When small quantities of analytical reagents are required to be measured,

A Note About this Web Page

SIMPLIFIED QUALITY CONTROL AND QUALITY ASSURANCE SAMPLES

BLANKS, CALIBRATION CHECK, (CALIBRATION CHECK SAMPLES), METHOD BALNKS, MATRIX SPIKES, SPIKE RECOVERY, AND DUPLICATE CALCULATIONS EXPLAINED

INFORMATION PRESENTED FOR THE BENEFIT OF STUDENTS ASPIRING TO WORK IN THE ENVIRONMENTAL TESTING FIELD AND ALL OTHERS WHO NEED TO LEARN HOW TO DO BLANKS, SPIKES, MATRIX SPIKES, AND OTHER TYPES OF QUALITY CONTROL / QUALITY ASSURANCE (QA/QC) SAMPLES IN A REGULATED ENVIRONMENT. HOW TO CALCULATE SPIKE RECOVERIES, ETC

SIMPLE AND EASY CALCULATIONS WITH EXAMPLES

This information posted by:

Kevin Olsen

Passaic River Institute

Montclair State University

Normal Ave

Montclair, NJ, 07043

OlsenK@Mail.Montclair.edu

Scope and Introduction

The purpose of a quality control sample is to provide independent verification that an entire analytical system is performing correctly. Good instrument maintenance, proper calibration, analyst training, sample preparation procedures, correctly made standards, and good quality reagents are some of the components that must all work together if high quality analytical results are to be generated. Quality control samples can also be used to establish both intra-laboratory or analyst-specific precision and bias.

This web page summarizes the requirements for QA/QC samples (other than duplicate analysis requirements) as set forth in the New Jersey Administrative Code Section 7:18. Other states, the EPA, and various regulatory agencies have similar requirements.

QC/QC Samples are required for all parameters run by the laboratory. For multi-element or multiple analyte determinations check samples must be prepared that include all of the analytes.

Types of Quality Control Samples

Blanks Blank samples are used to confirm that the analytical system is free of analyte contamination or interfering substances.

There are several types of blanks and these are explained in the Frequency of Quality Control Sample Analysis section of this document. The NJAC code has specific instructions for using the results of method blanks. Please refer to the section on Data Management in this SOP.

Calibration check standards, Usually required for metals analysis but may be used for any type of analytical system.

After creating a calibration curve, two check standards are used to verify the accuracy of the calibration. One sample is prepared so that it is at the low end of the concentration range and the other is at the high end.

Matrix Spikes Unlike the Quality Control (QC) check sample which uses an uncontaminated sample matrix, this type of sample consists of the original sample matrix to which has been added a known amount of analyte.

Quality control (QC) check This type of sample us used to verify the accuracy of the analytical system. It consists of an uncontaminated sample matrix (water, soil, sediment, etc) to which as been added a known amount of analyte (a.k.a. the spike). The spike is prepared from a source other than that of the calibration standards.

NOTE: When it is necessary to prepare a check sample using reference standards using a source other than that of the calibration standards, a traceable standard from another manufacturer should ideally be employed. If however there is only one supplier of a certain standard, then a reference standard from a different lot must be used.

Frequency of Quality Control Sample Analysis

Each time a calibration curve is prepared, at least one reagent blank is used. A reagent blank consists of the same solvent and dilution scheme that were used to prepare the calibration standards but without any analytes.

Method blanks are defined as a sample of the solvent system, reactants, dilution schemes and other preparation steps in the published method. This blank does not contain any analytes.

At least one method blank for each run of a sample preparation method. If the number of samples exceeds 20, an additional method blank for each batch of twenty or fewer samples is required. For example, two method blanks would be required for 21 samples and three method blanks would be required for a set of 59 samples.

At least one quality control (QC) check samples is analyzed for each batch of real samples. If the number of samples exceeds 20 per calendar month then one is analyzed for each group of twenty samples.

At least one pair of matrix spike samples are analyzed for each batch of client samples. If the number of samples exceeds 20 per calendar month then one pair is analyzed for each group of twenty samples.

Procedure for Laboratory Samples

Reagent blank Using the same glassware, solvents, and dilution scheme as would be used to prepare calibration standards, prepare a solution as if it were a calibration standard but do not add any analyte.

Note that this procedure ideally would give us a solution in which the concentration of the analyte is zero. But if there are any interfering substances in the solvents or glassware, the results of this blank are not going to be zero.

Method blanks Prepare this blank sample using the same glassware, solvent system, reactants, dilution schemes and other preparation steps through which an analytical sample is taken, but without adding any analytes.

This type of sample is not exactly representative of the sample matrix. For example, suppose a procedure called for digestion of surface water with nitric acid followed by dilution. In this instance, distilled water is substituted for surface water but the procedure for digestion and dilution would be the same as the real sample. In a case where a soil sample is acid digested and then diluted, an empty digestion vessel is used but with the same quantities of acid and diluent as with the real sample. The glassware, heating and cooling regimes, and all other reagent additions are the same as would be used with a real soil sample.

Quality control (QC) check samples Unlike the method blank, the quality control check sample must utilize uncontaminated sample matrix. The amount of analyte added to this sample must be at the low end of the anticipated concentration range.

This type of sample may be purchased commercially. Certificates of analysis for these samples are required.

If a commercial sample is unavailable, obtain a pure representation of the sample matrix and add a known amount of analyte to it. The analytes added to this sample must be reference standards from a source other than was used to prepare the calibration curve.

Clean samples of seawater, soil, sediments, and other matrices are available commercially. After the addition of the known amount of analyte the sample must be completely and thoroughly mixed.

Example calculation:

Suppose we have a set of soil samples in which we expect to find 10 ppm of nitrates. We would prepare a calibration curve from 1 to 20 ppm nitrate and now must prepare a QC check sample. Since we want this sample to represent the lower end of the concentration range, we decide to add a known amount of nitrate so that the final concentration is 5 ppm.

Suppose we anticipate processing 10 grams of each real sample. In that case we would first weigh out 10 grams of the clean soil purchased from a scientific supply house.

Since the final concentration we want is 5 ppm, and parts per million is defined as mg per kg:

(0.05 mg / 0.01 kg ) = 5 ppm

Note that we would need to weigh out less than 1 mg of nitrate. It would be extremely difficult to weigh out such a small amount of anything!

Now suppose we had a commercially prepared, and certified, solution of nitrate at 10 ppm. Since we need to add 0.05 mg of the nitrate ion to the clean soil:

(10 mg / Liter) ( 0.005 Liters) = 0.05 mg

Note that we are working with units of liters since the definition of ppm when working with liquids is mg / liter. Converting 0.005 liters to milliliters we see that the all we have to do is pipette 5 ml of the solution into the clean soil. Of course the soil and the solution must be mixed before we can use it.

Now suppose we do not have a commercially prepared 10 ppm nitrate solution. Remember we need to have 10 mg of the nitrate ion in the solution. So if we simply weighed out 10 mg of KNO3 and added it to a liter of water, the actual concentration of the nitrate ion would be:

KNO3 = 101 grams per mole.

NO3 = 62 grams per mole

K = 39 grams per mole

39 / 62 = 0.61

or 61% nitrate in potassium nitrate

(10 mg KNO3) (61% NO3) = 6.1 mg

To correct for the presence of the potassium we need to go through an additional calculation step

(16.4 mg KNO3) (61% NO3) = 10 mg NO3

Thus we need to weigh out 16.4 mg of KNO3 to obtain a 10 ppm concentration of NO3.

After the analysis it is necessary to calculate the percent recovery of the QC check sample. This is a simple procedure:

(Amount detected / Amount added to the sample) X 100 = percent recovery

In the example above we added enough nitrate ion so that the final concentration of the soil sample would be 10 ppm. Suppose we detected 9.5 ppm:

(9.5 / 10) X 100 = 95% recovery

Acceptable ranges of recoveries are usually found in the published methods.

To prepare a matrix spike sample it is necessary to start with an actual sample from the field. Since we do not know the concentration of the analyte, we have to make an educated guess as to the amount of spike material so that it does not exceed the calibration range.

Obtain two representative sub-samples from one of the field samples. The analytes added to this sample must be reference standards from a source other than was used to prepare the calibration curve.

Example Calculation

As before let us imagine that we are going to analyze a 10 gram soil sample for nitrates. We could simply add 5 ml of the 10 ppm nitrate solution to 10 grams of the soil, mix, and analyze.

But suppose we have a 100 ml sample of river water and we need to spike it with a known amount of chloride ion so that the concentration is about 20 ppm above what is already present in the sample. We can buy a NIST traceable chloride standard solutions at concentrations ranging from 25 ppm to 1,000 ppm.

Generally speaking it is always best to use the most concentrated spiking solution possible and add the smallest practical volume. Here is why.

Suppose we want to use the 25 ppm standard. Ignoring for the moment the amount of chloride already in the river water sample, we see that if we have a 100 ml (or 0.1 liter) sample and want to have a concentration of 20 ppm:

(25 mg / liter) (0.80 liters) = (20 mg/liter) (0.1 liters)

In other words, we would have to add 80 ml of the 25 ppm standard to 20 ml of our sample to get a concentration of 20 ppm plus whatever chloride is already in the river water.

Now you can see that this is going to be an huge problem since we have diluted the river water by a factor of 5. Thus the river water sample will no longer be representative of the conditions found in the field.

Suppose instead of the 25 ppm standard we were to use the 1,000 ppm standard.

(1000 mg / liter) (0.002 liters) = (20 mg/liter) (0.1 liters)

0.002 liters = 2 ml

In this case we only needed to add 2 ml of the standard to bring the concentration of the chloride to 20 ppm above what is already present in the sample. You can see that the dilution to the river water is minimal.

In the next step we will compensate for the addition of the extra 2 mL on the final volume of the sample.

The percent recovery calculation is slightly more complicated.

Suppose that the concentration of the unspiked portions of the original river water sample was found to be 50 ppm and the spiked sample concentration was measured at 65 ppm.

Because we measured it, we know that in each of the 100 ml sub-samples of the original unspiked river water there is 50 mg per liter. We can determine the milligrams of chloride in the 100 ml sample thus:

( 50 mg / liter ) (0.1 liter) = 5 mg

Now lets turn out attention to the spiked sample.

We had 5 mg of chloride in the sample before we spiked it.

We added 0.002 liters of standard at a concentration of 1,000 mg/liter

(0.002 liters) (1,000 mg / liter) = 2 mg

So, we have a total of 7 mg of chloride in the solution:

5 mg + 2 mg = 7 mg

And we have a total of 102 ml of spiked river water

100 ml original sample + 2 ml spike = 102 ml

Thus the final concentration is:

7 mg / 0.102 liters = 68.6 ppm

In this case the percent recovery of our spiked sample is:

(65 ppm measured / 68.6 ppm theoretical) X 100 = 95% recovery

Note that in this calculation we have assumed that the results of the original analysis of the unspiked sample are 100% correct. If the true concentration of the original sample was not in fact 50 ppm, then all of our subsequent calculations will be off. The spiked sample results will tell us if this assumption was or was not correct.

The purpose of the matrix spike sample is to reveal the presence of interfering substances in the sample matrix.

Data Management

Method Blank

Calculate the concentration of the analyte in the method blank. If the method blank result is greater than the detection limit or contributes greater than 10 percent of the total amount of analyte in the sample, then the source of the contamination must be investigated. Once the source of the contamination is found, steps must be taken to eliminate it.

Whenever contamination is found, the data must be qualified in the final report.

This type of error is usually the result of contaminated reagents or dirty glassware. If the sample is properly preserved and the holding times are not exceeded, then new measurements may be made with a new calibration curve, new reagents, and freshly cleaned glassware.

Some quality control regulations specify that the final report must indicate that the analysis was repeated because of contamination. A detailed explanation need not be made on the final report but the results of the investigation must be kept on file. The report must state that the original results were invalidated only after investigation and with full approval of the laboratory director. In other words, do not try to hide the problem.

QC Check Sample

Calculate the percent recovery (%R) for every analyte in the QC sample. Compare the results to the limits listed in the published method.

As with the method blank, then the source of the low recovery must be investigated. Once the problem is found, steps must be taken to eliminate it.

This type of error is usually the result of operator error or an equipment problem. If the sample is properly preserved and the holding times are not exceeded, then new measurements may be made.

Some quality control regulations specify that the final report must indicate that the analysis was repeated because of low recoveries. A detailed explanation need not be made on the final report, but the results of the investigation must be kept on file. The report must state that the original results were invalidated only after investigation and with full approval of the laboratory director. As before, dont try to hide the problem.

Whenever the published method does not give limits the laboratory must establish its own control limits. The procedure is as follows:

Standard deviations must be calculated for n-1 degrees of freedom where n = the number of samples for all of the percent recovery and relative percent difference of the QC Check Samples and the Matrix Spike samples. Ongoing data for at least 10 quality control samples of both types must be used.

For parameters in EPA methods in sections SDW02 or SDW04, the laboratory must use samples that have been prepared at the maximum contaminant level or MCL. For other parameters, the laboratory must prepare samples that are in the middle of the concentration range normally encountered in the analysis.

The next step is to prepare a control chart.

The X axis represents time. Values on the axis are the sequential numbers of analysis, 1 = first batch of analysis, 2 = second batch of analysis, 3 = third batch of analysis, etc..

Analytical results are on the Y axis. The theoretical or true value of the QC samples is an horizontal line parallel to the X axis. The value of the analytical results for all of the QC samples are plotted over time.

There are two other pairs of horizontal lines on the control chart, warning and control limits. Warming limit lines are plotted at the concentration value equal to 2 standard deviations above the true value line and 2 standard deviations below the true value line. Warning limit lines are plotted at the concentration value equal to 3 standard deviations above the true value line and 3 standard deviations below the true value line.

The chart with its horizontal lines will look something like this:

Warning limit (Upper)+++++++++++++++++++++++++++++

Control Limit (upper) *********************************

True value of the QC sample ----------------------------------------

Control Limit (lower) *********************************

Warning limit (lower)++++++++++++++++++++++++++++++

Recall that on a normal or bell shaped distribution:

68% of the observations fall within 1 standard deviation of the mean.

95% of the observations fall within 2 standard deviations of the mean.

99.7% of the observations fall within 3 standard deviations of the mean.

This type of control chart can tell us instantly if a sample result is outside a normal range of results.

Once the control chart is prepared the laboratory records the quality control sample results on the chart. If any percent recovery is not within three standard deviations of the limits, the laboratory must re-analyze the samples in question.