Constructing a Sample File and Performing Data Reduction

CONSTRUCTING A SAMPLE FILE AND PERFORMING DATA REDUCTION

Because the mechanics of data reduction are a direct result of the order that standards, unknowns, drift solutions, and other standard reference materials are analyzed, it is most appropriate to consider sample file and data reduction together.

There is no single correct method to reduce the data acquired during an ICP run. The discussion below outlines several alternatives and is not a comprehensive list. Each of these procedures has been demonstrated to work well. This section is intended to outline some of the general principles and goals of data reduction. We have provided to ODP specially designed Excel spreadsheets to perform these calculations in a fashion that is based on results from Legs 187-189 and at Boston University. These spreadsheets are located onboard the Resolution.

The one commonality to all data reduction methods is that the analyses of one or more SRMs in the analytical run, when treated as unknown samples, yields a result that is in agreement with the internationally recognized values. SRMs can be used in the analysis of rocks and sediments, and IAPSO seawater can be used (for some elements) in the analysis of interstitial waters. For Fe, Mn, and Ba (those elements not contained in IAPSO solution), accuracy can be assessed by comparison to a spiked IAPSO sample.

Sample File

A sample file is constructed by the analyst and is accessed by the ICP and the autosampler to analyze a series of analytes, including unknown samples, drift solutions, blanks, and any other solutions desired. In addition to specifying the physical position of each solution in the autosampler rack, the sample file tells the ICP the order in which these solutions are analyzed. A sample file is not used for calibration. For details on how to build a sample file using the JY software, please refer to the companion "Software Notes" that are located on the Resolution.

The critical component in the sample file is the order in which the samples are run, for this will affect the ability to perform an appropriate data reduction. A typical sample file may look like the following (note that this would be for a short run; a total of up to 30-40 items is more typical):

Drift 1

SRM-1

Drift 2

Sample A

Sample B

Drift 3

SRM-2

Sample C

Sample D

Drift 4

SRM-3

ZIP

Sample E

Drift 5, and so on...

There are several critical aspects to this sample file:

The run starts and finishes with a drift solution.
Drift solutions are run more frequently at the beginning of the run, when the ICP is more prone to instability.
Drift solutions should be run no less frequently than every 20-30 min of instrument time after the first few drifts have been run. Thus, the frequency of drift analyses depends on how many elements are being analyzed in that particular method.
The item identified as ZIP is a procedural blank. Avoid use of the word Blank or Blk in the Sample File, as well as of Standard or Std, as these at times have specific meaning to the JY software and may cause an unintended data manipulation.
Standards and check solutions are spaced throughout the run. For the analysis of rocks and sediments, these may be SRMs; for the analysis of interstitial waters, these may be calibration standards rerun as part of the Sample File, as well as Ba calibration standards.

Data Reduction

There are two main approaches to calibration and data reduction. We have tested calibration routines for both interstitial waters and igneous rocks using the calibration software (Option 1, below). Accordingly, the spreadsheets we have provided to ODP are tailored to Option 1.

Using the calibration feature. If a calibration is performed immediately prior to launching the Sample File, the JY software will provide concentration data for each item as the run proceeds. Thus, the first steps of the data reduction are performed by the JY software. Although the concentrations calculated during the run are preliminary because they have not been drift corrected, the analyst can immediately determine if the data appear to be reasonable as the run proceeds. If not, the run can be terminated and an explanation sought. The final data can be drift corrected using the Excel spreadsheets provided to ODP.
Without use of calibration feature. If the JY calibration software option is not selected, the analyst can construct a sample file similar to one shown above and perform all data reduction off-line using the following steps.

a. Drift correct all counts data, by assuming a constant linear change between drifts.
b. Blank subtract, by subtracting the counts of each element in the Zip item from each unknown.
c. Construct a calibration line for each element, based on the drift-corrected, blank-subtracted counts plotted on an x-y graph against the known concentration of the SRM (or in the case of interstitial waters of the synthetically constructed standard). Because the items were blank subtracted, the calibration line can be forced through the origin (0,0) if desired.
d. Calculate the concentration of each element in each unknown, using the equation of the calibration line derived above.

The main advantages of Option 2 are that it gives the analyst flexibility in all aspects of data reduction and it is not affected by the assumption of zero drift during the calibration because the SRMs used in the calibration are drift corrected along with the samples. In addition to being very time consuming, the main disadvantage to the manual data reduction of Option 2 is that the analyst can not assess if a run has been successful until the run is completed and all the data have been reduced. As solutions are somewhat precious and there is commonly a delay between a series of data acquisition runs and the associated data reduction, this often is not desirable.