Adulterant and Contaminant Screening

This is incoming-material and quality-control screening of legitimate research materials for a client under a written services agreement. It answers the question: does this lot of research material contain something it should not — from shared-equipment carryover, from an opaque upstream supply chain, or from deliberate substitution.

It is not a drug-checking service. We do not accept unsolicited or anonymous samples, we do not accept street samples, and we do not answer the question "is my sample safe." That is a categorically different activity with a categorically different legal footing, and it is not what this laboratory does.

The demand for the first question is real and boring, which is why it is defensible. Research chemicals, peptides, and bulk materials are frequently synthesized or milled in facilities with no segregation controls, and trace carryover from a prior campaign on the same mill, sieve, blender, or packaging line is a documented contamination route. At the potency of some contaminants, microgram-level carryover is analytically meaningful in a way that carryover of a benign excipient is not. Distributors also commonly sit three or more tiers downstream of the actual synthesizer, with incoming-material testing as their only real control point. And where a material is expensive and a cheap substitute exists, economically motivated adulteration occurs — the adulterant chosen specifically because it passes the routine test.

Arcadia Diagnostics is not registered with the DEA and does not accept controlled substances. This service screens non-controlled research materials for the presence of fentanyl-class and other adulterants as contaminants. Samples must be declared non-controlled at intake, and material known or suspected to be a controlled substance, an analog, or a listed chemical must not be sent. Where a screen indicates a scheduled substance, the sample is quarantined, testing stops, and disposition is handled under our intake policy — we do not proceed to quantitation of a controlled substance.

These techniques are not interchangeable. They answer different questions with different error structures, and the correct architecture runs them in sequence.

Tier 1 — presumptive screening.

Lateral-flow immunoassay strips are a competitive assay in which an antibody binds either the analyte or an immobilized conjugate, so a positive suppresses the test line — a result inverted relative to intuition, which is itself a documented source of user error. They are designed for urine at roughly a 20 ng/mL cutoff; use on bulk powder is off-label repurposing requiring enormous and unstandardized dilution. Documented failure modes include false positives from high-concentration interferents such as methamphetamine, MDMA, diphenhydramine, and lidocaine at or above about 1 mg/mL; non-monotonic high-dose behavior; substantial lot-to-lot variability in sensitivity to analytes and to interferents; and uneven cross-reactivity across the fentanyl analog family. Critically, nitazenes are 2-benzylbenzimidazoles — an entirely different scaffold from the 4-anilidopiperidine fentanyls — and fentanyl immunoassay antibodies do not reliably bind them. Under the SWGDRUG framework, immunoassay is Category C, the lowest discriminating-power tier, alongside colorimetric spot tests.

Portable FTIR and handheld Raman are Category A techniques, but they are bulk methods with detection limits typically in the low percent range — roughly 1–5% w/w depending on the component and the mixture. They identify the principal component and the major cut. They do not detect trace contamination, and Raman additionally suffers fluorescence interference from colored samples.

Tier 2 — confirmatory identification.

LC-HRMS, untargeted (QTOF or Orbitrap): full-scan accurate-mass acquisition, typically under 5 ppm mass error, with data-dependent or data-independent MS/MS. This is what finds the thing nobody was looking for. It supports suspect screening against a database of exact masses and predicted fragments, non-targeted characterization by accurate mass and isotope pattern, and — operationally the strongest argument for it — retrospective data mining, since an acquired file can be re-interrogated months later for a newly emergent compound without re-running the sample. The threat list changes faster than method development cycles do.

LC-MS/MS, targeted (triple quadrupole, MRM): the workhorse for trace confirmation and quantitation, with detection limits routinely in the low ng/mL and sub-ppm w/w in the original solid depending on dilution. A defensible identification requires all of: retention time matching an authentic reference standard within a defined tolerance; at least two independent MRM transitions, one quantifier and one or more qualifiers; the qualifier/quantifier ion ratio within a defined tolerance of the standard; and all of the above run in the same sequence as the standard.

GC-MS at 70 eV electron ionization: reproducible, library-searchable fragmentation, and strong for xylazine, caffeine, phenacetin, levamisole, local anesthetics, and most classic cutting agents. Weak for thermally labile and low-volatility compounds — nitazenes chromatograph poorly under GC conditions and can degrade in the inlet.

Confidence levels are stated, using the Schymanski scale. Level 1 is confirmed against a reference standard; Level 2 is a probable structure from library or diagnostic evidence; Level 3 a tentative candidate; Level 4 an unequivocal molecular formula; Level 5 an exact mass only. An HRMS hit without a standard is Level 2 at best and is never reported as an identification.

Isomerism is the core analytical difficulty for fentanyl-class analytes. Many analogs are exact-mass identical and produce near-identical MS/MS spectra — the ortho, meta, and para fluorofentanyl positional isomers are the canonical example, and cyclopropylfentanyl versus crotonylfentanyl is another. Accurate mass alone cannot distinguish them. Only chromatographic resolution against authentic standards can. Any claim to identify fentanyl analogs must be scoped to the analogs for which the laboratory holds standards and has demonstrated separation.

Nitazenes — isotonitazene, metonitazene, protonitazene, etonitazene, butonitazene, etodesnitazene, N-desethyl isotonitazene, and a continuously expanding set — are extremely potent, are not detected by fentanyl immunoassay strips, and behave poorly under GC conditions. The analog list turns over faster than reference-standard availability, which is exactly why untargeted HRMS matters here. A precursor-ion-scan approach on a triple quadrupole offers a class-level alternative: nitazene analogs share a common fragmentation core, and scanning for precursors of a small set of diagnostic product ions detects analogs for which no standard is held. That gives class-level detection of novel analogs — but class-level detection remains presumptive until a standard confirms the specific analog.

The moving-target problem, illustrated. Xylazine, an α₂-adrenergic agonist veterinary sedative, dominated as an adulterant through 2023–2024. Surveillance data from 2024–2025 show sharp regional substitution toward medetomidine and dexmedetomidine, which are far more potent — order-of-magnitude estimates in the 200–300× range versus xylazine — and therefore present at much lower concentrations. A method with adequate detection limits for xylazine may be entirely blind to medetomidine. That is the clearest available demonstration that detection limits are analyte-specific and that panels must be revised rather than inherited.

BTMPS makes the same point from the other direction. Bis(2,2,6,6-tetramethyl-4-piperidyl) sebacate is an industrial hindered-amine light stabilizer used as a plastics UV additive, with no human-use application. It appeared abruptly and at scale in the U.S. illicit supply in mid-2024 — in one surveillance dataset rising from 0% of samples in June 2024 to over a third by August 2024. It is not a drug, it was on nobody's panel, and it entered the supply faster than any targeted method could be developed and validated. That is the argument for untargeted acquisition and retrospective data mining, stated as a fact rather than as a promise.

Cutting agents and diluents more broadly — caffeine, phenacetin, acetaminophen, levamisole, mannitol, lactose, inositol, dextrose, sucrose, boric acid, benzocaine, lidocaine, procaine, tetramisole, quinine, dimethyl sulfone, creatine, diphenhydramine, and designer benzodiazepines such as bromazolam — are largely well handled by GC-MS with library matching. Sugars and polyols require derivatization or a dedicated method.

The targeted panel covers the fentanyl-class analogs, nitazenes, benzodiazepine-class adulterants, and common cutting agents for which we hold authentic reference standards, and it is named in full — compound by compound — in the quotation and again on the report. We deliberately do not publish an analyte count, because a count invites the assumption that anything unlisted was looked for. Where an isomeric set is on the panel, a result is reported to the individual isomer only where chromatographic separation has been demonstrated on our system; otherwise it is reported as the isomer group.

A targeted identification requires the retention time to fall within ±2% of the reference standard analyzed in the same batch, two MRM transitions to be acquired, and the qualifier-to-quantifier ion ratio to fall within ±20% relative of that standard. The criteria applied are printed on the report, and a feature that meets some but not all of them is reported as presumptive — never as confirmed.

A "not detected" result means, and means only: the analytes named in the report were not detected, in the aliquot analyzed, at or above the detection limits stated, by the methods stated, in the sample as received on the date received.

It does not mean:

• That the compound is absent. It is below the detection limit, which is not zero.
• That the batch is clean. One aliquot was tested. Trace contamination is rarely homogeneously distributed — where a contaminant is present as discrete particles, the subsample either contains a particle or it does not, and analytical precision is irrelevant to that outcome. Testing part of a unit tells you about that part; testing one unit tells you about that unit. Neither generalizes to the batch.
• That other units, lots, or shipments are clean. The result attaches to the sample tested.
• That compounds outside the panel are absent. A targeted method cannot exclude what it did not look for. Novel analogs and novel adulterants are simply unexamined.
• That a presumptive negative excludes the class. A negative fentanyl immunoassay does not exclude nitazenes, does not reliably exclude carfentanil, and does not exclude fentanyl below the strip's effective cutoff after dilution.
• That the material is safe, suitable for any use, or fit for any purpose. Contaminant screening is not a safety determination.
• That the material has not changed since sampling. Degradation, further handling, and repackaging all occur after the laboratory's custody ends.

Matrix effects are why one universal detection limit cannot honestly be quoted. In LC-MS with electrospray ionization, co-eluting sample components alter ionization efficiency — usually suppression. In a heavily cut or excipient-loaded powder the matrix is effectively the sample: mannitol, lactose, PEG, surfactants, and the bulk material all co-elute and compete for charge. An analyte genuinely present at five times its nominal detection limit can fail to be detected, and quantitation can be off by a large factor with a perfectly clean-looking chromatogram. Different matrices, different limits.

And a related integrity point that belongs here: an HPLC-UV chromatographic purity of 99% is an area percent of UV-absorbing species eluting under stated conditions. It says nothing about water, residual solvent, counterion, inorganic salts, or any co-eluting or non-chromophoric species. A material can be 99% chromatographically pure and well under 80% peptide by mass. Purity and content are separate measurements and are reported separately.