Вы находитесь на странице: 1из 8

Analytica Chimica Acta 432 (2001) 5966

Photocatalytic sensor for the determination of chemical oxygen


demand using ow injection analysis
Yoon-Chang Kim, Satoshi Sasaki, Kazuyoshi Yano, Kazunori Ikebukuro,
Kazuhito Hashimoto, Isao Karube

Research Center for Advanced Science and Technology, The University of Tokyo,
4-6-1 Komaba, Meguro-ku, Tokyo 153-8904, Japan
Received 9 May 2000; received in revised form 4 August 2000; accepted 15 August 2000
Abstract
A photocatalytic sensor for the determination of chemical oxygen demand (COD) with ow injection analysis (FIA) based
on the photocatalysis of organic compounds in the presence of titanium dioxide (TiO
2
) beads in a photochemical column
is described. The sensor was developed in conjunction with TiO
2
beads in the photochemical column and with an oxygen
electrode as the sensing part. The sensor signal was observed as a result of the detection of dissolved oxygen changes due
to photocatalytic oxidation of organic compounds in the sample solution. This sensor responded linearly to the COD
Mn
of
articially treated wastewater (AWW) in the range of 0.128 ppm. A complete analysis, including sampling and washing,
took about 10 min. The sensor was stable for over 15 days and has successfully been applied to the determination of COD in
lake samples. 2001 Elsevier Science B.V. All rights reserved.
Keywords: Photocatalytic sensor; Chemical Oxygen Demand (COD); Titanium dioxide (TiO
2
); Oxygen electrode; Flow injection analysis
1. Introduction
Pollutants in watercourses can be of several origins
and types, but some of the worst pollutants are those
that reduce the available oxygen content [1]. This loss
can have major adverse effects on natural ecosystems
if the oxygen content falls below the level necessary
to support aquatic life. Chemical oxygen demand
(COD), biological oxygen demand (BOD) and total
organic carbon (TOC) are three main indexes used to
assess this organic pollution in aqueous systems [2].
While the TOC analysis is valid for soluble organic

Corresponding author. Tel.: +81-3-5452-5220;


fax: +81-3-5452-5227.
E-mail address: karube@bio.rcast.u-tokyo.ac.jp (I. Karube).
compounds and the BOD reects the biodegradable
part of the pollutants, the COD analysis represents
the total pollution load of most wastewater discharges
[3]. As a pollution monitoring parameter, COD has
the advantage of speed and simplicity over BOD
[4,5], and requires less equipment compared to TOC
determinations. However, the COD procedure is very
time-consuming [3], and it demands high-quality
personnel and expensive or poisonous reagents
[6].
In conventional COD procedures, a known amount
of oxidant is added to a sample and the mixture is
boiled. After the oxidation has proceeded for a given
period of time, the initial concentration of organic
species can be calculated by determining the amount
of oxidizing agent remaining [7]. In the case of the
0003-2670/01/$ see front matter 2001 Elsevier Science B.V. All rights reserved.
PII: S0003- 2670( 00) 01145- 4
60 Y.-C. Kim et al. / Analytica Chimica Acta 432 (2001) 5966
dichromate method, the COD measurement procedure
is based on the observation that under acidic condi-
tions the dichromate ion (Cr
2
O
2
7
) oxidizes almost all
organic materials (C
x
H
y
O
z
) to CO
2
and H
2
O accord-
ing to the formula [5]:
C
x
H
y
O
z
+kCr
2
O
2
7
+kH
+
xCO
2
+
1
2
(y +8k)H
2
O +2kCr
+
3
(1)
where k = (2x/3) +(y/6) (z/3)
The chemical oxygen demand, however, represents
the oxygen required for complete chemical oxida-
tion of a sample [8,9]. Therefore we suggest that
the theoretical COD value of samples is related to
the determination of consumed oxygen concentration
in samples under chemical oxidation of the organic
compounds.
Photocatalysis is an efcient method for the degra-
dation and mineralization of organic compounds [10].
It has been the subject of numerous studies to obtain
a better understanding of the mechanisms of the re-
actions involved [1118]. A semiconductor particle is
believed to have a lled valence band separated from
a vacant conduction band by a gap whose energy
is E
g
. When irradiated with a light source, whose
energy exceeds E
g
, an electron is promoted from
the valence to the conduction band, leaving behind
a positive hole. Upon migration to the surface, the
electron would reduce any available species. In con-
trast, when the hole reaches the surface, it would
react with water to produce hydroxyl radicals. These
radicals play an important role in oxidizing organic
pollutants. Molecular oxygen dissolved is also an
essential component because it acts as scavenger of
the photogenerated electrons, forming a superoxide
radical ion, and further reacting with transient rad-
icals leading toward CO
2
formation [19,20]. It was
found that photocatalytic activity is almost com-
pletely suppressed in the absence of oxygen and the
concentration of oxygen has a profound effect on
the rate of photocatalyzed decomposition of organic
compounds [2123]. Many of photocatalytic pro-
cesses apply the TiO
2
as a photocatalyst because it
is non-photocorrosive, non-toxic, and capable of the
photooxidative destruction of most organic pollutants
[24,25].
It is well known that the TiO
2
photocatalysis leads
to stoichiometric photomineralization of organic com-
pounds as follows [16,26]
C
x
H
y
O
z
X
k
+x +

y k 2z
4

O
2
xCO
2
+kH
+
+kX

y x
2

H
2
O (2)
where X represents a halogen atom.
Therefore, this indicates that the COD of a given
sample can be assessed, by tracing the change of the
dissolved oxygen concentration under photocatalytic
conditions.
The objective of this study was to develop a simple,
fast, inexpensive, and safe photocatalytic sensor for
COD using the photocatalysis of TiO
2
. The technique
is based on the measurement of changes in oxygen
concentration resulting from photocatalytic oxidation
of organic compounds. The COD sensor was used for
the determination of COD in real samples from vari-
ous locations sources, and the results were compared
with those obtained by the Japanese Industrial Stan-
dard method [27].
2. Experimental
2.1. Reagents
The titanium dioxide beads (Type, ST-B11) used
in the Ishihara Sangyo, Ltd. (Japan). The tannic acid
and nitrohumic acid were purchased from Wako Pure
Chemical Industries (Japan) and the ligninesulfonic
acid (sodium salt) and sodium dodecylbenzenesul-
fonate were purchased from Tokyo Kasei Co., Ltd.
(Japan). The gum arabic (from the acacia tree) was
obtained from Sigma (USA). All organic compounds
used were reagent-grade, except for the nitrohumic
acid and gum arabic. The carrier solution used in
FIA experiments was deionized water. The deion-
ized water was puried in a Milipore-MilliQ system.
These reagents were used without further purication.
The solutions of articially treated wastewater were
prepared daily and stored at 4

C.
2.2. Preparation of articially treated wastewater
and lake samples
River water and wastewater efuents usually con-
tain highly stable organic compounds such as humic
Y.-C. Kim et al. / Analytica Chimica Acta 432 (2001) 5966 61
Table 1
Constituents (mg/l) of articial wastewater
Compound
Nitrohumic acid 4.25
Tannic acid 4.18
Sodium ligninesulfonate (NaLS) 2.43
Gum arabic 4.70
Sodium lauryl sulfate 0.94
Water quality analysis
COD (permangnate method) 5.89
BOD (5-days method) 3.77
TOC 9.12
acid and lignin, etc. which account for 50% of the
compositions of secondary efuents, according to
a previous report by Tanaka [28]. The constituents
of the articially treated wastewater are shown in
Table 1.
The water samples analyzed during this study were
collected from lakes all over Japan. Collection con-
ditions varied slightly, but all samples were taken at
a depth of 50 cm and were kept below 4

C until use.
Experiments were performed within a day or two after
collection.
COD values in lake samples measured with the
COD sensor were compared with those determined us-
ing the conventional methods (KMnO
4
and K
2
Cr
2
O
7
)
set forth in the Japanese Industrial Standards (JIS)
[27]. All experiments were performed on the day of
sample preparation.
2.3. Instrumentation
The set-up of the FIA system for COD determina-
tion is shown in Fig. 1. The ow carrier was deionized
water that was held in a reservoir and was propelled by
a peristaltic pump (Minipuls3, Gilson, France) through
a injector (7125NS, Rheodyne, USA), a photocatalytic
column, and an air damper, before being transported to
an oxygen electrode (BO-G, Able, Japan) and nally
to a waste tank. The photocatalytic column was made
from a Pyrex tube (1 ml) lled with 0.6 mm diameter
TiO
2
beads (ST-B11, Ishihara Sangyo, Japan) that
were held in position with Teon wool plugs. A UV ir-
radiator with 15 W UV lamp (XX-15 type, Funakoshi,
Japan) and built-in reector was located 0.5 cm from
the side of the photocatalytic column in the UV illu-
mination experiment. The temperature of the oxygen
Fig. 1. Schematic diagram of the apparatus: 1 =deionized water
supply; 2 =@pump; 3 =injector; 4 =reector; 5 =@photochemi-
cal column consisting of TiO
2
beads; 6 =UV lamp (
max
=
365 nm); 7 =air damper; 8 =oxygen electrode; 9 =integrator;
10 =thermostatic water bath; 11 =waste tank.
electrode was thermostated with a water jacket and a
temperature controller (SB-9, Eyela, Japan).
2.4. Procedure
The measurements were performed as follows.
First, the oxygen electrode was arranged in a tempe-
rature-controlled ow cell, and the responses were
recorded. When the UV lamp was turned on, the cur-
rent began to decrease until a steady state was reached
in 10 min. The following ve processes are considered
for the decrease of oxygen in an oxygenated TiO
2
aqueous system without organic compounds [29,30].
Reduction reaction:
O
2
+2H
+
aq
+2e

cb
H
2
O
2
(3)
Oxidation reactions:
H
2
O +2h
+
vb
2H
+
aq
+
1
2
O
2
(4)
2H
2
O +2h
+
vb
2H
+
aq
+H
2
O
2
(5)
Overall reaction of (3) and (4):
H
2
O +
1
2
O
2
+2(e

cb
+h
+
vb
) H
2
O
2
(6)
Overall reaction of (3) and (5):
2H
2
O +O
2
+2(e

cb
+h
+
vb
) 2H
2
O
2
(7)
62 Y.-C. Kim et al. / Analytica Chimica Acta 432 (2001) 5966
Here e

cb
and h
+
vb
are the photogenerated electron in
the conduction band and the hole in the valence band
respectively.
Then a sample was injected into the carrier stream,
and the oxygen electrode monitored the change in oxy-
gen concentration resulting from the photocatalytic
oxidation of organic compounds in the sample. The
peak-shaped currents from the oxygen electrode were
recorded as a function of time.
3. Results and discussion
3.1. Inuence of concentration of articially treated
wastewater
The oxygen electrode current decreased after the
addition of articially treated wastewater to reach a
peak (the maximum current decrease) and then re-
turned to the baseline. The relationship between the
maximum current decreases and the articially treated
wastewater concentrations was examined using a 2 ml
sample loop. Fig. 2 shows the inuence of articially
treated wastewater at concentrations ranging from0.12
to 23.6 ppm(COD
Mn
). The maximumcurrent decrease
reached a plateau at 23.6 ppm and did not change
Fig. 2. Inuence of concentration of articially treated wastewater.
Conditions: substrate, articially treated wastewater; light source,
UV lamp (
max
= 365 nm); temperature, 20

C; amount of TiO
2
,
0.9 g; sample volume, 2 ml; ow rate, 1.54 ml/min.
when the articially treated wastewater concentration
was increased to 30 ppm (data not shown). The stan-
dard deviation was 0.0090.036 for multiple injections
(three times) of articially treated wastewater when
tested in the concentration range (0.1223.6 ppm). The
detection limit of the linear AWW concentration range
(<8 ppm) was 0.12 ppm (signal-to-noise ratio = 3).
3.2. Optimization of the operating conditions for the
FIA determination of COD
3.2.1. Sample volume
The inuence of sample injection volume on the
determination of COD was investigated by measuring
the amount of current decrease with different volumes
of 4.098 ppm (COD
Mn
) articially treated wastwater.
The results indicated that the maximum current de-
crease increased rapidly with the increase of sample
injection volume from 0.02 to 3 ml. As can be seen,
a plateau in sensitivity can be observed above 3 ml
(Fig. 3). This is fully explained by both the larger
amount of the sample injected through a Rheodyne
injector and the longer residence time for the sample
in contact with the TiO
2
beads. Consequently, the use
of a large volume (>3 ml) of sample solution yielded
the same sensitivity. Therefore, the selection of the
Fig. 3. Inuence of sample volume. Conditions: substrate, arti-
cially treated wastewater (COD
Mn
= 4.098 ppm); light source,
UV lamp (
max
= 365 nm); temperature, 20

C; amount of TiO
2
,
0.9 g; ow rate, 1.88 ml/min.
Y.-C. Kim et al. / Analytica Chimica Acta 432 (2001) 5966 63
sample volume should reect a consideration of sen-
sitivity and analysis speed.
3.2.2. Flow rate
The effect of the ow rate of deionized water
(carrier) on the determination of articially treated
wastewater was investigated by measuring the amount
of current decrease with different ow rates of
4.098 ppm (COD
Mn
) articially treated wastewater.
In this study, the sample volume was kept at 2 ml.
Fig. 4 shows the changes in current measured with
different ow rates of deionized water. The maximum
current decrease increased dramatically with ow
rate in the range from 0.78 to 1.88 ml/min. and then
leveled off and even decreased slightly at rates higher
than 2.30 ml/min. This behavior might be explained
in terms of the dissolved oxygen concentration in the
sample and residence time of the sample in contact
with the TiO
2
beads. At lower ow rates, the peak
intensity is mainly governed by the dissolved oxy-
gen concentration in the sample. An increase in ow
rate leads to an increase in the amount of dissolved
oxygen. This is why larger signals were observed for
increased ow rates. At very high ow rates, however,
the residence time of the sample in the contact to the
TiO
2
beads becomes the limiting factor, as explained
by the fact less photocatalysis is detected by the oxy-
Fig. 4. Inuence of ow rate. Experimental conditions as described
in Fig. 3 and sample volume was 2 ml.
Fig. 5. Inuence of TiO
2
amount. Experimental conditions as
described in Fig. 3 and sample volume was 3 ml.
gen electrode during the shorter residence time of
the sample in the photochemical column, resulting
in a slight decrease in the peak height decreases. A
compromise between high sensitivity and fast analy-
sis time was considered desirable. Thus, a ow rate
of 1.88 ml/min. was selected for use in subsequent
experiments.
3.2.3. Amount of TiO
2
The effect of variable amounts of TiO
2
was tested
using a 3 ml sample volume of 4.098 ppm (COD
Mn
)
of articially treated wastewater injected into the sen-
sor system at a constant ow rate of 1.88 ml/min. The
response of the sensor to the amount of TiO
2
in Fig. 5
showed a very steep increase in signals from 0.11
to 0.9 g. The maximum current decrease decreased
at TiO
2
amounts above 0.9 g, therefore the optimal
conditions for using the COD sensor were established
as follows: ow rate, 1.88 ml/min., sample volume,
3 ml, and 0.9 g of TiO
2
amount.
3.3. Reproducibility and long-time stability
It is desirable for the COD sensor to have good
reproducibility and stability. The response repro-
ducibility of the sensor to 4.098 ppm (COD
Mn
) of
articially treated wastewater was studied using
64 Y.-C. Kim et al. / Analytica Chimica Acta 432 (2001) 5966
repeated determinations (n = 10), and evaluated by
the relative standard deviation of current changes.
This gave a relative standard deviation of 2.22%, in-
dicating that the sensor has a good reproducibility.
The stability of this COD sensor on the determi-
nation of articially treated wastewater (4.098 ppm
(COD
Mn
)) was investigated during a period of 15
days. The sensor response was tested daily over a
15-day period and on each occasion 24 samples
were injected, i.e. a total of about 45 samples. During
this period the response was almost constant. The
relative standard deviation obtained for successive
measurements was 4.40 %. Between measurements,
the sensors were stored in deionized water at room
temperature. The activated state of the COD sensor
appeared to be maintainable for more than 15 days
after inserting of the TiO
2
beads in the photochemical
column.
3.4. Real sample analysis
A calibration graph was constructed using chemical
deoxygenation with an oxygen scavenger sodium sul-
te (Na
2
SO
3
) test solution. The Na
2
SO
3
deoxygena-
tion is based on the following oxidationreduction
reaction [31]:
2SO
2
3
+O
2
= 2SO
2
4
After the oxygen electrode of the DO-meter (type
DO-25A, TOA Electronics, Ltd.) had been placed in
the batch cell, the background currents were allowed
to stabilize. The appropriate volume of Na
2
SO
3
so-
lution was injected into the cell, after which the
ppmtime response of the DO-meter was recorded
in a batch-mode fashion. The dissolved oxygen was
consumed by the added Na
2
SO
3
solution. The com-
mercial DO-meter displayed oxygen concentration
in ppm-units. The Na
2
SO
3
solutions, which were
measured using a batch mode, were injected via a
Redone injector and transported with deionized water,
pumped by a peristaltic pump. The oxygen electrode
was attached to an integrator and the current response
to the injected Na
2
SO
3
solutions was registered on an
integrator. The calibration curve of the COD sensor
to dissolved oxygen in solution was a second order
equation (Fig. 6). Examination of the calibration data
shows that the experimental data points t a curve in
Fig. 6. Calibration curve for COD sensor. Experimental conditions
as described in Fig. 5 and substrate was sodium sulte solution.
the range of 0.694.83 ppm O
2
concentrations. The
regression equation is
C = 0.00563O
2
2
+0.10064O
2
+0.00308
where C is the peak height decrease in A, O
2
is
the dissolved oxygen concentration decrease in ppm,
and the correlation coefcient is 0.99.
The COD sensor has been used to determine the
COD value in various real samples. Conventional
methods were also used to analyze these same real
samples. The COD values were directly evaluated
with a calibration graph for the oxygen electrode
and compared with those determined by conventional
COD methods. Figs. 7 and 8 show the relationships
between the results obtained by the FIA COD sensor
and conventional methods (dichromate and perman-
ganate): dichromate method, r = 0.98; permanganate
method, r = 0.95 for six lake samples. The linear
regression equations were y = 0.58x +0.36 (dichro-
mate method) and y = 1.21x 0.88 (permanganate
method), where x and y are the results obtained us-
ing conventional methods and the FIA COD sensor,
respectively. It can be seen that the results obtained
from the proposed sensor methods were in good
correspondence with those obtained by conventional
COD methods. When the CODs evaluated by the
conventional methods are compared, it can be seen
Y.-C. Kim et al. / Analytica Chimica Acta 432 (2001) 5966 65
Fig. 7. Relationship of COD sensor to conventional COD method
(dichromate). Experimental conditions as described in Fig. 5 and
substrates were lake waters.
that the absolute values obtained using the perman-
ganate method are about 46% of those obtained using
the dichromate method. This agrees with the fact that
the permanganate method gives oxidation ratios of
about 5070, while the dichromate method achieves
80100 % oxidation ratios [32]. In our former work
Fig. 8. Relationship of COD sensor to conventional COD method
(permanganate). Experimental conditions as described in Fig. 5
and substrates were lake waters.
[33], the COD senor for the batch type was presented
for the determination of COD using TiO
2
. In the two
correlation equations (conventional methods versus
the batch COD sensor), the slopes were in the ratio
of 1:2.04 (dichromate:permanganate). In this study,
the dichromate and permanganate slopes had a ratio
of 1:2.09. On the basis of the results presented, it can
be stated that a new COD sensor was invented using
a photocatalytic sensor to determine COD in real
lakes.
4. Conclusions
The objective of this work was to demonstrate
the viability of a photocatalytic sensor for chemical
oxygen demand determination based on FIA using
the photocatalyst, TiO
2
and an oxygen electrode as a
transducer. The main advantage of this photocatalytic
sensor for COD is its simplicity of preparation, low
cost in the manufacturing process for the sensor, fast
response time, acceptable lifetime, and potential for
automated monitoring. Additionally, the COD sensor
was in excellent correspondence with values obtained
using conventional methods (dichromate and perman-
ganate).
References
[1] C. Billings, M. Lane, A. Watson, T.P. Whitehead, G. Thorpe,
Analysis 22 (1994) 27.
[2] M. Novic, B. Pihlar, M. Dular, Fresenius Z. Anal. Chem.
332 (1988) 750.
[3] N.S. Abuzaid, M.H. Al-Malack, A.H. El-Mubarak, Bull.
Environ. Contam. Toxicol. 59 (1997) 626.
[4] K.L. Pamplin, D.C. Johnson, Electroanalysis 9 (1997) 279.
[5] D. Bilanovic, R.E. Loewenthal, Y. Avnimelech, M. Green,
Water SA 23 (1997) 301.
[6] N.U. Bykadorov, S.S. Radchenko, J. Anal. Chem. 50 (1995)
1994.
[7] J.S. Jerzy Zerbe, Chem. Anal. (Warsaw) 44 (1999) 263.
[8] I. Papaefstathiou, M.D. Luque de Castro, Int. J. Environ.
Anal. Chem. 66 (1997) 107.
[9] B. Vallejo-Pecharroman, A. Izquierdo-Reina, M.D. Luque de
Castro, Analyst 124 (1999) 1261.
[10] C. Richard, F. Bosqut, J.-F. Pilichowski, J. Photochem.
Photobiol. A: Chem. 108 (1997) 45.
[11] N. Serpone, R.F. Khairutdinov, Semicond. Nanoclusters 103
(1996) 417.
[12] K. Rajeshwar, J. Appl. Electrochem. 25 (1995) 1067.
66 Y.-C. Kim et al. / Analytica Chimica Acta 432 (2001) 5966
[13] G.A. Penuela, D. Barcelo, Trends Anal. Chem. 17 (1998)
605.
[14] J.K. Dohrmann, M. Chuan, N.-S. Schaaf, Solar Energy Mater.
Solar Cells 43 (1996) 273.
[15] J. Peral, X. Domenech, D.F. Ollis, J. Chem. Technol.
Biotechnol. 70 (1997) 117.
[16] M.R. Hoffmann, S.T. Martin, W. Choi, D.W. Bahnemann,
Chem. Rev. 95 (1995) 69.
[17] M. Bekbolet, G. Ozkosemen, Water Sci. Tech. 33 (1996) 189.
[18] R.A. Sierka, C.W. Bryant, Water Sci. Tech. 29 (1994) 209.
[19] E. Pelizzetti, C. Minero, Colloids Surf. A 151 (1999) 321.
[20] Y. Ohko, D.A. Tryk, K. Hashimoto, A. Fujishima, J. Phys.
Chem. B 102 (1998) 2699.
[21] I. Sopyan, M. Watanabe, S. Murasawa, K. Hashimoto, A.
Fujishima, J. Photochem. Photobiol. A: Chem. 98 (1996) 79.
[22] W. Choi, M.R. Hoffmann, Environ. Sci. Technol. 31 (1997)
89.
[23] A.J. Hoffman, E.R. Carraway, M.R. Hoffmann, Environ. Sci.
Technol. 28 (1994) 776.
[24] I. Poulios, M. Kositzi, A. Kouras, J. Photochem. Photobiol.
A: Chem. 115 (1998) 175.
[25] R.W. Matthews, Water Res. 25 (1991) 1169.
[26] S.T. Martin, H. Herrmann, W. Choi, M.R. Hoffmann, J. Chem.
Soc., Faraday Trans. 90 (1994) 3315.
[27] Japanese Industrial Standard Committee, Testing Methods for
Industrial Waste Water, JIS K 0101, JIS, Tokyo, 1991.
[28] H. Tanaka, E. Nakamura, Y. Minamiyama, T. Toyoda, Water
Sci. Tech. 30 (1994) 215.
[29] R. Cai, K. Hashimoto, A. Fujishima, J. Electroanal. Chem.
326 (1992) 345.
[30] H. Sakai, R. Baba, K. Hashimoto, A. Fujishima, J. Phys.
Chem. 99 (1995) 11896.
[31] H.R. Zhang, J. Zhang, Y.S. Wei, W.J. Jin, C.S. Liu, C.S. Liu,
Anal. Chim. Acta 357 (1997) 119.
[32] J. Tokuhira, J. Ui, S. Ichikawa, J. Kondo, J. Water Wastewater
(in Japanese) 12 (1970) 90.
[33] Y.-C. Kim, K.-H. Lee, S. Sasaki, K. Hashimoto, K. Ikebukuro,
I. Karube, Anal. Chem. 72 (2000) 3379.