1: Sample Collection Methods Free

Sample collection is probably the most important step of the whole analytical process because, if performed incorrectly, the whole of the rest of the analysis is pointless. As an example, if the air quality in the middle of the Atlantic Ocean is to be determined, then attaching a sampling device next to the funnel of a ship would lead to a gross error. This is because particulates and/or gases emitted from the funnel would obviously not be representative of the air as a whole. Obtaining a representative sample (i.e. a sample that has the same characteristics of the bulk sample) is the key factor for a successful analysis. Some sample types, e.g. air and water, that are fluid, tend to be more homogeneous than others, e.g. soils. This should mean that sampling is easier. Despite this, the above example of sampling the air over the Atlantic Ocean indicates that errors can still be made. A second example would be the (relatively) simple task of collecting tap water to ensure that it is safe to drink. If the tap is not left to run for at least a minute prior to collection, then it is probable that water that has been stationary in the pipes for several hours may be collected. This water may have leached metals from the pipes and pipe fittings and will potentially have much higher concentrations of analytes than the water that has been supplied from the reservoir. Analysing the unrepresentative water and then making a complaint to the local water company would be erroneous and embarrassing.

For solid samples, homogeneity can be much more problematic. For instance, when sampling a field of dimensions 100 × 100 m, then collecting 1 kg of material from one point in the field is exceptionally unlikely to give a reliable picture of the whole field. Certainly, if there is a corner where car batteries have been buried, then sampling a point perhaps 50 m away will not indicate any problem. Similarly, if the sample is collected from the top 10 cm of soil, then a very limited perspective of historical contamination will be obtained. Depending on why the field requires analysis, then this may or may not be a problem. For instance, if potatoes are to be grown, then soil that lies 1 m below the surface that has been contaminated by an ancient Roman smelting site is not likely to be terribly problematic. However, if building work to construct a new housing estate is envisaged for the field, then soil of at least 1 m depth is likely to be disturbed and brought to the surface so that suitable foundations can be built. This then means that the topsoil has become contaminated and may be unfit to grow vegetables and could become a potential health issue if passed from hand to mouth, smaller particulates being inhaled as dust, etc.

Another factor to be considered is the regularity of sampling. When analysing trace elements in a river, a single collection will potentially give a snapshot of the water quality at that instant but gives no indication of historical concentrations. Similarly, it will give no indication of water quality further downstream or, if there has been a single source of contamination upstream, where that may be, how much it has been diluted, etc. If the river quality is to be monitored regularly or continually, then different or specialised collectors may be required rather than a simple container.

It is therefore necessary to identify what is actually required from the analysis so that an appropriate sampling regime can be formulated. This chapter will attempt to highlight some of the errors and pitfalls of sampling different materials.

As with any analysis, contamination should be kept to a minimum. Therefore, collection devices should not contain significant concentrations of the analytes under investigation. Sample integrity should be maintained, i.e. steps may have to be taken to ensure that the analyte is not lost from the system, e.g. by becoming adsorbed to the collection container walls or lost via passing through low density polyethylene container walls (in the case of Hg). If a speciation analysis is to be undertaken to determine the overall toxicity potential of an element, then steps to ensure no changes in the oxidation state or to the chemical form in which that element is present should be taken. The majority of samples are stabilised in nitric acid because it does not cause analytes to precipitate. However, the trace metals analyst should recognise that many water chemists will also want to monitor the level of nitrates present. If such a requirement exists, then an alternative stabilisation acid will be necessary, so care should be taken to ensure that the analyte is not precipitated (or co-precipitated). In addition, an oxidising acid such as nitric acid may not be suitable to stabilise waters if a speciation analysis is to be undertaken. For example, a paper that discussed the stabilisation of As species in waters prior to analysis was published by Kumar and Riyazuddin.¹  Similarly, if organic analytical chemists wish to measure organic components, then plastic containers will not be suitable for them. Under such circumstances, two separate samples need to be collected: one in a plastic container (for trace metals) and the other in a glass container.

Collection and/or storage containers for the determination of trace metals should be inert, i.e. a pre-cleaned and rinsed container made from plastic. Glass is not suitable because it has active sites on its surface, and these sites can adsorb metal ions. The type of plastic used can also have a large effect. Polyethylene terephthalate (PET) contains significant amounts of Sb, so if this element happens to be one of the analytes, a PET container should not be used. Similarly, polyvinylchloride also contains several organometallic elements and so its use should either be avoided or it should be tested to determine if analytes are leached from it. Most frequently, the containers used are made from high density polyethylene and polypropylene. Different laboratories have different cleaning regimes, but most use a surfactant-based wash, rinsing with ultra-pure water, an acid soak (5–20% nitric acid) for a period of 24 hours followed by a last rinse with ultra-pure water. Lids should be treated in the same way. The containers and lids must then be left to dry – usually not using a paper towel as this can re-introduce contamination.

For river water collection where a limited number of samples intend to be collected, an analyst can enter the water then, facing upstream to ensure that their feet do not disturb sediment that can flow into the container, the analyst can turn the collection container entrance to face downstream so that the water flows around the container and then into it. This protocol minimises the amount of extraneous material, e.g. leaves, twigs, and weeds, that is inadvertently collected. It should be noted that although used commonly, this method is not really suitable for the collection of large volumes of numerous samples because of the problem of transporting possibly hundreds of samples each of a litre volume back to a laboratory. It should be noted that the container should be clearly labelled using a permanent marker pen or some other permanent methodology prior to it being used to collect the sample. This is especially true for the “permanent” marker pens that do not work properly on wet surfaces. Ideally, both the lid and the container body should be marked to ensure that accidental swapping of lids and hence cross-contamination and confusion does not occur. It should be noted that the water sample should be filtered prior to stabilisation using acid; otherwise, trace metals will be leached from any particulates present into the water, hence artificially elevating the analyte concentrations present. Depending on the local geology, if such an error occurs, it is very possible to inadvertently elevate the concentrations of Ca, Fe, Mg, Mn, etc.

Numerous other commercial water sampling devices have been developed including Kemmerer and Rhizon. A good overview of samplers for assorted water types was presented by the US Geological Survey.² 

The cost in terms of personnel, transport costs, etc. associated with the use of “grab” sampling discussed above can be appreciable. Therefore, automated samplers may also be employed. Examples include the Teledyne-ISCO series of samplers and others available from Sigma, Hach and Sirco. These can collect over 20 samples of volume 500 or 1000 mL and can provide high resolution sampling for several days or possibly weeks. However, they are not well suited for operation in freezing weather or for long duration unattended sampling at remote sites. Other examples exist, including a device developed by Chapin and Todd who used a mini-sipper to collect small volumes (∼5 mL) of acid mine water.³  This device did not use discrete containers. Instead, it collected sample, filtered and acidified it on-line and then inserted an air bubble to separate that sample from the next. Over 250 samples could be collected using the device before samples needed to be transported back to the laboratory. A brief review of some of the automated samplers available was presented by Chapin in 2015.⁴ 

In general, these types of samplers give a much better view of the state of the water course since they provide samples over lengthy time periods. If samplers are placed at intervals along a river, it may be possible to identify when and approximately where a contamination event occurs. This is far less likely to be possible with “snapshot” samplers, i.e. the ones that require an analyst to collect the sample manually.

For water collection at sea, Niskin bottles or Go-Flo bottles arranged in a “rosette” may be lowered over the side of a ship and the devices programmed to open and shut at the desired depth. Once the samples have been collected, they should be filtered and acidified as usual. Storage should normally be in containers that have been cleaned as described above.

For virtually all water sample types, care should be taken to ensure that minimal contamination occurs through sample handling. This is because most water types have high concentrations of “major” analytes, e.g. Ca and Mg (plus K, Na, S and others for seawater) but many of the analytes of interest are likely to be present at µg L⁻¹ levels or below. Being at such low concentration, they are at greatest risk of significant contamination. Even seawater that has percentage levels of Na and hundreds of mg L⁻¹ concentrations of other “major” elements, the other analytes may be present at concentrations significantly below the µg L⁻¹ level. Given the ultra-trace level of many potential analytes, the acid used to stabilise the sample should be of very high purity. Many laboratories that cannot afford the expense of ultra-pure acids resort to sub-boiling distillation of cheaper, poorer grade materials. When filtering the waters, the filters should be pre-cleaned prior to use by filtering some dilute acid through them. These washing can then be discarded and the filters rinsed with ultra-pure water. The containers used to collect the filtered samples should also be pre-cleaned to minimise contamination. Wherever possible, all sample manipulations should take place in either a clean room or a laminar flow hood (or both). Laminar flow hoods and clean rooms are areas where air is passed through HEPA filters which remove airborne particulates such as dust. Both are under positive pressure, i.e. the air pressure is slightly higher than that outside. This also enables “dust” to be blown off the analyst before they walk into the clean room and stand next to the laminar flow hood. A laminar flow hood placed in a clean room means that the air present would be doubly filtered. This could be useful since clean rooms vary in quality, with the cleanest being significantly more expensive than the least. Some workplaces may require the analysts to wear disposable suits and head and shoe coverings to minimise contamination further. The clean room often has an ante-room where the analyst enters, walks across a sticky mat to remove contamination from the soles of their shoes and a small bench where they can sit to put on the protective equipment. If a standard operating procedure is available at the workplace, then this should always be adhered to.

Numerous sediment samplers are available and the one to use depends on the amount of sediment required, the depth from which it must be collected, etc. The analyst should be clear of the requirements of the analysis to be undertaken. If a very limited study of the topsoil is required, then a garden trowel may be used to simply collect the top 2–5 cm of soil ensuring, as far as possible, that extraneous materials such as leaves and twigs have been removed prior to sampling. The soil collected may be placed in a labelled container to ensure that the place from where it was sampled (a GPS position or “what3words” would identify the position to within a less than a metre), the depth (if necessary) and other useful information such as the date it was sampled and the name of the analyst collecting it, can easily be identified. The sample collection can then be repeated at other sites over the area being studied. Some hand-held corers can only be used in relatively shallow water, whereas others may be lowered to the bottom of the seabed. Some, e.g. the hand-held corers, collect only small amounts of sediment, whereas some of the grabs are capable of collecting large amounts. Some are made of plastic whereas other are made of stainless steel. In general, the stainless-steel collectors can be used because the concentrations of analytes in the samples are significantly higher than in waters. It is therefore more difficult to contaminate at a significant level. Several soil corers are available commercially and differ in design, length and diameter. It may even be possible to hammer some rigid plastic piping into the soil to obtain a core. Alternatively, soil augers are also available commercially which, again, have different lengths and diameters. All of these samplers will provide a relatively small sample mass. A sediment grab is capable of a collecting much larger masses of material. This type of sampler has two large bucket-type jaws that can be forced together using a strong spring or, for the largest grabs, hydraulic pressure. The largest ones can collect huge amounts of samples. In areas of large rocks or boulders, the grabs have a tendency to become jammed open and their contents washed away during recovery to the surface. However, the hydraulically operated grabs are the most likely to be capable of recovering cobbles and small boulders than any other system. These huge samplers are most useful when geological exploration is being used to identify sources of economically useful materials.

When an extremely large sample size is taken, e.g. using a sediment grab, then the analyst may employ the method of coning and quartering to avoid systematic bias. Coning and quartering may be employed when even several tons of material have been collected and is where the whole sample is placed in an open space forming a mound (a cone). This is then flattened and divided into quarters and two of those quarters that sit opposite each other are discarded. The sample is then collected, mixed and undergoes the cone and quartering process again. The process continues until a manageable sample size is obtained. The method is regarded as being capable of providing a sub-sample that is most representative of the bulk.

The number of samples required to ensure that it is representative of the site should also be considered. An extremely useful guide to the number of samples to be taken for a given area as well as numerous other hints and tips were given in the book by Evans and Foulkes.⁵  Also covered in that text are some sample collection protocols, some sample preparation procedures, quality control and the use of reference materials. This book is not a normal textbook and would be an extremely useful read for students, inexperienced professionals and for even experienced analysts who are not used to collecting samples.

If temporal information is required, then a core should be obtained. Here, the older soil will be at the bottom of the core and the most recent at the top. Once collected, the core should be removed from the corer and wrapped in plastic wrap to maintain integrity. The plastic wrap should be labelled with the place of sampling, the top and bottom of the core and other potentially useful information such as date of sampling, the analyst, etc. It may then be transported back to the laboratory. Once returned, its treatment will depend on what the requirements of the analysis are. It may be carefully unwrapped, sliced into 1 cm sections representing different time periods, each individual section placed in a labelled container and then dried using one of the techniques described below. In that way, a depth profile analysis may be undertaken. It should be emphasised that each section should be carefully labelled with at least the name of the core from which it was taken and the depth.

The type of sediment or soil sampled may determine the size of the sections taken. Peat can be up to approximately 90% moisture. Therefore, once dried, a huge weight loss will result. This may leave a dry weight of only 0.1 g which is less than that normally taken for an analysis. This is a small but actually quite important point because if the analytical instrumentation available to a laboratory is only a flame atomic absorption or an inductively coupled plasma-optical emission spectrometry instrument, then sensitivity may be insufficient to determine all of the analytes. Alternatively, if a small volume of sample digest is produced, the flame-based instrument may not have enough sample to determine all analytes. It would therefore be advised to use a wide-bore corer so that the 1 cm slice actually has a much larger diameter and, hence, mass.

Once collected, sediment and soil samples should usually be preserved prior to analysis. This is to prevent biological activity transforming some analytes into different forms, e.g. bacteria transforming S into hydrogen sulfide. Depending on the amount taken, this may involve simply putting it into a labelled sediment bag, ensuring that the pen used to write on the bag does not become smudged/washed away when water drips out of the bag. The bags may then be oven dried where 110 °C is common for most analytes, but a much lower temperature may be employed for a much longer time if Hg is amongst the analytes to be determined. The drying period depends on the amount of sample to be dried and on how wet it is. Typically, drying 500 g samples of sediments could take at least 24 hours to dry at 110 °C. The analyst should remove the samples from the oven, allow them to cool to room temperature in a desiccator, weigh them and then place them back in the oven for a further period of time, e.g. one hour. The process should be repeated until no further loss of weight is obtained in any of the samples. Alternatively, excess water may be drained, the samples frozen and then the material freeze-dried. As the freeze drier starts to operate, the pressure in the sample chamber decreases as it becomes evacuated. The samples can be regarded as dry when the vacuum does not improve any further. The period required for drying in each case would depend on the number and size of the samples. In general, the freeze-drying process is likely to take longer than drying in an oven, but has the advantage of not running the risk of losing volatile analytes such as Hg. The drying processes are discussed at greater length in Chapter 2.

The two drying techniques described above are clearly only suitable for relatively small sample sizes. If a ton of material has been collected, it is wholly unrealistic to expect to be able to use these two methods for drying. It will be necessary to obtain a smaller sample size while ensuring that the subsample obtained is representative of the bulk sample. A sub-sampling technique such as coning and quartering could then be employed. Once the sub-samples of soil or sediment have been dried, they may be stored until further sample manipulation is required, e.g. grinding and sieving, acid digestion/fusion, etc.

It should be noted that coning and quartering is not limited to just soils and sediments. Consider, for instance, a large hopper load of animal feed that has been prepared and contains several mineral supplements and other materials of different particle size. It will be erroneous to just take a small sample from the bottom of the hopper because those particles of smaller size are likely to have slipped between larger particles and are present at higher frequency at the bottom compared with the top. Taking such as sub-sample would therefore be unrepresentative of the bulk material. Coning and quartering would help ensure that a more representative sample is obtained.

Airborne particulates are usually collected by using a pump to pass air through a filter at a known volume per unit time and for a known time period. Therefore, the volume of air passed through may easily be calculated. The filters and particulates on them may then be acid extracted or digested and the concentration of metallic contamination determined. It is important to note that not all filter types are metal-free, with glass fibre ones being renown for being “dirty”. These can potentially be “cleaned up” prior to use by soaking in acid and then rinsing with ultra-pure water and drying. This though is time-consuming, and hence, many analysts prefer to use cleaner filters, e.g. those made of PTFE. Other analysts use a ‘belt and braces’ approach and employ an acid clean-up of the cleaner filters.

Personal air samplers may also be employed if an occupational hygiene scenario is at play. These samplers may simply be clipped onto the lapel of workers as they go about their business at their workplace. They are demountable and so the filter inside may easily be removed and analysed. As with other filters, if they are touched, the analyst should use gloves or non-metallic forceps.

In both of the above cases, once removed from the holder, filters should be stored in clearly labelled containers that are unlikely to lead to contamination and may be sealed to ensure the filter is not lost. An example could be a Petri dish. Again, the label should be made indelible using either a permanent marker pen or by attaching a sticky label with the writing made using a means that cannot be smudged or erased easily. Both the base and the lid of the dish should be labelled to decrease the possibility of accidentally switching the lids between two different samples.

Gaseous organometallic vapours are rarely determined in air. However, if they are to be determined, then a container such as a Tedlar bag could be employed. Some of these come with metal components, e.g. a stainless steel valve. These bags should either be avoided or tests made to ensure that they do not lead to contamination/adsorption of the analytes prior to analysis.

The sampling of industrial samples will depend on the industry involved, the number of samples prepared per unit time, the standard operating procedures of the facility, etc. A factory producing a million individual blood collection tubes every day will not have the capacity to analyse even 1% of them. Instead, they may just collect about 5–20 samples every week and/or when a different batch of raw materials is delivered. If the analysis of a new batch shows no contamination, but analysis of samples from the same batch 3 or 4 weeks later does show evidence of trace elements, then this indicates that some problem has arisen. Further investigation will be required. It is possible that one (or more) of the machines used for the preparation may be starting to breakdown or wear, hence introducing the contamination. Under such circumstances, traceability is imperative, i.e. being able to identify which manufacturing machine made those samples. Additional samples may then be taken from that machine and tested further to confirm whether or not that machine needs to be shut down and repaired. Alternatively, contamination of the samples could have arisen through random error during either the sampling or preparation for analysis. This further testing would help clarify this. Analysis of new batches of raw materials would identify if any of these were contaminated at an early stage.

A similar scenario is applicable for other industries that produce samples in such numbers, e.g. the pharmaceutical industry. Other industrial manufacturers may produce only a few samples a day. For instance, a steelworks may have several furnaces, each producing different steels. It may be necessary only to collect one sample from each furnace per day. Many tons of steel may be produced from each furnace and so collecting a representative sample is imperative. Historically, this could involve collection of a tiny sample (often using a type of ladle on a long handle) which is then transported back to the laboratory and analysed using either arc/spark optical emission spectrometry on the solid material directly or acid digested prior to another atomic spectrometric analysis. Clearly, this involves an operator coming into fairly close proximity to extremely high temperatures and is time-consuming. The drive for increased efficiency and cost-effectiveness has driven the requirement for on-line analysis and so these sampling methods are increasingly becoming less necessary.

Numerous other industrial sample types exist and each will have their own sampling regime. It may be necessary to analyse the raw materials in the cases of glasses and ceramics when a new batch arrives. Analysis of the finished products will also usually be necessary for quality control purposes. Again, more than a few samples per day may well not be necessary.

In general, each workplace will have its own sampling regime and standard operating procedure detailing how frequently, how many and how to collect the samples for analysis. A junior analyst should stick rigidly to the SOP unless a lab manager or senior colleague says otherwise.