The purpose of the stationary phase is to effectuate the correct separation. The stationary phase is contained within the separation column. It is so-called because the stationary phase material remains immobilized during the GC process. The column contents are contained by tubing, sometimes made of glass or metal, but the dominant trend is towards fused silica because, unlike glass, it is physically flexible. Also, unlike metal tubing, it virtually never catalyzes with the sample.

There are three primary kinds of separation columns, each of which describes how the stationary phase is applied: in liquid form, solid form, and within a capillary tube. The nature of the stationary phase explains the formal designation for each process, namely, Gas-Liquid Chromatography (GLC), Gas-Solid Chromatography (GSC), and Capillary Gas Chromatography (Capillary GC). That designation refers to entire GC process involving that particular column. The difference between each process is essentially how and where the stationary phase is applied. GLC, the most common procedure, requires that an inert “support” substance be coated with the liquid stationary phase before being packed into the tubing. In GSC, the tube is also packed, but it is packed with stationary phase granules, which would be solid in form. In Capillary GC, an comparatively narrow tube is itself coated with the stationary phase. The substance is immobilized not by being packed into the tube, but by bonding to the inside of the tubing itself. The stationary phase adheres to the tubing either by the tubing’s physical properties or by an adhesive-like polyimide coating. The ideal stationary phase tends to have a heavy atomic weight and will result in the components of the sample to become distributed variously between the mobile and stationary phase.

The mobile phase consists of the carrier gas which must be inert to the column materials and the sample to be analyzed. The gases of choice are generally hydrogen and helium because they are typically inert in most situations. Also, they are ideal because they move efficiently due to a low ratio of diffusion to viscosity.

Beyond the specific properties for the column types described above, the operating principals are fundamentally the same among the various GC processes. The gas is already cycling through the GC system. First the sample is introduced into the mobile phase through the injection port. This was once almost exclusively done through a hypodermic-style syringe. Recently an automated form of injection is gaining currency in state-of-the-art labs to reduce error caused by improper operator technique. Moreover, such injectors permit the regulated injection of pre-heated samples and some even permit the regulated injection gasses. Next the components of the sample begin to separate. The components of the sample with more affinity to the stationary phase move at a slower rate relative to the flow of the mobile phase. Because there are different rates of movement, sample components separate into identifiable and measurable groups. Since the liquor or solid samples must be kept in a gaseous state after entering the GC process, the column is heated by ovens that maintain a suitable temperature. Sometimes the heating unit will be used to “ramp,” or gradually and systematically increase the temperature, which can cause the components with lower boiling points to exit the column sooner. Thus, in addition to the stationary phase, another resisting force causing the separation is heat. In due course, all the components exit the column and enter MS unit.

Another analysis measures the peaks in relation to one another, the tallest receiving 100% of the value, and the others receiving proportionate values—all values above 3% must be accounted for. The parent peak indicates the total mass of the unknown compound and reflects the highest mass on the spectrum. Based on this value, the analyst must attempt to fit all the total masses of the other particles into that value. Beyond that, a molecular structure and bonding must be defined, consistent with a substance with the characteristics recorded by GC/MS. This method too is best performed by computer because of possibility for human error and because the computer can exhaust all mathematical possibilities in making the analytical determinations. A computer can effectively and rather instantly do both processes with the lowest rate of error and the maximum confirmation.

A “full spectrum” analysis considers all the “peaks” within a spectrum. However, another method is selective ion monitoring (SIM), which looks only at a few characteristic peaks associated with a candidate substance. A lab does this on the assumption that a set of ions is characteristic of a certain subject. It is fast and efficient which makes the analysis less expensive for the lab, but this is a scientific hypothesis in itself.

If the heat is incorrectly applied, especially due to bad instrumentation, it can adversely affect the reading as well. If thermostat sensors in the heating unit become desensitized, things can be systematically overheated to the point of causing structural changes or decomposition in the sample. It is also important to ensure that any part of the GC/MS assembly is not contaminated by a sample fragment left behind by a prior test. This can happen when a stream is big enough to be accommodated by the GC system, but not by the MS system. This possibility could be detected by running a “blank” test, but huge laboratory backlogs make this problematic, given the length of time that GC/MS takes. Therefore, another way to address this problem would be to ensure that no more than a specified amount ever be placed into the machine. Of course, procedures don’t negate human error unless they are followed. Although more expensive, an automated sampling/injection system may eliminate most forms of human error related to intake. Because automatic injection units can typically be pre-loaded with a string of samples to analyze, this ultimately may lead to increased efficiency. Another, perhaps less costly option for labs that have several GC/MS apparatuses is, as standard procedure, to run confirmatory tests on positive samples on a different machine.

Basic attacks on equipment procedure and maintenance can also be made. Attention should be paid to cleaning procedures and whether they are in accordance with the manufacturer-specified cleaning procedure, in terms of both where to clean, frequency, and methodology. Similarly, one attacking GC/MS should follow up on the regular maintenance issues, such as parts replacements, regular service, and self-diagnostic techniques.

The accompanying mass spectography unit is in a new class of units designated “time of flight” (TOFMS). TOFMS makes its measurements by sending fragments down a low-pressure tube called a “drift tube,” where it hits the detector in order of their mass, and a measurement is made of the time in flight of each. The old class of units could only produce a few spectra per second at their top speed, and quite typically far fewer. 500 mass spectra per second can now be obtained using TOFMS. It also has a virtually unlimited mass range, whereas standard MS has more limitations. These rates allow for allow for accurate tracking of the considerably narrower peaks in generated by high speed gas chromatography (HSGC).

New software allows for “spectral deconvolution” which addresses the problem of overlapping chromatogram peaks where the overlapping components have different fragmentation patterns. However, when it comes to very complex mixtures, some chromatographic peaks may overlap to the point that deconvolution cannot rectify, even with the highest available rate of data acquisition. Luckily, this is limited to extremely complex mixtures (15+ primary compounds). Most useful is the automated capability to automatically define peaks—and in some cases not just relatively, but quantitatively.

GC/MS/MS functions just like GC/MS but with an additional MS step. Some particles will be analyzed with greater specificity in that final step. In the case of an actual two MS tandem, a vacuum pump would be required to transport particles from the first MS and place them into the second. In the case of an ion trap, the ions to be further tested collect in the trap until they are reanalyzed by the MS process.

The “second round” of MS involves isolating an ion of interest and routing it to a collision cell. Within the collision cell, the ion experience collisions from an argon or an inert gas. These “parent” ions are put through the mass analyzer and the resulting fragmentation produces “daughter” ions. This brings an ion of interest into the foreground even in an otherwise convoluted and noisy picture, clarifying the relationship of small ions to larger ions in the spectrum.

The EGIS-type can use an 8-pound portable vacuum system that sucks in air around luggage to be analyzed. The vacuum functions as the sampler of the GC/MS system. Alternatively, a cloth can be used to wipe surfaces and the cloth can be inserted directly into an injection port in the GC/MS system. The sample is analyzed with the push of a button. If the unit detects an explosive, it will activate an “audible, LCD, or LED” alarm.

On the morning of March 29, 1993, a United States Navy surveillance aircraft, a P3 Orion, was on a routine drug-interdiction mission in international waters off the coast of the Dominican Republic. Spotting a low-profile vessel in the waters below, the aviators identified the boat to be similar to the type of vessel commonly used in narcotics smuggling. After the Orion made several passes and witnessed the ship's crew members tossing bales overboard into the ocean, small arms tracer rounds came streaming towards the plane. Soon thereafter, Coast Guard officials aboard a Navy frigate, U.S.S. TAYLOR, intercepted and boarded the low-profile vessel. After an exhaustive search, the American officials were unable to locate any contraband on the boat or on the defendants.

Law enforcement officials, attempting to connect the bales of cocaine with the defendants, produced the Sentor, a state-of-the-art electronic device able to detect the faintest molecular traces of cocaine. The officers, using what looked like a large hand-held flashlight, approached the defendants and pointed the device towards their bodies. The machine began to vacuum in a large volume of air around the defendants bodies. The officers then took samples of the air on board the boat. Within thirty seconds, the drug-interdiction officials were able to detect trace amounts of cocaine on both the defendants and the boat. Based upon this and other evidence, the defendants were arrested and later convicted of smuggling narcotics.

Real scenarios like the one presented above are becoming increasingly common, as the drug-sniffing GC/MS-powered Sentor currently is in use by drug enforcement agents, the coast guard, and the border patrol. It both suggests the benefits of such a system, but also elicits worries. This particular scenario inspires the audience to want the test to confirm what we already “knew”: that the boat crew were ejecting bricks of cocaine into the ocean. We already knew they were guilty. In one scene we go from the Coast Guard officials take air samples from the boat and from the bodies of each individual to the happy ending. In a fast time frame for even Hollywood, less than a minute later, the aviators’ visual observations were “proven” through the magic of GC/MS and the machine called Sentor.

The Sentor is actually a derivative of the EGIS machine described in previous section, but instead of being optimized for explosives detection, it is optimized to detect drugs. The following technical description applies also the EGIS hardware.

Consistent with the GC/MS enhancements described earlier (in section 3.1.1), the unit employs high speed GC/MS accompanied by time-of-flight mass spectography. An interesting enhancement is the portable vacuum air auto-sampler, which was described above as resembling a large, handheld flashlight. Unlike traditional GC/MS where samples are injected with a syringe into the mobile state, the auto-sampler directly fits into an auto-intake/injection system.

The operation of the machine is simple. The auto-sampler is inserted into a machine resembling a refrigerator, the GC/MS process begins, and finally it displays the quantity and amount of drugs detected. Unlike the laboratory environment, there is no room to fine-tune. But there is also little possibility of operator error. This principal of operation, in the words of manufacturer Thermedics Detection, Inc., can be described as “simple one-person, one-button operation.” After all, this application was not designed for scientists.

The foregoing description describes Sentor in the criminal context, largely because of the particular agencies involved with the technology at present. However, Pinkerton Security and Investigation Services has entered a marketing agreement to administer the test in the workplace. It would be run by “specially trained Pinkerton guards” who would analyze everything and everyone in a 15,000 square foot business for approximately $5,000. In this civil context, Sentor-type technology raises some perplexing issues.

Another element for this discussion is that of Federal Rule of Evidence 403. Because of the ability to tie people to certain stashes of drugs based on their chemical makeup, there doubtlessly would be some instances where Sentor-type results would be tried to be introduced as evidence in itself. This certainly was the case for the drug traffickers off the coast of the Dominican Republic. The fact that this evidence “magically” discerns the presence of illicit substances might render it substantially more prejudicial than probative. Because the GC/MS configuration in the Sentor-type situation is essentially a “press here, dummy” configuration, one may be left with the impression that because it so simply indicates the presence of some drug residue, then a person who triggers its alarm must be “simply guilty.” The allure of the magic box, in conjunction with the misconception that people with drugs on them (however slight) are drug users, may cause an average juror to become prejudiced. While we know that jurors express a healthy skepticism to scientific evidence and expert testimony, in the case of this situation, it doesn’t necessarily have all the off-putting qualities of scientific machinery. For one, the Sentor and EGIS are very simple when it comes to their interface. However, this prejudice may be cured by effective lawyering on the other side, not in terms of the technicalities of GC/MS, but rather in showing that how common contamination is.

Questions of first impression about GC/MS/MS were raised in United States v. Campbell. In this case, Private Campbell’s urine tested positive for LSD only under GC/MS/MS at Northwest Toxicology Laboratories, but not under conventional tests at Fort Meade which included radioimmunoassay analysis (RIA). The RIA test indicated some LSD in the urine, but did not rise to the 200 picogram standard established by the military. Only with the GC/MS/MS test did the military discover 307 picograms of LSD, enough to convict Campbell for using LSD. The defense successfully maintained that GC/MS/MS is not reliable under MRE 702, primarily arguing that it did not meet the Daubert test. The U.S. Court of Appeal for the Armed Forces (CAAF) reversed the intermediate court, maintaining the trial court had correctly found that it did not meet Daubert standard for reliability:

“Of critical importance is that the Government did not prove the levels or frequency of error, which would indicate: (1) that the particular GC/MS/MS test reliably detected the presence of LSD metabolites in urine; (2) that GC/MS/MS reliably quantified the concentration of those metabolites; and (3) that the DoD cutoff level of 200 pg/ml was greater than the margin of error and sufficiently high to reasonably exclude the possibility of a false positive and establish the wrongfulness of any use. In particular, the Government introduced no evidence to show that it had taken into account what is necessary to eliminate the reasonable possibility of unknowing ingestion or a false positive.”

Falling far short of an in-depth treatment of treatment of Campbell, this introduction serves to highlight the fact that GC/MS/MS is not GC/MS, even though GC/MS/MS relies on GC/MS. While GC/MS’s is nearly-universally accepted as the “gold standard” in a broad array of applications, one ought not assume that this reverence extends much further.

The code section under which Private Campbell was prosecuted required “knowing use” and this knowing use could theoretically be established upon a certain level of narcotic in the bloodstream. One issue that surfaces in a number of novel GC/MS applications is the meaning of standards. It is troubling that, upon failing to achieve the threshold, one might turn to novel technology which can prove more of a substance is there. It is not automatically clear whether the first test understated the presence of a chemical, or the later test overstated its presence. But it does make sense that, as in Campbell, there should be burden to establish that a test proving the presence of a substance beyond established tests should be required to prove its quantification is reliable in its own regard. Otherwise the application of standards may not be a function of objective testing, but instead a function of “test shopping.”

Sentor, which relies on a technology already proven to be the “gold standard” for drug testing, may be open to a similar attack. Even though High Speed Gas Chromatography (HSGC)/ Time-of-Flight Mass Spectography (TOFMS) upon which Sentor relies is essentially a “souped up” version of first-generation GC/MS, one may question whether it is still essentially GC/MS, based on its hardware. A court may require a separate foundation apart from GC/MS. Adding hardware to increase precision such as an addition MS unit, as above, seems to make GC/MS/MS different enough from GC/MS to warrant independent establishment of its validity. In the case of replacing slower hardware with faster hardware (as with HSG/TOFMS), does that also require a separate showing of reliability from other GC/MS technologies?

The proponent of such evidence may argue that it does not warrant a separate showing because there are many variants among hardware configurations in GC/MS setups that have various characteristics—such as faster or slower speed—and such configurations have always been treated interchangeably in case treatments of GC/MS. Because the procedure is based upon the same essential technology, it could be argued, it should not therefore be treated differently than GC/MS has been. However, perhaps more powerfully, one could argue that Sentor is fundamentally different: HSGC/TOFMS is lower resolution, and while it is widely suggested that it makes up for that fact by performing 500 analyses per second, it remains important that an equivalency be evidenced.

Of course, would Sentor readings really be used for evidence? At first it might not seem so. It might merely be a screening test that gives rise to further confirmatory tests. But think above the scenario detailed above where the Coast Guard employs Sentor to link the accused to bricks of cocaine in the ocean. The fact that the Sentor reported a presence of cocaine, and a mixture of chemicals being identical to that of the cocaine in the bricks is the most incriminating fact in the scenario, as described. Certainly, the analysis of the cocaine in the bricks may be performed later in the laboratory to verify the results of the portable test. However, it is unlikely that the Coast Guard preserved an additional air sample. But this perhaps highlights another point, that even when the identification itself is not very suspect, the source of the sample might be suspect.

The Sentor (and the EGIS for that matter) sample imprecisely, especially when they are analyzing air. To suggest that there is much precision to vacuuming particles from around one’s person is quite suspect. Air is fluid and does not cling to the body like leaves hang on a tree. Contamination with an illicit substance may come from the environment. The particles that cling to someone may be there because they have come into contact with someone else. Individuals are constantly involuntarily and inadvertent bearers of other people’s particles. Along that line of thought, one ought to consider that, in order to activate a Sentor, one need only come in contact with an individual who had handled such drugs recently. Such an individual might be a police officer, a drug dealer, or a pharmacist. Certainly, only a minuscule amount of particles would transfer through casual contact. However, due to Sentor’s use of GC/MS, it is capable of discovering traces of narcotics like cocaine amounting to less than one billionth (0.000000001) of a gram. Moreover, most money in currency is contaminated with cocaine residue. Anyone who constantly handles money could theoretically set off a Sentor.

The concern for incidental contamination is not so far-fetched. It was a concern that was addressed by the Campbell court. Incidental contamination also poses an increasingly monumental danger when the otherwise insignificant contaminants become magnified along with same sample: what was once incidental looms quite large. It is essential to bring this problem into focus before the courts by suggesting, where appropriate, that mountains are being made of molehills. Even in proportional magnification, there is the danger that the sample itself is rather small, and it ceases to be representative of the whole. Imagine a sample from a haystack with a needle in it, consisting of 3 pieces of hay and 1 needle. Bear in mind that unlike a blood or urine sample, some of these more novel sampling techniques will feature clustering of a residue. A gust of wind might “link” someone to a massive drug operation, because a big fragment stuck to your lapel.

Standards can solve some of this problem. A standards-setting organization can determine at what level of reading something can be truly probative. How this would be determined is problematic. But assuming it can be done, perhaps one can come up with a standard that excludes the amount of particles that be merely blown or inadvertently rubbed onto a body. But can the same be said for GC/MS/MS? GC/MS/MS can really look behind a situation that is dominantly “debris” and see the gasoline that was used to burn the house down. But of what probative value is this, when perhaps GC/MS/MS detects not the accelerant in the fire but instead BBQ fluid that had spilled inadvertently into the carpet months ago on the Fourth of July? At some point, detecting too little may be too much for the courts, because alternative explanations can effectively rebut minute samples.