The New U.S. EPA Method TO-15A Blog Series-Part 6: Calibration Curve Fits

Our last several blogs covered the numerous ways to help clean up lab air to meet the new TO-15A guidelines for canister blanks. Absent from the previous blogs was how we were getting quantitative numbers, which is important given the new calibration guidelines in TO-15A. The original TO-15 method only mentioned using an average response factor calibration, with the acceptance requirement being the %RSD between points <30%, with at most 2 exceptions up to 40% (TO-15, section TO-15A adds in guidance for linear and quadratic fits, as well as alternate weighting models (TO-15A, section 15.2.3). The acceptance requirement for calibrations is either <30% RSD for average response factor or r2 ≥0.995, as well as each calculated concentration of each calibration point being within ±30% of the true value. This allows for more accuracy in your results, provided that the best curve fit can be found.

I won’t go deep into the math behind it here to keep this short, but equal weighted curve fits, be they linear, quadratic, or average response factor (RF), tend to be biased towards the higher concentration levels. Likewise r2, the coefficient of determination for linear and quadratic fits, also tends to be biased towards higher concentrations. This is because the calibration models are based on absolute concentrations (or in the case of r2, absolute error), so the bigger the difference between your high and low points the less your low points matter. This can be countered by either clustering more calibration points at the lower end, or using 1/x or 1/x2 weighting. The table below shows how the different curve weighting types can change an example linear calibration.


Concentration response Equal wt. absolute error % error
1 12 0.33 0.67 67%
10 110 10.74 0.74 7%
100 950 99.94 0.06 0%
r2 0.999657
total error 1.47 74%
%RSE 67%
Concentration response 1/x wt. absolute error % error
1 12 0.90 0.10 10%
10 110 11.14 1.14 11%
100 950 98.96 1.04 1%
r2 0.998526
total error 2.28 23%
%RSE 15%
Concentration response 1/x2 wt. absolute error % error
1 12 0.99 0.01 1%
10 110 10.69 0.69 7%
100 950 93.77 6.23 6%
r2 0.995172
total error 6.92 14%
%RSE 9%
Concentration response Avg. RF absolute error % error
1 12 1.11 0.11 11%
10 110 10.16 0.16 2%
100 950 87.72 12.28 12%
%RSD 11.62%
total error 12.55 25%
%RSE 17%

Table 1: Example comparison of curve fit types


As you can see, the equal weighted curve fit gives the best r2 and lowest absolute error, but has the highest total % error, off by 67% on the lowest calibration point. Due to the large error it would actually fail the ±30% criteria for calibration accuracy. As the weighting changes to 1/x or 1/x2 the total absolute error and r2 get worse, but the total % error improves. For comparison the average RF calibration has the worst absolute error and 2nd worst total % error.

What does this mean for your TO-15A analysis? Since the blank guidelines have dropped an order of magnitude from 200 pptv to 20 pptv, your error at the low end of your calibration could increase dramatically. In the above example the 1/x, 1/x2, and average RF calibrations all meet the TO-15A calibration criteria, so which would be the best choice? Accepting average RF as the default unless it fails the %RSD criteria can lead to rather large errors, as seen above. The same is true if r2 is used as a judgment of best fit. Calibration models should not be based solely on passing any specific QC, including blank values, so looking at the error on any one calibration point isn’t appropriate. Minimizing the total % error would be a good measure of the best calibration, in which case the 1/x2 weighted calibration above would be the most appropriate despite having the worst r2 value. Percent relative standard error (%RSE) is another good measure of overall calibration accuracy, and the NELAC Institute has a brief document on how to calculate %RSE at The calculation is shown below, where n= number of cal points, xi is the true value of the cal point, x’i is the calculated value for the cal point, and p=1 for average RF curves, 2 for linear curves, 3 for quadratic curves. In Table 1 you can see that the curve fits with lower total % error also have a lower %RSE.


Fig 1: Relative Standard error calculation


So, was I able to solve all of my blank issues through selection of the best calibration? As seen in the table below illustrating my problem compounds, I was not.


Compound pptv Curve fit
n-Pentane 22 Linear, equal weight
Ethanol ND Linear, 1/x
Acetonitrile 15 Linear, 1/x
Carbon disulfide ND Average RF
Isopropyl alcohol ND Average RF
Methylene chloride 21 Linear, equal weight
Acetone 12 Linear, 1/x
Hexane 162 Linear, 1/x
Tertiary butanol ND Average RF
Tetrahydrofuran (THF) 28 Linear, equal weight
2-Butanone (MEK) 30 Linear, 1/x
Toluene 12 Average RF
4-Methyl-2-2pentanone (MIBK) ND Average RF

Table 2: Selected TO-15A blank results humidified to 50% RH with boiled water. Calibration with lowest total % error used.

The 20 pptv cleanliness requirement is a challenging target to achieve and I imagine many labs are going to struggle with it, especially if your air lab shares space with heavy solvent users. Does this mean that running TO-15A is out of reach? Not necessarily. Remember that the TO methods are compendium methods meant to be used as a basis for laboratories to develop their own procedures, not as a strict list of requirements that must be followed. As long as regulatory limits and customer requirements are met, documenting some compounds as having blank or canister cleanliness limits above 20 pptv should be an acceptable deviance. Of course, labs will have to be prepared to defend this to their auditing agencies, and work to improve their blank results to be prepared for changing regulatory limits, customer requirements and potential competition from other labs, but a few outliers in your quest for <20 pptv cleanliness shouldn’t stop you from adopting TO-15A.

TO-15A blog series

Potency: Useful Techniques

The most labor-intensive part of determining the potency of samples containing cannabinoids is extraction. Due to a plethora of matrices, from candy, baked goods, lotions, and so on, selecting the most suitable extraction method can also be a challenge.  The following video and technical article will guide you through that initial process:

How Potent is My Sample? A Cannabinoid Q&A

Quantitative LC-UV Method for CBD in Topicals with Simplified Extraction of Lotions, Balms, and Creams

Let’s focus on the analytical part of the analysis.

Restek scientists showed that separation of regulated and some of the most frequent cannabinoids can be performed by a very simple isocratic analysis. Isocratic elution is typically effective in the separation of sample components that are very different in their affinity for the stationary phase. Raptor ARC-18 column chemistry compliments the cannabinoids’ structures, and the three analysis methods were developed on this column.  The main difference is column dimensions and particle size, which will affect the analysis time:

1. Starting point isocratic method.

2. Faster analysis using a UHPLC system.

3. Saving Solvent? The reduced ID column used in the application requires a lower flow rate.

A new Raptor ARC-18 column is stored in a reverse phase mobile phase 45/55 water/acetonitrile and should always be conditioned before use:

  • Our suggestion is to run approximately 10-20% less organic content than the storage solvent for about 10-20 column volumes. For example, if the HPLC column is stored in 45/55 water/acetonitrile, a good starting solvent would be approximately 60/40 water/acetonitrile. This step is crucial, mainly because our method mobile phase contains salts (buffer). Preconditioning the column will prevent salts from precipitating out of the solution, possibly clogging the column and increasing backpressure.
  • Then, introduce the method mobile phase containing buffer for 10-20 column volumes. You are now ready to start your analysis. Check out this video for more details: LC Column Conditioning

Tips and Tricks to help prevent variability:

  • Buffer Strength/pH: Buffering the mobile phase is crucial in keeping retention time consistent in this analysis. In our application, we used a 5mM ammonium formate buffer. It was found adjusting between 5mM and 10mM ammonium formate can fine-tune the separation. Our scientists didn’t see any noticeable difference (resolution/selectivity) in chromatography when switching from 5 to 10mM buffer concentration, you can adjust compounds like CBNA and THCA-A without losing the overall selectivity. Also, if you don’t use an acid modifier in the mobile phase, the retention time decreased drastically for acidic cannabinoids. Little chromatography change was noticed adjusting pH with 0.1% acetic acid or 0.1% formic acid.
  • Sample Solvent: Are you seeing poor peak shape, especially with the early eluting analytes? Injecting a 100% organic solvent sample will cause peak distortion, peak broadening, poor sensitivity, and shortening of retention times. This happens because some analytes will tend to travel too quickly through the column, instead of eluting in a symmetrical band. After all, they don’t mix evenly with the mobile phase. This is more pronounced in isocratic analysis than gradient analysis. The sample solvent should be as similar to the starting conditions as possible (aqueous/organic composition). Even if extracting in 100% organic solvent, consider doing a dilution into a more equivalent solvent to your starting conditions. In the applications mentioned above, we use 25:75 water/methanol as the diluent.
  • Injection volume: Seeing peak distortion? The injection volume leads to many of the same problems that the sample solvent can cause. Higher-than-recommended injection volume on an isocratic analysis can cause peak shape distortion. Please see this FAQ for the correct injection volume based on the column ID: HPLC FAQ’s
  • Guard columns: It is strongly recommended to use a guard column or pre-column filter for this analysis. This will help to increase column lifetime due to removing particulate contaminants and strongly retained compounds that are inherent with this analysis. Sample filtration is a step that shouldn’t be forgotten in the analysis. Sample-Filtration.
  • Column regeneration: How do I extend the lifetime of my HPLC column? At the end of the day or analytical run, best practice is to run an increasing organic gradient up to 95-100% organic solvent to dissolve the non-polar particulates trapped at the head of the column. If drastic increasing pressure and degrading peak shape is noticed the last resort before buying a new column, try a cleaning procedure (backflushing not recommended for Raptor ARC-18 columns); LC Cleaning Recommendations
  • Wavelength: Our applications use 228nm, but this is something that can be optimized instrument-to-instrument to get the best column sensitivity and low baseline distortion. Please note that interferences can come from possible terpenes in the matrix. Most terpenes do not respond well at UV wavelength ≥ 220nm. Minor terpene interferences could potentially impact the quantitation of CBGA and THCVA if present in high enough concentrations.

To conclude I would like to bring to your attention an excellent article about organic modifiers as it pertains to this analysis. Selecting a different organic modifier is a good way to change the selectivity and resolution of your analysis. The article shows the difference between a protic solvent like methanol (better resolution of later eluting cannabinoids) and an aprotic solvent like acetonitrile (better resolution of the earlier eluting cannabinoids), and how this can also change your chromatography.

Effect of Organic Solvent on Selectivity in LC Separations


FDA Warns Hand Sanitizers May Contain 1-Propanol

The Food and Drug Administration (FDA) has expanded their recall of hand sanitizers and found more chemical contaminants to include 1-propanol (n-propanol cas# 71-23-8). 2-propanol (isopropanol cas# 67-63-0) and Ethanol (cas# 64-17-5) are the two most widely used active ingredients in hand sanitizers. With this latest addition to the list, the main concern for exposure is ingestion (1). In a three-year period over 60,000 children under 12 years old ingested hand sanitizer (2). 1-propanol is metabolized via the enzyme alcohol dehydrogenase to propionic acid and thereby disrupts the pH balance of the blood, while 2-propanol is metabolized with the same enzyme, however, the product is acetone leading to ketosis. Ingestion of both alcohols results in 1-propanol slowing down the breakdown of 2-proponal to acetone prolonging the sickness (3).

Our previous blog addressed the recall of hand sanitizers that may contain methanol. Using the same method conditions it’s possible to analyze 1-propanol in addition to other compounds commonly found in adulterated hand sanitizer (Figure 1). Following the conditions below, we recommend eliminating 1-propanol as a co-solvent in the rinse vial and substituting acetonitrile instead. Figure 2 shows 1-propanol can be found as a contaminant in the blanks and samples.

Figure 1: The analysis of denatured alcohol with an overlayed chromatogram of 1-propanol. It is possible to analyze for a variety of alcohols in hand sanitizer at percent levels using the Rtx-VMS.

Figure 2: Trace amounts of 1-propanol can be found in samples and blanks when using 10% 1-propanol as a post rinse (see red text). Changing the co-solvent to a non-target compound is recommended. 


Related Reading Links:

Not Every Matrix-Matched Calibration Is Made Equal – Case of Spinach

Matrix-matched calibrations are a popular option for complex matrices, such as food commodities. When it is employed, one can assume that the majority of matrix effects will be accounted for and there won’t be much difference between calibration standards and the tested samples. But are all matrix-matched calibrations equal?

In the case of QuEChERS extraction, there are three places where the standards can be spiked into the matrix (Figure 1), after complete extraction (option A), before cleanup (option B) and before the extraction (option C). In LC analysis, the majority of matrix effects come into play in the ion source, while in GC analysis, the majority of matrix effects happen in the inlet. To simplify the situation, I’m going to use data collected using GC-MS/MS.

Figure 1: Options for matrix-matched calibrations

Read the rest of this entry »

Short Webinar on Calculating Column Length Using EZGC Flow Calculator

Determining the length of your GC capillary column is essential for accurate control of carrier gas flow rate.  Just because you purchased a 30 m column, does not mean that your column is exactly 30 meters long; in fact, columns will typically have a little extra tubing.  But ultimately, it is not the physical length tha matters as much as the total volume of the column.  Since columns can have slight variations in internal diameter, what we really want to find is the “effective length” of the column, which accounts for these variations.  The easiest way to do this is by injecting a non-retained compound and seeing how long it takes to elute from the column.  This is known as the “holdup time” or “dead time”.

Once we find a “holdup time”, how do we actually go about calculating column length?  Check out the following short webinar to see how you can use Restek’s EZGC Method Translator and Flow Calculator to do exactly this:


Derivatization techniques for free fatty acids by GC

Anyone who has tried to analyze free fatty acids by GC realizes they have to derivatize these compounds. The inherently low volatility of free fatty acids and the carboxylic acid (-COOH) interaction with the siloxanes in the stationary phase. If not addressed, the first problem leads to late-eluting peaks and the latter results in peak tailing due to secondary retention mechanism (see Figure 1).

Figure 1: Peak tailing due to multiple retention mechanisms [1]. Retention mechanism 2 represents the interaction between -COOH group and siloxanes in the stationary phase.

In this blog, I’ll go over the two main derivatization procedures I’ve seen most commonly used for free fatty acids – esterification with BF3 in methanol and silylation with BSTFA or MSTFA [2].

Note: Both of these methods are moisture sensitive! If the sample is aqueous, the water needs to be removed. For example, water can be removed by drying the sample or lyophilization.

Read the rest of this entry »

Preserve your SRM transition windows by replacing the same amount of column you remove during maintenance

I’ve been working on a detailed PCB congener analysis using the TSQ-9000 with the AEI. Typically this analysis is done by EPA Method 1668C, which uses high resolution MS. We get some specificity from the selected reaction monitoring, but there are some shortcomings when compared to high resolution data. Where the TSQ-9000 really shines is the sensitivity. We were able to achieve very linear calibrations (avg RF RSD% < 5.5%) for all target compounds using 6 levels (analyzed in triplicate) ranging from 40 ppt to 400 ppb. The 6 point calibration suggested by EPA Method 1668C ranges from 200 ppb (for sensitive instruments) to 2000 ppb.

After 6 weeks of constant running, it was time to perform some maintenance. We needed a new liner and septum, and we needed to trim the column to restore symmetrical peak shapes to the analytes. The SRM method we’re using has well over 500 transitions, many are close together with narrow windows, and the prospect of shifting multiple compounds out of their windows was alarming. Figure 1 shows the pentachlorobiphenyl homologue group. The comprehensive chromatogram would have some congeners from the 3-chloro, 4-chloro, 6-chloro, and 7-chloro homologue groups overlapping, as well as stable isotope labeled internal standards and surrogates.

Figure 1 – The complete pentachlorobiphenyl homologue group. Window defining and WHO toxic congeners are labeled with their ID in blue.


It’s possible to use the speed factor calculated by the EZGC method translator to calculate the new retention times, but it’s a manual process involving measuring a deadtime and calculating a new effective length, translating the method, and manually calculating and updating the new transition windows.

The more prudent course of action seemed to be removing the need to update the transition windows. Replacing the segment of column removed with the same length of the same column should result in virtually no change in elution times, eliminating the need to update the transition table. This is exactly what we saw when we replaced 1 loop of the 0.18 mm x 0.18 µm Rxi-PCB column with a fresh segment of the very same length, using a SilTite µ-Union to make the column connection. The before and after chromatograms for the WHO toxic congeners in the EPA 1668C calibration CCV are figures 2 and 3 respectively.

Figure 2 – Retention times for select toxic and window defining PCB congeners straddling the center third of the chromatogram


Figure 3 – Retention times for the same select toxic and window defining PCB congeners straddling the center third of the chromatogram following the replacement of the head of the analytical column with a fresh segment of Rtx-PCB.

Aaron Lamb from Thermo Fisher and I will be discussing persistent organic pollutant analysis this afternoon using the TSQ-9000. Following us, there will be a discussion on the alternative method of analysis – High resolution MS, specifically the Orbitrap. The webinar is titled ‘Persistent Ongoing Perfection: Optimization of a GC-MS/MS Method for the Analysis of POPs Plus an Alternate Approach Utilizing High-resolution Mass Spectrometry’. Click the link if you are interested. The webinar should be available for on demand viewing within a few days. There are a variety of on demand lectures to pick from on a variety of different analytical techniques.


The Foundation of Separation is Preparation

Restek is known as a leader in the field of chromatographic column technology.  The idea is built into our very identity – resolution technology is at the heart of Restek.  We also believe that proper sample preparation sets chromatography up for success, though, by reducing the incidence of confounding results and prolonging the lifetime of your chromatographic investments.  From dilute-and-shoot techniques to multi-step solid phase extraction methods, Restek is just as dedicated to helping you find the best preparation method and product for your sample.

While it is unlikely that we will change our name, we are nevertheless committed to providing analytical solutions to today’s toughest chromatographic challenges, from preparation through separation.  Check out Restek’s growing line of sample preparation products, and let us know how we can help.



Getting Your LC Up and Running Again

Welcome Back!  You and your lab have been through a lot these last few months and now it’s time to get your LC back up and running.  After it has been down for a few months, it likely needs a little TLC before it’s performing optimally again.  Just like a car that sits around for a long time without being used needs its fluids, filters, and tires checked and maintained, an LC has similar needs.

Let’s walk through the system and get you back up and running.  Before starting anything, you should consider performing preventative maintenance (PM) according to your manufacturer’s specifications.  Another excellent resource that can help is “Routine LC Maintenance: Simple Steps to Preventing Unexpected Downtime.”  After any PM or other maintenance that you decide to do is completed, you can move on to getting the rest of the system ready.

If there are old solvents sitting on the solvent rack, dispose of them all in the proper waste streams and replace them with fresh solvents.  This includes solvents for seal washes and needle washes etc. ALL solvents should be replaced.  At this point you should also replace your column with a union and divert your flow stream directly to waste, making sure the detector is disconnected from the flow path. You should also note right now whether any buffer was in the system.  If so, you’ll need to flush all pumps and lines containing buffer with water to prevent any salt precipitation in the system.  Your pumps will need to be primed and purged according to the LC manufacturer’s specifications. After this, start the pumps with low flow rates (0.2-0.4 mL/min) and purge any buffer from the lines with water for 10-20 minutes.  Also, be sure to rotate any valves in the system that might contain buffer.

Next, flush a strong organic solvent, like isopropyl alcohol, through the system by priming and purging the pumps.  Flushing the system for 10-20 minutes at 0.2-0.4 mL per minute should adequately flush the system.  Finally, you can load your desired mobile phase solvents onto your LC, though you should not incorporate buffers yet.  After all of these steps are completed, all buffers and organic contaminants should be flushed from your LC.

Now you are ready to reconnect the detector into the flow path.  Follow the same order of solvent flushing as above.  Start with aqueous and transition to organic.  With your entire system now flushed, you can begin running diagnostics.

At this point, installing an LC column will help you to identify any problems that might exist as many can show up when there is back pressure on the system.  Observing not only your baseline, but also your back pressure can help you determine if anything needs to be fixed or replaced.

Erratic baselines can indicate the presence of air bubbles.  To fix this, further purging of your pumps or entire system can fix this problem.  If you see systematic fluctuations in pressure profile or detector, this can be more indicative of a pump problem.  The culprit is likely a faulty check valve or a worn pump seal that requires cleaning or replacement.  For all these PM tasks or repairs, make sure to follow your LC system manufacturer’s guidelines.  Many standard parts are available on our website.

If everything checks out, then a diagnostic/system suitability tests with a column in place and an appropriate system suitability mix should be performed to see how well the instrument is performing.  Comparing the current results with previous tests can also be useful to make sure your instrument is performing optimally.

Good luck on getting your systems back up and running and if you have any questions, we are always here and happy to help.


Other Useful Resources:

LC Column Cleaning and Regeneration

Connecting Your Column

Preventing LC Column Clogs

LC Troubleshooting-Retention Time Shift

LC Troubleshooting-Baseline Problems

How can analyte protectants and matrix help improve peak shapes?

In my last blog, I presented a new technique called low pressure gas chromatography (LPGC, Figure 1). Just to recap, the LPGC system consists of a relatively short analytical column (10 – 15 m) with large ID and thick film (e.g. 0.53 mm and 1.0 µm, respectively) which is restricted with a narrow guard column (e.g. 5 m x 0.18 mm). The restrictor (guard column) allows for maintaining head pressure on the inlet, while the analytical column is under near-vacuum pressure.

Figure 1: LPGC schematics

Read the rest of this entry »