Workplace monitoring of hydrogen peroxide using titanyl-coated sorbents G. Hecht,* S. Aubert, F. Gerardin and M. Hery Institut National de Recherche et de Se�curite�, Avenue de Bourgogne, BP 27, F-54501 Vandoeuvre les Nancy, France. E-mail: hechtg@inrs.fr; Fax: +33 3 8350 2060; Tel:+33 3 8350 2000 Received 6th November 1998, Accepted 4th January 1999 The current methods of sampling hydrogen peroxide, based on bubbling in acid solutions (titanium tetrachloride or titanium oxysulfate), are unsuitable for personal sampling.An alternative medium has been developed: silica gel coated with titanium oxysulfate. Sample tubes of this new sampling medium are suitable for personal sampling. The sampling performance is the same as that of the methods based on bubbling, but the tubes must be desorbed as soon as possible, even if the spectrophotometric analysis can be postponed.Special emphasis has been placed on the possible interference by peracetic acid, which is currently used in conjunction with hydrogen peroxide in the food industry. (ii) Sulfuric acid (Merck, Merck Cle�venot, Nogent-sur-Marne, Introduction France, ref. 731.1000). Hydrogen peroxide is widely used for disinfection, particularly (iii) Aqueous solution of hydrogen peroxide (30%) (Prolabo, in the medical field and in the food industry.It is an irritant of Fontenay-sous-Bois, France, 23.615.245). the eyes, mucous membranes and skin. In humans, the inhalation (iv) Silica gel (0.2–0.5 mm) for column chromatography of high concentrations of hydrogen peroxide vapour or the mist (Merck, ref. 7733.1000). may cause irritation and inflammation of the nose and throat. Coating solution: 250 mL of H2SO4 are added to 21 g of Severe systemic poisoning may also cause headache, dizziness, TiOSO4. The solution is then heated, stirring continuously vomiting, diarrhoea, pulmonary oedema and unconsciousness.1 until it becomes clear. After cooling, this solution is added to TheAmericanConference ofGovernmental IndustrialHygienists 250 mL of deionised water and made up to 500 mL.This (ACGIH) recommends a TLV-TWA2 (threshold limit value-time solution is about 0.2 M. weighted average) of 1.4 mg m-3, very close to the corresponding Coating of silica gel: 100 g of silica gel and 25 mL of coating French limit value3 of 1.5 mg m-3. solution are mixed thoroughly.A method of sampling and analysing hydrogen peroxide Absorbing solution (for midget fritted-glass impingers): the was described by Pilz and Johann4 based on the formation of coating solution is diluted with deionised water to a concena coloured complex with titanium tetrachloride. Samples were tration of about 5 mM. collected by drawing known volumes of air through midget fritted-glass impingers containing an absorbing solution.After Materials sampling, the concentration of the complex formed was meas- (i) 4 mL capacity reservoirs (Bond Elut, Varian, Les Ullis, ured by spectrophotometry at 415 nm. The Occupational France, 1213–1008). Safety and Health Administration (OSHA) proposed replacing (ii) Porous polyethylene (20 mm pore size) frits (Bond Elut, titanium tetrachloride by titanium oxysulfate,5 and then substi- Varian 1213–1020).tuted a polarographic determination for the spectrophoto- Packing of sampling tubes: the reservoirs are packed with metric method,6 to obtain a better analytical detection limit about 0.8 g of coated silica gel placed between two polyethyland to minimise interferences. ene frits.The main advantage of the OSHA method, apart from its enhanced detection limit, is the better stability of the absorbing Analytical methods solution, which remains stable for several weeks without Midget fritted-glass impingers: 15 mL of the absorbing solution change, in contrast with the complex formed by the reaction are placed into the midget fritted-glass impinger. After sam- with titanium tetrachloride which can only be kept for one or pling, the absorbing solution is transferred to a vial.The two days. Both methods use midget fritted-glass impingers bubbler is rinsed with deionised water and transferred to the which make it diYcult to take personal samples. The aim of vial. The solution is made up to 25 mL. our study was to replace these sampling solutions by a medium Sampling tubes: after sampling, 8 mL of sulfuric acid (1 M) suitable for the requirements of occupational hygiene.Tubes are percolated through the coated silica gel. In our tests, the containing silica gel coated with titanium oxysulfate were used. desorption solution is then made up to 25 mL (for comparison The performance levels of this new medium and of the bubbling with samples taken by bubbling) or to 10 mL (for comparison in oxysulfate solution were then compared.of tubes with others). In both cases, the quantity of hydrogen peroxide sampled is determined by comparison with a calibration curve made Materials and methods up of standards prepared in the same way as the samples (midget fritted-glass impingers or tubes). The solutions are Reagents analysed by spectrophotometry at 410 nm.The response is linear for solutions of up to 30 mg L-1. Spectrophotometric (i) Titanium oxysulfate (Riedel de Hae�n, Hoechst, Paris La De�fense, France, ref. 14023). determination was chosen on account of both its shorter J. Environ. Monit., 1999, 1, 149–152 149analysis time and wider availability in laboratories, despite having a poorer detection limit and being more prone to interference than polarography. Tests in the laboratory Generation of controlled test atmospheres The controlled test atmospheres used in this evaluation were generated by volatilising and diluting hydrogen peroxide with air.A motorised syringe (KD Scientific, Bioblock, Illkirch, France) delivered a constant flow of an aqueous solution of hydrogen peroxide (about 0.2M).This solution was volatilised in a heated manifold and then mixed with air. The flow was again diluted with air to obtain the desired concentration. Eight samples could be taken simultaneously in the final chamber at flow rates of up to 0.8 L min-1. A diagram of the device is provided in Fig. 1. EYciency of the sampling method on tubes Six series of samples were taken, each consisting of four Fig. 2 Specific device for hydrogen peroxide vaporisation. samples obtained in midget fritted-glass impingers and four on tubes. The sampling rate was 0.8 L min-1 for both. The and at the analytical limit of the spectrophotometric method. results (arithmetic mean and standard deviation) of this test In the same way and on day D, the same quantities of are summarised in Table 1.There is a close agreement between hydrogen peroxide (34 mg and 272 mg) were injected directly the two types of sample. in 10 mL of absorbing solution contained in vials. The tubes and absorbing solutions were then analysed in Stability of samples batches of 6 or 12, on diVerent days from D+1 to D+61. It was not possible to use the above mentioned device to The results of these tests are given in Table 2.The reference generate samples on which a stability study could be performed 100 corresponds to fresh samples prepared on each day of as the quantity of hydrogen peroxide generated varies with analysis from hydrogen peroxide solution whose concentration time. A specific device, details of which are given in Fig. 2, had been checked just before preparation. For tubes analysed therefore had to be designed. A known volume of a known after D+4, the recovery percentage is under 90%, whereas for solution of hydrogen peroxide (about 0.1 M, titrated by the the absorbing solutions the recovery percentage is close to potassium iodide/sodium thiosulfate method6) was volatilised 100% for all the analyses performed from D+1 to D+61.As on sampling tubes. Two series of 30 samples were prepared the solutions were not stored in the dark, it is unlikely that on day D: one with 10 mL of the hydrogen peroxide solution the poor recovery of the hydrogen peroxide sampled on tubes (34 mg of hydrogen peroxide) and the other with 80 mL after several days is due to the destruction of the ‘peroxytitanyl (272 mg).The first value corresponds to a four-hour sample cation‘ by light.orbing solutions are in fact equivalat 10% of the limit value, whereas the second corresponds to ent to tubes which have been desorbed immediately after a sample of identical duration at a concentration of 1.1 mg m-3 sampling. Consequently, the tubes sampled must be desorbed as soon as possible, even if the analysis can be postponed.The poor recovery of the hydrogen peroxide after a few days is due to the migration of the complex sampled. This migration can be seen in Fig. 3. Only one direction is visible, but it seems reasonable to suppose that the migration develops in both directions. Interferences All compounds likely to interfere with the determination of hydrogen peroxide were not investigated in this study, which limited itself in scope to those liable to be used in conjunction with hydrogen peroxide in industry.OSHA reports permanganate and iodide ions as compounds which make the spectro- Fig. 1 Generation system for hydrogen peroxide production. photometric determination of hydrogen peroxide impossible, but they are rarely used jointly.On the other hand, peracetic Table 1 EYciency of the sampling method on tubes in the laboratory acid is often used as a sterilisation agent at ambient temperature, especially for fruit juice packaging, and peracetic acid Conc./mg m-3 (RSD) solutions contain hydrogen peroxide. To study how this substance interferes with the determination of hydrogen peroxide, Series N Sampling time/min Impingers Tubes the previously described method (see Analytical methods sec- 1 4 140 1.21 (±5.7) 1.22 (±3.9) tion) and a peracetic acid determination method by Pinkernell 2 4 130 1.43 (±4.2) 1.45 (±2.9) et al.7 were used.Peracetic acid (PAA) oxidises rapidly methyl- 3 4 80 1.54 (±6.0) 1.52 (±1.6) p-tolylsulfide (MTS) leading to methyl-p-tolylsulfoxide 4 4 100 1.90 (±2.1) 1.85 (±3.1) (MTSO): 5 4 135 2.84 (±2.9) 2.85 (±3.7) 6 4 60 2.74 (±2.8) 2.72 (±4.1) CH3C(O)OOH+CH3C6H4SCH3�CH3C6H4S(O)CH3 RSD, relative standard deviation.+CH3COOH 150 J. Environ. Monit., 1999, 1, 149–152Table 2 Storage data for the samples and the absorbing solutions Desorption Analysis N Reference % (RSD) N Sample % (RSD) Low level D+1 D+1 6 97 (±1.2) 6 103 (±1.5) D+8 D+8 6 89 (±5.2) D+20 D+20 6 107 (±0.8) 6 86 (±4.1) D+61 D+61 12 104 (±0.9) 12 71 (±6.1) D+8 D+20 6 104 (±3.3) 6 88 (±4.7) D+8 D+61 6 101 (±3.1) 6 89 (±6.5) High level D+1 D+1 6 98 (±0.7) 6 100 (±0.4) D+4 D+4 6 96 (±1.7) D+16 D+16 6 104 (±1.0) 6 90 (±1.4) D+57 D+57 12 101 (±0.8) 12 77 (±1.2) D+1 D+16 6 102 (±1.1) 6 99 (±1.2) D+4 D+57 6 100 (±1.2) 6 95 (±2.8) RSD, relative standard deviation.Fig. 3 D, Tube prepared on day D; D+6, same tube on day D+6. The reaction kinetics are very fast with peracetic acid, but very slow with hydrogen peroxide. The tests performed in our laboratory show that, after two days of reaction between MTS and hydrogen peroxide, the quantity of MTSO formed is 16 times lower than that observed using a solution of peracetic acid at the same concentration as the hydrogen peroxide Fig. 4 Comparison of tube and impinger sampling in workplace solution. The quantity of MTSO generated does not increase atmospheres. further later. A study was conducted to assess the interference due to peracetic acid in the determination of hydrogen peroxide. (i) Standards containing from 0.5 to 5 mM per sample in the absorbing solution were prepared from a 0.1M hydrogen peroxide solution.(ii) A 4% aqueous solution was prepared from a peracetic acid solution (CH3C(O)OOH#6 M, H2O22 M). (iii) Test A: 5 mL of MTS (Aldrich, Sigma Aldrich, Saint Quentin-Fallavier, France, ref. 623–13–2) (4 mM) were placed into three vials. Five mL of the 4% PAA solution were added to the first, 10 mL to the second and 20 mL to the third.After 5 min of reaction, the volume of the three vials was made up to 10 mL with the absorbing solution. The three solutions were analysed by spectrophotometry, using the calibration curve of (i) as reference. This test gave the quantity of hydrogen peroxide in the peracetic acid solution. (iv) Test B: the same experiment as in A, but with the Table 3 Interference of peracetic acid (PAA) in the determination of hydrogen peroxide Fig. 5 Comparison of successive samples versus single sample. ‘Real’ conc. of ‘Real’ conc. of ‘Apparent’ conc. PAA solution H2O2/mM PAA/mM ofH2 O2/mM added/mL (Test A) (Test B) (Test C) absorbing solution being replaced by a methanol–water 5 0.29 1.14 0.82 (75+25) mixture. The analysis of MTSO (Fluka, Sigma 10 0.60 2.23 1.67 Aldrich, Saint Quentin-Fallavier, France, ref. 69422) was 20 1.23 4.36 3.35 carried out by high performance liquid chromatography J. Environ. Monit., 1999, 1, 149–152 151is shown in Fig. 4. Like the laboratory test the results are satisfactory, with an underestimate of about 5% for the sampling on the tube compared with the sampling on the impinger. Considering this small diVerence, it does not seem necessary to use a correction factor.Comparison of single sample and successive samples The aim of this test was to check the ability of a tube to be sampled over an entire working shift. The result given by a sample taken for the whole working shift was compared with that given by several samples taken successively over the same duration. The correlation between the elements of the pairs of the results is shown in Fig. 5 and, as can be seen, is excellent, with a slight overestimate for the successive samplings. Fig. 6 Influence of the sampling rate. Influence of the sampling rate This test was performed in order to check if sampling could (HPLC). From these values of MTSO the concentration of be carried out at a low flow rate. Low flow pumps are useful: PAA in the peracetic acid solution could be determined.their weight and dimensions are smaller and they are better (v) Test C: 5 mL of the 4% PAA solution were placed into tolerated by the workers. Ten pairs of samples were taken for the first vial, 10 mL in the second and 20 mL in the third. The one to two hours: one sampled at 0.25 l min-1 and the other three vials were made up to 10 mL with the absorbing solution.at 1 L min-1. The correlation between the elements of the The three solutions were analysed by spectrophotometry, pairs is shown in Fig. 6. Although the correlation is good, for using the calibration curve of (i) as reference. This test low atmospheric concentrations (around one-tenth of the limit highlighted the influence of the presence of peracetic acid on value), the low flow samples are about 30% below the high the determination of hydrogen peroxide. flow samples.No comparison tests were made during a whole The results of these tests are summarised in Table 3 and working shift. show that peracetic acid interferes in the determination of hydrogen peroxide. From this series of tests, it would appear that 1 mM of peracetic acid corresponds to an apparent Conclusion 0.5 mM of hydrogen peroxide.The sampling medium developed in this study is a good This finding is not really surprising. Purnell et al.8 have alternative to sampling by bubbling in an impinger. It allows described a method of sampling and analysing methyl ketone industrial hygienists to take personal samples easily.If they peroxide in air which is based on the same principle as the are desorbed rapidly after sampling, the samples can be kept method by Pilz and Johann.4 It is therefore likely that all without loss for a long time (up to two months). This medium organic peroxides will lead to an overestimate of the quantity is suitable for the measurement of monopollution by hydrogen of hydrogen peroxide.peroxide, but it does not provide industrial hygienists with the The validation was not performed completely in accordance means of avoiding interferences inherent in spectrophotometric with the requirements of a test method such as, for example, techniques. From a practical point of view, the interference the EN 76 standard:9 generating a controlled hydrogen due to peracetic acid seems the most likely, because of its peroxide atmosphere is diYcult; we do not know of any direct widespread use in conjunction with hydrogen peroxide in the reading analyser for hydrogen peroxide food industry.Tests in workplace atmospheres References The method was tested in two factories, one packaging milk 1 American Conference of Governmental Industrial Hygienists and the other fruit juice.Both factories use the TetrapakA (ACGIH), Documentation of the Threshold Limit Values and packaging process. Three types of validation were performed. Biological Exposure Indices, ACGIH, Cincinnati, 6th edn., 1991. (i) Comparison of the concentrations measured by sampling 2 American Conference of Governmental Industrial Hygienists (ACGIH), Threshold Limit Values for Chemical Substances and on tubes containing coated silica gel and on midget fritted- Physical Agents, ACGIH, Cincinnati, 1998.glass impingers. 3 Ministe`re du Travail, Valeurs Limites D’exposition Professionnelle (ii) For the sampling on tubes, comparison of a sample aux Agents Chimiques en France, INRS, Paris, ND 1945, 1996. taken for 5 to 8 h with a series of samples taken at the 4 W.Pilz and I. Johann, Int. J. Environ. Anal. Chem., 1974, 3, 257. same time and place during successive periods of an hour. 5 Occupational Safety and Health Administration (OSHA), (iii) The influence of the sampling rate was investigated for Analytical Methods Manual, OSHA, Salt Lake City, 2nd edn., 1990, method VI-6. samples taken on tubes. Two samples, one at a rate of 6 Occupational Safety and Health Administration (OSHA), 1 Lmin-1 and the other at 0.25 L min-1 were taken simul- Analytical Methods Manual, OSHA, Salt Lake City, 2nd edn., 1990, taneously at the same place. method ID-126-SG. For the first two tests, the sampling pumps were calibrated 7 U. Pinkernell, U. Karst and K. Camann, Anal. Chem., 1994, 66, at very similar flows (about 1 L min-1) and the inlets of the 2599. impinger and of the tubes were sampling through the same 8 C. J. Purnell, G. L. Martin and H. Wolfson, Ann. Occup. Hyg., 1979, 22, 383. T-shaped aperture. 9 European Committee for Standardization, Workplace Atmospheres—Pumped Sorbent Tubes for the Determination of Comparison of tubes and impingers Gases and Vapours.—Requirements and Test Methods, EN 1076, CEN, Brussels, 1997. Fifty seven pairs of samples, one on a tube and the other on an impinger were taken in the two factories with atmospheric Paper 8/08686I concentrations of hydrogen peroxide ranging from 0.5 to 4 mgm-3. The correlation between the elements of the pairs 152 J. Environ. Monit., 1999, 1, 149–152