Early Detection of Issues

Save Time and Headache by Following These Easy Steps

Any flow analyzer using absorbance-based measurements is susceptible to contamination and

other chemical issues. Of course, if proper lab protocol is followed, these are very rare occurrences.

However, it is very important to routinely check for these types issues in order to address them in a

timely manner. This diligence can mean the difference between 3 hours vs. 30 minutes of time spent.

Early detection of issues will save time and headache. With this in mind, we have compiled a list of

simple checks to catch any issues before an entire sample rack must be rerun.

Check 1: Calibration Verification

Checking that the calibration curve yields a linear fit of the data is the first consideration. Remember

that the R^2 value must be as close to 1 as possible. One commonly used cutoff point is 0.995 or

greater. If a calibration point is off of the line, fix it! Ignoring the calibration point may yield a better R^2

value, but may be masking a larger issue. Re-pouring the calibration standard is a quick way to

determine whether the standard was off, or if further investigation is necessary.

Check 2: Sample Contamination in the Blank

It is easy to quickly “write off” blank response as carry over from a previous run. If this is thought to be

the case, the operator is better off rinsing the instrument and rerunning the blank. Neglecting blank

response may be neglecting contamination, especially if the calibration curve with the blank is linear.

This positively biased calibration curve not only yields inaccurate results, it can yield negative results.

Consider the following calibration run:

Figure 1: Calibration Run

The calibration curve appears to be very linear, no doubt passing the R^2 cutoff of 0.995. However,

upon inspection of the data, the extreme positive bias of the standards puts them much higher in

response than the data. This in turn yields negative results.

Figure 2: Positively Biased Cal Curve

You can see from the figures that a nice calibration curve does not always ensure quality data. It

becomes clear when comparing the responses of the data to the calibration standards. Many of the

responses are less than the blank. What’s more is that these data points have a discernable response.

Check 3: Drastic Changes in the Absorbance Scale

Always check to see that the absorbance scale is typical. If the response of the calibration curve is much

less than before, there could be an issue. For example, in nitrate measurements, if the response starts to

decrease, it is likely that the cadmium column is starting to foul, yielding a lower reduction efficiency. At

a low enough efficiency, the signal to noise becomes less and less favorable to the point where the peak

cannot be differentiated. A lower absorbance scale means less sensitivity at the low end. A snapshot of

this occurrence can be seen below:

Figure 3: Low Absorbance Calibration

Figure 4: 1 ppm Injection

The calibration curve in Figure 3 is clearly unfavorable, since the points do not fall on the line. The

reason becomes much clearer when viewing the circled calibration point’s absorbance vs time in Figure

4. There you can see that the response is jagged and close to becoming indistinguishable as a peak. The

absorbance scale in the calibration curve ties it all together. Make sure the absorbance scale is as

expected. If not, solving the problem will ensure you produce quality data.