Chlorine’s Unique Fingerprints: The April 7, 2018 Douma Incident Through A Chemistry Lens

The leak of an internal e-mail from an OPCW employee using the pseudonym “Alex” has stirred a controversy over the integrity of the OPCW Fact Finding Mission (FFM) into the April 7, 2018 chemical attack on Douma City.

Alex alleges that the evidence collected by the FFM was deliberately manipulated through omission, exaggeration, and misrepresentation to ultimately reach the unsupported conclusion that chlorine gas was used as a weapon. Alexander Shulgin, the Russian ambassador to the OPCW, cited this e-mail to challenge the conclusions of the FFM’s Douma report. He has accused the body of suppressing contradictory findings and “distorting reality.” The OPCW Director General, Fernando Arias, stated that he, “stands by the independent, professional conclusion [of the investigation].”

Alex’s e-mail presents two major concerns about inferences made from chemical analysis: 

  1. How molecular chlorine Cl2 can be distinguished among other chemical sources of reactive chlorine. 
  2. The trace quantities of chlorinated environmental residues, which were used to infer the presence of chlorine, may not be abnormal.

These criticisms referred to a redacted version of the FFM report that preceded the final FFM report by 9 months. The language of the final FFM report on the Douma incident does not contain language that he was critical of.

This article further considers these technical criticisms through the eye of a chemist, placing them within the context of available literature (open access, to the extent that this is possible), precedent, and common knowledge.

Not All Chlorine Is Alike

In his e-mail, Alex wrote, 

“The only evidence available at this moment is that some samples collected at Locations 2 and 4 were in contact with one or more chemicals that contain a reactive chlorine atom. Such chemicals could include molecular chlorine, phosgene, cyanogen chloride, hydrochloric acid, hydrogen chloride or sodium hypochlorite (the major ingredient of household chlorine-based bleach). Purposely singling out chlorine gas as one of the possibilities is disingenuous.”

There is a deeper analysis to be made, however, that distinguishes electrophilic chlorine (Cl+) from chloride (Cl-) and consequently strongly implicates molecular chlorine (Cl2). A familiar example of this distinction is that table salt (source of Cl-) brings out flavor in food while drinking bleach (source of Cl+) is bad for one’s health. This distinction is determined by assessing the polarization of bonds to chlorine, as determined by the electronegativity of the atom on the other side of the bond (see here). Higher electronegativity means a greater affinity for negative charge.

By example:

  • Hydrogen chloride: Hydrogen has an electronegativity of 2.1 and Chlorine of 3.0. Thus, a greater negative charge is placed on chlorine in hydrogen chloride, meaning it is Cl-. 
  • Hypochlorite ion: chlorine (3.0) is bound to oxygen (3.5). The bond is polarized to place negative charge density on oxygen rather than chlorine. Chlorine is thus Cl+.

The same treatment may be applied to the other chemicals suggested by Alex. These results are summarized in the following table.

There are important connections between these chemicals. Molecular chlorine partially dissolves in water to produce a mixture of hydrochloric and hypochlorous acids.

Although chlorine gas is reactive by itself, it is far more corrosive and aggressive when moisture is present. In a sense, distinguishing between chlorine and a mixture of hypochlorous/hydrochloric acids is meaningless: where chlorine goes, the latter two follow and vice versa. It should thus come as no surprise that elevated levels of chlorinated organic materials in Douma were found on moist wood.

Bleach solutions are made by dissolving chlorine gas into caustic solutions containing either sodium hydroxide or calcium hydroxide. These bases react with incipient hydrochloric acid and hypochlorous acid to form the respective sodium or calcium salts:

The dissolution of chlorine is much more complete in these circumstances. Sodium chloride and hypochlorite do not off-gas significant amounts of chlorine. Household bleach formulations have a pH close to 11. They are strongly basic and contain no free acid. This is important because free acid (i.e., the H+ in hydrochloric acid or hydrogen chloride) is what enables chloride to react with organic molecules such as pinene. So at pH 11, household bleach acts strictly as a source of reactive Cl+. 

Sodium chloride (table salt) is not a reactive chlorine source. This is demonstrated in a Chinese patent where sodium chloride is necessarily reacted with sulfuric acid to generate hydrogen chloride in order to convert pinene to bornyl chloride (source).

Annex 5 of the Douma final report lists a series of chlorinated organic chemicals that were found in environmental samples taken from the attack site. These are explained, with ample precedent, as chemicals that are found when wood and textiles are exposed to chlorine. Bornyl chloride and chlorophenols are formed via classic reactions in organic chemistry: hydrohalogenation of alkenes and electrophilic aromatic substitution. These mechanisms (the molecular acrobatics that explain a transformation) are illustrated in Scheme 1 below. The pieces of the mechanism in boxes are the salient points: one transformation requires Cl- while the other require Cl+.

Table 2 summarizes the chemical origins of the chlorinated residues at Douma city.

In environmental samples that contained bornyl chloride, 2,4,6-trichlorophenol was also present. Phenol and alpha-pinene are co-located in the same samples. This strongly suggests that they are precursors, rather than that the chlorinated compounds arose from an unrelated source. 

Converting both of those to their respective chlorinated products in situ requires both Cl+ and Cl- in the same place. The most self-contained explanation is that they contacted chlorine gas. Otherwise, two separate chemicals would have been required: hydrochloric acid and a source of hypochlorite. The caveat, though, is that this combination (acid and bleach) is well-known to produce chlorine gas anyway (video).

Bornyl chloride and polychlorinated aromatic hydrocarbons (PCAHs) were also found in hardwood samples from Sarmin, as described in the OPCW report on chemical incidents between 16 March and 20 May 2015. Bornyl chloride is not so ubiquitous that it could be found in any environmental sample. In paragraph 3.168 of the same report, solvent extractions from fir wood did not contain bornyl chloride. However, bornyl chloride was found in fir wood exposed to either chlorine gas or hydrogen chloride.

Paragraph 8.11 of the Douma report describes a similar experiment that detected chlorinated phenol after exposing a wood sample to chlorine gas.

This analysis and the studies performed by designated labs together provide both a theoretical and an experimental connection between chlorine gas and the observed environmental residues. 

What Is Known About Chlorination Of Environmental Materials?

The above conclusion is comfortably placed within extensive precedent from civil applications of chlorine gas. Chlorine gas reacts with large biopolymers such as lignin, cellulose, starch, humic acids, and fulvic acids to chew them to pieces. A series of depolymerization, oxidation, rearrangement, and chlorination events happen. This ultimately forms small, stable chlorinated molecules. 

Comprehensive studies of low molecular weight molecules resulting from the chlorination of humic and fulvic acids in surface water have painstakingly identified dozens of substances (source). Among the most important though, are mono-, di-, and trichloroacetic acids—chemicals found at Douma. 

From commercial wood and paper bleaching, we know that chlorine gas is very aggressive toward lignocellulose in wood (source). Under uncontrolled conditions of temperature and pH (e.g., in a chlorine gas attack), it can excessively degrade lignin and cellulose to small chlorinated molecules. More than 500 low molecular weight compounds have been identified in bleaching effluents from kraft pulp mills and the overwhelming majority came from lignin degradation (source). 

Thus, the signature of molecular chlorine and electrophilic chlorine is the observation/presence of many different small chlorinated molecules, regardless of the reaction medium or substrate. And these are seen exactly where we would expect them to be at the site of the Douma incident—on large organic polymers such as wood (lignocellulose); on cotton (cellulose) blankets, pillow cases, and clothing; and on concrete debris (presumably containing sawdust / aggregate). 

There is no single chemical that indicates the presence of chlorine gas. It is this combination of chemicals that is the signature and that disfavors the notion that any other substance, in isolation, was responsible for their presence at Douma. 

Trace Quantities Are Expected Quantities

So how much of these chlorinated organic residues would we expect to find in an apartment that was gassed with chlorine? To the best of the author’s knowledge, this is not a question that has been answered directly in the open scientific literature. However, studies on levels of chlorinated organic materials in bleaching processes (which deliberately expose lignocellulose to high concentrations of chlorine and hypochlorite) may be used to show what could reasonably be expected for chlorophenols and small chlorinated organic chemicals (e.g., chloroform, chloral, etc.). Bornyl chloride concentrations may be estimated from measurements of terpenes in hardwood.

A study on wheat straw pulp bleaching offers some guidance on expected phenolic concentrations. It found the following concentrations of chlorinated aromatic hydrocarbons effluent from the Chlorine-Alkali Extraction-Hypochlorite bleaching sequence, which deliberately exposes wood pulp to high concentrations of chlorine gas and hypochlorite. The results were reported in grams of chlorinated component per ton of pulp (1 g/t = 1 mg/kg = 1 ppm).

Each of these components is a family of compounds with different isomers and different degrees of chlorination (mono, di, tri, tetra, penta…). In total, 26 individual compounds were detected ranging in concentration from 0.1 ppm (3-chloroguaiacol and 4-chloroguaiacol) – 61 ppm (tetrachlorocatechol). The most abundant phenol was 2,6-dichlorophenol (5.8 mg/kg). Notably, an unspecified dichlorophenol was also found in Douma in concrete debris from the street outside Location 2 and on clothing from a victim. 

These are concentrations not far above “trace” levels ( 1 ppm), even with bonafide, purposeful chlorine exposure in liquid media. (N.B.: the tonnage involved still make these compounds possible pollutants of concern!) The concentrations of chlorinated organic residues from chlorine gas exposure at far lower concentrations and reaction times would undoubtedly be lower than these.

Bornyl chloride concentrations are somewhat more easily estimated. Bornyl chloride is not naturally found in wood. Alpha-pinene, its precursor, is found in the oil of coniferous trees. Pine oil is present at a level of 0.20% – 0.50% w/w in plant material (source) (source)

Of this oil, and depending on the exact species and age, alpha-pinene and beta-pinene together make up between 2% – 65% of the mass of this oil (source) (source) (source) (source) (source)

Thus by weight of wood, we expect between (0.20% x 2%) = 0.004% and (0.50% x 65%) = 0.325% w/w alpha-pinene. This is 40 – 3250 parts per million (ppm) in the wood.

Pinene closer to the surface of the wood will be most available for reaction with gaseous chlorine and liquid hypochlorite. Let’s assume this represents 1 – 10% of the available pinene. Now we’re considering 400 parts per billion (ppb) – 325 ppm. 

From the time of cutting the tree, the concentration of volatile terpenes like pinene is continuously decreasing. One study on emissions from hardwood building materials finds that old fir wood contains 1/60th of the concentration of volatile terpenes as fresh wood (source).

If all surface accessible pinene were to react, we would then expect concentrations of bornyl chloride between 7 ppb – 6 ppm. These are undoubtedly trace concentrations.

Chlorinated Organic Materials Are Not Quantitative Measurements Of Chlorine Gas Exposure

These may encapsulate the chemicals that were detected by the FFM, but it’s well-known that where these molecules appear together that others do too. It’s not just likely, but indeed probable that other chlorinated organic chemicals were present but undetected due to peculiarities of the investigation.

According to a study by Reeves and Weishar (Journal of Pulp and Paper Science, 1990, Vol. 16, p. J118-J125 –not available online), fully bleached hardwood pulp contained between 200 ppm – 2300 ppm bound chlorine by weight in oven dried pulp. Fully bleached softwood pulp contains an average of 460 ppm bound chlorine by weight of oven-dried pulp. Bound chlorine represents chlorine as part of organic molecules (source). 

Of this bound chlorine, only ~20% is typically low molecular weight material that’s easily identifiable and measurable by gas chromatography/mass spectrometry (GC/MS) (source), the technique used to characterize chlorinated organic compounds in environmental samples from Douma. The remainder is tied up in larger chunks of lignocellulose that are not volatile enough for GC/MS.

Of that 20% of volatile material, in the context of the FFM investigation into the Douma incident, much of it would be too volatile and non-persistent. Chloroform may be as high as 27% of the volatile material (source). The detected dichloroacetic acid (29%) and trichloroacetic acid (41%) may comprise 70% of the volatile material and they also have finite vapor pressures. 

It must be emphasized that the FFM was only able to reach the sites of interest starting on April 21, 2018—a full 2 weeks after the incident. In this time, much of the chlorinated organic materials would have been lost to evaporation. This is a principle reason that chlorinated organic chemicals are not considered a quantitative metric of chlorine exposure based on current knowledge. In a study of chlorine deposition in soils during the Jack Rabbit tests, the authors wrote, 

“As chlorine oxidizes organic molecules, the smaller—often chlorinated—organic products evaporate, and quantifying a myriad of non-volatile, chlorinated organic compounds is difficult. Thus chlorinated organic compounds do not provide acceptable quantitative chemical markers for Cl2 deposition.”

It would not be surprising if the OPCW’s designated laboratories were unable to identify the full share of chlorinated organic materials due to the technical challenge or loss of evidence from simple evaporation. Therefore scrutinizing quantitative data of chlorinated organic chemicals from environmental wipe samples (itself a semi-quantitative sampling method) is of dubious value.

But these environmental samples still have important qualitative value. They unambiguously show that chlorine gas had deposited on surfaces, reacted with biopolymers, and left its tell-tale signature of myriad small, chlorinated organic molecules.

This is further supported by the detection of elevated levels of inorganic chloride (not commented on in Alex’s e-mail). Inorganic chloride represents half of a chlorine molecule. Due to its low volatility and ease of measurement, it is a more convenient quantitative probe of exposure of materials to molecular chlorine (source). Unmistakably high (800 – 15,000 ppm) chloride levels were found on both gas cylinders as well as on paint chips. Chlorides are not desirable in carbon steels or paints due to its detrimental effects on their performance. Unspecified amounts of chloride were also found on a pillowcase on the bed with a chlorine cylinder. Benign as chloride is, its presence at percent levels on these surfaces is abnormal. And these concentrations are greatly in excess of natural chloride concentrations in plants, soils, or freshwater.

Additional Context And Reality Check

The terrestrial biosphere produces hundreds if not thousands of natural organohalogens and there is a finite background concentration of small, chlorinated organic molecules: chloromethane, chloroform, chlorophenols, and chloroacetic acids—these include some of the chemicals found at Douma. The environmental distribution of trichloroacetic acid has been widely studied (source) (source) (source) (source) (source).

There is an enormous body of research on the natural flux—biogenic sources and removal processes—of these molecules. Background levels of these range from single digit parts per trillion levels (e.g., snow in the Russian tundra) upwards to ca. 100 ppb (e.g., in the needles of scots pine).

It is nonetheless a specious notion that designated labs were simply detecting background levels of natural organohalogens. As mentioned previously, extraction experiments by designated labs on fresh fir wood samples failed to detect these compounds. Prior UN and OPCW FFM investigations into sarin incidents (e.g., Khan Sheikhoun, Saraqeb, E. Ghouta) have never recovered any of these chlorinated materials. And it’s not for lack of capability: environmental samples following the February 4, 2018 combined chlorine/sarin attack on Saraqib showed both sarin residues as well as chlorinated organic molecules.” (source). 

Autorefrigeration observed on the terrace cylinder at Location 2 in Douma is a strong, damning link between it and the incident (source) (video) (source). This shows that the cylinder contents are a liquified gas that flashed upon depressurization (i.e., when its valve was knocked off), and that the liquid’s boiling point is below 0°C (in order to have frozen water). These are physical properties consistent with chlorine gas but not cyanogen chloride or phosgene. The presence of frost, which can persist for some hours as the remaining refrigerated liquid slowly evaporates, also establishes the timing of the incident. 

It may be impossible to rule out cyanogen chloride or phosgene in addition to chlorine from these residues alone; but there is no reason to invoke them in the first place. Their presence would not offer additional clarity. They would neither affect the FFM’s mandate nor alter their conclusion that a chemical was used as a weapon. Both are schedule 3 substances according to the Chemical Weapons Convention with no domestic uses. Hydrogen chloride is likewise a toxic industrial gas with no direct consumer uses except as a nuisance emission from hydrochloric acid. 

In Douma, chlorinated organic residues and high levels of chloride are seen across multiple rooms and levels as well as in the adjacent street. Corrosion of metal fixtures like a ceiling chandelier is seen. Liquid and solids, such as hydrochloric acid or bleach, could not be directly responsible for this. Nobody bleaches their chandelier. Only a gas has the requisite mobility to cause contamination across multiple levels like this. And sometimes there actually is a smoking gun—or fuming, frost-coated gas cylinder.

Summary

  • Chlorinated organic compounds found at Douma require electrophilic chlorine (Cl+), chloride (Cl-), and acid to form. Chlorine is unique in being able to provide all of these.
  • Bornyl chloride is expected to be present at trace concentrations. Its concentration at Douma was nonetheless above background levels, as determined by prior experiments.
  • Chlorophenols, chloroacetic acids, chloral, and other small chlorinated molecules are downstream products of electrophilic chlorine (Cl+). These may be naturally occurring, but their combined presence from biogenic sources across both natural and synthetic surfaces at Locations 2 and 4 would be excessively serendipitous. 
  • Chlorinated organic compounds provide qualitative information: they tell us that Cl+ was present.
  • Inorganic chloride is of quantitative value, and high ppm levels tell us that a very high amount of molecular chlorine was present.
  • Phosgene, cyanogen chloride, hydrogen chloride, hydrochloric acids, and bleaches may be rejected as chlorine sources at Douma on the basis of their chemical reactivity and physical properties (boiling points with respect to autorefrigeration and frosting; states of matter).

Thanks to Clyde Davies, Andrea Sella, Cheryl Rofer, Hamish de Bretton-Gordon, Dan Kaszeta, and others for their help with this article.