How small a signal counts as real

Science · Chemistry · ages 18-19

Name ______________________ Date ____________

Ten blank readings scatter with a standard deviation of 0.02 signal units. The calibration slope is 0.5 units per ppm. What is the limit of detection in ppm?
Answer: ______________
How do you estimate the noise of an instrument?
- From a single blank reading
- From the steepest calibration slope
- From the spread of repeated blanks
A tight cluster of readings far from the true value is accurate.
Circle one: True False
How large must a signal be to count as detected?
- Three times the noise
- Ten times the noise
- One times the noise
An analyst averages 16 repeated scans instead of using a single scan. By what factor does the signal to noise ratio improve?
- 4
- 16
- 2
- 8
Using the same blanks and slope as before (noise 0.02, slope 0.5 per ppm), what is the limit of quantitation in ppm? Use 10 times the noise.
Answer: ______________
You average four repeated scans. What happens to the noise?
- It doubles the noise
- It halves the noise
- It removes the noise entirely
Blanks scatter with a standard deviation of 0.02 units. The slope is 0.5 units per ppm. What is the detection limit in ppm?
Answer: ______________
Why is a result below the detection limit never reported as a number?
- Noise alone could explain the reading
- Small numbers are always rounded to zero
- Instruments cannot display small values
A sample reads above detection but below quantitation. How do you report it?
- Report the exact number with confidence
- Report zero concentration
- Report detected but below quantitation

Answer key

For grown-ups. Fold this page away before handing over the rest.

How small a signal counts as real W1-mt_NLtCWo_qlB-s1

0.12 · Three times the noise is 0.06, and 0.06 divided by the slope 0.5 gives 0.12 ppm.
From the spread of repeated blanks · Wandering blanks reveal the noise floor.
False · Tight clusters can sit far from the truth.
Three times the noise · Only signals standing clear of the blank scatter count as real.
4 · Averaging n scans improves the ratio by the square root of n, and the square root of 16 is 4.
0.4 · Ten times the noise is 0.2, and 0.2 divided by the slope 0.5 gives 0.4 ppm.
It halves the noise · Averaging buys the square root of the repeat count, at the price of time.
0.12 · Three times 0.02 is 0.06, and 0.06 divided by 0.5 gives 0.12.
Noise alone could explain the reading · Honest chemistry reports the limit, not a lucky digit.
Report detected but below quantitation · Between the limits, presence is certain but amount stays guesswork.