Category: Lab Analysis

For over forty years, methamphetamine-using communities have been speculating why some batches of meth seem qualitatively different than others. When we started working in this field in the 1990s, the claim was that l-meth, solvents, and synthesis impurities were the culprits.

Since around 2010, the DEA’s policies have resulted in most methamphetamine inside the United States coming from large manufacturers outside the US. In the ensuing years, some have claimed that the unpleasant effects of street meth in the US are a result of meth containing N-Isopropylbenzylamine (isopropylbenzylamine; N-IPBA; “N-iso”) [PubChem]. It is a common enough claim that many skeptics have called N-iso the new meth boogeyman.

The most common claims about isopropylbenzylamine are that it is present in combination with meth, and causes more paranoia, psychotic ideation, and worse hangovers. It’s also blamed for effects atyptical for stimulants, such as lethargy and “brain fog”, 1-3 days into a meth-using session.

We’ve been asked about N-iso many times between 2019 and 2022. We purchased the certified reference lab standard for isopropylbenzylamine twice in the last two years in order to run experiments to check our methods. We have repeatedly confirmed that zero (0) meth samples analyzed by DrugsData to date have contained it:

https://drugsdata.org/results.php?substance1=94&substance2=2036

Erowid’s DrugsData reported those findings to the people submitting the samples in question, and on the corresponding entries on DrugsData.org. Up until now, we haven’t specifically pointed out publicly that multiple submitters have claimed that isopropylbenzylamine could be present in their methamphetamine and that our findings have not substantiated their claims.

Our results caused others to ask us more and more pointed questions about how sure we were that N-iso wasn’t in the meth we tested. So we dug deeper. After re-analyzing and examining the GC/MS data for more than a dozen samples submitted to DrugsData for which the submitter was certain their meth sample contained isopropylbenzylamine, we’ve still seen no results where this is the case.

Out of 271 samples containing methamphetamine analyzed between January 2019 and July 2022, none contained isopropylbenzylamine:

https://drugsdata.org/results.php?s=methamphetamine&search_field=substance&y1=2019&y2=2022

Re-examining Our Findings

In late 2020, we ran several samples of alleged meth + N-iso and they came back meth only. We bought the reference standard and tried slightly-modified procedures using this standard. It seemed to us that meth + isopropylbenzylamine resolved easily via GC/MS. That is to say, meth and N-iso were easily differentiated in our lab. They are close, but we deal with much harder problems all the time.

In May 2022, D.M. contacted us to point out a paper that said it is possible that mixed samples containing both meth and isopropylbenzylamine could cause analysis to fail to see one or the other. Essentially, they claimed our previous findings might be wrong because one of the two could hide inside the GC curve of the other and then elute (come out) at such a similar time into the Mass Spectrometer that our software would report only one of the two chemicals.

That paper is Luo Y, et al. (2021) “Simultaneous Determination of Methamphetamine and Its Isomer N-Isopropylbenzylamine in Forensic Samples by Using a Modified LC-ESI-MS/MS Method”. (ResearchGate Link) The authors write:

“However, the two compounds [methamphetamine (MA) N-isopropylbenzylamine ([N-IPBA]) ] were hard to be effectively discriminated by GC/MS when there was a large concentration difference between them. Because the retention times for MA and [N-IPBA] chromatographic separation were very close due to their high similar chemical structure, the compound with high concentration would interfere with another one with low concentration as the two compounds yield similar ion fragments for detection [25].”

Note the relevant claim in their paper is actually cited to someone else and is not something these authors themselves demonstrate in their article. The original citation (Xuan J et al 2015) is a paper in a Chinese journal that we’ve been unable to locate.

Given this new, reasonably specific claim from a 2021 paper, despite having done it before, we purchased a new isopropylbenzylamine reference standard, this time from a different chemical supplier. Unsurprisingly, it matched exactly the previous standard and also matched the GC/MS data in the main public/research/commercial/forensic libraries.

A Series of Experiments

We mixed pure d-meth with N-IPBA at 1:1, 1:10, 1:100, 10:1, and 100:1 ratios. In all of the conditions, our setup showed isopropylbenzylamine as clearly distinct from methamphetamine. They would not be mistaken for one another or lost, even way below 1:100. Our standard procedure involves methanol run through an Agilent GC/MS 5973 MSD with the GC column being an HP-5ms Ultra inert (5%-phenyl)-methylpolysiloxane. Unless you’re a lab tech, that won’t mean anything to you, but it’s a fairly normal setup for doing drug work like this and it’s well suited to analyzing chemicals of this type.

The GC output pictured above is from N-IPBA (“N-iso”) (1 part) mixed with meth (100 parts). The other main peak (9.972) is a calibration chemical. The slightly messy baseline to the right of the meth peak is related to the way that the salt versions (Meth HCl, for instance) of the two drugs differ in their elution times in the GC column. This example graph is after several tests in a row using different ratios of meth and isopropylbenzylamine. It is common when running methamphetamine salts to end up with a little right-side, baseline noise after the sharp freebase meth peak.

Methamphetamine and isopropylbenzylamine do elute at similar times, but using our procedure, they are clearly distinct. Note in the GC image the sharp valley between the N-iso and meth peaks: N-iso at 3.4777 minutes and meth at 3.687 minutes in this run.

And they always have clearly distinct Mass Spectra (MS), so they simply don’t get confused at our lab. If a lab were running a different column and procedure that isn’t targeted for doing work on methamphetamine and related drugs, it’s easy to imagine other procedures and rigs where an analytical chemist could confuse one with the other.

As of August 2022, DrugsData’s lab has found isopropylbenzylamine in eight samples total, ever, and two DEA-tested samples are republished in our database:

https://www.drugsdata.org/results.php?search_field=substance&s=Isopropylbenzylamine

It is Erowid’s view that most negative effects from meth use are a result of lack of sleep combined with irregular water and food consumption. People mistakenly attribute differences in experience from time to time to differences in impurities in the drugs, instead of other factors such as diet, mood, context, electrolyte levels, and physical rest.

It’s certainly possible that a sample analyzed in DrugsData’s lab in the future could contain both methamphetamine and isopropylbenzylamine. We feel certain that for such a sample, lab results would clearly show this to be the case.

—earth, Sylvia, Fire, Roi

This is a description of how we identified the unidentified substance in a fentanyl sample (#12495) analyzed in March 2022.

The unidentified substance has mass spectrum major ions at 96; 217.1; 174.1 with an elute time at around 10.7 in our main setup.

This small sample of white powder in a blue bindle was submitted to DrugsData via research partners we’re working with to do lab confirmatory testing. Besides Fentanyl and 4-ANPP, it contained a third chemical we were initially unable to identify.

We publish the Mass Spectrum (MS) images for substances we need help identifying. A colleague at UNC’s drug checking project examined this substance’s image and reached out with some clues, which put us on the path of figuring out what its structure is.

One of our top volunteer analytical chemistry experts, Eddee, found a close match in Wiley’s 2020 “Designer Drugs” library, a library that our DrugsData lab doesn’t have. The close (though not perfect) match is for ethyl-despropionyl-fentanyl (ethyl-4-ANPP). Our experts (thanks, Koby!) guess that this is likely to be much less potent than fentanyl and might not be very active, similar to despropionyl-fentanyl (4-ANPP).

There’s a PubChem page for it, but no CAS number yet:

https://pubchem.ncbi.nlm.nih.gov/compound/156346345

Eddee speculated that the difference between our sample and the Wiley library match might not be meaningful. In the following image, which is pretty complex to look at, the top chemical is our DrugsData sample’s unidentified substance. The lower chemical is the Mass Spectrum for the presumptive ethyl-despropionyl-fentanyl. The middle part of this image is a comparison of the two, with our sample on the top (lines going up) and the proposed match on the bottom (lines going down). You’ll need to open it in a new tab to see the detail, it’s dense stuff.

What you’re looking at are the relative heights of the largest peaks (vertical lines), aka the “ions”. Mass spectrometry (MS) relies on breaking up a chemical with a high energy stream of electrons; Erowid’s DrugsData lab uses an “electron spray” method. The resulting bits are highly charged ions that get spun through a magnetic whirlwind inside a specialized detector. The heights of the lines indicate how many of each ion was detected for this chemical using specific equipment and methods. Perfect matches usually require using the exact same equipment at the same settings, but there’s a lot of similarity when using equivalent machines.

Looking at the middle graph, where the two are compared against each other, you can see there’s a short line on our sample at 199 that doesn’t exist in the lower sample. And the relative heights of some of the key ions are different between the two. That doesn’t mean it’s not a match, but it isn’t a perfect match.

So Eddee checked and his lab did have a tiny bit of despropionyl-Fentanyl (aka 4-ANPP) left in their Fentanyl Analog Screening Kit (FAS Kit, sometimes referred to as a Traceable Opioid Material Kit, or TOM Kit).

There wasn’t much left, but he decided to try wet chemistry “derivitization” (a simple synthesis) using iodoethane, and was able to get a tiny amount of product he believes to be ethyl-despropionyl-fentanyl. He then ran that new product through a GC/MS and got the following output.

As above, in this image our unidentified substance is on the top; Eddee’s new ethyl-despropionyl-fentanyl is on the bottom.

Again, if one looks closely, there are some important differences between our sample (in blue on the top of the middle graph) and the newly synthesized chemical. There are several complexities we can’t completely account for. First, Eddee had only a teeny tiny amount of his synthesized chemical and sometimes “very low signal” amounts of a drug can have different Mass Spectrum profiles. Usually this doesn’t make a difference in which ions show up (except at the shortest peaks), but it can cause the relative ratios to be slightly different. Second, Eddee isn’t using the exact same GC/MS brand, model, and components as we have.

We might be able to change our GC coil and run parameters to better match this, but it’s so close, we’re going to consider this matter closed.

If you’re interested, you can check out the unidentified substances detected by DrugsData in 2021-2022 that haven’t been solved yet. This list changes as we make identification breakthroughs, thanks to tips from the Erowid Expert Network and others.

In an amusing postscript, one day after Eddee finalized this identification (Apr 4, 2022), DEA Special Testing and Research Laboratory (SFL1) wrote that they had come to the same identification of the so-called “308-G impurity”.