Gas Chromatograph-Mass Spectrometer

Determination of Caffeine in Beverages using SPME-GC-MS


pp. 109-110, 565-577, and 584-586 in Harris text



This experiment will be an introduction to quantitative analysis using gas chromatography/mass spectrometry (GC-MS). The GC-MS is one of the most powerful tools available for the analysis of complex organic and biochemical mixtures.  It is simple to use and provides both qualitative and quantitative information.

As the name implies, GC-MS is a combination of two instruments.  Gas chromatography (GC) separates the components of a mixture, and mass spectrometry (MS) helps to identify these components.  The diagram below shows the basic design of the GC-MS instrument.
Figure 1.

As with all chromatographic techniques, GC separations are based on compounds in a mixture being partitioned between two phases – one mobile and one stationary.  The sample mixture is injected into an oven where it is vaporized.  The mixture is then swept up by a stream of helium gas (the mobile phase) and carried to the coiled GC column.  This column is a 30-meter long thin tube made of fused silica (SiO2), the inner walls of which are coated with a film of liquid polymer (the stationary phase).  As the mixture of compounds moves through the column, the individual components separate from each other as some compounds spend more time in the stationary phase and others more time in the mobile phase.  This partitioning between phases is typically based on the relative polarity of the compound versus the polarity of the stationary phase.  The more polar the compound, the more time it will spend in a polar stationary phase, and the less time it will spend in a non-polar stationary phase.  Also, small compounds tend to move through the column faster than large compounds, just as you would expect.

The GC column is housed in a programmable oven, and we can also use temperature to change the amount of time it takes a compound to move through the column.  The warmer the column, the less time a compound will spend in the stationary phase.  Oftentimes, we can speed up a separation by slowing warming the column after the sample has been injected.  This is called temperature programming.  We start at a relatively cool temperature to allow the compounds to interact with the stationary phase and separate from each other.  Then we increase the temperature of the column to move the separated components through more quickly.

As the individual compounds emerge (elute) from the GC column, they are carried to the mass spectrometer.  As shown in the figure below, the mass spectrometer consists of three distinct regions: ionizer, mass analyzer, and detector.  Compounds enter the ionizer and encounter a beam of electrons emitted from a filament.  The electrons bombard the sample, breaking the molecules into fragments, and turning these fragments into ions.  The fragmented ions then encounter the quadrupole mass analyzer.  The quadrupole consists of four parallel metal rods to which both a dc (direct current) voltage and an oscillating radiofrequency voltage are applied simultaneously.  This causes ions entering the poles to move in an oscillating path.  If an ion has the correct mass-to-charge ratio (m/z), it will follow a stable (resonant) path between the rods.  Ions without the correct m/z, however, will follow a non-resonant path and collide with the rods.  By very quickly changing the values of the voltages applied to the rods, an entire mass range spanning several hundred mass units can be scanned in milliseconds.  Resonant ions emerge from the quadrupole and hit the face of the detector where they eject electrons from the detector surface.  These ejected electrons are carried through the detector, continuing to collide with the detector walls, creating even more ejected electrons and resulting in a measureable electrical current (i).

Figure 2.

The mass spectrometer can be operated in different modes.  In the total ion current (TIC) mode, all m/z values within a given range are monitored following injection of the sample into the GC.  The chromatogram is simply a plot of the TIC versus time. Sometimes, however, it is better to simply monitor one or more selected m/z values.  The chromatogram is then just a plot of a single m/z versus time.  This mode is called selective ion monitoring (SIM) and is especially effective if two compounds with different m/z values are not separated and come off the column together.  We’ll be using the SIM mode for our determination of caffeine (more on that below).

Below is a picture of our GC-MS instrument.  (The helium tank is housed in an adjacent room.)  The column length is 30 m, with a diameter of only 0.25 mm.  The thickness of the stationary phase on the inner wall of the column is 0.25 µm and consists of a film of polydimethylsiloxane in which 5% of the methyl groups in the polymer have been replaced with phenyl groups.  This is typically referred to as 5% phenyl-polydimethylsiloxane.

Figure 3.

Solid-Phase Microextraction (SPME):

To measure the amount of caffeine in a beverage, we must obviously get the caffeine into the GC.  It is generally considered a bad idea to inject aqueous samples directly into the GC.  A number of problems may arise, including backflash of the vaporized solvent, peak shape problems in the chromatogram, salt accumulation in the column, and eventual degradation of the column stationary phase.  We will instead use a technique called solid-phase microextraction (SPME) to remove (extract) the caffeine from the beverage sample.  SPME is a simple method for extracting compounds from solution without the use of more solvents.

The SPME apparatus and process is described in the figure below.  A very thin (and fragile) fused silica fiber coated with a liquid polymer (like a GC stationary phase) is initially protected inside the needle.  The needle is immersed in the sample solution, and the plunger is pushed down and locked in place to expose the fiber to the solution.  Solutes in the solution (including caffeine) adsorb to the fiber surface.  After a set amount of time, the fiber is retracted back into the needle and placed in the GC injection port.  The heat from the injection oven desorbs the solutes from the fiber and the helium mobile phase carries them to the GC separation column to begin the separation. 

Figure 4

Internal Standards:

Since we are interested in determining the amount of caffeine in our beverage, we’ll have to generate a calibration curve for our data.  The simplest way to do this is to prepare a number of solutions with known concentrations of caffeine, and then use SPME to extract the caffeine from the beverage and inject it into the GC.  We would then measure the area of the caffeine peak and construct a calibration curve by plotting peak height versus caffeine concentration. 

There is one serious problem with this simple approach, however.  It is almost impossible to ensure that we can reproduce the amount of caffeine adsorbed on the fiber from a solution.  If we were to run the same sample over and over, we would invariably end up with very different caffeine peak areas for each run.  We overcome this problem by adding an internal standard to our solutions.  An internal standard is a substance we add to the solution in a known amount that gives us a consistent reference for comparing different standard and sample injections.  Now imagine doing multiple runs with a solution containing caffeine and an internal standard.  Both compounds will adsorb to the SPME fiber.  While the absolute amount of each compound adsorbed may differ from run to run, the ratio of the signals from each compound should be a constant.  So when we prepare our solutions, we will add a constant amount of internal standard to both the standard solutions and the sample solution.  The calibration curve is then simply a plot of Sanalyte/SIS versus the concentration of the analyte, where Sanalyte and SIS are the signals from the analyte and internal standard, respectively.  Similarly, we will calculate the same ratio for our sample solution and fit it to our calibration curve.

It can be difficult to identify a good internal standard for a given analysis.  Ideally, the internal standard should give a signal comparable in magnitude to the signal from the analyte.  It must also not be present in the sample matrix.  In GC-MS, it is common to use an isotopically-labeled version of the analyte.  So in our analysis, our internal standard will be caffeine-trimethyl-13C3.  (It costs $650 per gram, so please be very careful with it!)  As shown in the figure below, three carbon-12 atoms in normal caffeine are replaced with carbon-13 atoms.  This gives a molecular weight increase of 3 amu.  Except for this mass difference these two molecules should behave identically, showing identical adsorption behavior on the SPME fiber.  Furthermore, there will be a very small probability of finding the isotopically-labeled form present in the sample matrix (1/100 × 1/100 × 1/100 = 1/1000000).

Figure 5.

Since both forms of caffeine have identical chemical and physical properties, it should make sense that they will not be separated by a GC column. They will “co-elute” (emerge from the column together).  So why isn’t this a problem?  This is where the beauty of the MS detector comes in.  We’ll use the SIM mode of detection described above.  We’ll monitor mass 194 for regular caffeine and mass 197 for the labeled form.  So even though they’ll come off the column at the same time, the MS will allow us to obtain distinct signals for each.