Analyse this!

Analysis procedure

Published 27th February 2017 by Andy Connelly. Updated 9th May 2017.

Introduction

For the majority of analyses there is a procedure you need to follow to ensure you get the best data. Getting this procedure right will give you confidence that the data you have are real and reliable.

Table 5 shows what I would suggest is an ideal run for a set of analyses. In reality, you may not have time for all of these aspects of the analytical procedure. However, you must have good reason to reject any step.

Analysis procedure
Table 1: Table showing an “ideal” example run sequence.

DISCLAIMER: I am not an expert on analytical chemistry. The content of this blog is what I have discovered through my efforts to understand the subject. I have done my best to make the information here in as accurate as possible. If you spot any errors or admissions, or have any comments, please let me know.

Analysis procedure

Blanks

I have written about blanks before. In summary, you need to measure blanks partly so you can check the Limit of Detection. It can also give you an indication of sample contamination.

On top of that, you can add extra blanks (normally a “true” blank – e.g ultra pure water) within with analysis to check for carry over or analysis contamination. This can be especially important in column techniques and where automatic sampling systems are in use as these are analysis situations where carry over contamination can be an issue.

Blanks can also be occur as a point on your calibration curve but only if they fit in with the other values. If your standards are 1001, 1002, 1003, etc. then a blank at 0 would not be appropriate in a calibration curve.

Calibration standards

I have written about calibration standards and calibration curves before. In summary, you need to ensure your calibration standards have the same matrix as the samples and over the appropriate concentration for the samples. Unless the technique is very well understood and defined you need you have at least 5-6 calibration standards.

Accuracy confirmation standards (aka reference material)

A reference material is a material for which you have a reliable value. They allow you to confirm your results are accurate and also that you are measuring what you think you are measuring (selectivity – see below). I have split them into two classes:

  • Certified Reference Materials (CRMs) – CRMs must be traceable. They normally come with a certificate which will give a value and usually also an uncertainty at a stated level of confidence.
  • Reference Materials (RMs). These are sometimes purchased from a manufacturer (e.g. 1000ppm Fe solution) or can be prepared in your laboratory (e.g. NaCO3 powder made into a solution). RMs are materials whose property values are sufficiently homogeneous and well established to be used for:
    • the calibration of an apparatus,
    • the assessment of a measurement method,
    • for assigning values to materials.

Whether it is a CRM or RM they should be as similar as possible to your samples. For example, if your sample is sea water with ~200ppm nitrate your reference should also be sea water with ~200ppm nitrate.

Reference materials also allow you to check how selective your method is, but also how accurate your measurements are. Your results should come out within the uncertainty quoted on the certificate that comes with the CRM. Examples of CRMs include:

  •  Check weights for a balance;
  • A solution of known concentration (e.g. sea water that has been analysed for nutrients);
  • A rock sample of known composition.

If a CRM is not available that is sufficiently similar to your samples then you may have to develop your own reference material. This will be something you can run in every experiment and check that your data is consistent over time. However, it will be difficult to guarantee accuracy with such a standard. Alternatively, you can use spike recovery.

Spike recovery

This method can be used to check for matrix issues in analysis and to check for recovery in extractions (i.e. whether you are getting all of the analyte you are trying to extract from a sample). It is especially useful if no Certified Reference Materials (CRMs) are available. The basic principle is very simple:

  1. Take a sample (ideally a process blank or a sample you have already analysed) and “spike” with a known amount of the analyte of interest
    1. e.g. add 0.1ml of 100ppm Ca standard to river water
  2. Extract and/or analyse the sample and the spiked sample as per the method you have developed
  3. Calculate how much of the spike you have recovered from the difference between the concentration of the two samples
    1. This is normally expressed as a percentage –higher the better [1].
    2. In reality anything between 95-105% is normally very good.

It is a very quick and simple method for checking problems with an analytical method and also for extractions as you can see if the spike carries through the extraction.

Repeats

Ideally, you would calculate an uncertainty for every analysis. The easiest way to do this is to take one representative sample and repeat it 5-6 times or more. This gives you enough data to calculate a statistically valid uncertainty. Clearly, it would take way too much time measure every sample this many times but it may be useful to measure every sample 2-3 times to check for random errors. However, you cannot calculate a reliable uncertainty from 3 measurements.

Samples

Simple, measure you samples. However,…

Check standards

As your analysis goes on it is important to put control checks within your samples. These are to confirm that the initial calibration is still valid and that nothing has change in the system. These can be:

  • Repeats of the calibration standards themselves -ideally these standards would be separately prepared in case you have made an error in your initial calibration standard preparation.
  • Reference material – this is the best option. However, this could potentially be very expensive and may not be possible depending on the analysis.

If the check standard is outside the acceptable range then the system should be recalibrated. The ‘acceptable range’ will depend on the technique and the samples; it is often the tolerances quoted for the CRM or based on the uncertainty of the measurement (e.g. 5%). It is a good idea to add a check standard every 10-20 samples.

In some techniques where variation with time is unavoidable standards are tested regularly and used to ‘bracket’ samples giving regular recalibration as the experiment continues.

Secret standards

If you are not analysing the samples yourself it is a good idea to put secret check standards (e.g. CRMs) in the samples given to an analyst. This allows you to check the work of the analyst and check that your preparation methods are appropriate.

Summary

Following the analysis procedure above can make analysis a slow and long winded process. However, if you do not follow this procedure, or something similar, then you will not have the same level of confidence in you data and nor will the person reviewing your paper!

References

[1] http://www.handymath.com/cgi-bin/recovery2.cgi?submit=Entry

Further reading

  1. Statistics and Chemometrics for Analytical Chemistry, Miller & Miller, 5th ed. Pearson (2005)
  2. Data analysis for chemistry: An introductory guide for students and laboratory scientists, Hibbert & Gooding, 2006
  3. Statistics: A guide to the use of statistical methods in the physical sciences. Roger Barlow, John Wiley & Sons, 1989.
Advertisements

6 Comments

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s