Phosgene discovery and uses

Phosgene was first synthesised by John Davy1 by exposing equal volumes of carbon monoxide and chlorine to sunshine for 15 min. He noted that the mixture contracted to half the original volume and the chlorine colour disappeared to form a colourless gas. He named it phosgene, from the Greek φως, phos meaning light, and γινομαι, gene meaning to produce,1 born of light. Its odour was recklessly described thus:

Thrown into the atmosphere, it did not fume. Its odour was different from that of chlorine, something like that which one might imagine would result from the smell of chlorine combined with that of ammonia, yet more intolerable and suffocating than chlorine itself, and affecting the eyes in a peculiar manner, producing a rapid flow of tears and occasioning painful sensations.

The odour has been variously described as of green corn or musty hay,2 or as new mown hay.

Phosgene is a highly reactive compound with the formula COCl₂. It is 3.5 times denser than air, with a boiling point of 7.4°C and a critical temperature of 182°C.3 It exists as a vapour at normal temperature and pressure.

Kirby Jackson4 described four chief methods for producing phosgene:

1. The photochemical combination of carbon monoxide and chlorine.

2. The oxidation of chlorinated hydrocarbons with chromic acid.

3. The interaction of sulfur trioxide or oleum with chlorinated hydrocarbons.

4. The combination of carbon monoxide and chlorine in the presence of a solid catalyst.

It is also produced in the thermal decomposition of certain chlorinated hydrocarbons.5 6

Phosgene is widely used commercially in the production of many chemical compounds. It reacts with a multitude of nitrogen, oxygen, sulfur and carbon centres, as well as with a variety of other inorganic compounds, by acylation, chlorination, decarboxylation and dehydration. It is an important chemical intermediate in many manufacturing processes.7 The majority of phosgene is used in the production of isocyanates for the synthesis of polyurethanes; other uses are in the production of polycarbonates, and chloroformates used to make pharmaceuticals and pesticides.8

Phosgene in World War 1

Gas attacks in World War 1 began on 22 April 1915 with the release of chlorine from vast arrays of cylinders, such as those in Figure 1, by the Germans. Phosgene was used to devastating effect during World War 1. The first recorded use was at Ypres on 19 December 1915 in combination with chlorine. Allied intelligence was such that respirators had been developed containing cotton impregnated with sodium phenolate, and later sodium phenolate in combination with hexamethylenetetramine.2 However the troops at Ypres on that occasion were inadequately protected. The Allies had adopted gas warfare in response to repeated German gas attacks by the end of September 1915, and this included the development of personal protective equipment (PPE) and tactics, techniques and procedures. Thus the gas attacks became less effective as time went on.2 9 Because of its relatively high boiling point of 7.4°C, phosgene could not be used alone in cylinders and was usually combined with chlorine in a ratio of 1:3 or 1:1. It could be used as a sole agent in large projected shells, sometimes adsorbed into pumice.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

A typical cylinder release gas attack. It was possible to use 3000 cylinders across a front of 3000 yards.9 https://commons.wikimedia.org/wiki/File:Poison_gas_attack.jpg.

The Germans had an established industrial use for phosgene in the manufacture of dyes.2 Chlorine had established peacetime uses but carbon monoxide had not. It became necessary for large quantities of carbon monoxide to be produced in order to manufacture phosgene. The Germans and Americans favoured reacting carbon dioxide with carbon, usually charcoal, but the French and the British used incomplete combustion of coke. All sides combined chlorine and carbon monoxide passing over a carbon catalyst and then dried before storage.2

Accurate meteorology was vital for the deployment of phosgene. Aside from the obvious wind direction and chance of change, the presence of water in the atmosphere would greatly diminish the effectiveness of phosgene. Phosgene reacts with water to produce carbon dioxide and hydrochloric acid:

This reaction was exploited in early PPE by keeping it moist, in addition to the countermeasures discussed above. The last gas cloud attack occurred on 8 August 1916 and attention was focused on firing shells, including ones containing phosgene.9 In April 1917 the British introduced the projector which could fire a drum of pressurised liquid into enemy territory where it would rupture and leak large quantities of vapour.9

Vedder10 describes the symptoms of phosgene poisoning as quite different from other agents. Even at high concentrations the inhalation does not cause irritation of the upper airways. Immediately following exposure, the victim is largely asymptomatic. Some soldiers reported a change in the taste of cigarettes to an unpleasant flavour, often reminiscent of rotten eggs.11 12 Later, and often after exertion, victims develop dyspnoea, which is often followed by cyanosis and death.10 Many authors subdivide cases of phosgene-induced pulmonary oedema into two groups10 13 14:

1. Venous engorgement with cyanosis. Congested plum to blue face with tachypnoea and increased respiratory excursion. These patients may be coughing up large volumes of pulmonary oedema. Pulse is typically 100 beats per minute.

2. Grey pallor. These patients are collapsed with ashen, grey lips. Their breathing is rapid and shallow, and pulse may be up to 140 beats per minute. Cough is often absent.

Most phosgene casualties are in group 2, and those in group 1 can progress to group 2. Severe cases in either group exhibit extreme anxiety, reduced level of consciousness or delirium.13 Survivors often have no recollection of their illness, even those who held normal conversations at the Casualty Clearing Station.14

Physiologist John Scott Haldane witnessed first hand the effects of phosgene and delivered a lecture on the subject at the Royal Army Medical College on 8 October 191914 . He described multiple stupefied casualties in respiratory distress, with deeply cyanosed, plum-coloured lips and distended neck veins. One of the casualties died while he was visiting a clearing station, and a postmortem was immediately carried out by a Dr McNee. ‘ The lungs were voluminous and much congested. Albuminous liquid could be squeezed from them in abundance. The bronchi and alveoli were inflamed, and a great deal of emphysema was present ’. He concluded that the cyanosis was due to anoxaemia and the distended neck veins secondary to an increase in pulmonary vascular resistance and distension of the right side of the heart. He noticed the characteristic dose response to phosgene, whereby a large inhalation causes immediate severe signs and symptoms, often resulting in death. At lower doses, exposure to a low concentration for a prolonged period or a relatively high concentration for a short duration, the signs and symptoms are delayed by hours and often precipitated by exertion. He goes on to describe how the administration air with supplemental oxygen titrated to effect can reverse the cyanosis. It did not, however, reverse the tachypnoea, which he surmised was driven by arterial carbon dioxide content. Oxygen was recommended in a dose of 2 L/min, and up to 5 L/min in severe cases. Both Haldane and Galwey15 recognised the toxicity of oxygen in higher concentrations, and Haldane’s16 oxygen apparatus could deliver oxygen with entrained air up to 10 L/min,16 although the logistic burden must have been considerable. Venesection was another treatment that proved effective in cases where venous congestion was prominent. Galwey15 states that the intervention is to reduce right heart strain and is achieved by the removal of 650 mL blood over 20 min. This removal reduces capillary pressure, causing uptake of interstitial fluid from the peripheries, possibly including the lungs and bringing the haematocrit closer to normal. He thought that the rise in haematocrit phosgene causes could be ameliorated by the intravenous injection of saline.

Vedder recognised that it was difficult to determine the pathology of phosgene in human because when weaponised it was commonly mixed with chlorine. However he reported that lung weights were increased, with the right lung typically weighing 1000 g and the left 875 g, and often cover the heart when the chest is opened.10 The lungs were grossly abnormal with dark discolouration and exuding frothy serous fluid when sectioned, and although full of this frothy serous fluid, the upper airways exhibited little or no inflammatory change. Fluid from brown discoloured lung was acidic when tested with litmus paper, and the right heart was generally dilated and often with petechial haemorrhages beneath the endocardium. Many of the organs show venous congestion. On microscopic examination of the alveoli, he noticed that they are full of fluid with desquamated alveolar epithelium, white cells and red cells. Fibrin was observed crossing the alveoli and the capillary beds. He concluded that this was the basis for obstruction of the pulmonary circulation and right heart strain, and therefore a mechanism for generalised venous engorgement. ‘ Thus the important acute changes caused by phosgene poisoning are practically limited to the lungs’.10

Further inflammatory changes are noted with disease progression, with patches of bronchopneumonia. The majority of cases recover if they survive the first 48 hours.10

By the end of World War 1, there were 180 983 British ‘ gas casualties ’, including 6062 deaths. These official figures are a gross underestimate; records begin in 1916 and they do not include the missing or captured.17 These are victims of chlorine, phosgene and mustard. Of the three, total phosgene production was the smallest, yet it was responsible for more devastation, completely out of proportion to the other agents.

Early evaluations of the pathophysiology of phosgene

It has long been recognised that phosgene acts primarily at the alveolar level. Hill18 showed in a series of elegant experiments that the poisoned lung, homogenised, extracted with Ringer’s solution and injected into a healthy animal, caused no adverse effect. Likewise the oedema fluid of poisoned animals injected into healthy subjects and plasma through which phosgene has been bubbled produced no adverse effects. He went a stage further and showed that by isolating a cat lung in vivo using a bronchial blocker, and then allowing the anaesthetised animal to breathe phosgene, only the exposed lung suffered the effects. Cyanosis and cardiovascular variables returned to baseline once the animal breathed air on removal of the blocker. Thus Hill18 demonstrated that phosgene acts only locally with no direct systemic effects. It was well known that the loss of oedema fluid was significant to cause a rise in haematocrit, and Hill18 suggested that this could be corrected by the intravenous infusion of gum saline. Gum saline is a solution of gum acacia of 6%–7% in 0.9% saline and acts as a colloid.19 This was rejected as a therapy on the grounds that it may worsen pulmonary oedema. Hill18 was unable to explain the selective alveolar damage and wrongly assumed that this damage was caused by hydrochloric acid formed by the reaction of phosgene with water.

Barcroft20 summarised his work on the pathophysiology of phosgene poisoning in a lecture given at the Royal Army Medical College in 1919. His work was based on animal experiments conducted at the Royal Engineers’ Experimental Ground at Porton. His chamber experiments were the first large-scale exposures to known concentrations of phosgene vapour. Histological examination showed that increased lung damage occurs with increasing doses, as expected. Macroscopic changes were not uniform, and the microscopic changes included damage to the alveolar epithelium, causing oedema formation and capillary changes including thrombosis. Although hypoxic pulmonary vasoconstriction was first recognised in 1894,21 it was not eponymised as the Euler-Liljestrand mechanism until 1946.22 23 Barcroft20 observed normoxic goats with lungs four times the normal weight. Rather than using hypoxic pulmonary vasoconstriction as the explanation, Barcroft20 first assumed that that oedematous alveoli were not perfused as a result of capillary thrombi. After developing an in vitro model to test the effects of raised pulmonary vascular resistance, he went on to test his hypothesis in vivo. By introducing a needle into the right ventricle of a goat, it was possible to measure and trace the pressure transmitted via the needle (Figure 2).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Schematic produced by Barcroft illustrating the method of recording intracardiac pressure by direct puncture developed in conjunction with Barcroft.20

These goats were then exposed to phosgene, and he observed the typical findings illustrated in Figure 3. Soon after the animals were exposed to phosgene, respiratory rate and right ventricular pressure increase, in line with the anticipated increase in pulmonary vascular resistance. Later in the experiments the right ventricular systolic pressure normalises; however, the diastolic pressures are much lower, reflecting hypovolaemia. Given that the pulse pressures remain similar, postexposure, the increased pulmonary vascular resistance is likely to persist. Vedder10 noted that a man’s lungs could contain up to 2 kg of oedema, yet Barcroft paid little attention to the effects of hypovolaemia in his analysis and made no measurement of hypoxaemia to correlate with minute volume, linking minute volume with right ventricular systolic pressure. He went on to show that in experiments of oil or starch embolism, there is a transient rise in right ventricular pressure, but minute volume would either increase or decrease. To further evaluate the relationship between minute volume and right ventricular pressure, he showed that right ventricular pressure falls during an increase in minute volume in the exercising goat.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

Tracings of respiratory rhythm above the right ventricular pressure pre-exposure and postexposure to phosgene (1.25 and 19.25 hours postexposure). MV, minute volume; P, pulse rate; R, respiratory rate. The y-axis is the right ventricular pressure measured in cmH₂O. D, mean diastolic pressure expressed in mm Hg; S, mean systolic pressure expressed in mm Hg. The bottom scale is time, measured in seconds.

While he recognised that there is a vasomotor component to phosgene lung injury, Barcroft summarised that the pulmonary capillaries are compressed by oedema fluid in the alveoli. On exertion these capillaries dilate under vasomotor influences, creating a large physiological shunt that worsens hypoxaemia. He demonstrated this in a series of experiments using oximetry of blood in phosgene-exposed goats at rest and during exercise, unfortunately not linking the data to right ventricular pressure.

Analysis of early conclusions

The classical descriptions of early and late effects were made during World War 1. Their importance to the management of a large number of casualties persists to this day. Barcroft’s correct observation of increased pulmonary vascular resistance was not underpinned by the hypoxic pulmonary vasoconstriction mechanism to minimise shunt. However the sudden collapse of patients yet to experience severe intoxication is reasonably explained by his hypothesis that it is pulmonary vasodilation known to occur on exertion that is the causative factor.

Many have classified the casualties into groups 1 (venous engorgement with cyanosis) and 2 (grey pallor), and this warrants further discussion. It was known that hypoxia and increased arterial carbon dioxide content will increase respiratory rate, and there was considerable debate over the relative contributions made by hypoxia and carbon dioxide, including the thought that elevated carbon dioxide contributed to the venous engorgement by contributing to vasodilation. Arterial carbon dioxide was not measured, and elevated carbon dioxide seems unlikely when considering the alveolar air equation:

PA_{O 2} = PI_O2 – (PA_CO2 / R)

Alveolar oxygen is calculated by the partial pressure of inspired oxygen; note that it is less than atmospheric after humidification, with the partial pressure of alveolar carbon dioxide subtracted and divided by the respiratory quotient (R). Thus, alveolar oxygen can be increased by reducing the alveolar carbon dioxide, most simply by hyperventilation. In fact the explanation may be even simpler. Group 1 casualties are yet to experience hypovolaemia secondary to fluid losses from extreme pulmonary oedema; their veins are engorged by increased pulmonary vascular resistance and they appear cyanosed from hypoxaemia. Group 2 casualties have significant fluid losses as evidenced by their rapid HR and pallor from vasoconstriction and appear grey because they are cyanosed. It has been stated that a patient’s disease may progress from group 1 to 2, and this is likely due to circulatory fluid loss. For all his astonishing work, Barcroft neglected to note the reduced and often transiently negative diastolic pressures measured in severely intoxicated goats.

Upper airway sparing and Hill’s work elegantly demonstrated that phosgene is decomposed by lungs, limiting its effect to that organ system. The cardiac effects follow from that.