Chemical Weapons Production and Storage
There are thousands of toxic chemicals that could be used in chemical weapons. Depending on the type of agent to be produced, there can be technical hurdles that must be overcome. "Classic" agents can be manufactured using existing chemical infrastructure, and most have legitimate commercial uses. Likewise, vesicants are not technologically complicated. The production of the nerve agents, however, requires significantly more sophisticated chemical processing. Some production processes require strict temperature control. Containment of toxic substances and gases can pose problems. Depending on the immediacy of use, purity of product can add a difficult dimension to production. In some cases, special equipment or handling is required to prevent corrosion of equipment and/or rapid deterioration of the product.
These hurdles can be overcome. If sufficient purity cannot be attained, an agent can be manufactured and used immediately. This presupposes the capability to manufacture a sufficient quantity in the time allotted. If special, corrosive-resistant equipment cannot be obtained, corroded equipment can be replaced when necessary or only a limited amount can be produced. If nerve agent production is technologically infeasible for a proliferant, a simpler agent (vesicant or classic agent) can be produced. Alternatives can entail increased costs, increased munition requirements, or reduced CW capability.
Classic and Vesicant Agents
Some of the simpler classic chemical agents can be manufactured using existing chemical infrastructure. For example, phosgene is manufactured internally within chemical plants throughout the world for use as a chlorinating agent. Chlorination is the most common of chemical intermediate reactions in the chemical process industry. A reasonable size phosgene facility could be purchased with an investment of $10-$14 million. Similarly, hydrogen cyanide is currently manufactured worldwide as an intermediate in the manufacture of acrylic polymers and could be diverted for other uses or separately manufactured with about the same investment. In either instance the technologies are simple, well known, and require no specialized equipment. These CW agents require high munitions expenditures and are easily defeated by a gas mask, so that use would most likely be against unprotected populations and/or poorly equipped combatants.
Almost all proliferant states since World War I have manufactured vesicants, principally sulfur mustard, bis (2-chloroethyl) sulfide. There are several routes to this compound, none of which require sophisticated technology and/or special materials. The earlier producers favored the Levinstein Process, which consists of bubbling dry ethylene through sulfur monochloride, allowing the mixture to settle and (usually) distilling the remaining material. More recent production has involved chlorination of thiodiglycol, a relatively common material with a dual use as an ingredient in some inks. This method does not result in the solid byproducts of the Levinstein Process and can be more easily distilled. Drums of thiodiglycol, produced in the United States and illegally diverted from their intended recipients, were found by international inspectors after the Gulf War at Iraqi CW production sites. The principal problem experienced by initial manufacturers of sulfur mustard has been the insidious nature of this material. Virtually all those producing mustard have experienced a large number of industrial accidents resulting in casualties from mustard burns. Nitrogen mustards have been synthesized only in pilot plant quantities, but did not require any unusual processes or materials. Lewisite was produced by both the United States and the Soviet Union during World War II. The plants were quite small and unsophisticated by today's standards. Lewisite is an arsenical and as such would require unusually large amounts of arsenates in its production.
Production of the nerve agents requires significantly more sophisticated chemical processing. In a majority of these materials, there are corrosive chemicals in the process that require specialized corrosion-resistant construction materials. With the exception of GA (tabun), manufactured by the Germans in World War II and the Iraqis during the Iran-Iraq war, G-agent production involves both chlorination and fluorination steps. Both of these steps require special and expensive construction materials. Reactors, degassers, distillation columns, and ancillary equipment made of high nickel alloys or precious metals are needed to contain the corrosive products and by products. Only the last step of the process involves the highly toxic material, so that special air handling equipment would be needed for only a small portion of the facility.
There are many process routes for producing the G- and V-agents; the majority involve the synthesis of methylphosphonic dichloride (DC) at some stage. The United States designed and built plants for four different processes for producing DC. Two were used in the stockpile production of GB (sarin), a third represented an upgrade of the stockpile production process to minimize waste, and the fourth represented a newer method used in producing material for binary weapons. The Soviet Union used a still different process to make DC and Iraq one similar to the last U.S. process. DC is a relatively easy material to store and to ship and need not be produced at the same site as the final product. It is very stable and has been stored for over 30 years with little deterioration. The size of the facility required to produce DC in militarily significant quantities ranges from very large down to room sized. A facility to produce DC with ancillary support would cost approximately $25 million not including pollution and environmental controls and waste treatment. Modern waste treatment and pollution abatement to U.S. standards would more than double the cost, although it is doubtful that a proliferant would build to these standards. The various DC production processes require some special corrosion-resistant equipment, generally glass-lined reactors and storage tanks, although not the ultra-expensive equipment required for later stages. DC has limited commercial use.
In the actual production of G-agents, the partially fluorinated DC (now a transient mixture called Di-Di) is reacted with an alcohol or alcohols and the product degassed and usually distilled. As noted previously, this is the toxic step of the reaction which requires air handling and filtering and also part of the highly corrosive portion that requires high nickel alloy (such as Hastelloy C) equipment and piping or precious metals (such as silver). The U.S. stockpile of GB was produced in this fashion and the former Soviet Union stockpiles of GB and GD (soman) by a variation of that process.
The Iraqis used a somewhat over-fluorinated DC and mixed alcohols to produce a GB/GF mixture which was inherently unstable.
Most of the alcohols involved in producing G-agents have large-scale commercial use. An exception is the alcohol for producing GD, pinacolyl alcohol, which has very limited pharmaceutical use.
Two principal general methods have been employed for V-agent production. The process used in the United States was called the Transester Process. It entails a rather difficult step in which phosphorus trichloride is methylated to produce methyl phosphonous dichloride. The material is reacted in turn with ethanol to form a diester and this material then transesterified to produce the immediate precursor of VX. The product is reacted with sulfur to form V-agent. This process has the advantage of being straightforward and producing high quality product. Conversely, it has the disadvantage of some difficult chemical engineering steps. The V-agent formed exclusively in the United States was VX.
The former Soviet Union, the only other known producer of significant quantities of V-agent, did not produce VX per se, but rather a structurally different variant with the same molecular weight. The Soviets designed their process to make maximum use of production capability already available. The DC of the G-agent process was used in an Amiton process conducted in solution. The technique has the advantage of producing a single intermediate (DC). Disadvantages include the need to recover a highly toxic material from solution and to handle large quantities of contaminated solvent.
In general, the V-agents are not easily distilled, and it is unlikely that a final purification process can be developed.
Incapacitating agent production is similar in many ways to the manufacture of pharmaceuticals, since these compounds are normally variations or derivatives of compounds used or postulated for use as pharmaceuticals. Since most pharmaceuticals are produced in relatively small quantities, production would entail a scale-up to an unusual process size for the type of reactions entailed. Moreover, virtually all candidate incapacitating agents are solids at room temperature and would require drying and grinding to an inhalable particulate. Given the tendency of many compounds to acquire a static charge and agglomerate, the grinding is a nontrivial manufacturing problem. The problems associated with manufacture (and use) of solid lethal agents (such as carbamates) are analogous to those experienced with incapacitants.
As a consequence of the diversity and complexity involved, it is difficult to provide any generic insights to toxin production. The only toxin to exist naturally in large quantities is ricin. It is a byproduct of castor oil production. Production of ricin is a physical separation. There are weak parallels with plutonium extraction or uranium isotope enrichment in nuclear processing. Toxin separation is much easier, less expensive, and requires smaller equipment. These advantages might make a toxin attractive to a poor, proliferating country. Most other toxins must be laboriously extracted in small quantities from the organism that secretes them. While synthetic toxins are possible, they are complex molecules, the synthesis of which in any significant amount would be difficult. Biotechnology may enhance the ability to produce toxins that were previously difficult to obtain in significant quantity.
Production of chemical agents in the past has anticipated their long-term storage since (in the instance of United States at least) they were viewed as deterrent weapons and by policy would not have been employed except in response to aggressor use. This also meant that the agents and/or their weapons of employment might be stored for extensive periods of time. The life span of chemical weapons was first expected to be a decade. The requirement was later increased to 20 years when it became clear that munitions were likely to be stored at least that long.
Chemical agents can either be stored in bulk quantities or loaded into munitions. With the nerve agents in particular, the quality of the initial material must be excellent and they must be stored under inert conditions with the absolute exclusion of oxygen and moisture. Generally an overlay of dry helium was employed to leak check munitions. A small amount of stabilizer (2-4 percent) was also used to extend agent life span. The United States stored agent in both bulk containers and in munitions. In the latter instance, the munitions were normally stored in revetted bunkers. This was particularly true when explosives and propellants were uploaded in the munitions. Storage of agents in explosive, uploaded munitions has both advantages and disadvantages. The principal advantage is speed of use when the munition is needed. There is no labor-intensive or time-consuming uploading process, and most munitions can be handled and shipped as if they were conventional munitions. The principal disadvantage is that explosives and propellants become part of the "system," and their storage and deterioration may complicate the handling of the chemical weapons. An illustrative case is seen in the 115-mm M55 rockets where burster, fuse, and rocket propellant cannot be easily and/or safely separated from the agent warhead before demilitarization. As a consequence, demilitarization is far more complicated and costly than it would be otherwise.
Agents stored in bulk in the United States are now stored entirely in large cylindrical "ton" containers similar to those used to store and ship many commercial chemicals. The procedure for the former Soviet Union's stockpile appears to have been to upload their stocks of nerve agent into munitions when produced, but to store them without the bursters or fuses. These munitions were then themselves stored in more conventional warehouse-like structures. Conversely, the older stocks of vesicants (i.e., mustard, lewisite and mustard-lewisite mixtures) are stored in bulk, apparently intended to be filled in munitions a short time before use. Bulk storage of the vesicants by the Russians is in large railroad-car-size tanks again located in warehouse-like structures.
When the Iraqis produced chemical munitions they appeared to adhere to a "make and use" regimen. Judging by the information Iraq gave the United Nations, later verified by on-site inspections, Iraq had poor product quality for their nerve agents. This low quality was likely due to a lack of purification. They had to get the agent to the front promptly or have it degrade in the munition. This problem would have been less severe in their mustard rounds because of less aggressive impurities. The problem of degradation inhibited their ability to deploy and employ nerve weapons but probably did not have a great effect on their use of mustard. Using their weapons soon after production probably worked well in the Iran-Iraq War, where the skies over Iraq were controlled by the Iraqis. Unfortunately for the Iraqis, loss of air control in the Gulf meant the weapons could never reach the front. The chemical munitions found in Iraq after the Gulf War contained badly deteriorated agents and a significant proportion was visibly leaking.
Binary munitions were once intended by the United States as a means of retaining a retaliatory capability without the necessity of an agent stockpile. The relatively nontoxic intermediates could be stored separately and not placed in proximity to one another until just before use. This requires some human engineering to ensure the munitions designs permit simple, rapid mating of the ingredient containers and production of the lethal agent en route to the target. The binary system was envisioned almost exclusively for application to the standard nerve agents. Although at least three types of binary munitions were planned, only one (155-mm artillery shell) was in production when the United States ended CW production. The Russians claim to have considered binary munitions but not produced any. The Iraqis had a small number of bastardized binary munitions in which some unfortunate individual was to pour one ingredient into the other from a Jerry can prior to use.
Release of agent by enemy action during shipment is a disadvantage of unitary chemical munitions. The sinking of the U.S. cargo ship John Harvey in the harbor at Bari, Italy, during World War II and the subsequent unwitting release of a large quantity of mustard gas is a case in point. Mustard escaped from damaged munitions contaminating those escaping the sinking ship and civilians on shore.