Deformulation: Reverse Engineering Chemical Formulations

Deformulation is the process of taking a finished product apart, chemically speaking, to figure out exactly what went into it. Think of it as reverse-engineering a recipe: instead of combining ingredients to make something, you start with the final thing and work backward to identify every component, its quantity, and sometimes even the steps used to put it all together. The practice spans industries from pharmaceuticals to perfumery, plastics recycling to competitive intelligence, and the toolkit has grown dramatically in the last few decades. What makes deformulation genuinely difficult, and genuinely interesting, is that commercial products are designed to blend their ingredients into seamless wholes, not to be easily read.

Why Anyone Would Want to Reverse-Engineer a Formula

The motivations behind deformulation are varied, and they shape the approach taken. A generic drug manufacturer trying to develop a bioequivalent tablet needs to know not just the active ingredient but also the fillers, binders, coatings, and disintegrants in the branded product, along with their approximate proportions. A materials scientist investigating why a rubber seal failed in the field needs to understand the polymer’s original formulation and how it changed over time. A fragrance house analyzing a competitor’s perfume wants to identify dozens or even hundreds of volatile compounds and their relative concentrations. And increasingly, recycling operations need to separate additives from plastics before those materials can re-enter the supply chain.

In each case, the starting point is the same: a finished product whose composition is not publicly disclosed, or whose label only tells part of the story. The challenge is also broadly similar. Commercial formulations are complex mixtures, and the manufacturing process itself can alter the chemical identity of some components through reactions like cross-linking, curing, or coating. Deformulation has to account for all of that.

The General Approach

Most deformulation work follows a two-phase logic, regardless of the industry. The first phase involves physically or chemically separating a product into simpler fractions. For an emulsion-based product like a cutting fluid, this means splitting the aqueous and organic phases apart. For a solid like a pharmaceutical tablet, it might mean dissolving certain layers, extracting specific compounds with solvents, or using heat to drive off volatile components. The second phase is detailed analysis of each separated fraction to identify and, when possible, quantify its individual compounds using spectrometric or chromatographic methods.

1Journal of Oleo Science. Review on Cutting Fluids: Formulation, Chemistry and Deformulation

This two-step pattern holds whether you’re dealing with a lubricant, a drug, a polymer composite, or a cosmetic. What changes from one product to another is which separation methods are chosen, which analytical instruments are pointed at the fractions, and how many layers of complexity you have to peel back before you get to clean identification.

The Analytical Toolkit

Deformulation relies on a broad set of instruments, and the choice of tools depends heavily on what the product is made of and what questions need answering. No single technique does everything, so analysts typically combine several methods, each revealing a different dimension of a sample’s identity.

Chromatography

Chromatographic techniques separate mixtures into individual components based on how those components interact with a stationary phase as they’re carried along by a mobile phase. Gas chromatography coupled with mass spectrometry (GC-MS) is a workhorse for identifying volatile and semi-volatile organic compounds. Liquid chromatography paired with mass spectrometry (LC-MS) handles compounds that are less volatile or thermally fragile. For polymers, size-exclusion chromatography (SEC) separates molecules by their physical size, which correlates with molecular weight. Sometimes one dimension of separation is not enough. A study of complex star block copolymers demonstrated that neither gradient liquid chromatography nor SEC alone could correctly resolve a 16-component mixture, but combining both methods in a two-dimensional system separated all 16 components in a single analysis and even revealed previously undiscovered end groups in polyester samples.

2ACS Publications. Two-Dimensional Chromatography for the Deformulation of Complex Copolymers

Coupling chromatographic separation with continuous infrared spectrum acquisition offers another powerful combination. This approach resolves polymer populations into discrete identifiable components and can determine how the mass of each component is distributed across the chromatographic profile. Applications range from analyzing additives in polymers to characterizing copolymer compositions and identifying degradation byproducts.

3International Journal of Spectroscopy. Polymer Characterization by Combined Chromatography‐Infrared Spectroscopy

Spectroscopy

Spectroscopic methods probe how a sample interacts with electromagnetic radiation, and different types of spectroscopy reveal different structural features. Infrared spectroscopy, including techniques like attenuated total reflectance Fourier transform infrared (ATR-FTIR) spectroscopy, identifies chemical bonds and functional groups. Raman spectroscopy provides complementary information and is useful for mapping spatial distributions of components within a solid product. Nuclear magnetic resonance (NMR) spectroscopy gives detailed information about molecular architecture. In the deformulation of liquid silicone rubbers, for instance, silicon-29 NMR was used to deduce polymer architecture while proton NMR quantified the relative concentrations of reactive groups and their stoichiometric balance.

4PubMed. Looking over liquid silicone rubbers: (1) network topology vs chemical formulations

Thermal Analysis

Thermal methods measure how a material behaves as it is heated. Differential scanning calorimetry (DSC) tracks heat flow into or out of a sample, revealing melting points, crystallization events, and glass transitions. Thermogravimetric analysis (TGA) monitors weight loss as temperature rises, indicating when components evaporate or decompose. Used together, DSC and TGA create a thermal “fingerprint” of a formulation. A forensic pharmaceutical lab demonstrated this by analyzing atenolol tablets: DSC and TGA together allowed qualitative identification of individual excipients, such as sodium starch glycolate, by comparing the finished tablet’s thermal profile to those of its known components.

5PubMed. The combined use of DSC and TGA for the thermal analysis of atenolol tablets

X-Ray Diffraction

X-ray powder diffraction (XRPD) is considered the definitive technique for identifying crystalline phases and polymorphs in solid materials.6PubMed. Quantifying API polymorphs in formulations using X-ray powder diffraction and multivariate standard addition method combined with net analyte signal analysis Polymorphism, where the same molecule can crystallize in different structural forms, matters enormously in pharmaceuticals because different polymorphs can dissolve at different rates and have different bioavailability. When researchers compared XRPD, DSC, and ATR-FTIR for quantifying two polymorphic forms of the cancer drug imatinib mesylate in both the raw material and finished tablets, XRPD proved the most accurate and was selected as the preferred validated method.

7PubMed. Quantitative determination of two polymorphic forms of imatinib mesylate in a drug substance and tablet formulation by X-ray powder diffraction, differential scanning calorimetry and attenuated total reflectance Fourier transform infrared spectroscopy

Pharmaceutical Deformulation

Drug products are among the most carefully deformulated materials, and the stakes are high. When a generic drug company wants to produce a version of an off-patent medicine, it needs to match the branded product’s performance in the human body. That means understanding more than just which active ingredient is present. The excipients (the inactive ingredients that make up most of the tablet’s weight), their particle sizes, their spatial distribution within the tablet, and even the crystal form of the active ingredient all influence how the drug dissolves and gets absorbed.

Raman mapping combined with statistical analysis methods has shown the ability to distinguish even small differences between tablets, including variations in component distribution and particle size, both of which are critical parameters when developing generic products that must match originals closely.8PubMed. Application of reverse engineering in the field of pharmaceutical tablets using Raman mapping and chemometrics The spatial arrangement of ingredients inside a tablet is not something a simple chemical analysis reveals. Techniques like Raman mapping generate a detailed picture of which compounds sit where, and statistical tools then quantify the differences in a way that removes human subjectivity from the comparison.

Deformulation in pharma extends beyond tablets. Packaging itself can introduce unwanted chemicals into a drug product. Elastomeric closures, the rubber stoppers in vial caps and syringe plungers, can leach organic compounds into injectable medications. Identifying these “extractables and leachables” typically requires LC-MS, GC-MS, or high-resolution mass spectrometry capable of detecting both known and unknown compounds at trace levels.

9ResolveMass Laboratories. Extractables and Leachables from Elastomeric Closures

Polymers and Materials Science

Polymers pose particular deformulation challenges because their properties depend not just on what monomers are present but on how those monomers are arranged, how long the chains are, how much cross-linking has occurred, and what additives have been blended in. A cured thermoset plastic, for example, cannot simply be dissolved and separated the way a mixture of small molecules can. The cross-linked network has to be broken apart under controlled conditions.

Pyrolysis-GC/MS offers one solution. By heating a polymer to the point where it thermally decomposes and then analyzing the volatile fragments, researchers can identify the starting materials and track structural changes. Work on a thermally cured polyimide system identified pyrolysis products characteristic of both precursor components and the cured polymer itself, making it possible to assess the systematic changes that occurred during the curing process.10Journal of Analytical and Applied Pyrolysis. Pyrolysis gas chromatography/mass spectrometry investigation of a thermally cured polymer In other words, pyrolysis-GC/MS can read not only what went into a polymer but how far the manufacturing process progressed in transforming it.

For uncured or soluble polymer systems, the combined chromatography-infrared approach mentioned earlier is often enough. But for heavily cross-linked or filled composites, analysts frequently need to layer several destructive and non-destructive techniques, building up a composite picture from partial information.

Perfume and Fragrance Deformulation

Few deformulation challenges are as romantically messy as reverse-engineering a perfume. A fine fragrance can contain hundreds of individual raw materials, including natural extracts that are themselves complex mixtures of dozens of molecules. The industry has relied on GC-MS and GC-FID (flame ionization detection) for roughly 40 years, but the traditional approach has a significant weakness: it relies on identifying individual “marker” molecules that point toward specific raw materials, a process that is manual, subjective, and unreliable due to the sheer variability and chromatographic overlap involved.

11Flavour and Fragrance Journal. Unraveling ingredients in complex mixtures by chromatographic spectrum recognition: Application to perfume deformulation

An alternative approach involves treating the entire chromatographic output as a kind of spectrum and matching it against reference libraries of known ingredients. Rather than picking out individual peaks and guessing which raw material they came from, this method compares the overall chromatographic “shape” of an unknown mixture to stored patterns. The shift is conceptually similar to moving from identifying a song by humming individual notes to using an app that recognizes the whole audio signature. It reduces the subjectivity inherent in marker-based identification, though building comprehensive reference libraries for every commercially available fragrance ingredient is a substantial undertaking in its own right.

Traditional Medicine and Natural Products

Deformulation takes on a different flavor when the “product” is a traditional medicine made from plant materials. Traditional Chinese medicine (TCM) prescriptions, for instance, combine multiple herbs in specific ratios, and verifying what is actually in a preparation is both a quality control and a safety concern. A binary-code approach to metabolite sequencing has been demonstrated for authenticating medicinal plants, performing preliminary chemical characterization, and deformulating TCM prescriptions by translating each sample’s diagnostic metabolome into a searchable digital sequence.12PubMed. Binary code, a flexible tool for diagnostic metabolite sequencing of medicinal plants The concept is clever: rather than trying to identify every compound in a complex herbal mixture, the method creates a simplified fingerprint that can be matched against a library. If the fingerprint of a supposed five-herb formula matches only four of the expected herbs, something is off.

The Legal Landscape

Deformulation inevitably raises questions about intellectual property. If a company can chemically dissect a competitor’s product, does that violate trade secret protections? In the United States, the answer is generally no, as long as the product being analyzed was obtained through legitimate means. Reverse engineering is widely recognized as a lawful method of discovering information, including under trade secret statutes.

The tobacco industry provides a striking illustration. Cigarette manufacturers routinely analyzed their competitors’ products in extraordinary detail for decades, generating extensive internal reverse-engineering reports. When these documents became public through litigation, it became clear that the compositions companies had claimed were proprietary were thoroughly documented by every rival. An analysis of these internal documents argued that because cigarette companies routinely analyzed competitors’ products in such detail, the formulation information was neither secret nor commercially valuable in a way that met the legal definition of a trade secret.13Tobacco Control. Cigarette company trade secrets are not secret: an analysis of reverse engineering reports in internal tobacco industry documents released as a result of litigation That finding had implications beyond the tobacco industry, because it highlighted a broader tension: if deformulation is routine and widespread in a given industry, it becomes hard for any company to claim that its formulation is truly secret.

This does not mean deformulation always gets a legal green light. If a product was obtained through theft, fraud, or breach of a confidentiality agreement, analyzing it can still constitute trade secret misappropriation. And patent protections operate independently of trade secret law. If a formulation is covered by a patent, reproducing it commercially could infringe regardless of how the knowledge was obtained.

Plastic Recycling and Environmental Applications

One of the more promising newer applications of deformulation thinking has nothing to do with competitive intelligence and everything to do with waste. Many plastics, particularly polyvinyl chloride (PVC), contain substantial quantities of additives like phthalate plasticizers that make the material flexible. These additives complicate recycling because they can leach out, contaminate recycled material streams, or pose health concerns. Deformulation in this context means separating those additives from the base polymer so that cleaner material can be fed back into production.

Accelerated solvent extraction (ASE) has been explored for separating phthalate plasticizers from PVC, with the explicit goal of obtaining materials more suitable for reincorporation into the value chain.14Industrial & Engineering Chemistry Research. Accelerated Solvent Extraction for Plastic Recycling: Modeling the Mass-Transfer Kinetics of the Separation of Phthalates from Poly(vinyl Chloride) ASE works by using solvents at elevated temperatures and pressures to extract target compounds faster and more completely than conventional extraction. The approach treats deformulation not as an end in itself but as a necessary preprocessing step that makes mechanical recycling viable for materials that would otherwise be downcycled or landfilled.

This environmental angle is relatively new but growing. As regulations around recycled content tighten and concern about legacy additives like certain phthalates increases, the ability to strip unwanted chemicals from plastic waste before recycling becomes commercially and regulatorily important. Deformulation, in this context, is less about figuring out what’s in a product and more about actively removing what shouldn’t be there when the material gets a second life.

Machine Learning and the Future of Spectral Interpretation

Across all of these applications, one of the persistent bottlenecks in deformulation is spectral interpretation: taking the raw data from instruments (mass spectra, infrared spectra, chromatograms) and correctly identifying which compounds produced which signals. When a mixture contains only a few well-characterized components, experienced analysts can do this manually. When it contains dozens or hundreds of overlapping signals, human interpretation becomes slow, subjective, and error-prone.

Machine learning is increasingly being applied to this problem. Spectral unmixing, the task of decomposing a complex measured spectrum into the individual contributions of its constituent compounds, is well suited to computational approaches that can learn patterns across large reference databases.15ACS Publications. Decoding Complex Spectra with Machine Learning: Advances in Spectral Unmixing for Chemical Analysis The promise is that algorithms trained on thousands of reference spectra could identify components in unknown mixtures faster and more consistently than human analysts, particularly for the kinds of complex, overlapping datasets that perfume analysis or polymer characterization generate.

The fragrance and pharmaceutical industries are both watching these developments closely. If spectral interpretation can be automated reliably, deformulation timelines shrink from weeks to days or even hours for routine products. The challenge, as with most machine learning applications, is ensuring that the training data is comprehensive enough and that the models generalize well to genuinely novel formulations rather than just recognizing things they’ve already seen.

What Deformulation Cannot Easily Tell You

For all its power, deformulation has real limits that are worth understanding. Knowing the ingredients and their proportions does not automatically reveal the manufacturing process. Two tablets with identical compositions can perform very differently in the body if one was made by wet granulation and the other by direct compression, because the process affects particle bonding, porosity, and dissolution rate. Deformulation can sometimes offer clues about processing (the crystal form of an ingredient might hint at what temperatures it was exposed to, for instance), but manufacturing know-how often remains opaque even after thorough chemical analysis.

Trace components present another challenge. Many instruments struggle to detect compounds that are present at very low concentrations, especially when those trace compounds are chemically similar to the major components and their signals overlap. Fragrances are a good example: a top-tier perfume might contain minute quantities of expensive natural extracts that profoundly affect the perceived scent but barely register in a standard GC-MS analysis. The analyst may detect the major components perfectly and still miss the touches that make the product distinctive.

Biological products, like vaccines, monoclonal antibodies, and cell therapies, push deformulation to its limits. These are not small-molecule mixtures but complex, heterogeneous macromolecules whose activity depends on three-dimensional structure, post-translational modifications, and aggregation state. Characterizing them requires an entirely different scale of analytical effort, and perfect replication is often impossible even with full knowledge of the starting materials, because the manufacturing process itself defines the product in ways that chemistry alone cannot capture.