PAPER www.rsc.org/jem | Journal of Environmental Monitoring

Publ

ishe

d on

08

Apr

il 20

10. D

ownl

oade

d by

Uni

vers

ity o

f W

arsa

w o

n 22

/10/

2014

12:

56:0

0.

View Article Online / Journal Homepage / Table of Contents for this issue

Development of a sensitive passive sampler using indigotrisulfonate for thedetermination of tropospheric ozone

Gabriel Garcia, Andrew George Allen and Arnaldo Alves Cardoso*

Received 12th October 2009, Accepted 15th March 2010

First published as an Advance Article on the web 8th April 2010

DOI: 10.1039/b920254d

A new sampling and analytical design for measurement of ambient ozone is presented. The procedure is

based on ozone absorption and decoloration (at 600 nm) of indigotrisulfonate dye, where ozone adds

itself across the carbon–carbon double bond of the indigo. A mean relative standard deviation of 8.6%

was obtained using samplers exposed in triplicate, and a correlation coefficient (r) of 0.957 was achieved

in parallel measurements using the samplers and a commercial UV ozone instrument. The devices were

evaluated in a measurement campaign, mapping spatial and temporal trends of ozone concentrations in

a region of southeast Brazil strongly influenced by seasonal agricultural biomass burning, with

associated emissions of ozone precursors. Ozone concentrations were highest in rural areas and lowest

at an urban site, due to formation during downwind transport and short-term depletion due to titration

with nitric oxide. Ozone concentrations showed strong seasonal trends, due to the influences of

precursor emissions, relative humidity and solar radiation intensity. Advantages of the technique

include ease and speed of use, the ready availability of components, and excellent sensitivity.

Achievable temporal resolution of ozone concentrations is 8 hours at an ambient ozone concentration

of 3.8 ppb, or 2 hours at a concentration of 15.2 ppb.

1 Introduction

Air pollution characterized by formation of ozone, produced

during NO2 photolysis at wavelengths <400 nm, is a common

urban and regional phenomenon in many parts of the world.1,2

Background concentrations of ozone appear to have been

increasing since the industrial revolution. An analysis of ozone

concentrations measured at remote European sites, from 1969 to

1989, indicates an average 1 to 2% annual increase.3 In outdoor

ambient air, the increase is due to increased traffic4 and conse-

quent emissions of NOx and volatile organic compounds.

Indoors (for example, in offices), ozone is produced during use of

laser printers and photocopiers.5,6 Outdoor ozone levels, which

are low in the early morning, increase significantly until around

noon and decrease thereafter, however ozone peaks can also

occur during the afternoon or early evening, due to transport

effects and mixing between different layers of the troposphere.7

Analytical Chemistry Department, Chemistry Institute, Sao Paulo StateUniversity, CP 355, CEP 14800-900, Brazil. E-mail: acardoso@iq.unesp.br; Fax: +5516 33016692; Tel: +5516 33016612

Environmental impact

Ambient ozone plays a key role in oxidation processes, air quality an

ozone have increased since the industrial revolution. Knowledge of

that control its tropospheric behavior, and in assessing progress

extensively used to this end, with advantage over active sampli

measurements during periods as short as a few hours, permitting

design includes a large cross-sectional area and a short diffusion pa

commercially available and inexpensive components, enabling rapi

This journal is ª The Royal Society of Chemistry 2010

The National Ambient Air Quality Standards for Ground-level

Ozone, established by the US EPA, stipulates a primary ozone

standard of 75 ppb (3 year average of the fourth-highest daily

maximum 8 hour average ozone concentration over a year). This

standard is currently (March 2010) under review.8 The WHO Air

Quality Guidelines 2005 Global Update sets the guideline value

for ozone at 100 mg m�3 (about 47 ppb) for a daily maximum

8 hour mean.9 In Brazil, the ozone standard established by

IBAMA (the Brazilian Environmental Agency) is 75 ppb, with no

more than one annual exceedance.10 When present at high

concentrations, ozone causes a range of adverse environmental

impacts on human health, crops, natural vegetation and outdoor

materials.11–15 It is ranked as the third most important greenhouse

gas, after carbon dioxide and methane.11 Increasing concentra-

tions in urban areas raise the risks to human health, materials and

cultural artifacts, justifying the development of new methods to

measure ozone in outdoor and indoor environments.

For assessments of environmental quality, long-term obser-

vations are often needed to obtain an accurate perspective of

pollutant behaviour in a given area. To this end, diffusive

(passive) samplers have been extensively used, with advantage

d atmospheric radiation transfer. Background concentrations of

ozone concentrations is essential in understanding the processes

in management of this pollutant. Passive samplers have been

ng. This article introduces a new sampler/sorbent for ozone

characterization of diurnal concentration trends. The sampler

th length. In contrast to other designs, this sampler is based on

d assembly and deployment.

J. Environ. Monit., 2010, 12, 1325–1329 | 1325

Publ

ishe

d on

08

Apr

il 20

10. D

ownl

oade

d by

Uni

vers

ity o

f W

arsa

w o

n 22

/10/

2014

12:

56:0

0.

View Article Online

over active sampling. These devices were originally developed in

the 1970s for personal exposure monitoring purposes in indus-

trial and indoor environments,12 and were later also used in

studies of open atmospheres. The most obvious advantages of

diffusive samplers include small size and weight, low cost,

minimal maintenance, and lack of any need for an air pump. A

disadvantage is that they are normally unsuitable for the deter-

mination of short-term pollutant concentration fluctuations. The

concentration of the analyte is integrated over the entire expo-

sure time, so that if the sampling time is longer than one day, it is

not feasible to measure daily variations of pollutant concentra-

tions. This can be critical for ozone, since concentrations can

fluctuate widely on a daily time scale, due to the influences of

changing climatic conditions or emissions of precursors.

Passive samplers employ controlled gaseous diffusion as

a means of analyte collection. The process is described by an

expression derived from Fick’s first law of diffusion:

mA ¼ A � Dx(C0,x � Cx)t/z (1)

where mA is the mass of analyte collected during time t, A the

cross-sectional area of the sampler, D the molecular diffusion

coefficient of gas x, z the diffusion path length, measured

between the sampler open end and the sorbent bed, Cx the gas x

concentration to be measured, and C0,x the gas x concentration

at the surface of the sorbent. For an ideal sorbent, C0,x is equal to

zero. The performance of the sampler depends on its geometry,

sorbent efficiency, and the diffusion coefficient (D). The sorbent

material should have sufficient capacity for collection of the

target pollutant over the desired time period and at the concen-

trations encountered, without overloading, and be sufficiently

stable to avoid any significant degradation (except due to reac-

tion with the target pollutant(s)) during the time period between

preparation and final analysis, taking account of the time

required to distribute samplers to field sites and for subsequent

retrieval.13 Samplers often show non-ideal behavior in relation to

Fick’s law. Pitombo and Cardoso14 observed that, under condi-

tions of high wind speeds using open-ended tubes, gas turbulence

within the sampler changes the effective diffusion path length,

due to disruption of the internal stagnant air layer. Conversely,

under conditions of low wind speed, external gas molecules close

to the end of a diffusion tube may not possess sufficient velocity

to replace the molecules that have already entered the sampler, so

that the effective diffusion path length may be increased.15

The use of tubes possessing small cross-sectional areas and

long diffusion path lengths minimizes the turbulence effect, but

increases the residence time of the gas inside the tube, conse-

quently increasing sampling time as well as the possibility of gas

phase reactions, for example between ozone and nitric oxide, that

can lead to artifacts.16 Since the materials used in construction of

samplers are usually not transparent to UV light, the photo-

stationary equilibrium is affected and the new condition is not

Fig. 1 Reaction of ozone w

1326 | J. Environ. Monit., 2010, 12, 1325–1329

representative of ambient air. This is another possible source of

error in ozone measurements.16

A variety of diffusive samplers designed to minimize the effect

of gas turbulence have been reported. Porous polyethylene

membranes have been used to cover the sampler inlet,17,18 so that

within the sampler convective movement is minimized and

transport is mainly due to molecular diffusion, in accordance

with Fick’s laws.19

This article introduces a new sampler/sorbent combination for

sensitive ozone measurements during measurement periods of as

little as a few hours, hence permitting characterization of diurnal

concentration trends. The passive sampler design includes a large

cross-sectional area, a short diffusion path length, and a porous

Teflon membrane filter at the inlet. In contrast to other recently

reported designs, this sampler is based on commercially available

and inexpensive components, enabling rapid assembly and

deployment. For ozone, the sorbent used was indigotrisulfonate,

retained in impregnated cellulose filters.20,21 In the reaction with

the sorbent (Fig. 1), ozone adds itself across the carbon–carbon

double bond of the sulfonated indigo dye. The concentration of

ozone is determined from the degree of decoloration of the indigo

reagent.

The new passive sampler was tested in an environmental

application, where outdoor ozone concentrations were mapped

and monitored using a meso-scale network of sites in the vicinity

of Araraquara city, Sao Paulo State, Brazil. This is a region of

intensive industrial-scale agriculture, where biomass burning

(during the sugar cane harvest) generates large quantities of the

precursors (NOx and hydrocarbons) necessary for ozone

production.

2 Experimental and results section

2.1 Materials and methods

2.1.1 Ozone sampler construction. The passive sampler is

illustrated schematically in Fig. 2. Components of a 37 mm

polycarbonate filter holder (Millipore catalogue no.

M000037A0) were used as both sampler body and end caps. The

PTFE membrane filters employed as turbulence barriers were

0.45 mm pore size, 175 mm thickness and 37 mm diameter (Mil-

lipore catalogue no. FHP 03700). The diffusion path length was

9.0 mm. Sorbent supports were Whatman No. 41 cellulose filters.

A major advantage of this sampler is its ease of use and the ready

availability of all components which, with the exception of the

absorbent, can be recycled and reused many times.

2.1.2 Coating solution. Reagent grade chemicals were used

throughout this work. Deionized water (18.2 MU cm), produced

by a MilliQ system (Millipore, Bedford, MA), was used to

prepare all solutions. For the coating solution, the indigo reagent

was prepared by adding 12.4 mg of potassium indigotrisulfonate

ith indigotrisulfonate.

This journal is ª The Royal Society of Chemistry 2010

Fig. 2 Schematic diagram of the passive sampler: (a) dissembled

sampler, C ¼ inlet end cap, T ¼ Teflon membrane filter, S ¼ cellulose

filter impregnated with indigotrisulfonate; (b) assembled sampler with

inlet end cap removed, D ¼ sampler diameter ¼ 37 mm, P ¼ diffusion

path length ¼ 9 mm.

Publ

ishe

d on

08

Apr

il 20

10. D

ownl

oade

d by

Uni

vers

ity o

f W

arsa

w o

n 22

/10/

2014

12:

56:0

0.

View Article Online

(Aldrich) to a 10 mL volumetric flask containing 5 mL of

ethylene glycol, stirring the mixture, and diluting to 10 mL. The

working solution was prepared daily.

2.1.3 Preparation of the sorbent surface. Cellulose filters

(Whatman No. 41) were cut into 37 mm diameter circles. Each

filter was impregnated using an 80 mL aliquot of indigo solution

(2.0 � 10�3 mol L�1), applied dropwise to the center of the filter,

with the solution spreading throughout the filter by capillary

action. During impregnation, the filters were supported on small

plastic rods to avoid any loss of reagent.

2.1.4 Experimental protocol. For each individual ozone

measurement, three passive samplers and one field blank were

used. The field blank was a passive sampler maintained sealed

during the sampling period, and subsequently processed using

identical techniques as for the exposed samplers. The following

experimental protocol was used: (a) the filter impregnated with

indigotrisulfonate was loaded into the sampler; (b) the passive

sampler was exposed in the field for 8 hours; (c) after sampling,

the filters were extracted in situ by shaking with sequential

Table 1 Recoveries for the solubilization procedure

Added volume/mL

Filter papers

Average absorbance SD

40 0.118 9.81 � 10�4

80 0.234 1.53 � 10�3

This journal is ª The Royal Society of Chemistry 2010

aliquots of approximately 1.0 mL of water. The extract solution

was collected in a 10.0 mL volumetric flask, through the central

opening at the sampler base. This was carried out at least nine

times to ensure complete dissolution and total removal of ana-

lytes; (d) the absorbance of the solution was measured using

a Hitachi U-2000 spectrophotometer, operated at 600 nm; (e) the

initial indigo amount was calculated by reproducing steps (c) to

(d) using blank impregnated filters.

The concentration of ozone was determined from the degree of

decoloration of the indigo reagent. The analytical signal was

obtained using Ai � Af, where Ai is the blank filter absorbance

and Af is the mean of the absorbances for the three replicate

filters.

2.1.5 Indigo analytical curve. The analytical curve describing

the influence of dye concentration on the absorbance signal was

determined using solutions of indigo at concentrations varying

from 2.59 to 20.7 mg L�1. The absorbance signal was linear over

this concentration range:

A ¼ 14.384[Ind] + 0.003 (R ¼ 0.99997) (2)

where A represents absorbance and [Ind] the indigo concentra-

tion (mg L�1). The limit of detection (LOD) for indigo, considered

as 3 times the blank signal, was better than 0.46 mg L�1.

2.1.6 Stability of indigo working solution. The initial working

solution showed low temporal stability, of about one day, so

further experiments were undertaken to formulate a more stable

solution. 50 mL of the working standard (pH 4.8) were prepared,

and divided into three parts. To one part, formic acid was added

until pH 1.9 was reached, and to another part citric acid was

added until pH 2.1 was achieved. The solutions were stored in

a refrigerator. An aliquot of 80 mL of each solution was peri-

odically removed, placed in a 10 mL volumetric flask, and the

volume made up to 10 mL with water. The absorbances of the

solutions were measured during a period of five days. Best results

were obtained using citric acid, with the indigo reagent showing

less than 10% degradation after �70 hours. Although the solu-

tions with preservative were more stable, it was more difficult to

impregnate the paper with the dye, and to solubilize the dye after

sampling. The dye solution used in subsequent experiments was

therefore prepared without the addition of preservative.

2.1.7 Efficiency of the solubilization procedure. The analytical

signal was measured as the difference between the initial absor-

bance (Ai) and the final absorbance (Af) of solutions obtained by

dissolution of the dye impregnated in the filters. Recovery

experiments were performed to determine the efficiency of solu-

bilization of the dye. Two sets of six cellulose filters were

Direct addition

Recovery (%)Average absorbance SD

0.116 1.31 � 10�3 98.30.234 8.31 � 10�4 100

J. Environ. Monit., 2010, 12, 1325–1329 | 1327

Publ

ishe

d on

08

Apr

il 20

10. D

ownl

oade

d by

Uni

vers

ity o

f W

arsa

w o

n 22

/10/

2014

12:

56:0

0.

View Article Online

impregnated with either 80 mL or 40 mL of 2.0 � 10�3 mol L�1

indigo solution. The filters were extracted, and the solution

absorbances measured. Recovery percentages (Table 1) were

determined by comparing the absorbances with those of indigo

solutions prepared by addition of the same volumes of dye

directly into volumetric flasks.

2.2 Field validation

2.2.1 Calibration of passive samplers in ambient air. Cali-

bration was achieved during ambient sampling outdoors at the

Institute of Chemistry in Araraquara, central Sao Paulo State,

Brazil, from March to December 2008 (simultaneous measure-

ments using the passive samplers were also made at three other

sites, as described in Section 2.3.1). Over this period the relative

humidity varied between 21% and 99%, and the temperature

between 10.5 �C and 36 �C. A set of four passive samplers (three

exposed and one blank) were installed under a plastic shield (cut

from a polycarbonate bottle), to provide protection from rain

while maintaining free ventilation (Fig. 3). The inlet of

a commercial UV photometric ozone analyzer (Model 49C,

Thermo Environmental Instruments Inc.) was positioned

Fig. 3 Arrangement of samplers during sampling (3 exposed samplers, 1

blank).

Fig. 4 Results of passive sampler calibration, conducted for 8 hour periods

(9 a.m. to 5 p.m.) in ambient air, between March andDecember 2008. Range

of temperature: 10–36 �C. Range of relative humidity: 20–90%.

1328 | J. Environ. Monit., 2010, 12, 1325–1329

adjacent to the passive samplers, and measurements in parallel

conducted for 8 hour periods (from 9 a.m. to 5 p.m). The mean of

the three replicate measurements was calculated for each

sampling period. The average relative standard deviation of the

three samplers was 8.6%.

The average value obtained from the three replicate diffusive

samplers was plotted against the concentration of ozone

obtained using the UV photometric analyzer (Fig. 4). A linear

relationship was obtained, described by:

[O3]pass ¼ 0.88[O3]UV + 1.66 (r ¼ 0.957) (3)

where [O3]pass is the ozone concentration measured using the

passive samplers, and [O3]UV that obtained from the photometric

ozone analyzer. Significant correlation was obtained between the

two methods, with an angular coefficient of 0.88.

2.2.2 Limit of detection (LOD). The limit of detection

(LOD), the amount of ozone that the passive sampler is capable

of detecting, considered to be 3 times the blank signal, was equal

to 3.8 ppb for a sampling period of 8 hours (equivalent to

15.2 ppb for a 2 hour sampling period).

2.3 Environmental application

2.3.1 Mapping and monitoring of outdoor ozone concentra-

tions. The potential application of the proposed sampler for

environmental studies was evaluated during measurement

campaigns employing sampling sites in the vicinity of the city of

Araraquara. Here, the objective was to be able to successfully

map and monitor ambient ozone levels, identifying any meteo-

rological, seasonal, or anthropogenic influences on tropospheric

ozone formation in this region of Brazil.

Measurements were made between March and December

2008, during the four seasons of the year, under variable condi-

tions of humidity and temperature. Four sampling sites were

selected to represent different environments (Fig. 5). Site 1

(urban) was situated in the grounds of a house in the city centre.

Site 2 (suburban) was on the Institute of Chemistry campus,

downwind of the urban center of Araraquara and with a number

of trunk and minor roads in the surrounding area. At this site,

the passive sampler measurements were conducted in parallel

with the UV photometric ozone analyzer. Comparison of the

results from the active and passive techniques is shown in Fig. 4.

Fig. 5 Map showing the four sampling sites and the urban perimeter of

Araraquara.

This journal is ª The Royal Society of Chemistry 2010

Fig. 6 Results of ambient ozone measurements at four sites (1 ¼ urban,

2 ¼ suburban, 3 ¼ semi-rural, 4 ¼ rural). Samples collected over 8 hour

periods (9 a.m. to 5 p.m.) between March and December 2008.

Publ

ishe

d on

08

Apr

il 20

10. D

ownl

oade

d by

Uni

vers

ity o

f W

arsa

w o

n 22

/10/

2014

12:

56:0

0.

View Article Online

Site 3 (semi-rural) was located in the grounds of a municipal

water reservoir (Ribeirao das Cruzes), at the northern perimeter

of Araraquara, upwind of the city during prevailing winds. Site 4

(rural) was situated in the grounds of the Sao Paulo State

University (UNESP) campus, in an agricultural region (pasture

and sugar cane plantations) southwest and downwind of the city,

with only minor local vehicular traffic movement.

The results obtained (Fig. 6) allow two main observations to

be made. On 18 out of 21 measurement days (>85%) ozone

concentrations at the more remote sites (3 and 4) were higher

than in the central urban region (site 1). This can be explained by

two possible chemical mechanisms. Emissions of ozone precur-

sors are likely to be intense in the urban centre, however it has

been shown that ozone concentrations tend to peak as an air

mass moves downwind from high emission regions (such as

cities) and NOx becomes limiting.22 In addition, the urban center

is a strong source of nitric oxide (NO), so that under non-steady

state conditions ozone can be lost by rapid titration with nitric

oxide (NO).

The second observation is that the variation of the ozone

concentration over the year is very similar for all sampling sites.

It is therefore clear that seasonal factors, such as magnitude of

precursor emissions, relative humidity and intensity of incident

solar radiation, strongly influence the formation of tropospheric

ozone in this region. These results also further demonstrate the

repeatability of results, and hence satisfactory design of the

sampler as well as the analytical procedure.

3 Conclusions

A new passive sampler for determination of tropospheric ozone,

based on reaction of ozone with potassium indigotrisulfonate,

has been successfully developed and field trialled. The detection

This journal is ª The Royal Society of Chemistry 2010

limit of the device enables a temporal resolution of better than

8 hours at an ozone concentration of 3.8 ppb (or as little as

2 hours at a concentration of 15.2 ppb), so that it can be deployed

in both polluted and background continental air masses.

Measurements in parallel with a commercial ozone analyzer

demonstrated excellent accuracy and absence of artifacts. The

samplers were validated in field experiments at a network of sites

in a rural region of southeast Brazil, from which it was possible

to map the spatial variability of ozone concentrations, charac-

terize seasonal trends, and interpret atmospheric ozone forma-

tion and loss mechanisms.

Acknowledgements

The authors acknowledge the financial support of CNPq (Con-

selho Nacional de Desenvolvimento Cientıfico e Tecnol�ogico)

and FAPESP (Fundacao de Amparo �a Pesquisa do Estado de

Sao Paulo) (process no. 2009/07415-6).

References

1 B. J. Finlayson-Pitts and J. N. Pitts, Jr, Chemistry of the Upper andLower Atmosphere: Theory, Experiments and Applications,Academic Press, San Diego, 2000, pp. 1–13.

2 D. Kley, M. Kleinmann, H. Sanderman and S. Krupa, Environ.Pollut., 1999, 100, 19–42.

3 W. E. Janach, J. Geophys. Res., 1989, 94, 18289–18295.4 I. Colbeck; A. R. Mackenzie, in Air Pollution by Photochemical

Oxidants, Elsevier Science Ltd, Amsterdam, 1994, pp. 58–79.5 M. R. Cox, Environ. Pollut., 2003, 126, 301–311.6 S. Caballero, N. Galindo, C. Pastor, M. Varea and J. Crespo, Atmos.

Environ., 2007, 41, 2881–2886.7 T. R. McCurdy, in Tropospheric Ozone, Human Health and

Agricultural Impacts, ed. D. J. McKee, CRC Press, Boca Raton,1994, pp. 19–37.

8 EPA: http//www.epa.gov/groundlevelozone/.9 WHO Regional Office for Europe, Air Quality Guidelines, Global

Update 2005, World Health Organisation, Copenhagen, 2006.10 IBAMA: http://www.ibama.gov.br/licenciamento/index.php.11 R. Derwent, W. Collins, C. Johnson and D. Stevenson, Environ. Sci.

Technol., 2002, 36, 379–382.12 E. D. Palmes, A. F. Gunnison, J. Dimattio and C. Tomczyk, Am. Ind.

Hyg. Assoc. J., 1976, 37, 570–577.13 E. D. Palmes, Environ. Int., 1981, 5, 97–100.14 L. R. M. Pitombo and A. A. Cardoso, Int. J. Environ. Anal. Chem.,

1990, 39, 349–360.15 R. H. Brown, J. Environ. Monit., 2000, 2, 1–9.16 R. M. Heal and J. N. Cape, Atmos. Environ., 1997, 31, 1911–1923.17 Y. Sekine, C. Hirota and M. Butsugan, J. Environ. Chem., 2002, 12,

847–854.18 Y. Sekine, S. F. Watts, A. Rendell and M. Butsugan, Atmos. Environ.,

2008, 42(18), 4079–4088.19 S. Seethapathy, T. G�orecki and X. Li, J. Chromatogr., A, 2008, 1184,

234–253.20 E. P. Felix, K. A. D. de Souza, C. M. Dias and A. A. Cardoso,

J. AOAC Int., 2006, 89, 480–485.21 B. A. Scheeren and E. H. Adema, Water, Air, Soil Pollut., 1996, 91,

335–350.22 B. N. Duncan and W. L. Chameides, J. Geophys. Res., 1998, 103,

28159–28179.

J. Environ. Monit., 2010, 12, 1325–1329 | 1329