Carbonyl dihalides: Synthesis and spectroscopic characterization

Article abstract of DOI:10.1039/a604090j

New, or improved, syntheses of phosgene, carbonyl bromide chloride and carbonyl bromide fluoride have been elaborated. The NMR (¹³C, ¹⁹F and ¹⁷O) and electron impact mass spectra were recorded for COX₂ (X = F, C

Full text of DOI:10.1039/a604090j

View Article Online / Journal Homepage / Table of Contents for this issue
Carbonyl dihalides: synthesis and spectroscopic characterization
a
b
,c
Michael J. Parkington, T. Anthony Ryan and Kenneth R. Seddon*
a
School of Chemistry and Molecular Sciences, University of Sussex, Falmer,
Brighton BN1 9QJ, UK
b
ICI Chemicals and Polymers Ltd., The Heath, Runcorn, Cheshire WA7 4QD, UK
School of Chemistry, The Queen’s University of Belfast, Stranmillis Road, Belfast BT9 5AG, UK
c
New, or improved, syntheses of phosgene, carbonyl bromide chloride and carbonyl bromide fluoride have been
1
3
19
17
elaborated. The NMR ( C, F and O) and electron impact mass spectra were recorded for COX (X = F, Cl or
2
Br), COClF, COBrCl and COBrF; trends in the chemical shifts are analysed in terms of σ and π contributions to
the bonding. Attempts to prepare carbonyl diiodide in solution were unsuccessful, and the reasons for this are
analysed.
Whilst completing a monograph describing phosgene and the
related carbonyl halides we became aware of many lacunae in
were carried out in a well ventilated fume cupboard with a face
Ϫ1
1
velocity of > 0.75 m s , and in the presence of at least one
the literature. As part of our investigation into the electronic
other experienced research worker. The vacuum line was con-
structed within the fume cupboard. The atmosphere both inside
and outside the fume cupboard was constantly checked using
structure of the carbonyl halides, COX (X = F, Cl or Br) and
2
2
,3
COXY (X = F, Y = Cl or Br; X = Cl, Y = Br), it was necessary
to develop safe, reproducible laboratory-scale syntheses for
these materials, and these are described herein, along with pre-
8
Dräger tubes and detector tape (Rimon Laboratories Ltd). All
glassware used greaseless taps, and joints were lubricated with
Teflon sleeves. After use, the carbonyl dihalides were destroyed
by passage through a column containing moist activated char-
coal. The fume cupboard was fitted with an alarm system,
which was activated automatically if the extractor mechanism
failed, or manually in the event of an accident. After use, all
equipment was washed with an aqueous solution of sodium
hydroxide before removal from the fume cupboard. Samples for
spectroscopic analysis were of such a volume that even cat-
astrophic release of the entire sample would not raise the local
concentration above the OEL.
1
7
viously unreported O NMR chemical shifts for these simple
materials.
Carbonyl diiodide has never been isolated, and (until very
recently) the only report of its observation was as a possible
(but improbable) component of the essential oil of a Hawaiian
seaweed. In 1995, however, there was a report of the first obser-
vation of COI , formed in the gas phase by the reaction of
2
4
‘
heated’ tetraiodomethane with dioxygen. The evidence for its
existence rests entirely upon the observation of two infrared
Ϫ1
bands, at 1812 [ν (C᎐O)] and at 760 cm [ν (CI)]. The latter
1
4
band occurs, somewhat disturbingly, at a frequency higher than
that observed (at 747 cm ) for the analogous asymmetrical
These toxic materials, after being loaded into the appropriate
spectrometer cells, should be transported to the instrument
laboratory in a suitable closed, padded carrying case. Only the
instrument operator and one other experienced worker should
be in the laboratory when the samples are being studied. The
mass spectrometer pumps should be vented to an exhaust sys-
tem and not to the laboratory.
Ϫ1
ν (CBr) band of the well characterized carbonyl dibromide,
4
5
COBr . We thus attempted to prepare this thermodynamically
2
unstable material, with an unsurprising lack of success.
Experimental
CAUTION: No toxicological data have been recorded for
carbonyl bromide fluoride or carbonyl bromide chloride, and
no occupational exposure limits have been recommended. The
physiological effects of all the carbonyl halides used here were
assumed similar to those of phosgene, but clearly a modern
detailed evaluation is required if they are to be used more
widely. The following safety precautions were adopted on the
assumption that the toxicity of these compounds was similar to
that of phosgene.
Source and purity of carbonyl dihalides
Phosgene, COCl (BDH), carbonyl difluoride (Fluorochem),
2
and carbonyl chloride fluoride (Ozark-Mahoning) were all dis-
tilled in vacuo prior to use, and their purity monitored by gas-
phase infrared spectroscopy. Both phosgene and carbonyl
chloride fluoride revealed no impurity bands; a weak band due
to CO was observed in the infrared spectrum of COF , which
2
2
corresponded to an impurity level of р2%. Carbonyl dibro-
mide was prepared and purified by a method described
9
in the following paper.
Handling carbonyl dihalides
Phosgene is a toxic gas, with a permissible UK occupational
Spectroscopic measurements
Ϫ3
6
exposure limit (OEL) of 0.08 mg m of air (0.02 ppm v/v). In
the event of exposure the victim may experience chest pain,
coughing and rapid breathing associated with pulmonary
œdema, and it may take over 24 h for symptoms to appear.
13
17
The C and O NMR spectra were recorded on a Bruker
WM360 spectrometer operating at 90.55 and 48.82 MHz,
13
17
respectively. The C and O chemical shifts were measured
with respect to external tetramethylsilane and water, respect-
ively. Mass spectra were recorded on a Kratos MS80RF mass
1
There is no antidote to phosgene poisoning, and hence treat-
ment is usually directed to the main symptom, toxic pulmonary
spectrometer by the electron-impact method (ionizing voltage
7
œdema. Hence, all manipulations involving carbonyl dihalides
Ϫ17
7
0 eV, ca. 1.12 × 10
J), and infrared spectra on a Perkin-
Elmer 598 spectrometer. Gas-phase infrared spectra were
recorded using a 10 cm gas cell fitted with CsI windows, a typ-
ical pressure being ca. 0.5 kPa. All spectra were calibrated using
*
E-Mail: k.seddon@qub.ac.uk WWW: http://www.ch.qub.ac.uk/krs/
krs.html.
J. Chem. Soc., Dalton Trans., 1997, Pages 251–256
251

View Article Online
Ϫ1
polystyrene (1601 and 907 cm ) and indene (551.7 and 420.5
Carbonyl diiodide. Four unsuccessful attempts to prepare COI2
by halide-exchange reactions, under extremely mild conditions,
are described.
Ϫ1
cm ). The band assignments for the unsymmetrical carbonyl
dihalides, used here, are formally incorrect. The accepted con-
vention for assigning fundamental modes of the same sym-
metry is to number them sequentially with the highest fre-
quency first and the lowest last. However, the convention
From sodium iodide and carbonyl dibromide. Carbonyl dibro-
mide (0.2 g, 1.1 mmol) was condensed onto sodium iodide (0.13
g, 0.87 mmol) at Ϫ196 ЊC and the reaction mixture allowed to
warm to Ϫ78 ЊC, whence a purple coloration was observed. The
gas-phase infrared spectrum of the reaction mixture at this
1
adopted here, as elsewhere, uses the same assignments for
COXY (C ) as for COX2 (C ), allowing easier comparison
s
2ν
Ϫ1
between the symmetrical and unsymmetrical carbonyl dihalides.
temperature showed a strong band at 2143 cm due to carbon
Ϫ1
monoxide, along with strong bands at 1828, 788 and 747 cm
indicative of carbonyl dibromide. No bands which could have
been attributed to carbonyl diiodide or carbonyl bromide iod-
ide were observed.
Preparations
Carbonyl bromide chloride. Tribromochloromethane (20 g,
7
0 mmol) was placed in a three-necked flask fitted with a
From sodium iodide and phosgene. Phosgene (0.38 g, 3.8
mmol) was condensed onto sodium iodide (0.7 g, 4.6 mmol) at
Ϫ196 ЊC and the reaction mixture allowed to warm to Ϫ78 ЊC,
whence a purple coloration was observed. The gas-phase infra-
red spectrum of the reaction mixture at this temperature
dropping funnel containing concentrated sulfuric acid (22
cm ), and two condensers (one positioned above the other
3
and maintained at 10 and Ϫ78 ЊC respectively). The flask was
gently warmed until the solid had melted, and the sulfuric acid
was added dropwise. The reaction mixture was then heated
under reflux at ca. 130 ЊC for 3 h. The two condensers were
then quickly removed, and replaced with a fractional distill-
ation unit. The products were fractionally distilled and the
fraction boiling in the range 23–40 ЊC was collected. The deep
red impure distillate was then transferred quickly onto the
vacuum line. It was held at Ϫ95 ЊC and continuously evacu-
Ϫ1
showed a strong band at 2143 cm due to carbon monoxide,
Ϫ1
together with strong bands at 1827 and 849 cm due to phos-
gene. No bands which could have been attributed to carbonyl
diiodide or carbonyl chloride iodide were observed.
From tetraethylammonium iodide and phosgene. Phosgene
(0.25 g, 2.5 mmol) was condensed onto tetraethylammonium
iodide (0.8 g, 3 mmol) at Ϫ196 ЊC, and the reaction mixture
allowed to warm to room temperature. No colour change was
observed, and the gas-phase infrared spectrum of the reaction
ated for 1 h to remove SO2 and COCl . To remove the
2
considerable quantities of free dibromine, the product was con-
densed into an ampoule (fitted with a greaseless tap) contain-
ing mercury, and allowed to warm to room temperature. The
ampoule was then closed and removed from the vacuum line,
and vigorously (but carefully) agitated within the fume cup-
board for 5 min. The ampoule was then reconnected to the
vacuum line, and the liquid distilled into a storage bulb. The
straw-coloured liquid was then distilled into an ampoule fitted
with a greaseless tap, and stored at room temperature in the
absence of light. The purity of the product was checked by
Ϫ1
mixture showed only strong bands at 1827 and 849 cm (due to
phosgene). No bands which could have been attributed to car-
bon monoxide, carbonyl diiodide or carbonyl chloride iodide
were observed.
From tetrabutylammonium iodide and phosgene. Tetrabutyl-
ammonium iodide (0.95 g, 2.6 mmol) was dissolved in dichloro-
3
methane (2 cm ). Phosgene (0.28 g, 2.8 mmol) was condensed
onto this solution at Ϫ196 ЊC, and the subsequent reaction mix-
ture allowed to warm to Ϫ78 ЊC, whence a purple coloration
was observed. The gas-phase infrared spectrum of the reaction
gas-phase infrared spectroscopy. Yield (based on CBr Cl)
3
Ϫ1
2
.98 g (30%).
mixture at this temperature showed a strong band at 2143 cm
due to carbon monoxide, together with strong bands at 1827
Ϫ1
and 849 cm due to phosgene.
Carbonyl bromide fluoride. A three-necked flask was fitted
with a dropping funnel containing tribromofluoromethane (50
g, 180 mmol), and two condensers (one positioned above the
other and maintained at 10 and 0 ЊC respectively). Sulfur()
oxide (200 g, 2.5 mol), carefully warmed using a hot-air blower,
was then poured into the flask under an atmosphere of dinitro-
gen. The CBr F was then added dropwise to the SO over 45
Results and Discussion
Carbonyl bromide chloride
Two methods are described in the literature for the synthesis of
3
3
COBrCl. The first, from phosgene and aluminium() brom-
1
0
min at 35–45 ЊC. The evolved gas was passed through tri-
ide, involves the use of high pressure and was therefore not
3
5
chloroethene (300 cm , 30 cm height) under UV light (supplied
adopted. The other method, described by Overend and Evans,
involves the reaction of bromotrichloromethane with concen-
trated sulfuric acid. However the use of CCl Br is more likely to
by a 100 W UV lamp) to remove dibromine, and then through
3
concentrated sulfuric acid (300 cm , 30 cm height) to remove
3
SO . The product was then collected at Ϫ196 ЊC, and rapidly
generate COCl rather than the desired COBrCl, and so a more
2
2
transferred onto the vacuum line, where the small amount of
logical choice of starting material would be CBr Cl.
Although the basis of the reaction is the same as that des-
cribed by Overend and Evans, the method employed here used
3
CO present was removed by trap-to-trap distillation. The col-
2
5
ourless gas was then distilled into a glass bulb (2 l) fitted with a
greaseless tap, and stored at room temperature. The purity of
the final product was checked by gas-phase infrared spec-
CBr Cl: as well as the products shown in equations (1) and (2),
3
troscopy and gas chromatography. Yield (based on CBr F)
3
CBr Cl + H SO → COBrCl + 2HBr + SO
(1)
(2)
3
2
4
3
1
3.59 g (58%).
2
HBr + SO → SO + H O + Br
3
2
2
2
Phosgene. In this novel preparation, hydrogen chloride (4.2 g,
15 mmol) was condensed onto N,NЈ-carbonyldiimidazole (2.5
1
there were also considerable quantities of phosgene in the
product mixture. Conventional trap-to-trap distillation was
ineffective at completely separating the COCl2 from the
g, 15 mmol) at Ϫ196 ЊC. The reaction vessel was warmed to
Ϫ63 ЊC, and connected to a trap held at Ϫ196 ЊC. The vessel
was opened and the excess of hydrogen chloride condensed into
the Ϫ196 ЊC trap. The gas-phase infrared spectrum of the con-
tents of the Ϫ63 ЊC trap showed strong bands at 1827 and 849
COBrCl (COCl , b.p. = 7.48 ЊC; COBrCl, b.p. = 25 ЊC). How-
2
ever, keeping the product mixture at Ϫ95 ЊC, and continuous
evacuation for 1 h, gave a purity of у 95% (determined from
Ϫ1
13
cm due to phosgene, as well as a much weaker band at 2886
the solution C NMR and gas-phase IR spectra of the gaseous
Ϫ1
cm
due to hydrogen chloride. Yield (based on N,NЈ-
product). The gas-phase infrared spectrum of the vapour above
carbonyldiimidazole) 0.74 g (49.8%).
the liquid COBrCl showed the impurities to be COCl (ca. 6%)
2
2
52
J. Chem. Soc., Dalton Trans., 1997, Pages 251–256

View Article Online
and COBr (ca. 3%). The concentration of the more volatile
prepared here showed no sign of disproportionation, even after
2
phosgene is exaggerated using this technique, and shows that
care must be taken when using gas-phase infrared spectroscopy
as an assessment of the purity of a liquid. The exaggerated
concentration of phosgene was also observed in the mass spec-
trum of the COBrCl (since here, again, it is the vapour which is
being analysed). However, in the photoelectron studies
10 months of storage in a glass bulb at room temperature in
normal light. The use of brass needle valves also caused yellow-
brown discoloration of the COBrF, presumably also due to
reaction (4). It is therefore critical that all manipulations of
COBrF use either glass or stainless-steel apparatus, neither of
which was observed to induce disproportionation.
3
described elsewhere, where continuous evacuation is occurring,
The gas-phase infrared spectrum of COBrF showed the fol-
lowing bands: ν , 1868; ν , 721; ν , 398; ν , 1068; ν , 335; and ν ,
no such problem was encountered. The synthesis of COBrCl
was performed twice, giving an average yield of 30%.
1
2
3
4
5
6
Ϫ1
620 cm . The frequencies observed here are in excellent agree-
5
Ϫ1
The synthesis employed by Overend and Evans is very poor-
ment (±1 cm ) with the recent values of Zhao and Fran-
1
5
ly described, and no account is given of the scale or yield of
their reaction. It is, however, apparent from their gas-phase
cisco, except for the value for ν , a weak, poorly resolved
3
band, which agrees with the observation of Patty and
1
6
infrared spectrum that the COBrCl contains COCl (ca. 25%)
Lagemann. Our value for ν , however, exactly corresponds to
1
15 Ϫ1
2
and COBr (ca. 10%). Even allowing for the fact that the gas-
that of Zhao and Francisco, being 6 cm lower than the
16
2
phase infrared spectrum will exaggerate the concentration of
phosgene, their sample of COBrCl is significantly less pure than
that prepared here. The synthesis described here is therefore a
significant improvement, both in yield and purity, to that
earlier value.
A novel route to phosgene
The rapid reaction of N,NЈ-carbonyldiimidazole with carb-
oxylic acids to generate N-acylimidazoles is well docu-
5
described by Overend and Evans.
The gas-phase infrared spectrum of COBrCl showed the fol-
1
7,18
mented.
It was decided, in the hope of synthesizing COCl2,
Ϫ1
lowing bands: ν , 1828; ν , 517; ν , 806; ν , 372; and ν , 547 cm
;
1
2
4
5
6
to perform the analogous reaction, substituting hydrogen
chloride for the carboxylic acid. Although this method
ν was too weak to observe, even at 46.6 kPa. The frequencies
3
Ϫ1
observed here are in excellent agreement (±2 cm ) with the
only values in the literature.
5
affords a new and very straightforward synthesis of COCl2,
the low yield of the reaction (50%) together with a purity of
only ca. 94% (ca. 6% HCl impurity determined from the gas-
phase infrared spectrum) indicates that it is unlikely to pro-
vide an alternative source of phosgene. Moreover, phosgene is
available very cheaply from commercial suppliers (synthesis
from carbon monoxide and dichlorine affords a yield of ca.
Carbonyl bromide fluoride
Three syntheses of COBrF are reported in the literature. Two
1
1
require the use of special equipment, since they use difluorine
1
1,12
or bromine() fluoride
as fluorinating agents. The other
method, which was used here, equation (3), was similar to that
1
00%).
Even the analogous reaction with hydrogen bromide, which
CBr F + SO → COBrF + Br + SO
(3)
3
3
2
2
in theory should yield COBr , is unlikely to replace the con-
2
9
ventional synthesis described elsewhere. However, N,NЈ-
1
3,14
described by Siegemund
and could be performed using con-
carbonyldiimidazole may be potentially useful in the synthesis
of unsymmetrical carbonyl dihalides, COXY, by successive
reactions with stoichiometric quantities of HX and HY, and
may offer a route to COXI.
ventional glassware. However, the purification of the COBrF
employed here was significantly different to that used by Siege-
1
3,14
mund,
who made no attempt to remove the SO formed in
2
the reaction, and consequently reported that 29.5% of his
product was SO . Here, the majority of SO was removed by
2
2
Attempted syntheses of carbonyl diiodide
passing the product mixture through concentrated sulfuric acid,
which achieved a purity of ca. 93% (as determined by gas-phase
Of the four possible carbonyl iodides, only COFI has been syn-
infrared spectroscopy). The impurities present were SO (ca.
thesized [by the reaction of iodine() fluoride and carbon
2
1
1,19
4
%) and CO (ca. 3%), and this was confirmed by gas chroma-
monoxide in an autoclave].
is reported to decompose at Ϫ20 ЊC.
It is a low-boiling liquid, which
2
20
tography. The results from these independent gas chroma-
tography studies help to justify the use of infrared spectroscopy
as a reliable means of assessing purity. As a result of the more
extensive purification procedure employed here, a lower yield of
a purer product was obtained, viz. 58%, compared with 64%
Limu kohu (Asparagopsis taxiformis) is an edible red seaweed
which is highly appreciated, for both its taste and smell, in
the American state of Hawaii. This alga, whose Hawaiian
name means ‘supreme seaweed’, contains a wealth of halo-
genated compounds in its essential oil, many of which are
normally regarded as acutely or chronically toxic. The volatile
oil of A. taxiformis (obtained by condensing onto a Ϫ78 ЊC
finger in vacuo) contains iodine compounds. Separation of
the oil using chromatography (on silica gel at 5 ЊC) and
analysis by gas chromatography–mass spectrometry reveals a
trace of material (< 0.1%) the mass spectrum of which is not
1
3,14
reported in the literature.
The alternative syntheses of
1
1
11,12
COBrF using difluorine and bromine() fluoride
give
reported yields of ca. 30 and ca. 90% respectively. Although
the latter synthesis affords an excellent yield, the reported
method of purification was complex, and unsuccessful in
removing the large amounts of COF2 present. The method
described here thus gives the highest-purity sample of COBrF
ever prepared.
2
1
inconsistent with that expected for carbonyl diiodide.
The purified COBrF was stored in a glass bulb (2 l), since it
was found that storage in a steel cylinder caused a yellow-brown
discoloration of the product after 1 month. The gas-phase
infrared spectrum of this yellow-brown product showed strong
Although CHI , and several other iodinated compounds, were
3
readily identified by both NMR and mass spectral analysis of
the crude oil obtained from A. taxiformis, many of the ori-
ginal iodine-containing compounds were found not to survive
the chromatographic separation. Carbonyl diiodide was there-
fore postulated as an artefact that results from the decom-
bands due to COF2 and COBr , and indicated a complete
2
absence of COBrF. It was therefore apparent that storage in a
steel cylinder was unsuitable and caused disproportionation,
according to equation (4). Disproportionation is also reported
2
1
position of CHI3 during the chromatographic process.
4
Recently, there was another report of the ‘first’ observation
of COI , formed in the gas phase by the reaction of ‘heated’
2
2
COBrF→COF + COBr2
(4)
tetraiodomethane with dioxygen: here, the only evidence
offered was the gas-phase infrared spectrum of the reaction
mixture.
2
to occur if COBrF is stored in a glass bulb at room temperature
1
2
for several weeks. However it should be noted that the sample
The reactions described here, all performed at Ϫ78 ЊC, were
J. Chem. Soc., Dalton Trans., 1997, Pages 251–256
253

View Article Online
even in this reaction, carbon monoxide and iodine are formed,
thus implying that COI is unstable above Ϫ78 ЊC. This conclu-
2
2
7
sion is supported by Staudinger and Anthes, whose attempts
to prepare ethanedioyl diiodide at Ϫ78 ЊC resulted in its
decomposition to carbon monoxide and iodine (presumably via
COI ).
2
It would appear, on the basis of the evidence presented and
discussed here, that COI is unstable in solution, not surprising
2
given that the estimated standard change of free energy associ-
Ϫ1
ated with dissociation (7) is Ϫ68.3 kJ mol , corresponding to a
COI (g)
CO(g) + I (g)
(7)
2
2
—
᭺
11 25
standard dissociation constant, K , of 9.2 × 10 .
The observations that COI is detected in the mass spectrum
2
2
1
of the oil extracted from the alga Asparagosis taxiformis and
in the gas-phase infrared spectrum of the reaction mixture pro-
4
duced by oxidizing CI with O must be regarded as unproven.
4
2
It is unsound scientific practice to base identification of a newly
claimed molecule upon a single spectroscopic technique, and
neither mass spectrometry nor infrared spectroscopy is capable,
alone, of producing definitive evidence of the existence of COI2,
especially under the conditions reported. As in law, proof
should be established beyond reasonable doubt. To quote the
Duke of Venice, ‘To vouch this is no proof, without more wider
and more overt test than these thin habits and poor likelihoods
of modern seeming do prefer against him’ (Shakespeare, Oth-
ello, Act I, Scene III). That is not to say that, under conditions
of low pressure in the gas phase, COI may not exist, simply
2
that convincing evidence is still lacking.
NMR studies
1
3
17
19
The C and O NMR spectra (and F when relevant) were
recorded for all the known carbonyl dihalides, except COFI.
The chemical shifts, coupling constants (where appropriate)
1
3
Fig. 1 The C NMR spectra for (A) COF , (B) COClF and (C)
2
1
7
13
17
COBrF, and the O NMR spectra for (D) COF , (E) COClF and (F)
and solvents are listed in Table 1. The C and O NMR spectra
of the three fluorinated carbonyl dihalides studied are shown in
Fig. 1. The chemical shifts and coupling constants recorded
here are in good agreement with the more restricted set of
2
COBrF
designed to see if the carbonyl iodides could be prepared under
mild conditions. The reactions of sodium iodide with either
COCl or COBr , or of [NBu ]I with COCl (in CH Cl ), pro-
2
8–32
19
values reported in the literature,
the exception being the F
chemical shift for COBrF. The value reported here is δ 84.7,
compared with a value of δ Ϫ83.7 obtained by Krannich and
2
2
4
2
2
2
duced free I , with evolution of carbon monoxide; treatment of
2
32
Sundermeyer, both shifts being measured with respect to
[
NEt ]I with phosgene produced no observable reaction. Thus,
in none of the reactions was there any evidence for the forma-
4
CCl F. It is clear that Krannich and Sundermeyer were not
3
following the IUPAC recommendations of sign convention,
and moreover they did not state which sign convention they
were using. A similar problem exists for some of the data for
tion of COI , or the more probable COClI and COBrI, despite
2
the temperature of reaction being Ϫ78 ЊC. There was no reac-
tion between tetraethylammonium iodide and phosgene, due to
the insolubility of the former in the latter. Earlier attempts to
prepare COI from phosgene and either hydrogen iodide or
aluminium() iodide had also been unsuccessful, yielding
only CO and iodine. The apparent instability of COI is sup-
ported by thermodynamic calculations, viz. equations (5) and
6), calculated from the JANAF Thermochemical Tables
COF amongst extant literature.
2
2
2
As is apparent from Fig. 2, there is no simple correlation
2
13
2
3
between C chemical shift and Pauling electronegativity (con-
2
8
,33
firming Gombler) or π(CO) bond order.† Clearly, the σ and
π effects are working against each other, and to different
extents, in each compound, the result being to nullify any cor-
relation that might exist between the chemical shift and either σ
2
2
4
(
2
8
or π effects. As noted by Gombler, the pattern revealed in Fig.
(a) is indicative of the presence of strong CF π bonding, and
2
NaI(s) + COCl (g) → 2NaCl(s) + CO(g) + I (g);
2
2
2
Ϫ1
∆
G298 = Ϫ107.4 kJ mol
(5)
this is supported by the reflection of this pattern in Fig. 2(b). The
1
J(CF) coupling constant should increase as the amount of
C᎐F π bonding increases: COF has the lowest π(C᎐F) bond
2
NaI(s) + COCl (g) → 2NaCl(s) + COI (g);
2
2
2
order (0.172 versus 0.199 and 0.192 for COClF and COBrF
Ϫ1
∆
G298 = Ϫ39.1 kJ mol
(6)
1
respectively), and hence it has the smallest J(CF) coupling con-
stant (310.3 Hz). The difference in coupling constants between
2
5
and estimates of the thermodynamic stability of COI2.
Reaction (5) is so facile at room temperature that it has been
used in the quantitative determination of phosgene.
2
6,27
The
†
The π(CO) bond orders, calculated on a simple overlap basis from ab
driving force for these reactions is the formation of NaCl; thus
substitution of tetrabutylammonium iodide for sodium iodide
would be expected to increase their free energy and therefore
initio calculations using an STO-3G basis set within the Gaussian 82
package, are: COF , 0.742; COClF, 0.750; COBrF, 0.751; COCl ,
2
2
0
.780; COBrCl, 0.775; COBr , 0.772. The π(CF) bond orders are:
2
3
3
possibly encourage the formation of COClI or COI . However,
COF , 0.172; COC1F, 0.199; COBrF, 0.192.
2
2
2
54
J. Chem. Soc., Dalton Trans., 1997, Pages 251–256

View Article Online
1
9
17
a
Table 1 Carbon-13, F and O NMR spectroscopic data for the carbonyl halides
1
3
b
19
c
17
d
1
e
2
f
δ( C)
δ( F)
δ( O)
J(CF) /Hz
J(OF) /Hz Ref.
COF2
134.2
Ϫ23.0
Ϫ20.4
309
310.3
28
1
34.1
259.5
37.6i
This work
29
30
30
g,h
+
21.54
308.4
h,j
+
+
23.4
22.5
h,k
COClF
COCl2
140.5
39.9p
141.8
59.7
60.0
366
366.4
28
1
375.1
37.4
This work
28
31
This work
28
This work
32
This work
28
This work
1,m
1
1
42.1n
n
n
42.8
483.6
526.5
401.9
549.2
COBrCl
COBrF
COBr2
125.0
n
1
27.2
h,m
Ϫ83.7
+84.7
1
27.61
404.2
41.7
103.4
1
n
n
06.9
a
b
c
Dissolved in trichlorofluoromethane, and measured at Ϫ50 ЊC, unless otherwise stated. Measured with respect to SiMe . Measured with respect
4
17
d
e
13
f
g
to CCl F. Measured with respect to external water. Measured from the C NMR spectra. Measured from the O NMR spectra. Value
extrapolated to infinite dilution, temperature not quoted. Presumably measured with a reversed (and unstated) sign convention. Measured in the
neat liquid. Measured in diethyl ether. Measured in fluorobenzene. Measured at room temperature. Measured in tetrachloromethane. Meas-
ured in dichloromethane.
3
h
i
j
k
l
m
n
Fig. 3 Oxygen-17 chemical shifts of COXY as a function of (a) the sum
of the Pauling electronegativities of X and Y, and (b) the π(CO) bond
order
Fig. 2 Carbon-13 chemical shifts of COXY as a function of (a) the sum
of the Pauling electronegativities of X and Y, and (b) the π(CO) bond
2
8
order
Examination of Fig. 3 shows there to be no meaningful cor-
17
relation between O chemical shift and π(CO) bond order, yet a
1
7
COC1F (366.4 Hz) and COBrF (404.2 Hz) must be due to elec-
tronegativity effects, since the π(C᎐F) bond orders predict
COClF would have the larger coupling constant.
very good correlation (r = Ϫ0.998, n = 6) exists between the O
chemical shift and the sum of the Pauling electronegativities of
the halogens. This suggests that the inductive effect dominates
J. Chem. Soc., Dalton Trans., 1997, Pages 251–256
255

View Article Online
a
Table 2 The El mass spectrometric data for the carbonyl halides COXY
+
+
+
+
+
+
+
+
+
[
COXY]
[COX]
[COY]
[CO]
[XY]
[CX]
[CY]
X
Y
b
COF2
73
22
100
100
100
57
92
100
—
—
—
100
100
13
30
24
4
7
8
5
17
12
—
—
—
COCl2
23
37
46
86
16
90
34
—
—
COBr2
COClF
COBrCl
COBrF
64
14
37
18
10
12
23
4
a
b
The relative % intensities are listed, normalized to the strongest peak in each spectrum. Data taken from ref. 35.
the determination of terminal oxygen chemical shift (as appears
8 K. Leichnitz, Dräger Detector Tube Handbook, Drägerwerk, Lubeck,
1979.
9 M. P. Parkington, T. A. Ryan and K. R. Seddon, J. Chem. Soc.,
Dalton Trans., following paper.
1
9
also to be the case with the restricted data set from the
chemical shifts), whereas the chemical shift of the central
F
2
8
carbon atom is determined by a balance of opposing σ and π
1
1
0 A. von Bartal, Z. Anorg. Allg. Chem., 1907, 55, 152.
1 W. Rüdorff, in FIAT Review of German Science (1939–1946):
Inorganic Chemistry, Part 1, ed. W. Klemm, Office of Military
Government for Germany Field Information Agencies Technical
2
effects. The J(OF) coupling constants listed in Table 1 are so
similar that it would be redundant to attempt to account for
their differences.
(FIAT), Weisbaden, 1948, p. 239.
1
2 W. Kwasnik, in Handbook of Preparative Inorganic Chemistry, 2nd
edn., ed. G. Brauer, Academic Press, New York, 1963, vol. 1, p. 210.
3 G. Siegemund, Angew. Chem., Int. Ed. Engl., 1973, 12, 918.
Mass spectrometric studies
3
4–36
1
The mass spectrum of COF is well documented,
and was
35,37
2
14 G. Siegemund, Ger. Pat., 2 261 108, 1974.
15 Y. Zhao and J. S. Francisco, Mol. Phys., 1992, 77, 1187.
therefore not recorded. The mass spectra of COCl ,
2
3
5
32
COClF and COBrF have been reported in the literature
previously, and these results are in reasonable agreement with
1
1
6 R. R. Patty and R. T. Lageman, Spectrochim. Acta, 1959, 15, 60.
7 R. Paul and G. W. Anderson, J. Am. Chem. Soc., 1960, 82, 4596.
those obtained here. The mass spectra of COBr and COBrCl
18 R. Paul and G. W. Anderson, J. Org. Chem., 1962, 27, 2094.
2
are previously unreported. As is apparent from Table 2, with the
19 W. Kwasnik, unpublished work, cited in ref. 11.
2
2
2
0 W. Kwasnik, in Handbook of Preparative Inorganic Chemistry, 2nd
edn., ed. G. Brauer, Academic Press, New York, 1963, vol. 1, p. 211.
1 B. J. Burreson, R. E. Moore and P. P. Roller, J. Agric. Food Chem.,
exception of COBr , all the carbonyl dihalides give well defined
2
molecular ions and show similar fragmentation patterns. The
absence of a molecular ion for COBr is not too surprising,
2
1
976, 24, 856.
since it is the least stable of the isolated carbonyl dihalides, and
2 A. Besson, Compt. Rend., 1896, 122, 140.
readily dissociates to CO and Br , with a dissociation constant
23 V. Tishchenko, J. Russ. Phys.-Chem. Soc., 1887, 19, 479.
2
3
8
of 12. There was no evidence of any associated species in any
24 M. W. Chase jun., C. A. Davies, J. R. Downey jun., D. J. Frurip,
R. A. McDonald and A. N. Syverud, JANAF Thermochemical
Tables, 3rd Edition, J. Phys. Chem. Ref. Data, 1985, 14 (Suppl. 1), 1.
of the mass spectra, this being particularly relevant to COCl ,
2
1
the dimer and trimer of which are well known.
2
2
5 A. L. Horvath, unpublished work, 1985, cited in ref. 1.
6 M. Sartori, The War Gases: Chemistry and Analysis, 2nd edn.,
J. & A. Churchill, London, 1939.
Acknowledgements
2
7 H. Staudinger and E. Anthes, Ber. Dtsch. Chem. Ges., 1913, 46,
1426.
We are grateful to the EPSRC and ICI plc for the award of
a CASE studentship (to M. J. P.), to the EPSRC and Royal
Academy of Engineering for the award of a Clean Technology
Fellowship (to K. R. S.), and to Drs. A. K. Abdul-Sada and
A. G. Avent for spectroscopic assistance.
28 W. Gombler, Spectrochim. Acta, Part A, 1981, 37, 57.
2
3
9 A. J. Downs, J. Chem. Soc., 1962, 4361.
0 M.-K. Lo, V. W. Weiss and W. H. Flygare, J. Chem. Phys., 1966, 45,
2
439.
3
1 N. Walker, W. B. Fox, R. A. D. Marco and W. B. Moniz, J. Magn.
Reson., 1979, 34, 295.
3
2 H.-J. Krannich and W. Sundermeyer, Z. Anorg. Allg. Chem., 1979,
References
4
50, 45.
1
T. A. Ryan, C. Ryan, E. A. Seddon and K. R. Seddon, Phosgene and
Related Carbonyl Halides, Elsevier, Amsterdam, 1996.
33 M. J. Parkington, D.Phil., Thesis, University of Sussex, Brighton,
1987.
2
M. J. Parkington, E. A. Seddon, K. R. Seddon, J. G. Stamper,
A. M. Sterne and T. A. Ryan, J. Electron Spectrosc. Relat. Phenom.,
34 J. L. Huston and M. H. Studier, J. Fluorine Chem., 1979, 13, 235.
35 Z. Karpas and F. S. Klein, Int. J. Mass Spectrom. Ion Phys., 1976,
22, 189.
36 J. C. J. Thynne and K. A. G. Macneil, Int. J. Mass Spectrom. Ion
Phys., 1970, 5, 95.
37 M. A. Henderson, Y. Zhou, K. G. Lloyd and J. M. White, J. Phys.
Chem., 1989, 93, 7191.
38 B. K. Dunning and H. O. Pritchard, J. Chem. Thermodyn., 1972, 4,
213.
1
986, 37, 395.
3
4
P. Sherwood, E. A. Seddon, M. F. Guest, M. J. Parkington, T. A.
Ryan and K. R. Seddon, J. Chem. Soc., Dalton Trans., 1995, 2359.
I. Barnes, K. H. Becker and J. Starcke, Chem. Phys. Lett., 1995, 246,
5
94.
5
6
J. Overend and J. C. Evans, Trans. Faraday Soc., 1959, 55, 1817.
Occupational Exposure Limits 1995, Guidance Note EH40, Health
and Safety Executive, 1995.
7
W. F. Diller, Toxicol. Ind. Health, 1985, 1, 93.
Received 11th June 1996; Paper 6/04090J
2
56
J. Chem. Soc., Dalton Trans., 1997, Pages 251–256