What is the limit of detection?

The concept of "limit of detection" (LOD) applies when taking any measurement. The LOD is the smallest measurement that you can distinguish from zero reliably.  When you collect filter samples of particulate matter (PM) and analyze those samples for PM mass, the LOD is the smallest mass of PM that you can be confident is greater than zero. For example, you might not be confident that a sample mass of 10 μg is different from zero, but you might be confident that a sample mass of 50 μg is different from zero. Your LOD will be influenced by: (a) your sample handling, storage, and transportation practices as well as (b) the precision with which the analyte of interest (e.g., PM mass) can be quantified during laboratory analysis. 

You'll want to select the flow rate and duration for each PM sample that you collect so that there is a high likelihood that the sample mass will be above your LOD. If the sample mass is below your LOD, you won't be able to have a high level of confidence in the measurement.

The remainder of this post includes guidance for determining your LOD. This post focuses on PM samples that are being collected onto filters, using Ultrasonic Personal Air Samplers (UPAS), and analyzed for PM mass by pre- and post-weighing each filter on a microbalance. However, the general concepts and procedures described below also apply to PM samples collected using other samplers (for example, Home Health Boxes) and analyzed for other pollutants (for example, black carbon).

What is my limit of detection?

If an external laboratory is analyzing your samples for PM mass, that laboratory might have a predetermined LOD. In that case, the laboratory will tell you what their LOD is. Furthermore, that laboratory might not report the actual values of PM mass measurements below their LOD.  For example, if the laboratory LOD is 50 μg and the laboratory measured only 20 μg of PM on your sample filter, the laboratory might just report the measurement to you as "below LOD" instead of "20 μg." 

Alternatively, you might need or want to calculate your own LOD. I think this approach is preferable, because your LOD will be influenced by more than just the precision of the measurements made during laboratory analysis of your samples. Your LOD will also be influenced by the sample handling, storage, and transportation practices that you use during your study. For that reason, many researchers will calculate a unique LOD for each experiment, field sampling campaign, or study that they complete. To determine your LOD, you will need to collect "blank" filter samples.

What is a "blank" sample?

A "blank" is a filter that is handled, transported, stored, and analyzed just like sample filters; however, no air gets sampled through a blank filter. You should collect blanks during every laboratory experiment or field sampling campaign that you complete.

Blanks are collected and analyzed to evaluate two things:

1. How much sample filters are contaminated during normal handling, transport, and storage.
2. How much uncertainty exists in the measurement procedure used to analyze the sample.

If you are analyzing your samples for PM mass by weighing each filter on a microbalance before and after sample collection, filters that are going to be used to collect blanks will be "pre-weighed" along with all of your other filters before sample collection occurs. The filters that are going to be used to collect blanks will be handled, transported, and stored along with the filters to be used for actual sample collection. In fact, before sample collection occurs, the filters that will be used to collect blanks do not need to be (and, ideally, should not be) distinguished in any way from the filters that will be used to collect actual samples. After sample collection is complete, blank filters will be stored, transported to the analysis laboratory, and "post-weighed" along with all of your actual sample filters. 

How many blank samples should I collect?

In general, you should either:

1. Collect blanks at a rate of 10%,
2. Collect one blank every day that you set up "actual" samples, or
3. Collect enough blanks to calculate a standard deviation (absolute minimum of 2 or 3 but preferably 7 or 8).

Out of options 1-3 listed above, you should select whichever option will result in the largest number of blanks being collected. Additional information and examples related to each of these three approaches are provided below. 

Collecting blanks at a rate of 10%

How many actual samples do you plan to collect during your study?  Divide that quantity by 10. You should collect at least that many blanks. For example, if you plan to collect 100 actual samples, you should also collect at least ten blanks, and you will need a total of 110 pre-weighed filters. 

Collecting one blank every day

I'll explain this approach using a real example: On each day during each field sampling campaign associated with this study, study personnel visited 1 to 3 participating homes and started 2 to 12 UPAS samples. On each day that we visited participating homes, we collected one blank with a UPAS. We collected approximately 10 UPAS samples per day, on average, so the quantity of blanks we collected during each campaign was always at least 10% of the quantity of samples. Additionally, each campaign included more than 8 days of sample collection, so we always collected at least 8 blanks per campaign.

Collecting enough blanks to calculate a standard deviation

If you are only collecting a small number of samples during your study, you might need to collect blanks at a rate of more than 10% or collect more than one blank per day in order to collect enough blanks to calculate the sample standard deviation for the change in blank mass.

How do I collect blank samples?

To collect a blank sample with a UPAS, start with any filter that has been pre-weighed by your sample analysis laboratory. Install the blank filter in a filter cartridge whenever and wherever you install the filters that will be used to collect actual samples in cartridges. All of your filters might have already been installed in cartridges when they were pre-weighed by your sample analysis laboratory; in that case, proceed to the steps outlined below.

If you are preparing samplers at a central site and then transporting those samplers, with filter cartridges already installed, to sampling sites

1. Select a filter cartridge and note in your laboratory notebook or field log that the selected filter serial ID will be used to collect a blank.
2. Install the blank filter cartridge and a size-selective inlet on a UPAS whenever and wherever you install filter cartridges and size-selective inlets on UPAS that will be used to collect actual samples.
3. It's a good idea to charge the UPAS that contains the blank filter and to program the sample settings on that UPAS the same way you program your actual samples. That way, if something goes wrong with one of your other UPAS, you can collect an actual sample using the UPAS that was originally intended to collect a blank.
4. If you perform a pre-sample flow check on each UPAS that has been set up to collect an actual sample, you should also perform a pre-sample flow check on the UPAS with the blank filter installed.
5. Transport the "blank" UPAS to your sampling site(s) along with your "sample" UPAS.
6. After you've set up and started your actual sample(s), transport the "blank" UPAS back to your central site.
7. Remove the blank filter cartridge from the UPAS, place the filter in it's storage case, and store it along with all of the other "used" filters from your study.

If you are transporting filter cartridges to your sampling sites, where samplers are already installed, and installing filter cartridges on-site

1. Select a filter cartridge and note in your laboratory notebook or field log that the selected filter serial ID will be used to collect a blank.
2. Transport the blank filter cartridge to your sampling site along with the cartridge(s) that will be used to collect the actual sample(s).
3. Once you arrive at your sampling site: Install the blank filter cartridge on a UPAS, along with a size-selective inlet, the same way that you would install a filter cartridge that was going to be used to collect an actual sample.
4. If you usually preform a flow check before starting an actual sample, perform a flow check on the UPAS with the blank filter installed.
5. Remove the blank filter cartridge from the UPAS and place it back in it's storage case.
6. Transport the blank filter cartridge back from your sampling site(s). 
7. Store the blank filter along with all of the other "used" filters from your study.

Note that, if your study is designed such that you will ask participants to install filter cartridges in UPAS and start samples, you might use a slightly different strategy to collect blanks.

How should I calculate the LOD for my study?

When analyzing filter samples for PM mass, the LOD is calculated based on the changes in mass that blank filters experienced between pre- and post-sampling analysis. More specifically, the LOD is typically calculated as the mean change in mass of a blank filter (or zero, whichever is greater) plus three times the sample standard deviation of the changes in mass measured for all of the blank filters (US EPA, 2016): 

LOD = max(x_blank, 0) + 3s_blank

In the equation above, LOD is the limit of detection, x_blank is the mean change in mass across all of the blanks, and s_blank is the sample standard deviation of the changes in mass measured for all of the blanks. LOD, x_blank, and s_blank should all be expressed in the same units (for example, μg).

What are typical LODs for PM mass?

A typical LOD for gravimetric analysis of PM mass collected on a Teflon™ membrane filter is approximately 30 μg. More specific LODs reported for prior studies in which UPAS were used to collect PM samples are summarized below:

• Tanner et al. (2025) reported monthly LODs ranging from 10 μg to approximately 30 μg for gravimetric analysis of PM samples collected onto Teflon™ membrane filters using UPAS v2.1 PLUS.
• Li et al. (2025) reported LODs ranging from 17 μg to 26 μg for gravimetric analysis of PM samples collected onto Teflon™ membrane filters, using UPAS v2.1 PLUS, during three different field sampling campaigns.
• Pillarisetti et al. (2019) reported an LOD of 27 μg for gravimetric analysis of PM samples collected onto Teflon™-coated glass fiber filters using UPAS v2.0.
• Burrowes et al. (2020) reported an LOD of 28 μg for gravimetric analysis of PM samples collected using UPAS v2.0.
• Benka-Coker et al. (2020) reported an LOD of 32 μg for gravimetric analysis of PM samples collected onto Teflon™ membrane filters using using UPAS v2.0.

To ensure that the vast majority of the PM samples collected during your study will be above your LOD, you might want to assume that your LOD could be as high as 50 μg or 100 μg. In other words, you might want to select sample flow rates and durations that you expect will result in at least 50 μg or 100 μg of PM being accumulated on each filter.