Haems as potential reagents for carbon monoxide

Tolantn. Vol. 24. pp. 291.-296. Pergamon Press. 1977, Prrnled in Great Britain.

HAEMS AS POTENTIAL REAGENTS CARBON MONOXIDE

FOR

S. H. MEHDI and A. CORSINI@ Department of Chemistry, McMaster University, Hamilton, Ontario, L8S 4Ml. Canada (Rtwiwd

1 October 1976. Accepted 27 October 1976)

Summary-A study of the potential of a myoglobin model system based on dipiperidyltetraphenylporphinatoiron(I1) for the spectrophotometric determination of carbon monoxide is reported. At room temperature, oxygen does not bind strongly at the iron centre and does not interfere. Nitric oxide, hydrogen sulphide and carbon dioxide interfere and must be removed before the determination if present in significant amounts. The main disadvantage of this novel method is the low sensitivity (about 300 ppm CO). Possible means of modif~ng the system to enhance the sensitivity and to further reduce the number of interfering species are discussed.

The haem group in haemoproteins shows a high degree of selectivity in reacting with gaseous ligands. Stable complexes are formed only with molecular oxygen, nitric oxide and carbon monoxide.’ Haem compounds might thus serve as selective probes for these gases, especially for carbon monoxide. Although such a possibility had been recognized earlier by Adler ef al.,* there is no evidence in the literature that serious attempts have been made in this direction. Because of their high affinity for oxygen, the haemoproteins themselves are unsuitable for the determination of carbon monoxide in oxygen-bearing atmospheres. A better understanding of the binding of gaseous ligands to haems has been made possible through recent studies on a variety of myoglobin mode1 systems. 3-8 Of special interest are the models based on the dipiperidine complex of meso-tetraphenylporphinatoiron(II), Fe(II)TPP(pip)2 ,3*7,8 and on the monopyridylpropanol ester of mesohaem.” Both classes of model not only exhibit a much lower affinity for oxygen than the haemoproteins but bind carbon monoxide more strongly than oxygen. In the present study, an evaluation of the first model as an analytical reagent for carbon monoxide is reported. Although the second model system appeared to have even more desirable characteristics for this purpose, experimental work on it was terminated because of the high expense and difficulty in the synthesis of the compound in high purity. The structure of meso-tetraphenylporphinatoiron(H), Fe(II)TPP, is shown at the top of the next column. The reaction of its dipiperidine complex, Fe(II)TPP(pip)z, with carbon monoxide involves the displacement of one of the axial piperidine ligands: WII)TPP(pip),

+ CO = Fe(II)TPP(pip)(CO)

+ pip (1)

Nitric oxide and oxygen can also bind at the iron centre. As shown later, interference by oxygen is elim291

inated under the experimental conditions but not that of nitric oxide. More serious interferences, however, result from carbon dioxide and hydrogen sulphide which interact not by simple co-ordination at the haem centre but in a more complex manner. This interaction necessitates the prior removal of such gases in samples. In the absence of interfering gases, the haem reagent can be used to determine carbon monoxide in air-samples in the range 0.03% (3oOppm)4/, (v/v) with an error of 25%. Ahhough the sensitivity is inadequate for most analytical purposes, the potential does exist for improvement of the sensitivity of haem systems towards carbon monoxide, as discussed below. EXPERIMENTAL Reagents Solvents. Toluene was distilled over sodium and stored over Linde 4A molecular sieve. The solvent was saturated with dry nitrogen before use. Piperidine was distilled over sodium hydroxide and the fraction boihng at 105-106” was collected and used immediately, or stored over Linde or Davison 4A molecular sieve in tightly capped bottles, which enhanced the stability of the reagent for up to two days. TPP and Fe(lll)TPPCl. Meso-tetraphenylporphine (TPP) was prepared and purified by the method of Adler

292

S. H. MEHDI and A C0~stNt

et u/.“’ Fe(III)TPPCl was prepared by refhrxing ferrous chloride with TPP in dimethylformamide as suggested by Adler et cd.” Solutions of this compound in chloroform showed little or no fluorescence, indicating the absence of any appreciable contamination by free TPP. It is useful to note that both TPP and Fe(III)TPPCl are available commercially from various sources. Stanl~~~~l stock so~ufj~lls. Standard stock solutions of Fe(III)TPPCl (3-5 x 10~5Mj in toluene were prepared. Since some porphyrins are photosensitive. the stock solutions were wrapped in aluminium foil to protect them from light and were found to be stable indefinitely.

Apparatus

Spectrophotometric measurements were made on a Bausch and Lomb Spectronic 600 double-beam instrument. Gas-line assmrnhly. A gas-rack assembly was constructed to incorporate two Z-litre gas-mixing bulbs with side septum ports to sample the gas mixtures for testing of the method. The gas-line also incorporated a conventional open-ended mercury m~ometer and a silicone-oil (Dow Corning 200 Fluid, 100 cs; density 0.93 + 0.01 g/ml at 23“) manometer. The more sensitive oil manometer was used to monitor the addition of small amounts of carbon monoxide to the gas-line (lO.Omm of oil pressure corresponded to 0.682 + 0.004 mmHg). Oil levels were measured with a cathetometer. Test mixtures of known amounts of carbon monoxide in air were prepared by introducing the carbon monoxide into the evacuated line and then diluting slowly with air passed through a tube of “Ascarite” to remove carbon dioxide. until the final pressure in the bulbs was one atmosphere. Samples were withdrawn from the bulbs by means of a IO-ml gas-tight syringe (Hamilton).

f H b 0

2 0.

0.c

I n

”

10

43% .__

Wavelength

(nm)

Fig. 1. Spectra of various iron-TPP species and effect of 0, on the spectrum of Fe(II)TPP(pip),.

RESULTS AND DISCUSSION The following procedure was used to obtain ~libration curves and the equilibrium constant, K, for reaction (1). To a IO-ml volumetric flask were added 1.00 ml of the Fe(III)TPPCI stock solution and I.00 mf of piperidine and the mixture was diluted to the mark with toluene. The solution was set aside for 15 min to allow complete reduction of the Fe(II1) by the piperidine. Next, 3 ml of this solution were drawn into a lo-ml gas-tight syringe. The gas-sample was then drawn into the same syringe (so that it bubbled through the solution) to a total volume (solution plus gas) of 10 ml. The syringe was shaken for 1 min and set aside for 9 min to allow equilibrium between the gas and solution phases to be reached A I-cm cell (fused silica) was then completely filled with the solution and quickly stoppered. and the absorbance measured at 420 nm. The solution remaining in the volumetric Rask served as the reference. This procedure was also used for the determination of carbon monoxide in sample unknowns. For those samples at atmospheric pressure and containing normal concentrations of carbon dioxide (about 0.0330/;,by volume) and similar or lower amounts of hydrogen sulphide, the sampling technique involved simply opening the evaculated 2-litre sampling flask. fitted at the neck with an “Ascarite” scrubber. to the atmosphere. The prior removal of small amounts of the interfering gases does not materially affect the concentration of carbon monoxide. If appreciable amounts of the interfering gases are present, these must be removed inside the Z-litre sampling flask, by exposure to “Ascarite“ or a dilute solution of sodium hydroxide at the bottom of the flask. The amount of carbon monoxide lost to the “Ascarite” or the solution is negligible. A volume correction should be made if the “Ascarite” or solution exceeds 2”,, of the flask volume.

The compounds F~III)TPPCl~ Fe(II)TPP(pip)* and Fe(II)TPP(pip)(CO) exhibit absorption maxima at 417,424 and 422 nm, respectively (Fig. I). The difference spectrum of Fe(II)TPP(pip), and the CO complex shows an absorption maximum at 420 nm and all measurements were made at this wavelength. The formation

of Fe(ll)TPP(pip),

The basic requirements for the binding of gaseous ligands to iron porphyrins are that the iron should be in the ferrous state and a base co-ordinated at one of the axial positions on the iron, The gaseous ligand then binds at the other axial site. Piperidinc conveniently serves to reduce Fe(Iti)TPPCl to Fe(II)TPP and to co-ordinate at the axial positions of the reduced species. Reaction with carbon monoxide then proceeds similarly to reaction (2). Prior reduction of the iron is necessary if other bases such as pyridine are to be used. We found that the reduction of iron done in the usual fashion with sodium dithionite’ or calcium hydride” resulted in substantial losses of the haem either in the solvent interface or though adsorption on the solid material. The rate of reduction of F~III~TPPCl by piperidine depends upon the concentration of the latter. Under the conditions described in this work. the reduction was complete within 15 min. Prior dilution of the

Haems as potential reagents for carbon monoxide

293

The presence of hydrogen sulphide caused very erratic behaviour, however, leading to both high and low recoveries of carbon monoxide (Table 1). Initially, in the presence of carbon monoxide, there is an increase in the absorbance at 420 nm to above the expected value, followed quickly by a decrease to below that value. Hence unless the measurement is taken at a precise time, the reproducibility is poor. Although the manner in which hydrogen sulphide interferes is not known, it may well involve interaction between the iron centre and sulphide. Recently, Chang and DolphiniS reported evidence for mercaptide (RS-) co-ordination at the iron centre when Stability of’ the rruymt towards osidatiorl reduced haem, thiol, carbon monoxide and a strong base are mixed together. Solutions of Fe(II)TPP(pip), are irreversibly oxiIn view of the fact that sulphur dioxide does not dized in air to form the p-oxo-dimer (IV) via the oxygen complex (HI). Kinetic studies7** have led to interfere, the interference by carbon dioxide (Table 1) is surprising. The spectrum of Fe(I1) TPP(pip)~ is the ~stulation of the follo~~~ng mechanism for the oxidation: I (2) 1 Fe(II)TPP(pip), ;t: pip + Fe(II)TPP(pip) 2 Fe(II)TPP(pip)(O,)

piperidine, lower concentrations of it or the use of impure piperidine resulted in a reduced rate of reduction or incomplete reduction. Reproducible results were obtained only if the piperidine was purified and used as described above. A final piperidine concentration of 1.u was found to be most satisfactory in yielding a reasonable rate of reduction and retarding the rate of aerial oxidation (uide injk) without greatly lowering the effective stability of the carbon monoxide complex. The mechanjsm of the reduction is not known, although free radical species are thought to be involved.” 3

I

III

II fast

TPPFe(III)- O-Fe(III)TPP IV The pentaco-ordinate species (II) can also react with carbon monoxide or nitric oxide. Binding of oxygen and oxidation to compound IV presents a potentially serious problem in the determination of carbon monoxide in air. However, although Basolo et ~1.~ have shown that Fe(II)TPF(pip)(O,) is readily formed at -79”. we found that at room temperature the compound is considerably destabilized, particularly in the presence of 1M pi~ridine. The CO complex remains stable under these conditions. Thus, bubbling oxygen through a solution of the reagent does not cause immediate formation of the oxygen complex (III) and its oxidation product (IV) (Fig. 1). A slow oxidation does occur but toluene solutions of compound I that are 1M in piperidine are stable in air for more than 90 min.

The idea1 haem compound for application to the determination of carbon monoxide would be one that binds no gaseous ligands but carbon monoxide at the iron centre. However. no haem compoLlnds, natural or synthetic, are known that bind only carbon monoxide. Most form stable complexes with nitric oxide, carbon monoxide and oxygen, the stabilities decreasing in that order. As shown above, with Fe(II)TPP(pip)z, the interference by oxygen is effectively eliminated at room temperature. Nitric oxide, however, was found to bind strongly and, as expected, caused a low recovery of carbon monoxide (Table 1). Since nitric oxide is rapidly oxidized to nitrogen dioxide, its background level in air is only 0.0002-0.002 ppm14 and is not a serious problem in the determination of carbon monoxide.

unaffected when the reagent solution is saturated with carbon dioxide, suggesting that a complex at the iron centre is not formed. Nevertheless, when carbon monoxide is determine in the presence of carbon dioxide, the absorbance at 420 nm is lower than expected, although no band shifts or new spectral structure are observed. The ultimate source of the interference may involve reaction of carbon dioxide with the piperidine. in a manner similar to the reaction of carbonyl suiphide with piperidine to produce piperidine oxythiocarbamate.’ 6 Evaluation of the equilihrim

constant

The equilibrium constant for reaction expressed as the dimensionless quantity:

(I)

is

If the concentration of carbon monoxide in solution is expressed through Henry’s law as the partial pressure (I’,,), then a conditional constant, K’, can be defined as

~‘2!L3:__[Fe(II)TPP(pip)(CO)] [PiPI

CFe(II)TPP(pip),lP,o

(4)

where S is the solubility of carbon monoxide in mole.l-‘. mmHg_‘. At 23”‘, for toluene as solvent, S = 8.6 x 10m6 mole.l~1.mmHg-i.i7 Under the experimental conditions, the initial concentration of piperidine is large (l&Q, so that [pip] is effectively constant (t&f). Evaluation of K’ requires knowledge of E,, the molar abso~tivity of the CO complex. Both E, and

S.

294 Table 1. Determination

Gas

H.

3.42 3.96 3.64 4.37 3.81 4.95 4.82

CO, H,S NO*

and A.

CORSINI

of carbon monoxide in air-samples in the presence of qther gases

Pgdq,nlmfh

so*

MEHDI

P,, taken, mmHg

P,, found, mmHg

Recovery, %

3.95 4.09 4.06 5.43 3.36 4.06 4.09

3.96 4.05 3.74 4.25 2.51 4.56 3.22

100 99.0 82.1 78.3 74.6 112 78.7

* Experiments carried out in a nitrogen atmosphere. K’ were evaluated by the method of Gans and Irving.” The method is based on a computerized search technique and requires no constraints on the relative concentration of the reactants. For h = lc - Q,, where E, is the molar absorptivity of Fe(II)TPP(pip), , a value of 1.25 x lo5 1. mole- ‘. cm-’ was obtained. The average value of K’ was 0.071 mmHg- ‘, with a standard deviation of 0.003. The value of K was calculated to be 8300. Calibration curves The lowest value of PC0 taken in this study corresponds to a concentration of carbon monoxide (in the solution) that is already higher than that of the curve reagent, Fe(II)TPP(pip), , and a calibration expressed as AA versus P,, shows curvature with increasing P,,. However, since the solubility of carbon monoxide in toluene is low, the partial pressure PC0 in the gas mixture remains constant to a good approximation and on this basis, the following linear expressions can be derived to serve as calibration curves:

P

AA = A - A,, = K&C,

0.3

-

0.2

-

Co 1 + K’P,,

(5)

and P,,

= L K’ AA,,,

AA

where AA,,, = kCO, A, is the absorbance of the reagent alone, A is the absorbance of the reagent in the presence of carbon monoxide and C, is the initial concentration of the reagent. The first expression relates AA to a function of P, and the second relates P,, to a function of AA. In both cases, the x-scale is compressed at the higher end, leading to inaccuracies at high levels (PC0 > 76 mmHg) of carbon monoxide. A set of twenty data points was used to evaluate the calibration curves. Slopes of 0.031 + 0.002 and 14.1 k 0.9 were found for equations (5) and (6) respectively, with C, = 3.51 x 10m6M. The scatter in the data is indicated in Fig. 2, the calibration curve based on equation (5).

Accuracy, precision and sensitivity The accuracy and precision were tested with nine air-samples containing carbon monoxide in the range 0.24% (v/v). The results are presented in Table 2. The average error and mean relative deviation are

F&(mm

Hg) 35

0

l

2

3

4

5

(6)

- AA

6

7

e

9

IO

45 II

60 I2

PC0 I+K

PC0

Fig. 2. Calibration curve for the determination of carbon monoxide (absorbance measured at 420 nm, Co = 3.51 x 10e6M).

Haems Table

2. Determination

as potential of carbon

Taken P,,,

mmHg 1.43 1.83 3.67 3.98 7.58 9.34 14.65 21.06 28.78

* Average t Relative

reagents

monoxide

for carbon

295

monoxide

in air-samples

at 1 atmosphere

pressure

Found* %, vlv

P co> mmHg

0.188 0.241 0.482 0.524 0.997 1.23 1.93 2.77 3.79

1.50 1.76 3.53 4.32 7.37 9.00 14.71 19.66 26.78

%. VIV

Error,

7;

Precision,t

4.9 5.9 3.6 8.5 2.8 2.7 0.4 6.7 7.0

0.197 0.226 0.464 0.568 0.970 1.18

1.94 2.59 3.52

%

11.2 0 4.6 2.3 0 6.3 5.1 3.9 0

of two determinations. average deviation.

each 5% over this range. Although not exhaustive, the results show that the model haem systems have the potential to yield results of sufficient accuracy and precision for analytical purposes. In addition to the interferences, the other serious shortcoming of the method is the lack of sensitivity. For example, with C, = 5 x lo- 6M and the remaining experimental conditions as specified, if AAlimit = 0.01, then the limit of sensitivity as calculated by equation (5) or (6) is 0.23 mmHg of carbon monoxide. For a gas-sample at one atmosphere pressure, this corresponds to 0.03% (300 ppm). This sensitivity is far too poor for use in the determination of carbon monoxide in cities, where the median annual concentrations lie in the range 3-13 ppm.19

and would seem to be ideal for analytical exploitation. Unfortunately, the synthesis of monopyridinemesohaem in high purity is both difficult and expensive and thus not practical. An easily and inexpensively synthesized compound is the dipyridine complex of Fe(II)TPP. This compound appears to have a greater affinity for carbon monoxide and a lower affinity for oxygen than the dipiperidine complex.’ Further, the use of pyridine instead of piperidine might eliminate the interference caused by carbon dioxide. The potential of the dipyridine complex as a reagent for carbon monoxide is now being investigated.

Possible

The present study represents an initial attempt to demonstrate the suitability of a model compound for a biological process as an analytical reagent. Although solutions of Fe(II)TPP(pip), can be used for the deter-

modifications

to the reagent

system

To be acceptable as an analytical procedure and competitive with other methods, for example, the method of Lambert and Wiens,” the sensitivity must be improved by a factor of 1000. Two ways by which this may be brought about are: (a) to improve the spectral resolution or (b) to increase the value of K. First, a more favourable separation of the Soret bands of the carbonylated and the uncarbonylated species than that shown in Fig. 1 will enhance the sensitivity. This resolution may be brought about by co-ordination of sterically-hindered bases (e.g., cc-picoline or 2,6-lutidine) to the axial positions of the iron(I1) to keep the iron in a predominantly pentaco-ordinate high-spin state, thus shifting the Soret-band position from that of the hexaco-ordinate low-spin carbon monoxide complex. *’ Preliminary studies on such a system, however, indicate that it will be difficult to stabilize the system with respect to aerial oxidation. Second, in the search for haem systems with larger K values, it should be noted that natural or related haems such as haematohaem have higher binding constants for carbon monoxide than Fe(II)TPP but are much more susceptible to aerial oxidation and thus are unsuitable for analytical purposes. The pyridine-mesohaem mode1 system reported by Chang and Traylor,’ however, has a high affinity for carbon monoxide and a surprisingly low affinity for oxygen

CONCLUSION

mination of carbon monoxide, the analytical system described lacks sufficient sensitivity. Also, interference by carbon dioxide and hydrogen sulphide causes a decrease in selectivity. Modifications involving the use of different axial ligands could well increase the sensitivity to useful levels and improve the selectivity. Acknowledgements-The authors wish to thank Mr. 0. Herrmann for the preparation of TPP and Dr. 0. E. Hileman, Jr. for use of the gas-rack assembly. The authors also gratefully acknowledge financial support of this work from the National Research Council of Canada. REFERENCES

and M. Brunori, Hemoglobin nnd Myoglobin in their Reactions with Ligands, pp. 20-33. North Holland, Amsterdam, 1971. A. D. Adler, V. Varadi and N. Wilson, Ann. N.Y. Acad. Sci., 1975, 244. 685. D. V. Stynes, H. C. Stynes, B. R. James and J. A. Ibers, J. Am. Chem. Sot., 1973, 95, 4087. C. K. Chang and T. G. Traylor, Proc. NatL Acad. Sci. USA, 1973, 70. 2647. J. E. Baldwin and J. Huff, J. Am. Chem. Sot., 1973, 95. 5757. J. P. Collman, R. A. Gagne, C. A. Reed, T. R. Halbert, G. Land and W. T. Robinson, ibid., 1975, 91. 1427.

1. E. Antonini

2. 3. 4. 5. 6.

296

S. H. MEHDI and A. CORSINI

7. C. J. Weschler, D. L. Anderson and F. Basolo, ibid., 1975, 91. 6707. 8. D. V. Stynes and B. R. James, Chrm. Cornmun., 1973. 325. 9. C. K. Chang and T. G. Traylor, J. Am Chem. Sot., 1973, 95. 8477. 10. A. D. Adler, F. R. Longo, J. D. Finarelli, J. Goldmacher, J. Assour and L. Korsakoff, J. Org. Chem., 1967. 32. 476. and J. Kim, 1 I. A. D. Adler, F. R. Longo, R. Kampas J. Inorg. Nucl. Chem., 1970, 32. 2443. 12. W. S. Brinigar. C. K. Chang. J. Geibel and T. G. Traylor, J. Am. Chrm. Sot.. 1974, 36, 5597. 13. J. Del Gaudio and G. N. LaMar. ibid., 1976, 38, 3014. 14. G. Robinson and R. C. Robbins. J. Air Pollut. Contr. -I.\\OC~..I970. 20. 303.

15. C. K. Chang and D. Dolphin, J. Am. Chrm. Sot., 1975, 91. 594x. 16. F. J. O’Hara, W. M. Keely and H. W. Fleming, Anal. Chem.. 1956, 28. 466. Solubilitirs cf Iuorgarlic cl& Mcvul Orgunic Compounds, 3rd Ed.. Vol. I. p. 218. Van Nostrand.

17. A. Seidell,

New York, 1940. 18. P. Cans and H. M. N. H. Irving. J. Irmr