Method Detection Limits
by Tim Loftus
Why would you ever want to determine a detection limit for a procedure in your laboratory? Don’t the analytical instrument makers tell you what the detection limits are for their equipment? So why waste valuable time figuring out detection limits?
There are several different types of detection limits depending what is being defined. However, the method detection limit (MDL) is the only one designed to be determined in your laboratory using your chemicals, equipment, and technicians. (The other detection limits are either determined by the instrument manufacturer or are calculated from the MDL. While each has its place, not all are “useful” in your day to day analyses.)
Determining the MDL puts on paper certain practical limitations of a test method. It takes into account the instrument’s lamp having 2000 hours of on-time, scratched sample cells, films that have developed on the optics, different lots of reagents, the sample matrix, and the technician performing the analysis. It can tell you, for instance, whether the 26.4 microgram/L chlorine residual limit in you NPDES permit can be accurately analyzed using your laboratory equipment and personnel. MDLs can guide you in determining how much to “spike” a sample for QC known additions and to help define the lower range of the test method.
The EPA defines MDL as “the minimum concentration that can be determined with 99% confidence that the true concentration is greater than zero.” This procedure is outlined in 40 CFR 136. Basically you make a solution of the analyte that is one to five times the estimated detection. Test this solution seven or more times, then determine the standard deviation of the data set. The method detection limit is calculated according to the formula: MDL = Student’s t value x the standard deviation.
For a real-life example, let’s look at a method detection limit determination I recently did for phosphate analysis using our laboratory’s spectrophotometer.
An estimated detection limit of 0.045 mg/L for this method was printed in the instrument manual. (For those manuals that do not list their estimated detection limits, 40 CFR offers several ways to determine one.) I made a solution of 0.175 mg/L phosphate, which is approximately four times the estimated limit. This solution was analyzed nine times over the course of several days. The results were 0.194, 0.166, 0.174, 0.149, 0.183, 0.153, 0.144, 0.173, 0.190 mg/L. The standard deviation of this data set calculated to be 0.018. Substituting the numbers into the equation: MDL = Student’s t value x std. deviation = 2.896 x 0.018 = 0.052 mg/L (The student’s t value changes with the number of data points. If you cannot find a Student’s t chart, call me for one or follow my example using the same amount of data points.)
I got an MDL of 0.052 mg/L for my phosphate method. What does this tell me?
First, it indicates what the lower limit, often called the limit of quantitation (LOQ), of the test method is. The method detection limit is not the lowest limit I can accurately test to during routine analyses – the lowest limit is actually 2.5 times higher than the MDL. The lower limit in my example would be 0.13 mg/L (0.052 x 2.5 = 0.13). When I test phosphate following this method, any value below 0.13 mg/L will most likely be inaccurate - even though my “detection limit” is 0.052 mg/L. As you can see, not all detection limits are created equal!
Second, the MDL is also used to determine an amount to “spike” a sample for a known addition. Generally, you spike a sample at 5 to 50 times the MDL. For this test method, I would spike my sample with a phosphate solution so that an increase of 0.26 mg/L to 2.6 mg/L is expected. From this, a percent recovery can be calculated.
Third, if the MDL is much higher than the estimated detection limit, or if over the course of time the MDL steadily increases, it may indicate problems with the procedure or with the analytical equipment. Some things to check for may include improper lamp alignment, low lamp output efficiency, expiration dates for chemicals, improperly made reagents, worn sample cells, etc.
Determining method detection limits for your laboratory is really an easy thing to do. It provides some practical applications and it helps to assure your regulating authorities that your numbers are accurate.
The recommendations here for the method detection limit are very general. Always check your state and local regulations. You may have additional requirements to meet.
If you have any questions, suggestions, or comments, please contact LPC Chair Paul Fitzgibbons at (401) 222-6780 ext. 118 (firstname.lastname@example.org) or Tim Loftus at (508) 949-3865 (email@example.com). You can also visit our website at newea.org. Once on the website, press the Lab Program