Use the “Subscribe” link at right. It just takes a few seconds!
It has been over 3 months since the last electronic cigarette blog. I will spare you the excuses. However, I can proudly say that we will be releasing a full application note on electronic cigarettes in the very near future (i.e., 1 to 2 months hopefully). In the meantime I feel obliged to respond to all the various inquiries I have received from colleagues about the presence of formaldehyde in electronic cigarettes. Lately I hear things like “oh I just heard on NPR about formaldehyde in electronic cigarettes… is this true… did you know that?” Well… obviously you already know my answer to this hot topic question. So allow me to indulge…
In parts I II III and IV of this blog series we covered analyzing the e-juice for nicotine and impurities. In the last e-cig blog we saw that electronic cigarette solutions contained numerous impurities. At that point in time we did not go into a lengthy discussion on what compounds were found and their potential implications for human health. Why you ask? Well because theoretically no one is drinking the e-juice… nor is anyone bathing in the e-liquid… and lastly I doubt anyone is injecting the e-solution. The main route of exposure to e-cigs and any compounds of interest is the direct result of the vaporization process, whereby with the use of a heated filament, the liquid is turned into a vapor which the end user inhales. Despite the aforementioned, the majority of e-cig research has focused on analyzing the e-juice. However, now that the knowledge of formaldehyde in the vapor has hit the mainstream media, perhaps that paradigm will change. FYI – Anyone interested in further reading or references should just reach out. Now… after much delay and teasing here is what we have seen in the e-cig vapor:
Based on the above chromatogram our e-cig vapor has a lot more constituents than the manufacturer’s juice listing of propylene glycol, glycerin, and nicotine. No surprise here, we already saw this trend when analyzing the raw solutions. NOTE: these vapor results are for the exact same e-juice results presented in our last e-cig blog. So what are all of these peaks? Well… the following table will hopefully help break down all the unknown compounds into several more digestible sections. Note: We only included tentative compounds with a mass spectral quality of 80% or greater according to the NIST 2005 database. Compounds with 100% quality have been confirmed with standards.
We found 82 unidentified and identified (some only tentatively) compounds in the e-cigarette vapor for this manufacturer. Of particular interest was the presence of formaldehyde, acetaldehyde, acrolein, as well as several siloxanes. All of which we did not find in the raw e-juice.
So… what does this all mean? Oh… and how did we measure the vapor? This time I promise to come back sooner than three months to answer these questions.
Restek’s EZGC Online Suite that includes the Method Translator and Flow Calculator, and a Chromatogram Modeler, wins a TASIA Award – Chris Nelson, One of the Suite Builders
The Analytical Scientist is a very smartly produced scientific magazine full of interesting articles, including many on chromatography. I’ve had the pleasure of working with two of the minds behind this publication, Rich Whitworth and Frank van Geel, on a TAS GCxGC contribution, and have been impressed with the volume of quality work they’ve put out. I’m sure they’d be the first to admit that it’s not just the two of them; it’s the people that work with them that make their efforts stronger. And that’s one of the reasons I’m posting this blog, to acknowledge the contribution of Restek’s Chris Nelson, one of the chief architects behind our EZGC™ Online Suite, which recently won one of The Analytical Scientist Innovations Awards for 2014.
Restek’s EZGC™ Online Suite, which includes the Method Translator and Flow Calculator, and a Chromatogram Modeler, speeds gas chromatography method development, and we’ve benefitted greatly from these tools in our own work as you can see from the list of ChromaBLOGraphy posts below. We couldn’t have done any of this without the efforts of Chris Nelson to make the program accurate and user-friendly. So…many thanks to Chris for his work on EZGC™ and to The Analytical Scientist for recognizing the power of this GC method development platform. Try the tools in your lab today and let me know what you think of them. You can get some ideas from the posts below (note that those are links that you can click in to get the post).
I am often asked for QuEChERS product recommendations. Usually the request is something like, “I need to test pesticides in _____” and that blank can be just about anything…sometimes not even food.
Unless it is something I have tested before, I usually take a quick look online to determine the approximate composition of the food. I do this because standard QuEChERS methods work best for high water, low lipid, and low carbohydrate commodities. The general recommendation is that the sample should contain about 80% water. If this isn’t the case, then adding some water is helpful and often needed for the extraction to work properly. In addition, high lipid samples can be challenging and you want to have a high level of C18 in your dispersive cleanup step.
The USDA Nutrient Database is a great resource for finding this information. Many foods are included and values for water, protein, lipid, carbohydrate, fiber and sugar levels are typically listed. The really great part is that these values are listed on a 100 g basis so the values are the same as percent weight.
I started checking this every time I test a different food because looks can be deceiving. The first time I tested spinach, I thought it looked “dry”. When I checked the USDA nutrient database, I saw spinach was listed at ~90% water. I was surprised. I did not add any water and the extraction worked perfectly.
In addition to water content, I take a look at the lipid and sugar values. If the lipid is “high” (above ~5%), I want to make sure to have some C18 sorbent in my dSPE tube to help remove coextracted lipids. Sugar is removed by PSA sorbent so knowing this value will help determine the amount of PSA in your cleanup.
U.S. Department of Agriculture, Agricultural Research Service. 2014. USDA National Nutrient Database for Standard Reference, Release . Nutrient Data Laboratory Home Page, http://www.ars.usda.gov/nutrientdata
*Amazing food art found at http://myhoneysplace.com/food-art-pictures/food-art-22/
This is a common GC troubleshooting question. Since determining the source of the ghost (contamination) peaks can be difficult, I decided to write this post to (hopefully) help our customers.
GC column carryover
When I operated an instrument on a daily basis, it was only a matter of time before something caused ghost peaks in my chromatogram. So how did I narrow down the source of the ghost peaks? First, I would cool the injection port. Then, I would program the GC so that I could simply press the Start Button to acquire data for an analytical analysis without injecting any sample, standard, solvent blank, etc. In other words, with a cold injection port, and without injecting anything, I would acquire some data while heating the GC oven. In most cases, there were no ghost peaks in my chromatogram, or if there was, they were much, much smaller. This told me that the ghost peaks were not a result of my GC column. However, if the ghost peaks remained, I would trim 1-loop (approximately 0.5-meters) from the inlet side of my capillary column and perform a 10 minute bake-out by setting the GC oven to 20°C less than the column’s maximum isothermal temperature.
Injection port septa
If the GC column is not causing ghost peaks, where should you look next? One of the most common sources is injection port septa. I wrote the following post to help identify and fix this issue.
To test, replace the injection port liner after thoroughly cleaning the injection port. Then replace your injection port septa and make sure your septum purge is working properly (by measuring the septum purge flow with a flow meter). Keep the injection port cool and without injecting anything, obtain some data like described in the GC column carryover paragraph above. This same test can also identify potential off-gassing contamination from the injection port O-ring (if applicable to your instrument).
If the baseline does not contain ghost peaks, heat the injection port and acquire an analytical analysis without injecting anything. If the baseline still does not contain ghost peaks, inject a solvent blank into a heated injection port. If you observe peaks, your syringe, blank solvent, and/or rinse vial solvent may be the problem, and you should review the posts below. Or, your injection port may not be as clean as you think it is.
Another potential source of ghost peaks is from your sample/standard vial septa. So how do you test for this? I used to remove the septa from the vial cap and make an injection (manually or using an autosampler). Do you still see peaks? If not, these septa are likely the cause. If you do, then switch to a septa made from different material and repeat the experiment. Remember to only inject once from a vial, as piercing the septa multiple times may lead to ghost peaks.
If nothing listed above helps you determine the source of the ghost peaks, take a closer look at your instrument and/or carrier gas. Testing carrier gas is relatively easy; cool your injection port and GC oven temperature to 35°C while letting the carrier gas flow for several hours (do not cool the detector). With the GC oven and injection port still at 35°C, start acquiring data with the instrument. Then, increase the injection port and oven temperatures. If you notice ghost peaks, then suspect the carrier gas. (Example of a carrier gas filter trap shown below.)
Split Vent line and trap
If you normally operate the GC in splitless mode, you need to check one more thing. The ghost peaks may actually be coming from the split vent line and/or split vent trap. Remember that the split vent line is seldom heated, so material (compounds) introduced into it may be condensing out, and vapors may be working their way back into the injection port. To test, set the injection mode to split (2:1 split should be fine). Inject a few solvent blanks. Without letting the instrument switch to splitless mode, acquire data for a sample, standard, or solvent blank. If the ghost peaks disappear while in the split mode, then the split vent line and/or trap are likely the problem, and both should be replaced. (Example of a split vent trap shown below.)
I hope something here has helped you identify and fix the source of your ghost peaks. For additional information on troubleshooting ghost peaks, I suggest you read Jaap de Zeeuw’s blog post Poster on sources for “Ghost-Peaks” in Gas Chromatography which contains links to five articles he has written about ghost peaks.
A while back I was doing some reading and came across an application of GC-MS in food safety that caught my attention. The analysis of food products, specifically soy sauce, for contamination with chloropropanols. So how do chlorinated alcohols end up in food?
During the production of a food ingredient known as hydrolyzed vegetable protein (HVP) various vegetable protein feedstocks such as corn gluten, wheat gluten, and soybean meal are combined with dilute HCl and heated at around 100° C. This process causes hydrolysis of the peptide bonds within the bulk protein leading to a mixture of short peptides and free amino acids. We perceive the presence of free amino acids as an umami flavor and so HVP is often added to foods in order to give a more rich or meat like taste.
Problems can arise when process parameters like the strength of the acid, reaction time, temperature, and the level of residual lipids in the feedstock are not well controlled. Residual lipids can undergo the following reactions in which the released glycerol is chlorinated.
Most analytical work and regulations in this area focus on the compounds 3-chloro-1,2-propanediol (3-MCPD) and 1,3-dichloro-2-propanol (1,3-DCP) although other structural isomers are known to occur as well as species in which one or more fatty esters remain bound. The European Commission has set an upper limit of 0.02 mg/kg for 3-MCPD in soy sauce and HVP on a dry matter basis and requires analytical methods to achieve an LOQ of 0.01 mg/kg. Assuming our soy sauce under test is 40% dry matter, the LOQ would be 4 ppb on a liquid basis.
The state of analytical methods for these compounds in soy sauce and other food matrices has been advancing in recent years, however many still rely on large volumes of extraction solvents which can be expensive and unfriendly to work with. My goal for this project is to develop a method for determining chloropropanols in HVP based soy sauce using a QuEChERS extraction with acetonitrile followed by dispersive sample cleanup, derivatization with heptafluorobutyric anhydride (HFBA), and finally GC-MS analysis. I also hope to incorporate the shoot and dilute concept which entails running the GC inlet in split mode to reduce buildup of nonvolatile sample components on the inlet liner and column.
A tall order indeed! I began by extracting 10 G portions of both traditionally brewed and HVP based soy sauce using the standard unbuffered QuEChERS procedure. Traditionally brewed soy sauce is made by fermentation and should be free of any incurred chloropropanols while HVP based products have the potential for contamination. The traditionally brewed sample will serve as a blank matrix for calibration and method development.
Giving those extracts the highly scientific eyeball test hints that we are in for quite a challenge. Keep following my series of blogs on this project as things progress and feel free to comment or contact me if you are interested in this analysis or perhaps have even run it yourself.
Here are some references for those interested in reading up on the issue.
EFSA Journal, 2013, 11(9), 3381
AOAC official method 2000.01
Journal of Chromatography A, 952 (2002), 185–192
Food Control, 18 (2007), 81–90
The benefits of superficially porous particles are without question. Who wouldn’t want to perform faster separations without the need for expensive UHPLC instrumentation? It sounds too good to be true. There must be some drawback – right?
The solid, impermeable core present in Raptor™ particles increases column efficiency by decreasing the diffusion path at the cost of reduced surface area typically available in traditional fully porous material. There is a potential concern that arises from a reduction in surface area: column loading ability. Column overloading (mass overload) occurs when the amount of material injected onto the column exceeds the available active sites of the stationary phase.
Figure 1: Classically observed peak shape due to column overload
The solid core in the Raptor™ 2.7 μm particle has a diameter of 1.7 μm with a porous layer that is 0.5 μm thick while the core in the Raptor™ 5 μm particle has a diameter of 3.4 μm with a porous layer that is 0.6 μm thick. After doing some simple math, the volume of fully porous material is reduced by ~25% in the 2.7 μm particle and by ~40% in the 5 μm particle compared to fully porous particles of the same diameter (along with a proportional decrease in surface area). Will this limit sample loading capabilities compared to fully porous silica? Let’s check by looking at data collected by injecting a standard volume (1 μL) across a concentration range (0.01 – 100 μg on column) of a neutral probe (biphenyl).
Figure 2: Mass on Column vs. Peak Width (w0.5) Normalized to w0.5,min
The peak width at half height remains consistent until the mass on column approaches ~10 μg for both fully porous material and Raptor™ (Figure 2). This corresponds to a 1 μL injection at 10 mg/mL. For most modern detectors and assays, this is orders of magnitude higher than is typically injected. How can we lose so much surface area, but still maintain loading ability at relevant concentrations?
The volume ratio of porous to non-porous material utilized in Raptor™ serves to reduce the diffusion pathway and maintain detection-appropriate loading abilities. Reducing the size of the solid, impermeable core would increase the loading ability, but it would also decrease efficiency. Likewise, increasing the size of the solid, impermeable core would increase your efficiency, but the loading ability would no longer be practical. As manufactured, Raptor™ provides speed and efficiency without compromising the loading ability required for most modern methods.
Internal Standard versus External Standard Quantification in Medical Cannabis Potency Analysis with GC-FID
The ChromaBLOGraphy series continues for the use of internal standards with medical cannabis potency testing by GC-FID (I’ve listed the first two parts in the series immediately below as web-links). This third part demonstrates the positive impact an internal standard can have on quantitative accuracy.
The chromatograms below show a calibration standard at 50 ng/µL cannabinoids and internal standards (Phencyclidine, or PCP, was used for calibration and quantification) analyzed during initial calibration on a Friday and then again on Monday after sitting on the autosampler all weekend. Now first, I would never leave calibration standards of any type out all weekend and plan to use them again, except in this case the evaporative loss of solvent and resulting concentration of analytes serve to illustrate one of the main benefits of an internal standard: taking care of a proportional loss or gain in peak areas so that quantitative accuracy is not compromised. A peak area gain or loss could arise from injecting too much, injecting too little, or evaporative loss of solvent over the course of analyzing a large sample queue.
As you can see in the orange chromatogram, especially since I marked PCP and delta-9-THC with green lines, the peak heights (and thus areas) are higher across the board for this standard, which is the same one analyzed previously. However, when the internal standard technique is used for quantification, the analyzed values are very close to the expected values. If internal standard (ISTD) quantification is not used, in favor of external standard (ESTD) quantification, the values are not accurate.
The above situation could be considered an extreme case, so I put together a table that compares ISTD and ESTD values for check standards (those standards analyzed during/after a sample queue to make sure the calibration is holding) and standards analyzed after sitting on the autosampler for the weekend. In almost all cases the ISTD quantification method produces a better value for the standards. Hopefully this motivates you to consider using internal standards in your medical cannabis analyses.
The following video is step-by-step tutorial on assembling a Restek Veriflo® flow controller and then leak checking the controller:
For information on calibrating Restek flow controllers (following assembly) and collecting a sample with a Restek flow controller, be sure to check our part III of this blog series
*I said 90 inch-pounds for the face nut torque, but it should be 90 foot-pounds.
Yesterday’s ChromaBLOGraphy post concerned the use of internal standards (ISTDs) for GC-FID potency testing of medical cannabis. In that post I defined desirable characteristics for an ISTD and said that one of the benefits of ISTD use is better quantitative accuracy. Good quantitative accuracy starts with good calibration, which I demonstrate in this post by showing a calibration curve for delta-9-THC that is based on ratios generated through PCP internal standard use. For each standard of THC, ranging from 0.1 to 100 ng/µL, PCP and Prazepam were added at 50 ng/µL each. The peak area of delta-9-THC divided by the peak area of PCP represents the “Ratio” shown on the calibration curve for each concentration level.
As previously, cannabinoids were analyzed using a 15m x 0.25mm x 0.25µm Rxi-35Sil MS GC column with FID (see chromatogram below). Evaluation of the initial calibration for cannabinoids can be done in several ways, including by visual inspection of calibration curves. If you look below you will see the calibration curve for delta-9-THC, which stretches from 0.1 ng/µL to 100 ng/µL in the first figure. I zoom in on the lower points in the next figure so the reader can see how close to the curve the points are, which is an indication of linear fit.
In addition to the “eyeball test”, calibration curve quality can be viewed in terms of Average Response Factor (Avg RF) and Response Factor % RSD (% RSD), which are shown in the table below. It’s up to the individual user to determine what meets their data quality objectives, but achieving % RSD values less than 20% is generally very good for such a wide calibration range as I show here. For the ISTD cannabinoid calibrations in this work, the % RSD values are much lower than 20%.
Correlation coefficients (CCs) are commonly used to determine linearity for calibration curves, and those are also shown in the table below. Every cannabinoid calibration curved generated using the internal standard technique shows a 0.9998 correlation coefficient, which is excellent.
One thing to point out is that I am using split injection with our Sky Precision split inlet liner with wool at a split ratio of 20. This means that on column amounts for the calibration range are ~5 pg to 5 ng. Not only is the FID our most linear detector, well demonstrated here, but it is a very sensitive detector, too, something shown by the last chromatogram below for ~10 pg of each cannabinoid on column. My recommendation to those that are doing dirty samples for potency determination, whether they are chlorophyll-laden flower extracts, or lipid-rich food extracts, or feedstocks, is to take advantage of that FID sensitivity and inject as little sample as possible while still being able to do accurate quantification for all the cannabinoids of interest. You’ll keep your GC system up longer by injecting less “dirt”. Accuracy will be helped by using an internal standard, too.
I am often asked about internal standards for use in medical cannabis potency testing with gas chromatography. I finally got some time in the lab to check this out and came up with a couple of possibilities after testing numerous compounds for favorable retention times versus typically analyzed cannabinoids.
Internal standards are mostly used by adding to calibration standards and sample extracts at known and consistent concentration levels immediately prior to GC of those standards and extracts. Internal standards are especially helpful to indicate when an injection malfunction has occurred, including injecting too much or too little versus what is typical. In either of these cases, the cannabinoid compounds of interest show the same proportional peak area increase or decrease, as would an internal standard. By using response factors (peak area of compound of interest divided by peak area of internal standard), quantitative accuracy will be better, even when relatively large errors occur in injected sample volumes. Modern GC data processing software allows easy calibration and quantification with internal standards.
Desirable qualities in an internal standard include:
- Easy to chromatograph.
- Does not react with any compounds in standards and/or samples.
- Well separated from any compounds in standards and/or samples.
- Not expected to be present in samples before its addition as an internal standard.
- Low cost.
The 15m x 0.25mm x 0.25µm Rxi-35Sil MS chromatogram below shows that phencyclidine (PCP) and Prazepam (Przpm) meet the “well separated” criterion for internal standards, bracketing the cannabinoids shown in my last ChromaBLOGraphy post. PCP and Przpm are extremely unlikely to be present in medical cannabis extracts. Restek offers PCP and Przpm at DEA-exempt concentrations appropriate for internal standard addition and, at reasonable costs.
The performance of both internal standards was reliable in recent testing work, which will be demonstrated in an upcoming blog post.