Measurement of Atmospheric Hydrogen Peroxide

7/29/2019 Measurement of Atmospheric Hydrogen Peroxide

1/10

Measurement of Atmospheric Hydrogen Peroxideand Hydroxymethyl Hydroperoxide with a DiffusionScrubber and Light Emitting Diode-Liquid CoreWaveguide-Based Fluorometry

Jianzhong Li and Purnendu K. Dasgupta*

Department of Chemistry and Biochemistry, Texas Tech University, Lubbock, Texas 79409-1061

Wedescribe a new automated instrument for measuring

gas- and aqueous-phase H2O2. The chemistry relies on

the hematin-catalyzed oxidation of nonfluorescent thia-

mine to fluorescentthiochromebyH2O2; this reaction is

25-fold more selective for hydrogen peroxide than its

nearest alkyl hydroperoxide congener, CH3HO2. The

optical characteristicsofthefluorescentproductaresuchthatit is ideallyexcitedbynewlyavailableGaN-basedlight

emitting diodes emitting in the near-UV. A stable long-

life miniature flow-through fluorescence detector based

on a transversely illuminated liquid core waveguide is

thus used for this purpose. The limit of detection (LOD,

S/ N) 3) for liquid-phaseH2O2 is 11 nM. A temperature-

controlled high-efficiency Nafion membrane diffusion

scrubber is used to collect gaseous hydrogen peroxide

with near-quantitativeefficiency with an S/N ) 3 LOD of

13.5 pptv. Thesystemresponds tohydrogenperoxideand

hydroxymethyl hydroperoxide but not to methyl hydrop-

eroxide. Theuseofveryinexpensiveand stable reagents,

highly sensitive detection, benign chemistry, and a fluo-rescence detector usinga solid-stateillumination source

results in aparticularly affordable automatedinstrument.

Design and performance details and illustrative results

from a1999 field campaign (AtlantaSupersiteStudy) are

presented.

Hydrogen peroxide is an important analyte i n clinical chem-

istry. I t also plays a central role in atmospheric studies due to its

ability to oxidize dissolved SO2.1 Hydrogen peroxide, methyl

hydroperoxide (M HP), and hydroxymethyl hydroperoxide (HM HP)

are of particular interest and detailed reviews are available.2

In the aqueous phase, H2O2 is the primary analyte of interest

since M HP displays a much lower Henrys law constant3

andHMHP T H2O2 + HCHO exhibits a dissociation constant of 6

mM at 25 C.4 Luminol chemiluminescence (CL) can be very

sensitive and has been exploited for the measurement of marine

H2O2.5 Many oxidants generate luminol CL; direct measurement

of H2O2(g) by luminol CL is difficult. Alternative CL techniques

for H2O2(g) have been developed.6 Few have attempted to measure

gaseous peroxides electrochemically7 because it is not sufficiently

sensitive. Introduced by Lazrus et al.,8 fluorometry dominates

present atmospheric peroxide measurements. There are pedagog-

ical merits to fluorometric H2O2 measurements as well.9

Typically,a nonfluorescent substrate is oxidized by H2O2 to a fluorescent

product mediated by a peroxidase enzyme, typically horseradish

peroxidase (HRP).8-10 Cost and stability issues of HRP have led

to attempts to use photocatalysis11 and other nonenzymatic

reactions.12 Synthetic metalloporphyrins have been extensively

studied as HRP mimics13 and shown to be effective.14 Phtha-

locyanine complexes are effective catalysts in both acidic and basic

solutions.15 Several Fe-porphine compounds, inexpensive com-

mercial products from animal blood, display peroxidatic activity.14

Bovine hematin displays a greater peroxidatic activity per unit

mass than any commercial HRP preparation at


2/10

fluorescent intensity, at 1%of the cost. However, al l of above

schemes require excitation between 300 and 326 nm. Narrow-

band GaN-based light emitting diodes (LEDs) emitting at375

nm are now available;18 and simple liquid core waveguide (LCW)-

based fluorescencedetectors havebeen described.19 A LCW-based

solid-state fluorometer for H2O2 will be attractive.

With metallophthalocyanine or hemin as a catalyst, thiamine

(vitamin B1), can be used to determine H2O2.15,20 It is oxidized to

fluorescent thiochromewith an excitation maximum in our desired

range.21,22

Generally, the analytical methods above do not discriminate

between H2O2 and the two principal organic peroxides, M HP and

HMHP. The enzyme catalase or a MnO2 catalyst has often been

used to selectively destroy H2O2 and thus establish differential

measurements.8,23 Both approaches have problems. Extending the

data obtained with pure peroxides produces errors with mixtures

of H2O2 and organic peroxides in the catalase system.24 With

MnO2, there may be some destruction of organic peroxides.25 Our

experience with continuously operating MnO2 reactors indicates

that they lose activity and must be replaced on a frequent basis,

requiring recalibration. In a different approach, the Fenton reaction

has been used to measure H2O2 + HM HP (40%) + M HP (12%)

with a low-pH collector, H2O2+ HMHP+MH P (12%) with a high-pH collector, and H2O2 + HMHP + M HP (60%) with a high-pH

collector.26 All gaseous H2O2 measurement techniques that involve

aqueous scrubbing have some degree of O3 interference since

O3-water reactions at surfaces produce H2O2.27 Such reactions

are facilit ated at a higher pH,28 and the extent of O3 interference

with a high-pH collector needs careful investigation. At the present

time, the best method for the determination of individual peroxides

appears to be chromatography with postcolumn r eaction-fluorom-

etry.29 M easurement of peroxide gases by such techniques must

nevertheless involve collection. With aqueousscrubbing, collection

of MHP is incomplete, albeit subquantitative collection can be

calibrated for.30 On the other hand, cryogenic trapping is quantita-

tive but can lead to significant amounts of artifact peroxides.31

Interference from SO2 poses a problem in both methods and an

universally applicable solution is not available. Clearly, intercom-

parisons are necessary to clarify the strengths and weaknesses

of the different methods. Even if only H2O2, HMHP, and MHP

are pr esent, i t is mini mally desirable to be certain how much of

whatprecisely a given procedure actually measures; the response

coefficients must not depend on the exact composition of the

sample.

Peroxide gases have also been collected with the coil,8 the

diffusion scrubber32 (DS), the mist chamber, and, most recently,

the condensate trap.33 The condensate trap requires no power;

ice-filled test tubes are hung in a container where peroxides in

the ambient air condense on the outer walls. It requires carefulmonitoring of humidity and is inapplicable at low absolute

humidities. The mist chamber is an efficient collector but

obligatorily collects the aerosol. Interaction with the deposited

aerosol can lead to the loss of H2O2; interaction between ozone

and deposited organic aerosol can lead to the generation of H2O2.

Aerosol deposited in the sampling line has the same effect.

Periodic cleaning of all contact surfaces is necessary for this

reason. The DS permits simple strategies to preconcentrate.34 The

extent of aerosol deposition in a properly designed DS is very

low,35 but periodic routine cleaning is still necessary. The coil,

probably the most widely used, shows a modest amount of aerosol

loss. Insoluble particles can be carried by the liquid stream and

be problematic,25 but they do not accumulate in the collectionsurface. Dir ect intercomparisons of collection techniques have not

been frequent. In one such study on a coil-DS comparison, the

authors concluded that the DS is superior.25 The DS used in the

present work is of the straight inlet design and is a near-

quantitative collector of H 2O2 over a significant rangeof flow rates.

In the present report, we describe a fully automated instrument

that thus combines several chemical, material science, and

technological developments since our last effort on gaseous

H2O2,16 to collect H2O2 nearly quantitatively and measure [H2O2]

+ [HMHP]. Illustrative results from an extensive field study in

the summer of 1999 in Atlanta (the Atlanta Supersite Study36) are

also presented.

EXPERIMENTAL SECTIONReagents. Hematin (from bovine blood, Aldrich) stock solution

was prepared by dissolving 10 mg of hematin i n 100 mL of 0.1 M

NaOH. Refrigerated at 4 C, this solution is stable for at least 2

months. Thiamine stock solution (10 mM ) was prepared by

dissolving 337 mg of thiamine hydrochloride (Sigma) in 100 mL

of water. Refrigerated, this solution was stable for at least 1

months. Working hematin (nominally 10 M) and thiamine

solutions (100 M ) were prepared daily by diluting 6.3 mL of the

hematin stock to 100 mL with phosphate buffer ( 50 mM K2HPO4adjusted to pH 12.0 with 2 M NaOH) and diluting the thiamine

stock with water, respectively. Hydrogen peroxide stock solution

(1 M) was prepared by dilution of 30%solution (Fisher) with waterand standardized by titration with secondary standard KMnO4.

M HP was a gift from researchers at the National Center for

Atmospheric Research. HMHP was synthesized by the reaction

(18) www.nichia.com.

(19) Dasgupta, P. K.; Zhang, G.; Li, J.; Boring, C. B.; Jambunathan, S.; Al-Horr,

R. Anal. Chem. 1999, 71, 1400-1407.

(20) Zhu, Q.-Z.; Li, Q.-G.; Lu, J.-Z.; Xu, J.-G. Anal. Lett, 1996, 29, 1729-1740.

(21) Guo, X.-Q.; Xuo, J.-G.;Wu, Y.-Z.;Zhao, Y.-B.; Huang, X.-Z.; Chen, G.-Z. Anal.

Chim. Acta1993, 276, 151-160.

(22) Jie, N.; Yang, D.; Zhang, Q.; Yang, J.; Song, Z. Anal. Chim. Acta 1998,

359, 87-92.

(23) Dasgupta, P. K.; Hwang, H. Anal. Chem. 1985, 57, 1009-1012.

(24) Heikes, B. G. University of Rhode Island, personal communication, 2000.

(25) de Serves, C.; Ross, H. B. Environ. Sci. Technol. 1993, 27, 2712-2718.(26) Lee, J. H.; Tang, I. N.; Weinstein-Lloyd, J. B.; Halper, E. B. Environ. Sci.

Technol. 1994, 28, 1180-1185.

(27) Heikes, B. G. Atmos. Environ. 1984, 18, 1433-1445.

(28) Staehelin, J.; H oigne, J. Envir on. Sci. Technol. 1982, 16, 676-681.

(29) Hellpointner, E.; Gab, S. Nature1989, 337, 631. Kurth, H.-H.; Gab, S.;

Turner, W. V.; Kettrup, A. Anal. Chem. 1991, 63, 2586-2588. Hewitt, C.

N.; Kok, G. L. J. Atmos. Chem. 1991, 12, 181-194. Kok, G. L.; M cLaren, S.

E.; Staffelbach, T. A.; J. Atm os. Ocean. Technol. 1995, 12, 282-289. Sauer,

F.; Limbach, S.; Moortgat, G. K. Atmos. Environ. 1997, 31, 1173-1184.

Wang, K.; Glaze, W. H. J. Chromatogr. 1998, 822, 207-214.

(30) Lee, M.; Noone, B. C.; OSullivan, D.;Heikes,B. G. J. Atmos. Ocean. Technol.

1995, 12, 1060-1070.

(31) Staffelbach, T.; Neftel, A.; Dasgupta, P. K. Geophys. Res. L ett. 1995, 22,

2605-2608.

(32) Dasgupta,P. K.; Dong, S.; Hwang, H.; Yang, H.-C.; Genfa, Z. Atmos. Environ.

1988, 22, 949-963.

(33) Deforest, C. L.; Kieber, R. J.; Willey, J. D. Envir on Sci. Technol. 1997, 31,

3068-3073.

(34) Genfa, Z.;D asgupta, P. K.; Dong, S. Environ. Sci. Technol. 1989, 23, 1467-

1474. Trapp, D.; De Serves, C. Atmos. Environ. 1995, 29, 3239-3243.

(35) Zhang, G.; Dasgupta, P. K.; Cheng, Y.-S. Atmos. Environ.1991, 25A , 2717-

2729.

(36) http:/ / www-wlc.eas.gatech.edu/ supersite/ .

Analytical Chemistry, Vol. 72, No. 21, November 1, 2000 5339


3/10

of H2O2 and HCHO.37 The material was primarily an oil of HMHP

with some H2O2 and bis-HMHP (BHMHP). The H2O2 present in

the MHP or HMHP/ BHMHP solution was removed by passing

the solution through a MnO2 column,23 and the concentration of

the organic peroxide in the effluent was determined by measuring

the absorbance of I 3- at 352 nm produced by the oxidation KI in

the presence of acetic acid under a blanket of CO2.

Analytical System. The arrangement of liquid and gas flows

in the system is schematically shown in Figure 1a. Water wasaspirated by a peristaltic pump (P, Rainin Dynamax) thr ough the

diffusion scrubber (DS), which collects the H2O2 from the gas

phaseto the water stream. The aqueous effluent from the DS was

mixed with the thiamine reagent in a Serpentine-II knit 38 PTFE

reactor (L1, 0.65 mm i.d. 400 mm). The hematin stream was

then merged in and the mixed stream and reacted in another

similarly knit PTFE r eactor (L2, 0.65 mm i. d. 1700 mm,

residence time180 s) and then flowed through the LCW-based

fluorescence detector. Reaction coil L2 is potted in a low-melting

bismuth-tin alloy (Cerrobase 5550-1, Canfield Technologies,

Sayrevill e, NJ) in a poly(vinyl chl oride) container containing two

flexible siliconized heaters (Watlow, St. Louis, M O) i n series (22.5

W total). A 100- platinum RTD monitors the temperature, andtemperature control is achieved by a mi niature temperature

controller (type CN 1632 GNR, Omega Engg., Stamford, CT).

Unless otherwise stated, the reactor was maintained at 30 C. A

flush port F is provided for the detector via a tee-arm to remove

recalcitr ant bubbles or to wash and clean the detector (vide infra).

A miniature air pump (AP, UNMP30KNI, KNF Neuberger,

Trenton, NJ) was used to aspirate the sample gas through the

DS. The pump outlet was connected to a three-way valve V (M BD

002, Honeywell) via flowmeter FM . This valve was controlled by

an independent programmable timer (RS4A22, SSAC Inc., Bald-

winsville, NY). In one position, the pump exhaust was vented, and

in the other position, the pump exhaust was passed through a

column I filled with pelletized activated carbon (8 mesh, Fluka)

to remove any H2O2 in it and feed this zero gas to the DS. The

total aspiration by AP constituted that through the DS, plus a small

amount (2%of the total; total flow was typically 1.6 standard liters

per minute, SLPM) through a 23-gauge hypodermic needle (N)that was connected to the AP inlet by a tee. This design ensures

that when valve V is switched to provide zero gas to the DS inlet,

the airflow is sufficient and no ambient air is drawn through the

sample inlet; rather, a small amount of zero air back-flushes the

sample inlet line. Unless otherwise stated, valve V was pro-

grammed to sample the test gas for 2 min and zero gas for 8 min.

LCW Fluorescence Detector. The structure of the trans-

versely illumi nated LCW fluorescence detector is shown in Figure

1b; it is similar to previous reports19,39 except that an LED emitting

in the near-UV was used as the light source. An LED excitation

source is very stable, exhibits long life, and produces no heat.

The lateral distancebetween the excitation source and the detector

receiving fiber is 45 mm. Although the illuminated volume is notof major concern to this work, it is sufficiently small for the device

to function as a detector in high-performance liquid chromatog-

raphy. Large-core (1.5 mm) acrylic optical fiber collects the

emitted light from the end of the LCW AF (1.25 mm i.d.; Random

Technologies, San Francisco, CA) and conducts it to miniature

photomultiplier tube (PMT; Hamamatsu H5784). As an excitation

filter, we used two disks of blue plastic OF (type 861, Edmund

Scientific. Gloucester, NJ). The excitation light status is monitored

by a silicon photodiode (PD; BPW 34, Siemens AG).(37) Marklund, S. Acta Chem. Scand. 1971, 25, 3517-3531.

(38) Curtis, M. A.; Shahwan, G. J. LC-GC M ag. 1988, 6, 158-164. ( 39) L i, J.; D asgupt a, P. K .; Zh ang, G. Talanta1999, 50, 517-623.

Figure 1. (a) Instrument schematic. Key: P, peristaltic pump; DS, diffusion scrubber; L1, thiamine mixing coil; L2, main reaction coil, thermostatedat 30 C; F, flush port (normally closed); AP, air aspiration pump; FM, flowmeter; N, 23-gauge hypodermic needle (supplementary flow); V,

three-way solenoid valve controller by timer T; C, activated carbon column. (b) Transversely illuminated fluorescence detector schematic. Key:

T, entrance PEEK tee; LI, liquid inlet; FO, silica fiber optic; OF, colored plastic optical filter; PMT, miniature photomultiplier tube; AF, Teflon AF2400 liquid core waveguide; SSJ, stainless steel jacket, LED. UV emitting light emitting diode; PD, photodiode for monitoring source light intensity;U, outlet union; W, waste.

5340 Analytical Chemistry, Vol. 72, No. 21, November 1, 2000


4/10

Diffusion Scrubber. A new robust design of the diffusion

scrubber was developed for this work and is shown in Figure 2.

The DS was housed in a Plexiglas enclosure with inside dimen-

sions being 21/8 in. wide 201/8 in. long 1 i n. deep. Flexible

siliconized heater elements (2 20 in., Watlow, 50 W at110 V)

were glued by silicone rubber adhesive to the interior surfaces

of both the top and the bottom covers. The main DS unit

constitutes of Nafion tube N (wet dimensions 0.75 mm i.d., 0.90

mm o.d.) housed in a PTFE jacket tube J (3.2 mm o.d., 2.7 mm

i.d.) held in place by poly(vinyledene fluoride) (PVDF) tees T1

and T2. T1 is connected to 1/8-1/4-in. union fitting U1 with a

minimum of PTFE tubing to provide 1/4-in. tube connections to

the air pump. T2 is connected with a minimum of tubing to teeT3 that connects to the sample inlet fitting U2 and to an elbow E

by rigid stainless steel tubing SS. E is further connected to zero

air inlet U3 by another piece of stainless steel tube SS. Nafion

tube N contains inserted PTFE capillary tubing at its termini; these

are sealed at the tee ports of T1 and T2 with suitable sleeves.

The PTFE capillaries exit through the frame through suitable

sleeves and 10-32 thr eaded compression seals. The active length

of the Nafion membrane is 45 cm. A 100- platinum RTD was

placed on the midsection of the DS jacket and affixed in place

with PTFE tape. The RTD was connected in a three-wire config-

uration to a stereo mini-Jack M. This was connected by an

appropriate cable to a temperature controller (of the same type

as used with reactor L2). The heater leads were connected in

series to reduce the total heater power to 25 W. We investigated

a number of options for the filler material in the DS housing. The

desirable characteristics were low weight, good thermal conduc-

tivity, low off-gassing of any interfering vapor, and low cost. We

chose a decorative gravel for filli ng aquaria, available in most pet

stores. The filling material can be put in or removed through a

hole normally capped by a cap C.

Scrubber collection efficiency was determined by connecting

two identical diffusion scrubbers in series and comparing the ratio

of the observed signal in the upstream versus the downstream

scrubber. The scrubber positions are then r eversed, and the same

evaluation is repeated. This procedure does not need to assume

identical collection efficiencies of the two DS units, and the basic

theory has been previously described.32

Instrument Calibration and Standard Gas Generation.

Both gas-phase and liquid-phase calibr ations are of interest

because the instrument can be used for either gas-phase measure-

ment or for the measurement of dissolved H2O2 in rain or

cloudwater. Both types of calibration were therefore conducted.

The loop-type six-port injector valve (I, Figure 1a, P/ N V540,

Upchurch Scientifi c, Oak Harbor, WA) was provided with a loop

of 50-L volume for the purposes of liquid-phase calibration. The

gas-phase H2O2 generation source was set up and calibrated aspreviously reported; the r esults were in good agreement with

previously published Henrys law constant for H2O2.40 Gaseous

M HP was generated in a similar manner using an MHP solution

as the generation source solution. To inhibit the hydrolysis of

MHP, the solution was prepared in 1 mM HCl. The gas-phase

MHP concentration was calculated from the known Henrys law

constant3 and the measured aqueous MHP concentration.

Several factors complicate similar experiments with HMHP.

First, the stability of dilute HM HP solutions over a long period is

a concern.4 Second, the solution is not pure and selective transfer

of the less soluble H2O2 to the gas phase will alter the nature and

concentration of the gas-phase peroxide(s) over a peri od of time.

We chose to perform the HMHP experiments by depositing 20

L of a HMHP solution in a Teflon tube through which 3.5L/ min

air was flowing. The gas was sampled by the instrument with zero

on/ off cycling di sabled, the area under the entire response peak

was integrated to assess the response, and the results were

compared with an identical experiment with hydrogen peroxide

solutions.

Ozone was generated by a high-pressure mercury lamp. The

concentration of the generated O3 was determined by neutral

(40) Hwang, H.; Dasgupta, P. K. Envir on. Sci. Technol. 1995, 19, 255-258.

Figure 2. Diffusion scrubber and housing. Key: U1, U2, U3, PVDF 1/4-1/8-in. tubing union fittings; SI, sample inlet; ZA, zero air inlet; AP, toair pump; T1, T2, T3, PVDF tees; E, PVDF elbow; N, Nafion tube; J, PTFE jacket; SS, stainless steel tubes; WI/WO, water inlet/outlet (PTFE

capillaries); RTD, platinum resistance thermometer connected to miniature three-wire jack M; PF, Plexiglas frame; C, filler cap; H, heater (topand bottom); HL, heater leads.



5/10

buffered potassium iodide method (1%KI buffered with phosphate,

pH 6.8 ( 0.2) and measuring the absorbance of produced I3- at

352 nm. A diode arr ay spectrophotometer (Hewlett-Packard 8453)

was used for all spectrophotometric measurements. A photometric

ozone monitor (model 1003 AH, Dasibi Instrument Corp.) was

used as a secondary instrument to monitor the ozone concentra-

tion. The generated ozone was scrubbed with water to remove

H2O2 potentially present in the ozone stream. The scrubbed ozone

was diluted with a H2O2-bearing stream of known concentration

for testing.Sulfur dioxi de was generated with a Henrys law-based source

with a source solution containing 5 mM NaHSO3 in a 50 mM

potassium acid phthalate buffer at pH 4 and 20 C. The anticipated

primary source concentration was 22.3 ppm, based on published

equilibri um constants. The measured value (sampled into bubbler

containing dilute H2O2 and ion chr omatographic measurement as

sulfate) showed an initial value of 18.7 ppm; that decreased to

18.0 ppm by the end of the interference studies.

Field Studies. The instrument was deployed in the Atlanta

Supersite experiment36 from August 3, 1999 to August 29, 1999.

The site, located in midtown Atlanta(latitude 33.777 N, longitude

84.414 W, altitude 265 m M SL) provided an air-conditioned

shelter for the instrument. A Teflon-coated aluminum cyclone(designed for a 2.5-m cut point at a flow rate of 3 L/ min, URG,

Chapel Hill , NC) with a downward pointing i nlet 2 m above the

roof of the shelter (6 m above ground) was used. The inlet

conduit from the cyclone to the DS was PFA Teflon (5 mm in

i.d.). The total flow through the sampling system was 2.7 SLPM,

1.6 SLPM sampled by the present instrument and 1.1 SLPM

sampled by a second instrument designed to measure formalde-

hyde. The residence time in the sampling line was 2 s. The

sampling line was tightly wr apped with aluminum tape to prevent

exposure to sunlight. The aluminum tape was also grounded; in

our experience this reduces aerosol deposition. The instrument

sensitivity was checked daily with aqueous standards. The cyclone

and the entire sample inlet line was washed weekly with deionized

water and dried with compressed nitrogen to remove any

deposited parti cles. The instrument was then calibr ated fresh with

a standard gaseous H2O2 source every week. Any calibration shift

was compensated for by linear interpolation of the calibration

within the interval.

RESULTS AND DISCUSSIONSpectral Characteristics. Figure 3 shows the spectral details

relevant to this system. Thiochrome, the oxidation product of

thiamine, absorbs in a broad band (fwhm 52 nm) centered at 373

nm, which is ideally excited by the narrow-band (fwhm 12 nm)

GaN LEDs (NSHU 590E, Nichia America Corp.) with its emission

centered at 375nm. The emission of the product occurs maximally

in a broad band (fwhm 74 nm) centered at 440 nm. The reaction

medium containing the hematin catalyst itself acts as a filter

toward stray excitation light. Doubtless this stray light filtering

will be even more efficient if the distancebetween the light source

and the receiving fiber is increased. Presently two sheets of a

colored plastic, with a transmittance window centered near the

emission maximum of thiochrome (see Figure 3) suffice as

emission filters.

Optimization of Reaction Conditions. Choice of Bu ffer and

pH . These studies were carried out with aqueous H2O2 as the

test analyte. Buffering systems based on carbonate, ammonia, and

phosphate were tested. An anmmoniacal medium provided the

highest net sensitivity but also produced a higher blank back-

ground. While the difference is not dramatic, the sensitivity was

the poorest with carbonate. We also found that, in continued use,

hematin or some insoluble product therefrom slowly deposits in

the system, including on the optical components. When ammonia

or carbonate buffer was used, a significant decrease in t he signal

over 1 day was seen. With a phosphate buffer, the decrease inthe signal over 24 h of continuous use was


6/10

Effect of H emati n Concentr ation. The r esults of this study are

shown i n Figure 4b. The fluorescence signal from H2O2 sample

injections increased with increasing hematin concentrations,

becoming constant at concentrations ofg10 M; this was used

in further work. Because of t he optical filtering of the excitation

light by hematin (Figure 3), the blank background decreases

continuously with increasing hematin concentration.

Effect of Thi ami ne Concentr ati on. For 10 M H2O2 as sample,

the fluorescence signal increased with increasing thiamine con-

centration up to 30 M, decreasing slightly thereafter and

becoming constant (Figure 5a). The blank also increased slowly

with increasing thiamine concentration. To operatei n the constant-

sensitivity region, we chose a thiamine concentration of 100 M .

Effect of Reaction Temperature. In the tested range from 24 to

40 C, the fluorescence signal due to H 2O2 injection decreased

with an increase in temperature, decreasing by 10%from 30 to 40

C. Whether this is caused by a decrease in the quantum efficiency

of thiochrome or in the catalytic activity that leads to its

production, or enhancement of the production of an alternative

product, nonfluorescent thiamine disulfide,41 is not known. For

practical purposes, the temperature of the reactor was controlled

at 30 ( 0.5 C.

Effect of Flow Rate. The liquid flow r ates control the residence/

reaction time and, as such, have a significant effect on the

observed signal (Figure 5b). Even though low flow rates increase

the net signal, the blank background is also increased and thus

increases the noise level. Low flow rates also increasethe response

time. As a compromise between system response ti me and

sensitivity, a flow rate of 80-100 L/ min (per channel) appears

to be optimum, corresponding to a total residence time of 2.5-

3.1 min.

Effect of Gas Sampli ng Rateon H2O2(g) Collection Efficiency. The

collection efficiency of diffusion-based collectors decreases with

increasing flow rate. The collection efficiencies of two nominally

(41) Puiz, T. P.; Lozano, C. M.; Tomas, V.; Ibarra, I. Talanta1992, 39, 907-

912.

Figure 4. (a) Effect of the nature of the buffer and pH on net fluorescence signal; (b) effect of hematin concentration on the net signal and

background.

Figure 5. (a) Effect of thiamine concentration on a 10 M H2O2 signal and background fluorescence; (b) net signal from a 10 M H2O2injection and background as a function of flow rate.



7/10

identical DS units (A, B) were determined over a sampling rate

range of 0.76-2.94 SLPM. (Note that our local atmospheric

pressure is 680 mmHg and the laboratory temperature is 23 C;

if the same experiments were conducted at sealevel at 20C, the

same linear sampling velocities (on which collection efficiencies

ultimately depend)42 will correspond to flow rates of 0.86-3.32

SLPM.) The observed collection efficiencies fi (from98to88%

at the highest to lowest flow r ate studied) obeyed theoretical

expectations in that l n(1 - fi) was l inearly related with the

reciprocal of the sampling rate:43

where fA and fB are fractional collection efficiencies pertaining to

DS units A and B and Q is the sampling rate in SLPM. Because

the collection efficiency is high, the overall signal changes nearly

linearly with the sampling rate. As an optimum choice between

keeping the sensitivity high and achieving nearly quantitative

collection efficiency, we selected a sampling rate of 1.6 SLPM;

this permits a collection efficiency of94%in our laboratories

and 95% at the higher atmospheric pressures in the field

experiments in Atlanta.

InterferenceStudies. Sulfur D ioxide. Sulfur dioxide is highly

water-soluble and is readily oxidized by H2O2 and, as such,

presents a significant potential negative interference for themeasurement of H2O2. Thus far, the primary effort to minimize

SO2 interference has been pH control of the collection liquid and

the addition of HCHO to the collection liquid to tie up the S(IV)

before it can be oxidized by H2O2.26,30 The HCHO concentration

in the collection liquid and the pH are both important consider-

ations. Addition of HCHO at pH levels of g6 leads to artifact

production of HMHP;30 on the other hand, at low pH, significant

(even quantitative) interference by SO2 can occur whether or not

HCHO is added.26,30 It is obviously difficult to use collector pH

variation-based schemes26 to determine different peroxide species

if significant concentrations of SO2 will be present.

The Nafion membrane DS represents an i nteresting dilemma.Since the deviceis a diffusive collector and SO2 has a significantly

smaller diffusion coefficient than H2O2, this is an advantage. Since

Nafion is a cation-exchange membrane, it excludes S(I V) or other

anions derived from it to pass through it and this may also be an

advantage. However, if the exterior surface of the membrane is

effectively wet and r eaction can take place there, then the highly

acidic nature of Nafion may actually catalyze the H 2O2-S(IV)

reaction.

The actual interference study was conducted over several days.

These results, shown in Table 1, suggest that the more favorable

of the two possibilities listed above. As shown in Table 1, the

experiments were conducted both at a constant level of H 2O2 (1

ppbv) with variable levels of SO2 (0-200 ppbv) and at a constantlevel of SO2 (150 ppbv) with 1-5 ppbv levels of H2O2. The

interference equivalent for the entire data set averages 1.43 (

0.27 pptv H2O2 loss per ppbv SO2. Although this loss is not

insignificant at low levels of H2O2 and high levels of SO2, this

performance is far better t han other collection methods; if a total

collector was used, SO2 will interfere with H2O2 on a completely

equivalent basis.

Ozone. Ozone can produce H2O2 by surface-catalyzed reactions

with water.27 Some degree of ozone interference is therefore to

be expected. Ozone concentrations between 0and 180 ppbv (n)

5, at least six points at each concentration) were studied. The

resulting data showed a linear interference by ozone ( r2) 0.99)

with an intercept statistically indistinguishable from zero. Theozone interference equivalent was determined to be

All field data were corrected with this assumed degree of ozone

interference based on simultaneously measured ozone levels. This

may r esult in slight overcorr ection in that when ozoneinterference

was measured in the presence of 2.0 ppb H2O2 and 50-100 ppbv

O3, the observed interference was less than what would be

predicted by eq 3. In any case, the overall degree of interference(42) Dasgupta, P. K.; Lindgren, P. F. Envir on. Sci. T echnol. 1989, 23, 895-897.

(43) Dasgupta, P. K. ACS Adv. Chem. Ser. 1993, No. 232, 41-90.

Table 1. Results from SO2 Interference Study

concentration, ppbv

H2O2 SO2 signal, mV %interferenceinterference equiv

pptv H2O2/ ppbv SO2

1.0 0 441.3( 2.0 control1.0 25 426.1( 6.1 -0.034 -1.381.0 50 400.0( 2.0 -0.094 -1.871.0 75 381.7( 3.9 -0.135 -1.351.0 150 347.4( 1.6 -0.213 -1.421.0 200 334.5( 1.5 -0.242 -1.211.0 0 434.7( 8.8 control

1.0 150 366.4( 2.4 -0.157 -1.052.0 0 818.3( 11.3 control2.0 150 751.7( 5.0 -0.081 -1.093.0 0 1219.4( 1.7 control3.0 150 1125.5( 10.4 -0.077 -1.544.0 0 1685.0( 13.8 control4.0 150 1576.5( 8.6 -0.064 -1.725.0 0 2200.7( 36.3 control5.0 150 2093.5( 25.9 -0.049 -1.62

ln(1- fA) ) -(1.8276( 0.0543)/ Q- 1.4909 ( 0.04298,

r2) 0.9965 (1)

ln(1- fB) ) -(1.9294( 0.0613)/ Q- 1.4990 ( 0.04853,

r2) 0.9960 (2)

1 ppb ozone) 2.37 ppt H2O2 (3)



8/10

is not very high and is generally ameliorated by the fact that

concentrations of H2O2 are also typically higher duri ng high ozone

episodes.

M ethyl H ydroperoxide. With substrates other than thiamine it

has been previously observed that hematin is significantly more

selective for H2O2 than for alkylhydroperoxides.14 Potential

interference from MHP was studied both in the solution phase

and in the gas phase for the thiamine-hematin system. When a

standardized M HP solution (20.3and 40.6M), from which H2O2

had been freshly removed by passing through a MnO2 column,was injected in the analytical system, the response was found to

be 4.6( 0.1 and 5.3( 0.1%, respectively, of the same concentration

of H2O2. Additive tests indicated an even lower degree of

interference. For 2 M H2O2, the addition of 50.8 M M HP led to

5.5 ( 0.1%positive error in the determination. The same experi-

ment with 4M H2O2 showed that the addition of 101.6M MHP

resulted in a positive error of 9.5( 0.2%. The difference with the

M HP only experiments is likely due to the preferential binding

of H2O2 to hematin to produce the active intermediate. Such a

process would leave the catalyst unavailable for M HP and lead to

even lower interference than pure MHP experiments.

For the gas phase, the best-fit linear regression equations for

the response behavior were as follows over the ranges of 0-10.0ppbv H2O2 and MHP individually (5 concentration points each,

g3 measurements at each concentration):

The response of gas-phase MHP is thus only 1%of that of thesame concentration of H2O2. Thisis due to thelower Henrys law

solubility and diffusion coefficient of MHP relative to H 2O2 and

hence a lower collection efficiency (vide infra). Because of the

very low potential interference from MH P, mixed gaseous

standards of H2O2 and MHP, which for reasons outlined above

are likely to lead to an even lower interference index for MHP,

were not tested.

If HRP is substituted for hematin in the reaction system,

significant responses for MHP can be observed. The collection

efficiency of the present DS (water as scrubber liquid) to MHP

was assessed with a HRP( 32 purpurogallin units/ mL) -thiamine

(1 mg/ mL) reaction system with a reaction pH of 9.0 (phosphate

buffer). Both DS units show essentially identical behavior, exhibit-

ing a fractional collection efficiency of 0.318, 0.284, 0.275, 0.265,

and 0.214 at flow rates of 0.76, 1.07, 1.36, 1.74, and 2.40 L/ min,

respectively. The fit of these data to a diffusion-controlled collec-

tion equation (e.g., eq 1) is very poor (r2) 0.83), and if a diffusion

controlled collection regime is assumed nevertheless, the best-fit

diffusion coefficient is calculated to be a factor of 15 less than

that of H2O2; thi s is clearly impossible. The collection of M HP is

therefore controlled by factors other than diffusive mass transport,

notably the sink efficiency of the collection liquid and the solubili ty

of MHP in water. It is possible that a different collection liquid

(e.g., containing an alcohol or a surfactant) may improvecollectionefficiency for M HP; however, compatibili ty with the DS and the

analysis system must also be ensured.

Hydroxymethyl Hydroperoxide. In all of the following, by HM HP,

we mean a mixture that contains mostly HMHP. According to

our analysis and the literature,37 our best estimate for the compo-

sition of this mixtur e, the molar ratio of BHM HP:HM HP:H2O2 is

1:7.5:2.6. In terms of its aqueoussolution behavior, when a solution

of HMHP (or BHM HP) i s injected into a strongly alkalineaqueous

medium like the present analysis system, it should hydrolyze to

H2O2 very rapidly. When acrude HM HP mixture is freshly passed

through an MnO2 bed and the resulting effluent i s i mmediately

analyzed by injection into the present analysis system and by

iodometric analysis, the fluorescence results (tr eated on the basis

of a H2O2 calibration plot) account for 92-94% of the total

peroxides determined by iodometry. Given the difficulties of

accurate calibr ationsof mixtur es of different peroxides, we believe

these results largely confirm the theoretical expectation that, in

the solution phase, the system r esponds to any HM HP in the same

way as H2O2.

Regarding the behavior of gas-phase HMHP, in contrast to

MHP, the dependence of HMHP collection efficiency upon flow

rate does closely follow a diffusion-limited collection behavior (flow

rate 0. 76-2.4 L/ min, collection efficiency 88.5-55.7%):

In as much as HMHP displays higher solubility than H 2O2 (the

Henrys law coefficient of BHMHP is even higher4), it is reason-

able to assume that, like HM HP, it too has very high Henrys law

solubility), diffusion-limited collection behavior is expected. Fur-

ther, the ratio of the slope in eq 4 to the mean slope in eqs 1 and

2 is 0.79. Nominally, this is the ratio of the diffusion coefficient of

HM HP to H2O2. We can estimate the diffusion coefficient of

HM HP on the basis of t he weighted average of Grahams l aw-

H2O2 (peak height, A/ D count) )

3102( 53[H2O2, ppbv] - (148.5( 35.0),

r2) 0.9991 (4)

M HP (peak height, A/ D count) )

37.8( 1.3 [ M HP, ppbv] + 44.4( 8.6,

r2) 0.9967 (5)

Figure 6. Typical system output, gas-phase samples; concentrationas indicated; 2-min sample, 8-min zero.

ln(1- f) ) -(1.4863( 0.0774)/ Q- 0.2352 ( 0.0643,

r2) 0.9946 (6)



9/10

based diffusion coefficients of BHMHP, HMHP, and H2O2. On

the basis of the molecular weight of the individual compounds

and the molar ratio of the different compounds in the mixture,

the diffusion coefficient of HMHP is computed to be 0.80 relative

to H2O2 being unity. This is obviously in excellent agreement with

the experi mental observations.

Analytical Performance for Hydrogen Peroxide. For the

liquid-phase system, the observed fluorescence signal was linear

with H2O2 concentration in the tested range of 0-100 M :

Calibration linearity data for gas-phase H2O2 has already been

presented in the previous section. A typical chart output for system

response to 0-1ppbv H2O2(g) is shown in Figure 6. The limits of

detection (S/ N ) 3) for gaseous and aqueous H2O2 were

computed to be 13.5 pptv and 11 nM, respectively. The relative

standard deviation for the determination of 1 ppbv H 2O2(g) and 1

M H2O2(aq) were 1.8 and 1.4%, respectively (n) 7 each).

Field Study Results. We cannot obviously differentiate

between H2O2 and HMH P at this time. However, recent evidence

based on chromatographic analysis suggests that H2O2 and MHPare the only two important atmospheric peroxides;24,44 as such,

we cannot be in gross error in assuming what we measure to be

H2O2.TheH2O2 concentrations measured during the study ranged

from less than 200 pptv (usually shortly after r ainfall) to 4.5 ppbv

(dur ing a day when the ozone concentration exceeded 140 ppbv).

A portion of the H2O2 results is shown in Figure 7. Detailed results

wil l appear i n the NARSTO database.45 The characteristic diurnal

pattern for H2O2 is readily apparent in Figure 7.

Figure 8 shows a trace of maximum hourly ozone concentra-

tions (taken among the several measurement sites located in

metropolitan Atlanta)36 each day and the daily maximum in H 2O2.

There is obviously a striking similarity in these temporal patterns.

A detailed rationale is beyond the scope of this paper. Figure 9

shows a shorter span vignette of H2O2, SO2, and O3 data, along

with rainfall, during a mild rain episode. During August 8, the

H2O2, SO2, and O3 all peak at the same time, suggesting a common

plume source impacting the sampling location. Fossil fuel com-

bustion sources often result in the quixotic combination of both

elevated SO2 and oxidant levels; we have made similar observa-

tions in airborne monitoring of Ships plumes.46 Interestingly, rain

washes out H2O2 and SO2, the two water-soluble gases, quite(44) http:/ / euros.gso.uri.edu/ snow/ instrumentschem.html.

(45) http:/ / www.cgenv.com/ Narsto/ .

Figure 7. Hydrogen peroxide concentrations measured in Atlanta, GA, August 1999 In this and the subsequent figures, by hydrogen peroxidewe mean the sum of H2O2 and HMHP.

peak height, A/ D count )

2696( 25 [H2O2, M ] -9.2( 15.2,r

2) 0.9996 (7)

Figure 8. Daily ozone maximum (hourly maximum measured

among any of several metropolitan Atlanta measurement sites36) and

daily H2O2 maximum measured at the Supersite experimental site.



10/10

effectively, and ozone is removed to a much smaller degree. In

fact, ozone concentrations go up with the second burst of showers,

which was accompanied by significant lightning activity. OnAugust 9, SO2 and ozone both reach higher levels than the

previous day but H2O2 levels remain generally low and generally

anticorrelated to SO2 concentrations, as may happen when the

air mass has been processed at least partially through a cloud.

During a total deployment period of 39,300 min, the instrument

was collecting data 96.88%of the total time; calibration required

1.54%and reagent replacement/ cleaning requir ed 1.27%of the total

time. The instrument was inoperative 0.31%of the total time, due

to power failure resulting from rainwater running down the

exterior of the sampling conduit and shorti ng out a power socket.

The compact size of the instrument, largely due to the unique

fluorescence detector, (entire instrument is buil t within a tower-

style PC chassis) makes for facile fi eld deployment. The zero air

self-purification design was also particularly useful in field ap-

plications in that it obviated the need for compressed gas cylinders

altogether. It was found that the activated carbon column (1.5 cm

15 cm) still removes H2O2 quantitatively after more than 1

month of continuous use.

Future Work. It is not our intention to suggest that th is type

of analysis will replace chromatographic analysis of peroxides.

However, the present instruments are capable of simpler alterna-

tives. If indeed it is sufficient to determine H 2O2 and MHP, thiscan be accomplished by two parallel systems in which hematin

and HRP are respectively used and lead to signals that are

proportional to [H2O2] and [H2O2] + x[MHP], where x is of the

order of 0.2. It may be possible to increase the value of x, by

enhancing the collection of M HP thr ough the incorporation of a

modifier, e.g., a surfactant or an organic solvent in the scrubber

liquid. If HMH P concentrations ar eimportant, then both systems

will be measuring [H2O2] + y[HMHP] instead of [H2O2], where

yis of the order of 0.8. In this case, a third channel can be operated

with a ir on phthalocyanine catalyst that can work in an acid

solution15 and should thus providea method of determining H2O2individually without significant reaction of HMHP. However, the

attractiveness of dir ect chromatographic analysis obviously in-creases as the number of components increase.

ACKNOWLEDGMENTThis research was funded in part by the SOS/ Supersite

Research Program of the USEPA. The manuscript was not subject

to review by any of these agencies and no endorsement should

be inferred. We are grateful to W. L. Chameides, S. V. Hering,

and C. S. Kiang for asking us to be a part of this study. The help

of E. Edgerton, Jefferson street marshall, before, during, and after

the study, is gratefully acknowledged.

Received for review May 31, 2000. Accepted August 28,2000.

AC000611+(46) Genfa, Z.; Dasgupta, P. K.; Frick, G. M.; Hoppel, W. A. M icr ochem. J. 199,

62, 99-113.

Figure 9. H2O2, ozone, SO2, and rainfall data during a 2-day periodin Atlanta, GA, August 1999.


Measurement of Atmospheric Hydrogen Peroxide

Documents

Transcript of Measurement of Atmospheric Hydrogen Peroxide