Levels of Detection

[Joey]

I'm helping a young teen on a 4-H project. I hope you can help. His question is using GC/MS can levels of detection be set lower than 10 ug/g? If it's possible to test at a lower level, could this result in false readings (accuracy)? Thank you very much.
Joey

[tom jupille]

The answer depends on what you are trying to detect (as always, "the devil is in the details").

That said, 10 μg/g is approximately 10 parts-per-million (ppm). GC/MS is capable of much lower detection limits for many compounds. Just to put this in perspective, there has been a recent flap over benzene in soft drinks. The FDA does not have a limit, but I believe the EPA limit for benzene in drinking water is 500 ppb (parts per billion), which is 0.5 μg/mL.

[Joey]

Tom, thanks. That's a good start and will be helpful for him.

He wanted to select an issue that is currently in the news, so he chose the pet food issues. He's trying to look at all sides of the issue regarding testing levels. So what he would be most interested in is melamine, cyanuric acid and also acetaminophen. I think these are the main ones identified. Some info he's read suggested it could be detected in the food below 10ppm and others say it cannot. The other question might be (excuse our ignorance) is there some type of standard that suggests how low certain compounds can be detected?

Any thoughts would be appreciated.

[Schmitty]

I have been working on the "pet food issue" and I can most definitely say that the compounds can be detected in the food at 5 or 10ppm with the current FDA method (see the FDA website for the current release). The difference, as Tom started to point out, is in how you want to define the LOD.

If, as in the pet food example, we take a 0.5g sample and sonicate it with 20mL of the extracting solvent, then analyze the extract after derivitization results in the following equation:

(0.10 µg/mL x 20 mL) / 0.50g = 4ppm

Where 0.10 is the low standard of the FDA method. Typically you would like to bracket the levels you are trying to detect, so the method may/may not be applied too low (4ppm). In the case of some matrices, there is significant interference on some of the ions specific to each of the analytes specific to the melamine analysis.

So, the limit that is detectable in “foodâ€

[tom jupille]

I don't think there is any simple general explanation (if there were, you wouldn't need the equivalent of a graduate degree to figure out how to do it, and most members of the Forum would be unemployed! ).

First of all, the detection limit for most detectors is actually better expressed in mass terms (nanograms, for example) than in concentration terms. In principle, if you have a large enough sample, you can get down to some very low concentrations (not always easy in practice!). Just to take melamine, you should be able to go to a much lower level in pet food (where you can start with kilograms of sample) than you could in urine (where you might be limited to a few hundred milliliters) or in blood (a milliliter or so).

Next, you have to try explaining the concept of signal-to-noise ratio. It's always possible to amplify the signal coming out of any detector. The problem is that at some point, you're amplifying noise so that the signal due to the analyte peak gets "lost in the weeds". If you want an analogy, grab a screen shot of a satellite picture of a suburban neighborhood from Google maps. You can see cars, but have your student try to itentify the model. You can only magnify the image so far (TV crime shows notwithstanding).

By convention, "limit of detection" (LOD) is the amount of analyte that generates a signal three times bigger than the noise (signal/noise ratio = 3). You really can't do quantitation at that level. "Limit of quantitation" (LOQ) is usually given as the amount of analyte that generates a signal/noise ratio of 10.

Since detection limit depends on signal/noise ratio, there are two ways to improve (decrease) it: more signal, or less noise.

The amount of signal will be affected by the chemistry of the compound being analyzed and how the detector interacts with that chemistry. Mass Spectrometry (the "MS" in GC/MS) involves ionizing the analyte (making it charged). If a molecule is easy to ionize, then lots of those molecules will be detected. If it's hard to ionize, then only a fraction will be detected.

The amount of noise will be affected by the intrinsic characteristics of the detector, by how much "junk" (stuff you're not interested in) from the sample is in there, and by how well that "junk" responds.

[WK]

Tom,
Very neat - I hope you can paste that answer into the FAQ section!
WK

[DR]

Acetaminophen is a good one if the student wants to consider using HPLC. It (acetaminophen) has an enormoous extinction coefficient, meaning that it can be detected at ppm levels with a UV detector and just about any C18 column. Getting a "clean" sample from food products would be the biggest challenge as Acetaminophen is soluble in pretty well all of teh same things that most food related things are...

[MikeD]

With regards to benzene in drinking water all European Union member states must comply with a maximum limit of 1 ug/litre (1 ppb). The same will apply to drinks. As Schmitty says, 0.1 ppb is achievable, otherwise we could not demonstrate compliance.

[Mary Carson]

Tom, thanks. That's a good start and will be helpful for him.

He wanted to select an issue that is currently in the news, so he chose the pet food issues. He's trying to look at all sides of the issue regarding testing levels. So what he would be most interested in is melamine, cyanuric acid and also acetaminophen. I think these are the main ones identified. Some info he's read suggested it could be detected in the food below 10ppm and others say it cannot. The other question might be (excuse our ignorance) is there some type of standard that suggests how low certain compounds can be detected?

Any thoughts would be appreciated.

If the chemist looks hard enough, they can find very low levels of contaminants. However, for the analysis to be valid, they must also show that a truly negative sample (sometimes difficult to come by...) will not give a positive result. At sub-ppb levels of contaminant analysis, that can be a challenging task. Misidentification is possible, even with MS-MS, (folic acid and the rat poison aminopterin are quite similiar structurally). Interference testing is often part of method validation in order to minimize likelihood of both false positives and false negatives.

As to how low a compound CAN be detected--no there are no standards and it depends very much on the method and instrument. Perhaps the question really being asked is how low SHOULD a method be able to detect certain compounds? In some cases there are "standards" --tolerances or maximum residue levels set by national or occasionally international governments, and usually where there is an approved use and toxicology data to support a level at which there is very high confidence no harm will occur. However, in the case where compound is being used in an unapproved manner, there is often no agreed-upon target level for an analytical method.

[Peter Apps]

Of course, a detection or quantitation limit only applies if a sample is being analysed in the first place. In the pet food case no melamine or cyanuric acid was found, because nobody was looking for them, not because the methods could not have found it if it was there. The underlying problem is that the "protein" content of foods and feeds is measured as nitrogen content by non-specific methods, hence spiking raw materials with compounds that have a high nitrogen content gives an inflated value for protein content, and a higher price.

Peter

[WK]

Peter,
That goes for a lot of wet chemistry methods - its good feeling when we can use chromatography to unpick the titrations with a bit more detail.
This is a very serious problem for US as they contain the largest percentage of lawyers.
Seriously though - the head of China's pharma has been executed recently over a few "lapses" like this (deg in toothpaste).
Somebody may be able to tell us how much soft drink at 1 ppb benzene you need to drink to reach the level taken in during car re-fueling average once a week?!
WK

[MikeD]

Interesting comparison WK. The average benzene exposure of people actually working in petrol stations has been measured extensively for 20+ years, but rarely customers because their exposure is very short. However, assuming it's about 0.3 ug/litre air for 5 mins/week their uptake (not 100%)would be about 6 ug at resting rate breathing 20 litre/min. So the answer is 6 litres of the drink containing 1 ug/litre.

Now when the customer leaves and goes about his or her everyday life their average exposure might be about 0.001-0.002 ug/litre air equivalent to about 40-80 ug benzene/week body burden, not counting food and fluid sources.

[WK]

If I drank 6 litres of soft drink a week then I might be more worried about the quantity of carbon dioxide (or acid/sweeteners) than the benzene!

[Peter Apps]

6 l a week is less than a litre a day - three cans a day. Your teeth will all rot before you get cancer from the benzene, but apparently kids put coke on their breakfast cereal instead of milk so 6 l a week is pretty conservative.

What's the limit in beer ?

Peter

[Schmitty]

First Google hit: Soda consumption

Is that just over 5l per week in the US...?