear lenses, electric fields act on the charge of the ions to change their flight paths. Electric potentials between shaped metals in the ion path act as lenses and deflectors. Different from light, ion beams can be accelerated and decelerated. Ions with different masses can be separated from each other. Pulses of ions can be made and focused onto the sample.

Neutral species are not affected by the ion optics of the column. Since ions can be neutralized by collision in the column and neutrals will continue heading in the direction they were going when they lost their charge until they run into something, it is important that the ion column should not be perfectly straight and unimpeded. Ions accelerated near the source and neutralized should not be able to reach the sample. Such neutrals essentially add a DC noise into the spectrum, since their arrival at the sample would not be timed with the pulse of primary ions. LMIG guns shipped today are not sources of neutrals.

For use in a FIB tool, the LMIG tuning and usage is straightforward, far more so than when an LMIG is used in a TOF-SIMS instrument. There is a steady stream of ions emitted from the tip. All the ions emitted are given the same energy and are focused onto the sample. There are three major differences between this and what is needed for use in a TOF-SIMS operation when the more common instrument is used, one that cannot use DC primary ion beams unless you are pulsing the secondary ions instead.

1. The ion stream needs to be pulsed. To produce mass spectra, tightly spaced packets of ions are the ideal. The tighter the spacing in the time of arrival at the sample, the better the mass resolution of the instrument will be, at least at low mass. Typically, a voltage on one deflection plate is kept static so that without an additional voltage, the ion beam is deflected and will not reach the sample. A voltage pulse sent to a plate opposite the plate with the DC voltage swings the ion beam onto the path to the sample. An edge placed just by the ion path helps make the pulses crisp.

2. To produce short pulses that have a reasonable number of ions in each, the ion beam needs to be bunched. This way, a long pulse can be produced initially with a larger number of ions, with bunching used to shorten the time window in which they strike the sample. Bunching is accomplished by applying a pulsed voltage difference between two plates, one behind the pulse of ions and one in front. The voltage accelerates the ions at the back of the pulse more than the ions at the front. Tuned properly, the buncher arranges the ions in such a way that the ions at the back just catch up with the ions at the front when they all reach the sample. The down side is that the ions now have a distribution of energies, making it harder to focus them. This is known as a chromatic aberation.

3. For sources that produce more than one species of ions, it is necessary to separate these. If they are not separated, then ions of higher mass will arrive at the sample at a later time than the lighter ones, and you will have multiple peaks in the spectrum for each secondary ion species. This is, of course, another time of flight effect. The primary ions all have the same energy, so like all the secondary ions with nominally the same energy, their flight times to the sample will be distinct. In the LMIG, this time of flight difference for the primary ion species is what aids separation of the species. All that is needed is a pair of pulses. Ions are pulsed far up the column, and then pulsed again as described earlier. The difference in time between the two pulses is adjusted to match the flight time of the species wanted in the ion beam. All other species will arrive too early or too late for the second pulse and, therefore, will never make it through to the sample, impacting instead on a surface within the gun. As long as the elements in the source are monoisotopic, it will not take great mass resolution in the LMIG column to separate the different species.

With all these elements in place in the ion column and the extractor and suppressor elements for control of the source, deflectors, lenses, buncher plates, and finally stigmator elements for fine tuning the shape of the primary ion beam, the operator can tune an LMIG in a variety of ways. The beam can be optimized for the highest lateral resolution. Since the pulsing of the gun itself affects the spot size as does bunching, in this case the pulse width would be set to be long (reducing the proportion of the time of operation when the beam is subject to a pulse) and the buncher would be turned off, producing poor (up to 1 amu) mass resolution. Or, at the other extreme, the ion gun can be tuned to produce the optimum mass resolution with shorter pulses and with the buncher turned on. Other modes allow for higher currents with small spot size, or higher currents with short pulses and therefore high mass resolution. Finally, a compromise can be struck that gives decent mass resolution and submicron spot size. The following figure shows a schematic of an LMIG tuned in two quite different ways.

Additionally, the operator gets to choose the primary ion from those emitted by the source. In fact, most of the choices the operator makes when planning an analysis is in the use of the ion gun.

At the other extreme from the atomic and small cluster species produced in an LMIG, are the large clusters that can be produced by condensing Ar in a supersonic jet (Mahoney 2013; Kayser et al. 2013). The expansion of the gas stream leaving the nozzle directly results in a sudden cooling of the gas. The Ar gas gets so cold that it condenses into clusters of Ar atoms. Skimmers placed about the expanding jet allow for differential pumping and the introduction of the clusters into the vacuum system of the instrument. The skimmers also effectively collimate the beam. This continuous beam is then ionized by electron impact. The result is a beam of mixed ionized clusters with a variety of sizes and energies, along with neutral clusters.

The ions then need to be accelerated, separated from the neutrals and from each other by mass, focused, and, if being used analytically in a system that uses pulsed primary ion beams, pulsed. Of these tasks, the one that in particular varies from gun to gun is the mass separation. Unlike in the LMIG, the time of flight trick for separating these ions is impractical given the size of the ions and, therefore, their substantially lower speeds. The application of a Wien filter is one method that is used. A beam passing between two plates with a voltage difference between them will be deflected differently as a function of the mass to charge ratio. Simply placing a slit in the beam path allows some significant filtering. Ion-Tof sends the ions through the Wien filter for initial mass separation, and then when pulsing the beam for use as a primary ion source, adds further mass separation using a 90-degree bend (Kayser et al. 2013). The voltage on the bend is pulsed. The pulse has the effect of giving all the ions the same added momentum (and not energy), which means that the angle of the deflection will depend on the mass of the ion. These guns still end up providing a range of cluster sizes to the sample, but when multiple filters are used, the range is fairly narrow. In this way reasonable mass resolution can be achieved.

Focus is still an issue for these sources. The ability to focus is aided by the fact that the beam is somehwat collimated, and can be further aided by design of the source to reduce the size of the volume in which the electron impact ionization takes place. Spot sizes of a few microns have been achieved. A DC beam that can be used for spectrometry in the J series III can be focused down to 1.5 microns.

If a molecular species is volatile enough to produce sufficient vapor when heated in a vacuum (but nonvolatile enough to wait in the said vacuum system until heated), the vapor can be entrained into an electron beam, and then the ionized molecules can be accelerated and otherwise dealt with as described earlier for the previously discussed ion sources. While a variety of such molecular species have been tried, the C₆₀⁺ Buckminster Fullerine (also known as the “bucky ball”) molecule has been by far the most studied and the most commonly utilized molecule in such an ion source (Hill and Blenkinsopp 2004). C₆₀⁺ sources are in general use, as much for X-ray Photoelectron Spectroscopy (XPS) depth profiling as for Secondary Ion Mass Spectrometry (SIMS).

These sources also produce very different SIMS spectra, with intense higher mass fragments and molecular ions. As with the gas cluster sources, C₆₀⁺ can produce characteristic organic ions well beyond the static SIMS limit, although with C₆₀⁺ the effect is more material dependent than it is for the more massive cluster ions. Success with TOF-SIMS 3D depth profiling using C₆₀⁺ is very material-dependent (Rading et al. 2013).

On the other hand, C₆₀⁺ is a better source for sputtering inorganic materials; so on surfaces with inorganic features (such as in plastics with fillers) the C₆₀⁺ source may be preferred. However, the gas cluster sources can also be used in these situations if smaller cluster sizes are chosen.

Once the primary ions strike the sample, secondary ions are formed. The instrument needs to gather, focus, separate, and measure these ions. The ions are emitted from the surface with a variety of energies, and they fly off in varied directions.

Fortunately, their energy range is reasonably low; most have much less than 100 eV. Atomic ions have a larger energy distribution than inorganic clusters, molecular ions, and their fragments.

Differences in the energies of these different ion types can lead to slight differences in their mass calibration in the spectra due to the limitations of TOF-SIMS spectrometers energy/angular emission angle compensation.

The ideal scenario is that the secondary ion trajectory begins speeding along the normal to the surface, directly into the aperture through which it is being extracted.

From a flat surface, the emission tends to be in something like a cosine distribution relative to the sample normal.

That is, most ions will be emitted nearly normal to the surface, with an ever-decreasing number being emitted at higher angles. However, in the presence of the sample topography ion emission can end up being forced away from the sample normal to the surface.

Capturing these ions for measurement is a bigger problem.

The ideal spectrometer in a TOF-SIMS instrument would manage to capture all of the secondary ions produced at the sample, transmit them without losing one to the detector where every secondary ion is counted and perfectly resolved by mass. One of the strengths of the TOF-SIMS method is that the transmission and detection of secondary ions is very efficient. As a rule of thumb, modern spectrometers will typically transmit and detect on the order of 50 percent of the secondary ions formed at the sample, an impressive number, indeed. The tuning of the instrument can be varied to maximize transmission at the expense of other parameters such as mass resolution, and conversely, transmission can be sacrificed to improve mass resolution.

Mass resolution in the TOF-SIMS is not solely a function of the spectrometer, nor is it constant across the mass range. Mass resolution is defined as the ratio of the mass/charge ratio to the resolving power or resolution at that mass, m/∆m. This latter ∆m is defined in various ways by various people, as the width of the peak at the half way intensity (Full Width at Half Max or FWHM) or at some other named percentage of the peak maximum or the distance between 2 barely separated peaks when the valley between them is sufficiently low (10 or 50 percent). However one defines it, the TOF-SIMS mass resolution varies across the spectrum. At low masses, the mass resolution is primarily affected by the primary pulse width in time, a function of the ion gun and not the spectrometer. At high masses, the effect of the primary pulse diminishes as the absolute time of travel increases. At the highest masses, the spectrometer’s characteristics dominate the mass resolution.

Extraction of secondary ions into the spectrometer proper is done with the use of an electrostatic field, the result of a potential difference that is created between the sample and the extraction optics. The extraction optics are, in effect, a part of the spectrometer design, and so they are discussed, along with the rest of the time of flight mass spectrometers described in this section.

As noted in previously, once secondary ions have been accelerated so that they have approximately the same energy, all that is needed to measure their mass is to wait. Their times of arrival at the end of an evacuated flight tube can be converted to a mass scale. Unfortunately, as noted above, neither their energies nor their trajectories are exactly the same, and as a result, the mass resolution possible with a linear spectrometer is poor.

The commercial TOF-SIMS systems all have unique designs and differing characteristics that affect experiment design and usage. There is also a hybrid system with an add on Orbitrap spectrometer which, while not being a TOF instrument is nonetheless used in conjunction with one and thus is described below.

IonTof manufactures Reflectron type TOF-SIMS instruments (Niehuis et al. 1987). In concept, the Reflectron is the simplest TOF spectrometer next to the linear TOF. To improve mass resolution, the ions are reflected back using an electrostatic field; nearly directly back at the sample. Ions carrying more energy penetrate the field further before turning back, thus taking a longer path and a little more time. With the right tuning of fields, this extra time nearly exactly makes up for the fact that these ions are travelling faster than lower energy ions with the same mass to charge ratio, and thus all the ions with the same m/e arrive at the detector at the same time. A schematic of such a spectrometer is shown below.

One characteristic of a Reflectron is that most ions produced at the sample or that result from a secondary ion’s decomposition in flight will reach the detector. Ions that are the result of a secondary ion’s decomposition are known as metastable ions. Such ions may have reduced energy (since they share their energy with a neutral fragment of the original ion). Most of these ions will nonetheless reach the detector, producing peaks in the spectrum far from their proper m/e position, and thus such peaks can be difficult to interpret. Further, they may end up with a range of energies, so they may make broad peaks and extended features in the spectra. Ions that are so energetic that they penetrate the ion mirror far enough to hit the far wall inside the Reflectron are not detected. Elastically scattered and most inelastically scattered primary ions will fall into this category.

In Reflectron based instruments, the sample is kept at a ground potential, and the acceleration voltage is provided by the difference in potential between the sample and an extractor nozzle 1-2 mm above the surface. The primary ion beam skirts the extractor nozzle on its way to the sample, and thus the extractor affects the primary ions’ flight path, but not its energy. The extractor voltage is reversed when between positive and negative ion modes (as are most of the voltages on the ion beam optics within the spectrometer). The primary ion source is largely unaffected by the change in instrument polarity, except that the ion beam deflection needs to be shifted to make up for the reversed effect the extractor has on the primary ion beam.

Secondary ions, produced by primary ion impact, begin life directly beneath the extractor, and are accelerated by the voltage they see. For conductive samples where the surface potential is indeed at the ground level of the sample holder, the extraction voltage is the same as the acceleration voltage, as long as the spectrometer as a whole has that same voltage. IonTof reflections can be run in a variety of modes with different settings and their consequences.

The major tradeoffs are between mass resolution and transmission. If all ions had a low enough energy such that their energy differences could be compensated, and if they all left the sample exactly normal to it straight into the extraction nozzle, virtually all the ions would be detected. However, as noted above, secondary ions do not always leave normal to the sample. Most have some velocity that will take them out of the center line of secondary ion flight paths leading through the spectrometer, either due to their natural angular distribution, or due to sample topography. There is then a race between the secondary ions' divergence from the center path and their progress towards the detector. The slower the progress to the detector, the higher the probability that secondary ions will move so far out of the main flight path that they hit something and do not make it to the detector. Higher spectrometer voltage speeds the secondary ions along, reducing the chances of their being lost and increasing their chance of detection. Higher speeds, though, lead to lower mass resolution, since the secondary ions arrive at the detector less dispersed in time. The collection of ions that have strayed from the center line also reduces mass resolution, since any ion that takes a longer fight path to the detector will arrive later, broadening that species peak.

In summary, tradeoffs can be made with the combination of the extraction voltage and spectrometer (acceleration) voltage to preferentially improve mass resolution, lateral resolution, or transmission, just not all three at the same time.

Changes in the extraction voltage also affect the field of view, the maximum area from which ions can be extracted. Modern instruments are equipped with stage raster capability, and the software seamlessly stitches the data acquired at each stage position into a single data set. The adjustment of the extraction potential is therefore not needed for the analysis of larger sample areas.

Samples with significant topography are problems for TOF-SIMS instruments in general. There are several approaches to solving this problem.

• Discrete samples should be mounted in such a way as to produce an essentially flat surface from which to extract ions. When mounting magnetic recording heads for failure analysis, for example, a head of interest may be mounted adjoining a set of heads with the same height. In this way, the drop off at the head’s edge is replaced with an effectively continuous surface.

• Sample tilt can be employed to effectively extract ions from an edge or a topographical feature. This will tend to reduce the signal from the overall sample, so it is best employed when the signals to be obtained from a sample feature are the main points of the analysis.

• A pulsed late extraction field can be used to effectively eliminate the effects of topography. The trick is to have no extraction potential when the primary ions strike the sample. By allowing a delay before the extraction field is applied, secondary ions have time to drift away from the sample surface with the few eV they naturally have from the sputter process. There is also no need to use short primary ion pulses in this mode. Thus, one can use modes of the primary ion source that produce small spots and therefore high lateral resolution without bunching and the accompanying chromatic aberrations. When the field is finally applied, the secondary ions are not close to any sample topography and see the straighter field leading into the spectrometer exclusively. Setting up pulsed late extraction requires adjustment of the spectrometer since the ions will have moved some distance towards the extractor while there is no extraction field. When the voltage is switched on, the ions will see a smaller voltage being part way between sample and extractor. They will receive less energy unless the acceleration voltage is changed to compensate. They will also be starting later. Low mass peaks (< 40 amu or so) are lost when using this method. Mass resolution is intermediate, much reduced compared to the regular capabilities of the Reflectron, but much better than would be obtained normally with an un-bunched beam. There is also some loss of transmission in general.

• The latest M6 reflection spectrometer from IonTof has an added lens inside the spectrometer, placed just before the reflector. When activated, this "topography" lens acts to drive secondary ions that have strayed from the center line through the spectrometer back towards the center. This has the effect of improving transmission, especially in circumstances where surface topography has acted to drive secondary ion emission trajectories away from the sample normal. Capturing ions that stray is bad for mass resolution, however, since these ions arrive later and broaden peaks in the spectra.

The primary ion beam is pulsed at intervals that allow time for the slowest ion in the mass range desired to reach the detector before the next pulse is initiated. The smaller the mass range desired in the analysis, the faster the system can pulse, and, therefore, the greater the data rate and the quicker the analysis will be completed at any given primary ion flux. The mass range the analyst chooses may be less than what the sample will produce. To eliminate secondary ions with masses greater than the range chosen by the analyst, secondary ions with larger masses are deflected out of the flight path. Without this deflection, higher mass ions would appear in the next spectrum at a false position, creating false peaks in the spectrum.

A secondary electron detector is also present in the system. Because the sample is at ground potential, secondary electrons can leave the sample in either instrument polarity, but in negative ion polarity the extractor cone will attract them, and foil detection by the secondary electron detector. Electron images (most often used for tuning) are thus best obtained in positive ion mode.

The raster of the primary ion beam across the surface changes the origin of the secondary ions relative to the extractor ion optics. The resulting variation in the flight path can change the relative flight times. Deflection plates placed after the extractor optics in the flight path can be used to guide the secondary ion back down the center of the spectrometer. The voltages on those plates are varied with the predictable location of the primary ions within their raster. This trick, known as dynamic emittance matching (DEM), significantly improves the spectrometer’s mass resolution.

Once they are accelerated and lined up inside the spectrometer, the secondary ions are allowed to fly unaffected until they reach the electrostatic mirror that is the heart of the Reflectron TOF spectrometer. The modern Reflectron has more than one ring acting as an ion optic component, allowing for some fine-tuning of the energy compensation that the Reflectron provides. The Reflectron also needs some adjustment for use with the delayed extraction method described earlier and for analysis of insulators. The Reflectron is slightly canted so that instead of the ions being directly reflected back at the sample, their flight is shifted a few degrees, so that after traveling through another flight tube in which they pass unhindered, they reach the detector.

Before the detector, the secondary ions receive an added boost in energy known as postacceleration. The reason for this is that higher mass ions can end up reaching the detector with much lower velocities, and so even though their energies are the same as those with small masses, with their lower momentums they do not always create a pulse at the detector that is sufficient to be registered as a count in the spectrum. By applying anywhere from 5 to 20 KeV of extra energy (but most typically 10 KeV), the higher mass ions are not discriminated against at the detector, and thus the end up being detected as sensitively as lower mass ions.

The detector in the IonTof instruments is a channel plate and scintillator assembly that works in conjunction with a photomultiplier (Niehuis et al. 1987). The secondary ions hit the channel plate producing secondary electrons. The potential across the plate drives the electrons down the channels, accelerating the electrons as they go. The electrons cannot help but strike the walls of the channels, producing more electrons in a cascade that amplifies the signal. A pulse of electrons exits the plate and strikes the scintillator, producing a burst of light. The light exits the vacuum to reach the photomultiplier tube. The photomultiplier further amplifies the signal, transforming it into an electrical pulse that is subsequently captured by the electronics.

While improvements have been made to this system over the years, it is still true that there is a low but quite discernable background noise produced at this detector. For measurements requiring trace analysis and the best possible signal to noise ratio, this background noise can be significant. In these cases, a high primary ion current, which increases the signal leaving the noise unchanged, can be used to improve the signal to noise ratio. There is no upper limit on the secondary ion current this system will tolerate; so high primary ion currents can be used safely, although this will produce issues with instrument dead time (see below).

The detector is used in a pulse counting mode. The electrical pulse is first sent to a constant fraction discriminator (CFD), which determines whether a signal is intense enough to be counted (to represent a real secondary ion). Beyond checking to see if the signal meets a predetermined threshold, the CFD does not use the intensity of the signal. This is why whether one secondary ion reaches the detector or many do at the same time, the instrument will register only one pulse.

Furthermore, there is a brief period following the registration of a pulse during which the detector will be unable to register the next pulse. This is what is known as “dead time”. Up to a point, the data can be corrected for simultaneous arrivals of secondary ions with the same mass using statistical methods (Stephan 1994), although when the signals get so intense that a particular secondary ion is detected with most every pulse, no correction is possible. In such cases it is often possible to use a less intense secondary ion containing a minor isotope for the analysis. Alternatively, there is a hardware option that can solve this problem. Called “extended dynamic range” (EDR), the method involves the shunting of select secondary ions in the drift tube after the Reflectron so that they pass through a grid of known transparency. The secondary ions in a narrow mass range can then be attenuated by an exactly known factor of 10–100X. In practice, EDR is more useful in depth profile analyses, but it can also be used in situations where a major component needs to be monitored during trace analyses.

The characteristics of this detection system have implications for how the system tends to be used. The primary ion guns the instrument is paired with are optimized in many configurations for current, allowing rapid data collection. Because of the potentially significant background noise level, this also leads to the best signal to noise when using the system. The exceptions to this tendency towards running the system with high primary ion currents are when decent lateral resolution and high mass resolution are wanted simultaneously, or when the ultimate lateral resolution is needed. Running the system “hot” means planning acquisitions properly in advance. At high primary ion currents, the static SIMS limit is reached rather quickly. Dead time correction of the data is essential for almost any analysis.

Instead of an ion mirror, the triple focusing time of flight analyzer (TRIFT) spectrometer, used by Physical Electronics in their TOF-SIMS instruments has three electrostatic sector analyzers (ESAs) (Schueler, Sander, and Reed 1990a, 1990b). Although no longer used as an ion microscope in a commercial instrument, the TRIFT spectrometer has the characteristic that it projects an image of the sample onto the detector, thus allowing it to be used in an ion microscope. Like the Reflectron, it is used in Physical Electronics instruments as a microprobe (where the image information comes from the position of the primary ion raster instead). A schematic of the TRIFT spectrometer is shown below.

It provides energy compensation in a fashion similar to the Reflectron, with higher energy ions penetrating each 90 degree curve farther and thus taking a longer flight path than lower energy ions with the same m/e ratio, nearly exactly compensating for the higher velocities of the higher energy ions and thus producing identical flight times for identical secondary ions. A diagram showing how this energy compensation works is shown here.

Unlike the Reflectron, the TRIFT will not pass ions outside of a 240 eV or so range for which energy compensation is possible. This means that metastable ions are rarely detected in the TRIFT instrument. Since metastable peaks are more often a nuisance than not, this can be an advantage.

In the TRIFT instrument, the spectrometer is kept at ground potential, and it is the sample that is placed at a potential (usually 3 KeV) relative to the grounded extractor so as to provide for secondary ion acceleration into the spectrometer. The difference matters mainly with regard to situations in which that voltage needs to be pulsed. The time constant for the change in potential at the sample surface can vary from sample to sample, and in any case it involves charging up the entire sample holder, an object larger than the extraction element. The most common case in which such pulsing is required is for charge compensation. Variations in voltage have the effects noted earlier with the Reflectron, namely, altered flight times, mass resolution, transmission, and field of view.

When the TRIFT spectrometer is used as an ion microscope, the ions travel through the extractor directly into the grounded field free flight path leading to the first ESA. However, such a flight path length varies somewhat with the exact origin of the ions at the sample. Better mass resolution is obtained by using scanning plates to center the ions on a single path headed into the spectrometer, the same dynamic emittance trick (DEM) used in the Reflectron. This results in improved mass resolution. DEM can be turned off for instrument alignment and troubleshooting.

Secondary ions and electrons can be shunted into a detector by a voltage pulse placed on a deflection plate placed part way down the first part of the secondary ion flight path. The primary use for this detector is in tuning the primary ion source, but it can also be useful in efficiently capturing secondary electron images when the system is in negative ion mode. In positive ion mode, electrons are brought right back to the positively charged sample. Note that the polarity in which secondary electron images can be obtained is different for the two main spectrometer types (Reflectron and TRIFT).

While the TRIFT spectrometer has the advantage of better acceptance of ions with momentum in the plane of the sample as described above, the acceptance of such ions degrades the mass resolution. The user can choose to block these ions with more divergent flight paths by moving an aperture known as the contrast diagram (CD) into the flight path ahead of the first ESA. With the CD in place, mass resolution is improved with some transmission loss. On flat samples, this has a larger effect on atomic ions, which tend to have higher energy ranges to start with, and, therefore, represent a higher percentage of ions leaving the sample with significant velocities in the sample plane. Molecular species tend to be much less affected. Consequently it is also true that the CD improves mass resolution for atomic species more than it does for molecular fragments. As a corollary to this, the transmission loss when using the CD tends to be greater for atomic ions.

After rounding the 90-degree bend produced by the first ESA, the ion flight paths now vary with energy. The TRIFT has an optional slit available that can be moved in to partially block the beam, allowing secondary ions with more limited energy range to pass. You can block only the higher energy ions (which amounts to the same thing as reducing the Reflectron voltage allowing higher energy ions to hit the back wall), or you can move the slit into place selecting ions with higher energies. The energy slit has been used in studies of ion formation (Delcorte and Bertrand 1996), for example.

After the final two ESAs, the secondary ion beam is now pointed directly at the detector. The detector in the Physical Electronics TRIFT system is of a type that will be damaged by high secondary ion currents. A deflector plate placed along this line to the detector can be used as a high mass blanker, selectively removing very intense secondary ions that are of less interest (or which can be effectively replaced by ions containing less common isotopes), and thus allowing analyses with intense primary ion beams that otherwise will not be tolerated by the system.

Postacceleration is as necessary in the TRIFT as it is in the Reflectron. Similarly, it is placed just before the detector. Different though is the addition of additional raster plates in this region. To further reduce consumption of this more fragile detector, the secondary ion beam, which unaltered would be focused to a small spot on the detector, is rastered across the detector with a raster size and frequency that is unrelated to the primary ion beam raster at the sample. The result is that the secondary ions fall on the detector in a square region that can be steered independently by the user. Reduction of the raster size can improve mass resolution, but at the expense of the lifetime of the detector.

The first part of the detector is the same as in the Reflectron, a channel plate. But in the TRIFT, behind the first channel plate is a second channel plate, which magnifies the first pulse into a larger pulse of electrons, a pulse that can be directly evaluated by a constant fraction discriminator (CFD). As in the Reflectron, a pulse is counted as a single secondary ion, so the detector is effectively dead to further arrivals for a short time, secondary ions with the same mass to charge ratio arriving at nearly the same time. It is also the same for the binning of the pulses in time channels by the time to digital converter (TDC). For tuning and alignment purposes, a phosphorescent plate is also part of this detector assembly, along with a camera that can see the light it generates. The image of the secondary ions’ impacts on the detector allows the operator to ensure that the secondary ion raster at the detector is well centered on the “sweet spot” where mass resolution is best.

The double channel plate detector described here has characteristics that make for important differences in the way the TRIFT instrument is used in comparison with the Reflectron and its channel plate and scintillator detector. The detector can be damaged when secondary ion currents exceed 5 × 10⁴ counts per second. Consequently, the system is paired with primary ion sources that produce less current, and analyses tend to take longer than in the Reflectron. Dead time corrections of the data are less necessary and can be avoided entirely in this system where primary ion currents are routinely controlled to avoid high secondary ion currents. Lower data rates encourage real time monitoring of the data, since early hints in the ion images that, for example, the raster could be better centered on the area of interest, can be acted on while the analyst is still far from the static SIMS limit. Higher data rates can be obtained by blanking out the most intense ions. This is more of use when there are only a few of them, for example, the Si⁺ ion from a silicon wafer.

One of the perennial problems with using TOF-SIMS for depth profiling applications is that the pulsed primary ion beams used for analysis etch the sample only very slowly, precisely because they are pulsed. In any case, for inorganic depth profiling applications it is generally better to analyze with a preferred analysis beam while etching with a different beam. However, while the sample is being etched, the material removed is not analyzed, so in fact only a fraction of the material in the sample sees the analysis beam. When cluster beams showed promise for 3D molecular analysis, it still remained true that much of the sample being analyzed would have to be thrown away when a conventional TOF-SIMS instrument performs the analysis. This is particularly frustrating for analysts attempting to study biological samples, where there can be precious few analyte molecules present in the sample to start with.

The design of the J series instruments of Ionoptica, is such that continuous or nearly continuous beams can be used on the sample, and the pulsing needed for mass analysis is provided by the spectrometer itself (Fletcher et al. 2008; Rabbani 2010). This instrument effectively separates the creation and extraction of secondary ions from their analysis. The primary beam need not be pulsed, which can allow for better lateral resolution with the cluster ion guns that, when pulsed, are poor imaging sources. Even though these smaller spot cluster sources (with a specification as of this writing of 1.5 microns) have very low currents, the fact that they do not have to be pulsed still allows them to produce decent data rates. Topography at the sample does not play a role in the mass resolving power of the spectrometer. Ion extraction is optimized simply to extract secondary ions as efficiently as possible. A diagram of the first in the series, the J105 spectrometer, is shown below.

Source: Adapted from (Fletcher et al. 2008).

After extraction the secondary ions pass through an RF quadrupole filled with N2 gas, then through an electrostatic analyzer (ESA). The gas collisionally cools the secondary ions, reducing the energy spread induced by their extraction. The ESA filters the secondary ions so that their energy spread is reduced to less than 100 eV. This is clearly a step that can reduce the transmission of secondary ions. Further, the transmission effects will be mass dependent. If collisional cooling is optimized to allow transmission of high mass ions, low mass ions may not even make it to the ESA. If the cooling gas pressure is optimized to allow lower mass ions to exit the quadrupole, higher mass ions may not be cooled sufficiently to exit the ESA. In effect, this is one place in the spectrometer where the instrument needs to be optimized for the mass range of greatest interest in a given analysis. Note that the secondary ion beam remains continuous (or nearly so) at this point.

The ions then pass a deflector and enter the buncher. When the buncher is “filled,” it is fired. While it is being fired, the deflector temporarily prevents further ions from entering the buncher. This buncher is constructed with a series of electrodes. Each electrode is given a potential, 7 KeV to the electrode at the entrance to the buncher, down to 1 KeV at the last electrode before the exit. Like the buncher used in primary ion columns to reduce the pulse width of the ion gun, this buncher causes ions that have the same mass to charge ratio, which are caught within the buncher when it is fired, to arrive at a point past the exit of the buncher at a narrow point in time. Naturally, the same energies applied to ions with different mass to charge ratios will have them arrive at different times. The deflector goes off, and the buncher begins to fill again. The cycle repeats every ~100 μs or so, taking ~90 μs for the buncher to fill, and ~10 μs for the buncher to operate. The cycle time is also an adjustable parameter with implications for secondary ion transmission. The secondary ions all had 100 eV or so of energy to start with. Low mass ions will tend to be travelling faster than high mass ions. The first and fastest ions to enter the buncher at each cycle will overshoot the buncher before it is fired and be lost. The instrument could be cycled so fast that no low mass ions are lost, but then the mass range and sensitivity to high mass ions would be affected. Instead, the instrument, by design, is intended to maximize the transmission and detection of the higher mass ions that are more likely to be molecular ions or large fragments. The result is transmission loss of lower mass ions.

The secondary ions are now focused in time just at the entrance to the next element in the spectrometer, a quadratic field Reflectron. In fact, a detector in this position would record a complete mass spectrum, albeit one with less resolution than that obtained using the added Reflectron. Unlike the Reflectron described previously, this portion of the J series instruments are not just free field regions with electrostatic mirrors. Instead, the electrostatic field increases as the square of the distance from this Reflectron’s entrance. The quadratic field Reflectron (also called a Harmonic Field TOF Reflectron) is needed because the energy spread of the secondary ions at the Reflectron’s entrance is now 6 KeV (the difference in potential between the maximum and minimum voltages the secondary ions see inside the buncher).

The detector used in this instrument is similar to what is used with the more common Reflectron-based instruments; a multichannel plate, a scintillator, and a photomultiplier. Pulse counting is still the normal mode in which this detector is used, so in the case where saturation due to intense secondary ion yields is found, the solution is to reduce the primary ion current. The “DC” nature of this instrument increases the chance of such a problem, while the reduced transmission for lower mass ions reduces it.

IonTof in conjunction with the UK's National Physical Labs (NPL) has designed a hybrid instrument with two spectrometers on it, one a Reflectron TOF-SIMS, and the other an Orbitrap. The Orbitrap is not a time of flight instrument, but given that it is measuring secondary ions in an instrument that is otherwise a TOF-SIMS, it must be discussed here. The Orbitrap delivers orders of magnitude higher mass resolution and accuracy than is possible with any TOF analyzer, and since it does not require that the secondary ions be pulsed, it can analyze ions produced by a DC beam such as an otherwise optimized primary ion source can produce. This lends it particularly suited to depth profiling with a GCIB.

The Orbitrap is, however, also orders of magnitude slower, so that imaging in particular will take a good deal of instrument time. The 2 methods can be used in conjunction. An image can be produced with the TOF, and then a DC beam of a cluster beam be used to erode the sample, with the Orbitrap taking data during that phase, then an image can be obtained using the TOF on the next layer. However, if the Orbitrap's resolution is required for proper imaging of ions with similar mass, very long acquisitions will be required.

Ionizing radiation produces a charge in insulators that is not quickly dissipated without assistance. If the potential at the sample surface is not controlled, the extraction of secondary ions cannot be adequately controlled. TOF-SIMS analysis of insulators thus requires an added effort over the analysis of reasonably conducting samples.

Because pulsed primary ion currents are low, not a lot of conductivity is needed to make TOF-SIMS analysis possible without added measures. Many glasses, for example, can be analyzed without added charge compensation because the ionic conduction available in those materials is sufficient. Similarly, semiconductors rarely pose a problem. In fact, many surfaces that would be difficult to analyze by Auger, or without a coating by SEM, can be analyzed with little difficulty by TOF-SIMS.

There are two aspects to the compensation for charge in the TOF-SIMS instrument. The first involves the neutralization of surface charge. The second requires compensation for any differences in the surface potential of the possibly not completely neutral surface.

A beam of low energy charged species accomplishes the neutralization of charge (Briggs et al. 1990; Hagenhoff et al. 1988, 1989). In the Reflectron, a simple low energy electron beam does the trick. The electrons need to be low in energy so that they don’t themselves act as an ionizing radiation and damage the surface. The electrons basically pass the sample by, except when they become attracted to positive charge at the sample, at which point the incoming electron neutralizes the charge. Since the sample is grounded, the low energy electrons can reach the sample readily as long as the nearby extractor nozzle is pulsed to 0 V. In the TRIFT instrument, low energy Ar ions supplement the low energy electrons in newer versions of this instrument. As mentioned above, the sample in the TRIFT is raised to the accelerating potential when the primary ion pulse hits the sample. The sample then needs to be brought to ground before any low energy species can be brought to the surface and still remain at low energy. This intrinsically makes the elimination of charge in the TRIFT more difficult, but the addition of the low energy ions effectively overcomes this difficulty.

While a stable dynamic equilibrium between charge creation by the primary ion beam and charge neutralization by low energy particles can be created, it is generally true that the result is not a completely neutral surface. In both systems, the voltages need to be adjusted so that the actual extraction potential is what is intended. In the Reflectron, this means adjusting the acceleration voltage to compensate for the (stable) dynamic charging of the insulator surface. In the TRIFT, where the sample is not intended to be grounded, in addition to the charging of the surface, there is the issue of potential loss across the thickness of the insulator. If, for example, the extraction voltage is meant to be 3 KeV over a 2 mm gap and the sample is an insulator 1 mm think with a conductor beneath it, the conductor’s surface will be at 3 KeV, but the insulator surface, which is one-third of the way closer to the extractor, will be only at 2 KeV. By raising the sample voltage, the insulator’s surface can be adjusted to be at the correct 3 KeV. This obviously works best when the insulator is a flat material with a constant thickness over the area to be analyzed. Alternatively, a thin conductive grid can be placed on top of the sample. The grid itself will be at 3 KeV, effectively pinning the surface close to 3 KeV. Unfortunately, it can be difficult to analyze specific spots on the surface of an insulating sample with a grid.

MS/MS is the method by which the ions in a given m/e range in a first mass spectrum (MS1) can be selectively fragmented, and the resulting fragment ions analyzed to produce a second mass spectrum (MS2). At best, the exact mass of a peak in the MS1 spectrum can lead to the determination of that species' empirical formula. MS2 derived from the fragmentation of ions in that peak can usually be used to determine the original structure of the ions responsible for that peak. The MS2 spectrum is not the result of a sputter process, but is much simpler, usually involving a collision process, and thus produces spectra that, given an identical source ion, will give very similar spectra independent of sample, primary ion, and other variations of the SIMS technique. MS2 spectra are therefore more amenable to being matched to spectral databases. In fact, MS2 spectra obtained from SIMS instruments are sufficiently close to those oobtained from other mass spectrometers that the enormous databases that have previously been created can be used. The MS/MS method thus pretty much solves the problem of unknown species in SIMS spectra.

Once secondary ions have passed the reflector and are on their way down the last portion of their path to the detector, they have become nearly as separated in time as they are going to be, and a set of plates can be activated to divert some of them out of that path and around a 90 degree turn. This is now a pulse of secondary ions of similar mass and with the almost the same energy. These are passed through a collision cell with a low pressure of a gas of choice (typically He in standard operation). Ion fragments of the original species created via collisions in this cell continue along this path and separate again in time by m/e ratio. The instrument here uses a simple linear TOF system (just a detector at the end of the tube) to capture an MS2 spectrum. The mass resolution in this MS2 spectrum is generally poor but usually sufficient, given what is already known from the better resolved MS1 spectrum.

The TRIFT systems already had a deflector placed past the last ESA whose original purpose was to excluded very intense ions from the spectrum in order to reduce the total ion current arriving at that more fragile detector. In order to do MS/MS, this deflector is used to send ions in an ~ 1 amu mass range off at a 45 degree angle. A collision cell with an appropriate low pressure gas induces fragmentation, and the ions induced are separated with a linear Time of Flight spectrometer (a straight field free tube from this point on with a detector at the end). The 45 degree angle reduces the secondary ion energy from its original 3 KeV to 1.5 KeV. This is a higher energy than is used for most MS/MS experiments, but the results do not appear to be so different as to hurt efforts at matching MS2 spectra from this instrument with existing database entries.

Because the ions are focused in time even before the Reflectron, it is possible to do MS/MS experiments in this instrument. A pulse of a gas intersecting the secondary ion beam at just the point in time at which a particular secondary ion (ions with a specific mass) will cause collisionally induced dissociation (CID). The analysis of the resulting fragments can help with structural assignments for the original secondary ion.

The Orbitrap spectrometer is preceded by a quadrupole mass filter which can be used to narrowly or broadly filter the ions entering the spectrometer. The filter can be set to allow only a narrow range in mass of secondary ions. For the purposes of performing MS/MS, ions at a single amu can be selected, sent through a collision cell, and the fragments analyzed in the Orbitrap.

The spectra produced in a TOF-SIMS instrument from a given sample will vary substantially depending on the primary ion source used. Light monatomic ions produce spectra with relatively more fragmentation. Cluster ion sources produce spectra with relatively more intense molecular ions and higher mass fragment ions. The variation in relative ion intensities itself provides information about the sample.

The Gentle SIMS (G-SIMS) methodology exploits the differences between spectra obtained with different ion sources (Gilmore and Seah 2000). G-SIMS is based on the idea that the trend in spectra from ion sources producing more fragmentation to those producing less can be extrapolated to produce the spectra that would be produced in the scenario in which the sputter event was so gentle that molecular ions would be emitted from the sample with virtually no fragmentation at all. G-SIMS is thus intended to simplify the spectra and indicate the ions that are most indicative of the original state of the sample.

The G-SIMS methodology, which is described in detail in Chapter 6, magnifies differences in the two spectra obtained by two different ion sources to a large degree. The primary ions need to be quite different, preferably a low mass atomic ion coupled with a cluster ion. For example, Au1+ and Au3+ are not an ideal pair for G-SIMS analysis, nor are Bi1+ and Bi3+.

The G-SIMS spectrum is generated most successfully when the sample surface is so homogeneous that spectra taken in different locations are identical, and therefore the G-SIMS calculation will not be accidently magnifying differences in the sample itself, rather than the intended ion yield differences of the two sources. For a less homogeneous sample, the two spectra that will be the basis of the G-SIMS spectrum would need to be taken from the exact same location on the sample. This latter experiment is difficult when two different ion sources are used for the analysis.

A single source that can produce both a lower mass atomic ion and a cluster ion is the BiMn source, used in an LMIG (Green et al. 2008). Mn+ is the lower mass atomic ion emitted by this source, and the ever-popular Bi3+ the cluster source. Switching between the two requires only a change in timing within the gun, therefore the two sources will produce identical raster scans on the sample with ease. Not only does this allow the G-SIMS analysis on less than homogeneous samples but it also allows for the calculation of G-SIMS images. The analyst still needs to make the decision in advance to use some of the limited ion flux available in the analysis before the Static SIMS limit is reached for a less informative Mn+ spectrum, so the technique is more likely to be used when larger sample areas, where the static SIMS limit presents less of an issue, are analyzed.