Anyone dealing with gas chromatography knows that one has to cope with disturbances and defects. Some of these problems can be attributed to the equipment (e.g. defective parts), while other problems are caused by matters such as the carrier gas, the sample, or the column. The operator/analyst can also be a source of problems and errors, for instance, by handling equipment incorrectly or by inaccurate injection of the sample.
In addition, the various demands made of the instrument and the analysis are not always compatible, e.g. precision, durability, manageability, sensitivity etc. These factors are all interrelated and are considered in the development of the equipment; the designers have to find optima within the matrix of interrelated variables. Thus, every GC has its own particular “strong points”.
No matter how trivial it may sound, to resolve a problem, one must detect it first. If an error or malfunction is detected, one can then
Though many problems are fairly obvious, there are many more that are not so evident. Of course, the latter are more common. A malfunction in the GC is generally recognized in the long run, but other errors and sources of error in an analysis may go undetected for long periods of time. The development of the QC process within the lab is meant to catch even the latter, silent errors with its more comprehensive approach. Some examples:
|Sample pre-treatment:||Data processing:||Chromatographic errors:|
|measurement errors||integration errors||peak overlap|
|preparation errors||improper calibration||detector nonlinearity|
|addition of standard||incorrect chromatographic peak selection||injection errors (decomposition, discrimination)|
|cross contamination||change in detector sensitivity|
|instability/decomposition of sample components|
|variation in analyte recovery|
Errors are generally spoken of as either:
Obviously, any inability to recognize disturbances in results is a serious problem. Thus, it is important to develop effective QA/QC strategies within a laboratory. Quality systems and manuals detailing the steps for analyzing disturbances can be of great use. When documenting errors and troubleshooting analyses, the analyst must work methodically, generally paying close attention to his actions, documenting the steps taken and subsequent results, and editing the list of possible causes frequently. Once again, the whole process is simplified if the lab has a well-documented strategy for troubleshooting analyses.
The following are generally key parts of such a strategy:
In the event of disturbances or deviations, it is important to work systematically.
If this cursory check does not identify the source of the problem, a thorough, systematic search is required. The disturbance should be defined and then associated with the relevant system components or analytical procedures. Of course, the disturbance must be well-defined in order to initiate the search, first in terms of its nature (electronic, chromatographic etc.), then in terms of the part of the equipment or method. Try to find out which parts of a method or a system could be involved in the disturbance and which parts can be excluded.
It is seldom the case that only one part of the equipment is implicated via such a direct analysis; problems can generally be attributed to more than one component. In the latter case, a systematic elimination of the possible sources should be employed.
|column + stationary phase|
|chromatographic||carrier gas; dead volume; sample|
|thermal||injector; oven; detector|
|mechanical/pneumatic||flow control/pressure regulation; injector; connections|
|electrical/electronic||gas chromatograph; data system; printer|
|personnel||injection technique; sample preparation; general operation|
When looking for failures or disturbances, the systematic approach is of utmost importance. When the cause cannot be located after an initial check, the chromatographer must switch to the systematic approach. At first this seems time-consuming and academic, but practice proves that it is the only proper way of handling problems, especially complex problems.
There are a few “golden rules” of the systematic approach.
The golden rules of trouble shooting:
1. Try it one time, and then one more time
2. Install it again
3. Throw it away
4. Write down everything you do
5. Remove the root cause
When the disturbance is supposed to arise from a certain part of the instrument, this part should be replaced with one that is known to be working properly, followed by a test of the system. If the change resolves the problem, remove the malfunctioning part at once. Do not keep it in the laboratory, running the risk that it may be used again by an unsuspecting operator.
Document! It is important that all troubleshooting operations be well-documented, especially in the case of complex disturbances and problems. In such situations, troubleshooting and testing can extend over several days. If one does not record the results well, there is a risk of going around in circles, repeating measurements that have already been completed.
Make a written record of each action in the troubleshooting process in which the symptoms, the tests, the possible solutions, the final solution and all relevant information are recorded. This type of information is also profitable for colleagues (exchanging experiences) and for the future. Nothing is more frustrating than remembering that the current problem has occurred before, but not remembering how it was resolved: keep in mind that problems may recur. Along these lines, remember that every piece of equipment has its own manual, and that the manual should specify the expected lifetime of the equipment.
|gas||part of flow/flow controlpurity|
|injector||temperature setting; installation of liner, septum etc.|
|detector||temperature setting; adjustment; linearity etc.|
|column||min./max. temperature; age; connections|
|operation of the gas chromatograph|
|data processing||computer; data system|
Problem: baseline drifts in the course of a temperature programmed GC-analysis
2. What might be the cause?
3. What can be done to test this?
4. Counter indications?
|1.column bleeding|| |
1a. new column
|1a. column is new |
|Rising baseline in the chromatogram with increasing temperature||2. column |
2a. as 1.
|2a. nothing |
2b. see 1.b. Result is unpredictable
|3.injector contamination||3a. clean injector||3a. no primary cause, possibly other causes|
|4. contaminated |
|4a. attach or replace filter 4b. new gas bottle||4a. filters attached and not saturated |
4b. improbable that gas is contaminated
|The electrometer on the GC gives a rising signal during the program||broken electrometer||a.test electrometer |
|a.expertise missing |
b.no spare available
c. problem seems improbable
It is up to the operator to determine how completely he/she will fill in this scheme.
The advantage of such a large scheme is its thoroughness; the disadvantage is the time needed to fill it in. Note that column 4 can contain more than just chromatographic or technical “counter indications”—logistical and financial arguments (among others) are also relevant. Complex problems may require decisions from lab management or the company as a whole.
Once the scheme is completed in full, deal with the most likely and simplest problems first. These are obvious, as they are the tests without counter indications. If these do not solve the problem, one can attend to the more serious possibilities. Do not forget to report and to document! If needed, refine the scheme with additions.
When using this scheme, one is forced to keep a physical distance from a malfunction or a problem (i.e. one does not start to remedy it immediately), which helps the troubleshooter contemplate and analyze the problem more fully. This is, of course, of lesser interest for relatively simple problems, but is essential in complex matters. Keeping a distance broadens perspective and forces one to work systematically. Remember that efficient and successful trouble shooting is largely done behind a desk. The Problem-Analysis Scheme is a means to that end.
It is advisable to make a quick, cursory check first before delving too deeply into a problem or a system that does not perform adequately. It is essential to make a good survey of all instruments and parts which could provide information, including pressure gauges, control panels, and recorders. Actually, this does not apply only to solving problems, but also to the general use of equipment.
1. Electricity supply
A complete gas chromatographic system is electrically fed from the mains voltage via one or more power points. Check if all connections are correct and if the mains voltage matches the required electric potential. American apparatus nominally require 110 V and will no doubt be converted, so this is seldom a source of problems. English apparatus, adjusted to 240 V, typically do not give any problems at the continental European 220 V. Yet it is possible that this small difference could be a source of problems. In addition to the difference in voltage, there can also be a difference in AC frequency. Compare the mains frequency with the required frequency (usually listed on the back of the apparatus). Consult an electrician or the supplier’s service department in case there is a problem.
2. Gas supply
The gases needed for a gas chromatograph are supplied from cylinders or generators.
3. Temperature control
The temperature control unit consists of 2 or 3 parts viz. the oven, the injector and the detector, often provided with separate power switches and displays. A first check includes these panels: are the on/off switches in the correct position, what is the selected temperature and what is the actual value? A digital thermometer is handy for checking whether the measured values at various locations in the oven are correct.
In addition to the built-in thermostats, most gas chromatographs have built-in maximum temperature controllers for the injector, the detector and the oven. These do not function as thermostats, but as off switches. They serve as emergency protection for the instrument in the event of a problem with the primary thermostats.
In contrast to the injector and detector controllers, which maintain a fixed value, the oven temperature controller can be programmed. In practice the maximum allowed column temperature is dictated by the column’s sensitivity to overheating. Mind that the column can also be damaged if the temperatures of the injector and the detector are considerably higher than the oven, especially if they exceed the maximum column temperature. The stationary phase in the region of the column near the injector or the detector may become damaged, which can cause contamination of the detector (bleeding), among other problems.
Perhaps the most important indicator of problems prior to, during, and after the analysis is the data on the chromatogram. The absence or presence of peaks, the stability and regularity of the baseline, and the peak shapes should be noted any time a GC is used. Peak distortions, ghost peaks, and baseline perturbations occurring after a sample is injected are sometimes mistakenly attributed to the column or the equipment while the sample and/or standards might be the real cause of the problems. Independent test mixtures, with their well-known compositions, provide a simple solution for distinguishing between problems with the GC and problems with the injected samples/standards. If an independent test mixture produces a stable baseline and symmetrical peaks, it may be assumed that the problem is caused by the sample and the standards or by the interaction of sample components/stationary phase/apparatus rather than by the column or the apparatus itself. Column overloading is a prime example of sample problems, and leads to "tailing" peak(s) if the stationary phase is an adsorbent, or "fronting" peaks (pre-tailing) in the case of a gas-liquid partition chromatography separation.
The shape of the chromatographic peak is a valuable source of information. All peaks are symmetrical (Gaussian) and sharp (this is a function of their retention factor and of the column’s plate number) in a properly functioning and efficient system. If blunted or asymmetrical peaks are observed, the following questions should be addressed:
In the event that asymmetry cannot be attributed to overloading (which can be checked quickly by injecting a diluted sample), consider the following:
Tailing of certain (especially strongly polar) components can indicate active phases (adsorption) in the system, e.g. glass fracture or damage to the injector insert or the detector jet.
5. The analyst
Operator error is a frequent source of error in analytical methods. These errors can be quite accidental, and are frequently a variable function of operator motivation, self control, day, hour, weather, and the like. They can also be systematic, such as a certain improper injection techniques, measuring the peak areas by hand, etc. When these errors are minor, it does not matter as long as they fall within the expected variance of a measurement result. However, improper technique and misinterpretation of data can lead to useless results. Such errors are normally tracked down by having the analysis repeated two or three times.
Systematic personnel errors can be discovered by having someone else repeat the analysis. This should not be regarded, although some people do, as a "motion of no-confidence", but as a method to find errors and disturbances which, after elimination, provides better results. The management of a laboratory has the important job of seeing to it that the employees are sufficiently educated and motivated. There should be a personal development plan for each employee detailing the skills learned to date (as a result of internal and external courses and training), as well as the knowledge yet to be gained.
Both knowledge and experience are crucial for troubleshooting. Experience can be acquired slowly with time, but can also result from conversations with experts, like maintenance technicians or chromatography specialists.
6. Auxiliary equipment
If a first check does not solve the problem (or at least locate it), one must delve deeper. The systematic approach to trouble shooting has already been discussed. Isolating the location of the problem is a crucial part of the process. Work from large to small, or in other words:
Spare parts are important tools for isolating the problem. Replace a component of the entire instrumentation with another of the same quality and test the effect. Special additional substitutions can be made to speed the process- perhaps a different column or recorder might simplify a measurement. In addition to these components, the lab should have a supply of the smaller devices, including hypodermic syringes; septa; ferrules; a means to detect failures, such as Snoop or a gas leak detector; a soap film meter or flowmeter; and a thermometer. Finally, and it may seem unnecessary to mention it, it is vital to have good tools (e.g. wrenches of the correct American, English and European sizes) and an instrument manual. The table below gives a list of tools, replacements, and other items that are likely to be of use in a gas chromatographic laboratory.
Table "First Aid Kit for Trouble Shooting" in preparation
Suitable standards are an absolute “must” for trouble shooting-- test samples, test standards, blanks, calibration standards etc. These are all homogeneous and stable mixtures of components whose compositions and qualities are known. Since quality systems dictate that the analytical method used, the gas chromatograph, and the other material and equipment employed should be tested to guarantee the reliability of the analytical results, there should be a variety of standards on hand in the lab. The nature, composition, accuracy, age and the use of standards should be well understood and characterized, which implies adequate (previous) documentation.
Test and calibration mixtures have a certain hierarchy of accuracy and reliability. At the top are the certified test samples and the certified test solutions (test samples contain the analyte in a certain matrix, while test solutions lack a specific matrix beyond the solvent). These certified standards are primarily used to validate a particular analytical method. In addition they can be used in ring tests between laboratories. Certified test mixtures are commercially available to a limited extent. For this reason, when looking for method errors, laboratory test mixtures are more commonly used. These mixtures tend to be more specialized and relevant to the analysis and equipment of a particular lab.
Calibration standards and test standards generally consist of a solution of specific components in a carefully chosen solvent. When standards are to be made from concentrated solutions, keep in mind that concentrated solutions are less subject to deterioration than the diluted product. As such, more dilute solutions are useless as standards. Such deterioration is primarily the result of decomposition, evaporation and adsorption (e.g. on glass walls). The more a solution is diluted, the more significant and likely these effects become. Note that the concentration of the analyte can either decrease or increase. Evaporation of a relatively volatile solvent is typically to blame for the latter. Thus, it is best to work from fresh solutions in the case of volatile solution components (solvents or solutes). Always check the purity of the solvent used. A blank GC analysis gives a definitive answer.
Standards should be lightly shaken after they have been prepared, and then stored in well-closed glassware kept in a cool location. Silanized glass is preferred when glass comes into contact with standards or samples. If the glass is sealed by a septum, chose septa that are provided with a layer of Teflon.
Important Data for Characterizing Standards
The following test standards are recommended:
1. Test mixtures to determine the instrumental detection limit. This value is determined by measuring the signal/noise ratio. See to it that this measurement is carried out under the conditions specified by the manufacturer: column temperature, gas velocity, injection technique and the like are important.
2. Test mixtures to check the status of the analytical column: efficiency, inertness and separation power. Text mixtures may be provided by the column manufacturer, or may be generated in-house. A good test mixture contains a number of inert components, as well as some that are more prone to adsorption. Ideally, the latter should be selected so that they are relevant to the lab’s analyses.
GROB mix is a good test mixture for evaluating columns. This well-known mixture is available commercially and contains a range of components such that hydroxyl and aldehyde-adsorption effects, acid behaviour, separating power, and selectivity are all apparent in the chromatogram.
Problems in quantitation can be particularly troublesome. As usual, one cannot fix the problem unless one finds that there is a problem-- the analysis looks fine, except that the calculated concentrations are not correct. As with any problem, a logical, structured, and systematic approach is required for quantitative issues. This is particularly true, however, when defining the problem. The flow diagram mentioned previously is an excellent aid.
It is good to make a distinction between accidental, random errors and systematic errors. Accidental errors can always be diagnosed by repetition, unlike systematic errors. The latter arise from an incorrect (and repeatably thus) technique, apparatus, laboratory or method. As a result, systematic errors are much more difficult to sniff out. When a particular analyst provides systematically higher quantitative values, it is a personal systematic error and can be diagnosed quite simply.
The situation is more difficult in the case of a systematic error in a method. Even with a ring test, such a failure may not be exposed; if all participating laboratories operate according to the same method, a systematic error in the method goes unnoticed. Even when using different methods, the interpretation of any deviations is not straightforward. When eight laboratories provide a value that deviates from the values of two other laboratories, we may wrongly point an accusing finger at those two labs. Systematic method failures can only be located by means of certified reference materials. If the analytical values of such a standard deviate repeatedly and significantly from the actual value, we have to deal with a systematic error inherent in the method. A ring test employing multiple methods and a single, certified standard is useful for evaluating the performance of the different methods for an analysis.
If a quantitative analysis exhibits poor reproducibility, one should first check to see if there is a trend in the error, or if it is truly just a large random deviation around a correct average. In the latter case, the error is most likely in the injection: a poorly functioning autosampler, a poor syringe or bad manual injection technique, and may be sample dependent. For these kinds of problems, we refer the reader to the section "Syringes and injection".
"Sample chemistry" can cause certain non-random errors. However, integration and injector errors should be ruled out before investigating component-dependent errors. Integration problems can result from peak tailing or inadequate separation of peaks (both of which can result from the sample chemistry). The peak’s start point, end point, and baseline must all be chosen in such a way that the integration cannot become a source of error. If the integration is inaccurate or nonreproducible, the parameters must be adjusted. When in doubt, manual measurement of the peak height or area from a printout of the chromatogram is recommended.
Remember that matrix components deposited or condensed in the injector or in the first part of the column can form active sites onto which sample components can adsorb and decompose. Regular maintenance or replacement of the injector insert solves the problem on the injector side. In order to avoid problems with the column, a retention gap is recommended. This, of course, must be maintained regularly or replaced. Additionally, changes in detector sensitivity can cause variations in peak sizes. This is particularly problematic in Nitrogen-Phosphor-Detectors (NPD’s) and in Electron-Capture-Detectors (ECD’s).
An increase in the peak area can also be caused by the evaporation of the (volatile) solvent from the sample during storage. A previously added internal standard compensates for this effect.
In trace analyses, increases in peak area with repeated injections are sometimes observed. This occurs as active sites in the chromatographic system are passivated by the adsorption of analyte. Essentially, the analyte partitions into the adsorbed “phase” until equilibrium is reached. A pre-injection of a relatively concentrated sample will solve the problem-- adsorption balance has already been created prior to the analysis series.
We have already discussed procedures for finding difficult systematic errors. In the next section, widely occurring systematic errors within gas chromatography will be discussed, in so far as they have not been discussed before. For further sources of errors, refer to table (link to table)
The potential for concentration changes within standards and samples in the course of storage has been discussed above.
Systematic errors can arise from any injection technique, with the exception of on column injection. Such errors are often attributed to the various matrix effects, which can give rise to measured differences between the "dirty" sample and the "clean" standards. Addition of a known amount of analyte to the sample can give information about the effect of the matrix. Small amounts of residual water in organic sample solutions are found to be extremely problematic when using both split and the splitless injection as salt residues or chemical reactions can result. The matrix can also give enormous problems in headspace and purge-and-trap injections. Here, the technique of standard additions can be useful.
Co-elution of contaminants in the standard is a very important source of systematic error. Certainly with multi-component standards, one should be aware of this. The gas chromatographic purity of both the solutes and the solvent should be measured in advance in order to avoid this problem. The determination should be made under the chromatographic conditions of the analysis in question in order to provide the most relevant information. Co-elution is more likely to be a problem in separations employing selective detectors than in those using universal detectors (e.g. FID). If a selective detector is used (e.g. ECD), and a co-eluting pollutant in the calibration standard has a higher response than the analyte with which it coelutes, subsequent quantitative analyses will be systematically low. Co-elution tends to be less of a problem in universal detectors, where detector sensitivity is less analyte-dependent.
The extrapolation of calibration values above or below the range of standards is another source of systematic errors. Such extrapolation assumes that the chromatographic response is still linear at the extrapolated value. Practice teaches that this is a bad assumption. Though it is generally common knowledge that one cannot extrapolate to higher signals, extrapolation to lower signals can also lead to error.
In multi-calibration, quantitation should be limited to signals within the calibration range, unless the linearity is verified outside this region. Since a single point cannot test linearity, one-point calibration is a major source of errors. One simply assumes that the curve is linear and that it goes through the origin. Practice proves that one should make these assumptions very cautiously.
A multi-point calibration of at least five concentrations spanning several decades is preferred. If measurements show that the generated calibration curve is not linear, subsequent calibration within the linear range is required.
See to it that the injection of the calibration standards takes place in an arbitrary order. If the standards are injected in a certain order each time (e.g. from low concentration to high concentration or the reverse), problems caused by "sample chemistry" or cross-contamination can be masked. Random injection with blanks in between prevents these systematic problems.