Talanta 58 (2002) 995 /1001

www.elsevier.com/locate/talanta

Determination of residual chlorine in greywater using

o-tolidine
J.G. March *, M. Gual, B.M. Simonet
Department of Chemistry, University of Balearic Islands, 07071 Palma de Mallorca, Spain

Received 17 April 2002; received in revised form 16 July 2002; accepted 19 August 2002

Abstract

The determination of chlorine with o -tolidine in greywater has been studied, and a batch method and a sequential
injection method have been proposed. It was found that the reaction of o -tolidine with chlorine was slower in a
greywater matrix. Grey water samples must be filtered before analysis, or alternatively, a blank of sample must be
measured. The samples are very unstable, and after 2 h of storage in dark conditions at 4 8C the chlorine concentration
can diminish. The standard addition method, in some samples gave unsatisfactory results as a consequence of matrix
effects. It was concluded that samples with a total organic carbon (TOC) higher than 60 mgC l
1 can not be accurately
analysed using these methods. However, samples with a TOC lower than 60 mgC l
1 gave 1009/3% recoveries. The
linear range of the methods was 0 /3.0 and 0 /5.0 mgCl2 l
1, and the limit of quantification 0.2 and 0.5 mgCl2 l
1, for
the batch method and the sequential injection method, respectively. This is the first paper devoted to grey water from
the analytical chemistry point of view.
# 2002 Elsevier Science B.V. All rights reserved.

Keywords: Determination of chlorine; Greywater; Sequential injection; o -Tolidine

1. Introduction A useful strategy aimed at decreasing water

The Mediterranean coast of Spain, like many
other parts of the world, suffers from persistent
drought. This fact, together with the increase in
water consumption has led to a limited availability
of hydric resources that demand an improvement
in water management.
* Corresponding author. Tel.: /34-971-172405; fax: /34-
971-173426
E-mail address: joan.march@uib.es (J.G. March).
0039-9140/02/$ - see front matter # 2002 Elsevier Science B.V. All rights reserved.
PII: S 0 0 3 9 - 9 1 4 0 ( 0 2 ) 0 0 4 5 3 - 8
consumption involves: (a) the separation of waste-
water into two flows ‘greywater’ (washing water
from bathroom and laundry), from ‘blackwater’
(faecal water), (b) a simple greywater treatment (in
the same building where it is produced) involving
filtration, sedimentation and disinfection, and (c)
reuse of such water for non potable uses such as
toilet flushing.
Such a strategy has been assayed in familiar
urban residential properties [1] and also in high
rise buildings [2]. At present, we are involved in a
project focused on the study of greywater reuse in
a hotel.
996 J.G. March et al. / Talanta 58 (2002) 995 /1001
Greywater frequently contains levels of faecal
coliforms and other bacteria, which make it
unsuitable for use as toilet flushing due to a
certain risk of human contact (from splashing
and aerosols) [3]. For these reasons disinfection is
a key step, either from the sanitary point of view,
and also to prevent the appearance of undesirable
odour. An efficient and commonly used waste-
water disinfection technology is the chlorination
with hypochlorite [4]. According to preliminary
experiments carried out in a 200-bed hotel during
summer 2001, and considering recently published
results on risk assessment in small scale water
recycling systems [5], the safety and quality of the
reused water depends, mainly, on the correct
chlorination of the greywater circuit. Greywater
varies greatly in composition and generation flow
over a 24-h period, and consequently the chlorine
demand is also variable. Then, a constant chlor-
ination can cause problems due to excess (disin-
fectant odour and waste of hypochlorite) or to
defect (storage of undisinfected greywater for
more than 24 h can result in offensive odours).
Adequate chlorination requires a frequent analy-
tical control of samples from different parts of the
circuit. The aim of this paper is to study an
analytical determination of hypochlorite in grey-
water. According to reviewed literature, such a
matrix (greywater), in spite of its potential, non-
potable, reuse to flush toilets, and others, such as
garden irrigation [6], remains unstudied from the
analytical point of view. Normally, greywater
parameters are determined using methods for
wastewater. However, greywater matrix shows a
higher content of detergents and lower content of
organic matter, and this can lead to errors due to
matrix effects. Greywater should be studied in
Analytical Chemistry as a specific sample, differ-
ent to blackwater.
The analytical reaction selected to carry out the
determination reported here is the well-known
chlorination of o -tolidine [7 /9] due to its extensive
use in potable water control.
Besides, the automation of this analytical deter-
mination was also an objective of this paper, in
order to facilitate the routine analysis.
Flow injection analysis (FIA) and sequential
injection analysis (SIA) are two widely used
strategies in Analytical Chemistry [10 /12]. Basi-
cally, FIA allows a higher sampling frequency
than SIA, and SIA is less reagent consuming than
FIA. Considering that the required frequency of
chlorine analysis for a satisfactory control of the
chlorination process is a few samples per hour at
3/4 sampling points in a small scale greywater
recycling system for flushing toilets, and also
bearing in mind that o-tolidine is classified as a
toxic reagent, and consequently the minimization
of its consumption is advisable, a SI procedure was
considered adequate for the automatic determina-
tion of chlorine. Many flow injection procedures
for chlorine suitable for drinking water analysis
with good analytical characteristics with photo-
metric [13 /16], chemiluminescent [17 /19], electro-
chemical [20], and potentiometric [21] detection
have been published, but this is the first paper
devoted to a sequential injection procedure for
chlorine determination.
2. Experimental
2.1. Apparatus
The manifold components were an automatic
burette Crison 2031 and an eight port multiposi-
tion valve supplied by Crison Instruments, Barce-
lona, Spain. The burette was equipped with a 2.5-
ml syringe, which permitted a 1 ml minimum piston
stroke, and with a two way valve that permitted
the aspiration of carrier stream when necessary.
0.8 mm i.d. PTFE tubing was used in the
construction of the sequential injection system.
The burette and the multiselection valve were
connected to a personal computer via an RS 232
interface, and controlled by home-made software.
Total organic carbon was determined using a
TOC-5000-A Shimadzu analyzer.
A Shimadzu UV-120-02 spectrophotometer
equipped with a flow through cell of 10 mm light
pathlength and 18 ml volume was used as a
detector. Peaks were obtained using a pen recor-
der. The sequential injection system is depicted in
Fig. 1. The cell used in the batch procedure had
also a pathlength of 10 mm.
 J.G. March et al. / Talanta 58 (2002) 995 /1001
Fig. 1. Scheme of the sequential injection system. Standard or samples were arranged around the valve at valve position from four to
eight. AB, autoburette; MV, multiselection valve.
2.2. Reagents and solutions
o-Tolidine (3,3?-dimethyl bencidine), purchased
from Probus (Spain), and sodium hypochlorite 5%
solution, sodium thiosulphate, potassium iodate,
potassium iodide and starch, purchased from
Panreac (Spain), were used. All reagents were of
analytical-reagent grade. A 6.3 10
3 mol l
1 o-
tolidine reagent solution was prepared in 1.8
mol l
1 HCl. The carrier used was 1.1 mol l
1
HCl. Sodium hypoclorite was titrated before use
[9] and its concentration labelled as chlorine, an
hypochlorite working solutions were prepared by
dilution with water.
2.3. Procedures
2.3.1. Batch procedure
Suitable volume of hypochlorite standard or
sample containing from 0.6 to 9.4 mgCl2 and water
up to 3 ml were introduced in a 3.5 ml disposable
cuvette. Then, 150 ml of o -tolidine reagent were
added. The cuvette content was homogenized and
left to stand for 9 min. The absorbance at 437 nm
was measured. Measures must be obtained against
a blank of sample or, alternatively, the sample
must be filtered through 0.45 mm cellulose acetate
membrane.
2.3.2. Automatic method
The following sequence was used: (1) Aspiration
of 200 ml of sample (or standard) containing
between 0.5 and 5.0 mgCl2 l
1. (2) Aspiration of
50 ml of o-tolidine reagent. (3) Aspiration of 200 ml
of sample (or standard). (4) Propulsion of 500 ml to
997
the reaction coil. (5) Stop for 3 min. (6) Propulsion
of 2000 ml through the reaction coil to the
measuring cell. The absorbance at 437 nm was
recorded. Aspiration rate and propulsion rate were
16.7 and 2.0 ml min
1, respectively. 1.1 mol l
1
HCl was used as carrier. As in the batch method,
measurements must be obtained against a blank of
sample or, alternatively, the sample must be
filtered through 0.45 mm cellulose acetate mem-
brane.
3. Results and discussion
3.1. Batch method
o -Tolidine is widely recommended for spectro-
photometric residual chlorine determination in tap
water. Three minutes reaction time is recom-
mended for such samples. Nevertheless, the analy-
tical reaction was slower in a greywater matrix, as
can be seen in Fig. 2. This effect on the kinetics of
the reaction was probably due to the presence of
chloramines (with lower tendency to chlorate o -
tolidine) and to adsorption processes on micellar
particles from the matrix. Based on this data, a 9
min reaction time was recommended for grey-
water.
3.2. Automatic method
Several experimental variables affecting the
analytical characteristics of the method were
studied.
998 J.G. March et al. / Talanta 58 (2002) 995 /1001
Fig. 2. Kinetics of the analytical reaction. (a /c) in tap water,
(d /f) in grey-water. (a) and (d) contained 0.45 mg Cl2 l
1 (b)
and (e) contained 0.34 mg Cl2 l
1, (c) and (f) contained 0.25 mg
Cl2 l
1. The other variables are indicated in the procedure.
Greywater samples were filtered through a 0.45 mm cellulose
acetate membrane. TOC of samples used in this experiment
were in the range 40 /50 mgC l
1.
3.2.1. Study of the carrier
When recording the analytical peak using water
as carrier, it was found that the analytical peak
was surrounded by small peaks due to refractive
processes between zones with different HCl con-
centration. The presence of such peaks disturbed
the measurements of peaks corresponding to blank
of samples or samples with low chlorine content.
In order to solve this pernicious effect, HCl of
different concentration were used as carrier. Fig. 3
depicts several typical morphologies of peaks
obtained under different experimental conditions.
Fig. 3. Typical absorbance peaks recorded using (a) water as
carrier (b) 1.1 molHCl l
1 as carrier (c) peak corresponding to
a sandwich 200 mlH2O/50 ml 1.1 molHCl l
1/200 mlH2O using
1.1 molHCl l
1 as carrier.
As can be seen, minimum refraction peaks were
obtained when using 1.1 mol l
1 HCl as carrier.
For this reason it was recommended. Besides, HCl
contained in the o-tolidine segment also generated
a small peak, due to refraction processes (Fig. 3c)
and, moreover, o -tolidine showed a slight absor-
bance peak when the recommended procedure was
followed.
3.2.2. Study of the sandwich composition
The composition of the sandwich of reagents
notably affected the sensitivity and precision of the
method (Table 1). Binary and ternary sandwiches
were compared. As can be seen, for binary
sandwiches, the absorbance peak increased with
the volume of the segment (exp. 1/10). This fact
could be explained by diffusion of the carrier into
the overlapping reagent zone. For a longer reagent
segment a more concentrated overlapping zone
can be formed. Nevertheless, under our experi-
mental conditions and apparatus, satisfactory
precision (coefficient of variation in the range
1.4 /1.8%) was only obtained using binary sand-
wiches of 150 or 200 ml for each reagent (exp 3 and
4). Lower and higher volumes had a negative effect
on the precision. Moreover, it was found that the
sensitivity of peaks could be notably improved
with the same level of precision using ternary
sandwiches. Then, ternary sandwiches were stu-
died. Two sandwich compositions were compared,
‘o -tolidine reagent/standard/o -tolidine reagent’
and ‘standard/o-tolidine reagent/standard’ (exp.
11/16). It was found that much greater sensitivity
was obtained when using a sandwich ‘standard/o -
tolidine reagent/standard’ and, a good compro-
mise between sensitivity and precision was ob-
served when using a volume of 200 ml for each
reagent segment. Finally, the effect of the volume
of the o -tolidine segment in a ‘200 ml standard/o -
tolidine/200 ml standard’ ternary sandwich was
also studied (exp. 17 /22). For sensitivity and
precision reasons, sandwich reported in exp 19
was selected for the recommended procedure.
3.2.3. Study of the reaction time
As can be deduced from data reported in
Section 3.1, the analytical reaction can be classified
as a slow reaction. Consequently, when studying
 J.G. March et al. / Talanta 58 (2002) 995 /1001 999

Table 1
Study of the sandwich composition

Exp. Sandwich composition Volume of each segment (ml) Absorbancea cv (%)

1 o -Tolidine/Cl2 50/50 0.122 6.6

2 o -Tolidine/Cl2 100/100 0.262 3.1
3 o -Tolidine/Cl2 150/150 0.454 1.4
4 o -Tolidine/Cl2 200/200 0.619 1.8
5 o -Tolidine/Cl2 250/250 0.808 3.0
6 o -Tolidine/Cl2 300/300 0.884 3.0
7 o -Tolidine/Cl2 350/350 0.896 2.9
8 o -Tolidine/Cl2 400/400 1.026 6.3
9 o -tolidine/Cl2 450/450 1.311 8.3
10 o -Tolidine/Cl2 500/500 1.376 12.4
11 o -Tolidine/Cl2/o -tolidine 100/100/100 0.181 4.15
12 o -Tolidine/Cl2/o -tolidine 200/200/200 0.402 2.68
13 o -Tolidine/Cl2/o -tolidine 300/300/300 0.609 0.99
14 Cl2/o -tolidine/Cl2 100/100/100 0.445 0.94
15 Cl2/o -tolidine/Cl2 200/200/200 0.798 1.59
16 Cl2/o -tolidine/Cl2 300/300/300 1.106 2.62
17 Cl2/o -tolidine/Cl2 200/150/200 0.826 1.51
18 Cl2/o -tolidine/Cl2 200/100/200 0.899 1.15
19 Cl2/o -tolidine/Cl2 200/50/200 1.015 1.53
20 Cl2/o -tolidine/Cl2 200/25/200 1.056 2.09
21 Cl2/o -tolidine/Cl2 200/10/200 1.035 3.09
22 Cl2/o -tolidine/Cl2 200/5/200 1.030 3.99

A 3.2 mg Cl2 l
l standard was used. Other variables were indicated in the procedure.
a
Mean of three measurements.

the sequential injection method, after sandwich

formation the system was stopped for a time. On
the other hand, as was found in batch procedure,
the matrix of the sample affected the kinetics of the
reaction and for this reason the reaction time was
studied. Results obtained from four different real
samples are depicted in Fig. 4, and from this, a 3
min reaction time was recommended. It is worth
mentioning that the analysis of the same samples
following the batch procedure required 6 more min
for each sample. This fact could be explained
considering that working as described in proce-
dures, the molar ratio between chlorine and o- Fig. 4. Peak absorbance values corresponding to greywater
tolidine was, probably, lower in the batch method samples as a function of time. Experimental conditions were as
than in the overlapped zone into the coil. in the Section 2.3.2. TOC of sample (a) was 70 mgC l
1. TOC
of other samples were in the range 40 /50 mgC l
1.

3.2.4. Stability of the solutions

The influence of temperature, time and environ-
mental light on the stability of samples and
standards was tested. In general terms, an average
decrease of 4% in chlorine concentration in
standards per day was observed when they were
not protected from light. In dark conditions,
standards can be considered stable for 2 weeks.
Storage temperature in the range 4/25 8C did not
affect stability. Nevertheless, samples were clearly
less stable than standards. The maximum stability
1000 J.G. March et al. / Talanta 58 (2002) 995 /1001

was obtained in dark conditions and at 4 8C, and Table 2

in such conditions, after 2 h a significative loss in Analytical figures of merit of the methods
chlorine was observed in most cases. This fact Batch SI
reaffirms the convenience of an extensive chlorine
control in the reuse greywater circuit to guarantee Calibration linea A/(0.6699/ A/(0.289/
the right chlorine content in the system. 0.008)C/(
/ 0.01)C/(0.129/
0.0129/0.009) 0.01)
R2 0.9995 0.991
Sy/x 0.012 0.014
3.3. Analytical figures of merit Linear range (mg 0 /3.0 0 /5.0
Cl2 l
1)
The analytical figures of merit of the methods Limit of detectionb 0.06 0.2
are summarized in Table 2. As can be seen, the (mg Cl2 l
1)
batch method showed greater sensitivity, precision Limit of quantifica- 0.2 0.5
tionc (mg Cl2 l
1)
and limit of detection. The ordinate intercept of Sampling (h
1) 11
the calibration straight line of the SI method was
a
significantly higher, statistically, than zero. This A/(b9/S )C/(a9/S ); C represents mg Cl2 l
1, n/5.
b
blank is due to refractive processes and to Calculated as 3 Sy/x/sensitivity.
c
Calculated as 10 Sy/x/sensitivity.
absorption of o-tolidine. The sampling frequency
(11 samples h
1) is adequate to control the
chlorine concentration in greywater in order to recovery values (no sample gave a statistically
reuse it for flushing toilets in the same building significant recovery higher than 100%). This find-
where the greywater is produced. ing would indicate that, depending on the matrix
In order to check the accuracy of the methods composition, a negative error can be obtained.
the standard addition method and the recovery In order to investigate the origin of that inter-
obtained from 25 spiked with hypochlorite grey- ference, the total organic carbon (TOC) [7] of the
water samples were studied using both methods. samples was also determined. A graph of recovery
Most of samples gave a recovery in the 100% (%) versus TOC is depicted in Fig. 5. It seems that
order. Nevertheless, several samples showed lower samples can be roughly classified in two groups:

Fig. 5. Relationship between % recovery and TOC for several greywater samples. The chlorine concentration in the samples in this
experiment were in the range 0.6 /1.2 mgCl2 l
1. Each result is an average of four experiments.
 J.G. March et al. / Talanta 58 (2002) 995 /1001
samples with a TOC B/60 mgC l
1 which led to
satisfactory recovery values (in the order of 1009/
3%) and samples with TOC /60 mgC l
1 that
showed, in most cases, deficient recovery values.
These facts were, in part, due to the reaction of
added chlorine with (organic)compounds present
in the sample during the application of the
method. This affirmation is supported on the
diminution of chlorine concentration with time
that was observed from chlorinated samples in the
laboratory. In fact, several days after chlorination
of a sample, the chlorine concentration fell to
undetectable values. In samples with low TOC
these reaction can take place in a non detectable
extension (during the application of the method).
Taking these facts into account, it is recommend-
able, when doing routine work, to know the TOC
range of analysed samples. Then, samples with
TOC /60 mgC l
1 should be diluted before
analysis. It is worth mentioning that grey water
from the Hotel involved in this project gave, in all
cases, TOC in the 309/20 range (grey water
samples with higher TOC used in this work were
obtained from residential buildings).
The SI method and the batch method were also
compared using 10 different samples. The correla-
tion line found for the mean values obtained by the
automatic method (Y) versus the batch method
(X), corresponding to a 95% confidence level and 8
degree of freedom, was
Y /(
/0.099/0.02)/(1.0089/0.007)X, R2 /

1
0.9998, (X and Y are mgCl2 l ). This line
demonstrated the agreement between both meth-
odologies.
Acknowledgements
The authors acknowledge the financial support
from the Spanish Ministry of Science and Tech-
nology for the project on grey water reuse (REN
2000-0336-P4-02). Also, the availability of Hotel
Brasilia installation (Palma de Mallorca, Spain)
1001
and the collaboration and interest of Mr Domingo
Muñoz (manager of the hotel) in this project is
gratefully acknowledged. M. Gual thanks the
town council of Palma de Mallorca for a fellow-
ship.
References
[1] D. Christova-Boal, R.E. Eden, S. McFarlane, Desalination
106 (1996) 391.
[2] F. Kiya, H. Aya, Water Sci. Technol. 23 (1991) 2189.
[3] B. Jeppesen, Desalination 106 (1996) 311.
[4] Metcalf-Eddy, Tratamiento y depuración de aguas resi-
duales, Ed Diaz de Santos, Madrid, 1995.
[5] C. Diaper, A. Dixon, D. Butler, A. Fewkes, S.A. Parsons,
M. Strathern, T. Stephenson, J. Strutt, Water Sci. Technol.
43 (2001) 83.
[6] J.A. Andersen, Water Sci. Technol. 33 (1996) 37.
[7] APHA, AWWA, WCCF, Standard Methods for the
examination of Water and Wastewater, 16th ed., 1985.
[8] D.J. Leggett, N.H. Chen, D.S. Mahadevappa, Analyst 107
(1982) 433.
[9] D.F. Boltz, W.J. Holland, J.A. Howell, in: D.F. Boltz, J.A.
Howell (Eds.), Colorimetric Determination of Nonmetals,
Interscience, New York, 1978.
[10] M. Valcárcel, M.D. Luque-de-Castro, Flow Injection
Analysis: Principles and Applications, Ellis Horwood,
Chichester, 1987.
[11] J. Ruzicka, E.H. Hansen, Trends Anal. Chem. 17 (1998)
69.
[12] J. Ruzicka, G.D. Marshall, Anal. Chim. Acta 237 (1990)
329.
[13] M.C. Cheregi, A.F. Danet, Revista de Chimie 51 (2000)
566.
[14] F. Cañete, A. Rios, M.D. Luque-de-Castro, M. Valcárcel,
Analyst 113 (1988) 739.
[15] K.K. Verma, A. Jain, A. Townshend, Anal. Chim. Acta
261 (1992) 233.
[16] M. Catalá Icardo, J.V. Garcı́a Mateo, J. Martı́nez
Calatayud, Analyst 126 (2001) 2087.
[17] W. Quin, Z.J. Zhang, S.N. Liu, Anal. Lett. 30 (1997) 11.
[18] G.P. Irons, G.M. Greenway, Analyst 120 (1995) 477.
[19] G.P. Irons, G.M. Greenway, Anal. Proc. 31 (1994) 31.
[20] H. Iketake, A. Yamada, Bunseki Kagaku 48 (1999) 1123.
[21] A. Sakai, A. Hemmi, H. Hachiya, F. Kobayashi, S. Ito, Y.
Asano, T. Imato, Y. Fushinuki, I. Taniguchi, Talanta 45
(1998) 575.