Environ. Sci. Techno/.
1995, 29, 3007-3014
Chemical Oxidation by Photolytic
Decomposition of Hydrogen
Peroxide
C H I H - H S I A N G LIAO A N D
M I R A T D . GUROL’
Environmental Studies Institute, Drexel University,
Philadelphia, Pennsylvania 191 04
This paper describes a study of a chemical oxidation
process involving simultaneous application of
hydrogen peroxide solution and ultraviolet light
(H20dUV) for removal of organic pollutants from aqueous
solution. The process was investigated experimentally
in a continuous-flow stirred tank reactor (CSTR)
under various operational conditions, Le., H202 dosage,
UV light intensity, and liquid residence time. Synthetic
solutions of a model organic compound, n-chlor-
obutane (BuCI), were oxidized at various pH and in the
presence of various amounts of humic material and
carbonate/bicarbonate ions in order to examine the
effect of water quality on the process efficiency.
A kinetic model of the process, which was developed
based on HzOdUV-induced radical oxidation of organic
compounds, was successfully verified in pure water
as well as in synthetic solutions.
Introduction
Since the late 1960%numerous researchers have used the
H202/W process to oxidize various organic substances in
water. Recently, commercial units employing this process
have been developed for on-site oxidation of organic
contaminants in groundwater. The success of this process
has been generally atttributed to the stoichiometric forma-
tion of hydroxyl radical (‘OH) by photolytic decomposition
of H202. The hydroxyl radical is an extremely reactive and
nonselective oxidant and, thus, when produced in sufficient
quantities, can lead to complete oxidation of organic
compounds to carbon dioxide, water, and inorganic ions.
In spite of the commercial success of the H202/UV
process, the chemistry, kinetics, and engineering principles
associated with the process have not been elucidated
adequately. The recent careful studies by Glaze and Lay
(11, Yao et al. (21, and Wolfrum (3) provided a better,
although partial, understanding of the process fundamen-
tals. The present study attempts to add to the existing
literature through a systematic investigation of the effect
of the process parameters and water quality on the process
performance in terms of the rate of H202 photolysis and
the oxidation rate of the organic contaminant(s)of interest.
This objective is realized by experimental observations as
* Corresponding author e-mail address: Mirat.Gurol@CoE.Drexel.
Edu; fax: (215) 895-2267.
0013-936X/95/0929-3007$09.00/0 G 1995 American Chemical Society
well as by developing and testing a kinetic model of the
process.
The H2O21UVprocess was first investigated in “pure”
water, and then the results were applied to the effect of
naturally occurring species such as carbonate ions and
natural organic matter. A continuous-flow stirred tank
reactor (CSTR) was used for experimental observations as
well as for modeling purposes. The CSTR was preferred
over a batch reactor to obtain steady-state conditions, which
generally allow a more effective evaluation of the kinetic
systems.
Experimental Approach
The laboratory studies were conducted using a bench-scale
circular quartz reactorwith a 10-cm 0.d. and a liquidvolume
of 840 mL. The reactor was inserted into a chamber that
contained concentrically placed 16 removable low-pressure
mercurylamps that emit Wlight primarily at 254 nm (93%
of the light emitted in the UV range). It should be noted
that the quantum yield for H202 photolysis does not change
significantlywith wavelength (6).The reactorwas operated
in continuous flow mode and stirred vigorously to mimic
a CSTR. The schematic diagram of the experimental set-
up is shown in Figure 1. In order to avoid stripping of the
volatile compounds by mixing, no head space was allowed
within the reactor. The temperature of the reactor content
increased only slightly during the experiments because of
the presence of a continuously operating fan within the
chamber. The hydraulic characteristics of the reactor were
studied by applying a tracer (methyleneblue) in pulse mode
for various liquid residence times (r). The results of these
pulse tracer analyses showed than an ideal CSTR has been
approximated very closely under all the experimental
conditions (4). Hence, the performance of the reactor was
governed only by the chemical kinetics and the liquid
residence time. After reaching steady-state conditions, Le.,
after a time period equal to 3r, water samples were collected
from the effluent stream for analysis.
The oxidation efficiency of the process was determined
by monitoring the removal rate of n-chlorobutane (BuC1)
under various experimental conditions. Butyl chloride was
used in this study as a probe compound for ‘OH; its reaction
with H202 and direct photolysis by UV radiation are
negligible. Hence, the conclusions drawn from this study
would be restricted to the compounds that can react only
with the primary oxidant *OHin the H202/UVsystem.
Materials. The 30% and 50%certified solutions of H202
were purchased from Fisher, and peroxidase type VI (POD
reagent) was purchased from Sigma Chemical. n-Chlo-
robutane (BuCl, 99.9%) and pentyl chloride (PeC1, 99%)
were obtained from Aldrich Chemical Company. Potassium
trioxalatoferrateand certifiedACS sodium bicarbonate were
received from the Alfa Company and Fisher Scientific,
respectively.
The stock solution of humic acid was prepared by
dissolving solid humic acid obtained from Fluka AG in a
0.1 N NaOH solution and then filtering it through No. 1
Whatman filter paper. The stock solution was diluted at
desired proportions to be used in the experiments.
Analyses. BuCl in aqueous samples was analyzed by
static head space gas chromatography (GC) using pentyl
VOL. 29, NO. 12, 1995 / ENVIRONMENTAL SCIENCE &TECHNOLOGY 3007
 measure the absorbance of water samples and H202
solutions at 254 nm.
L
Reaction Mechanism in Pure Water. The reaction steps
I I 1 1-4 presented below are accepted in the literature to
represent the mechanism of hydrogen peroxide photolysis
in pure water (12-15). Furthermore, the quantum yield
for the process, which can be defined as the number of
moles of hydrogen peroxide decomposed per mole of light
photon absorbed, has been estimated as 1.0 for the overall
quantum yield (@I) (12,14, 15) and as 0.5 for the primary
quantum yield (ak,) (12-15), Le., for the initiation reaction,
photoreactor for relatively high light intensity, and for low peroxide
feeding tank chamber concentrations.
FIGURE 1. Experimental setup.
light absorptionlinitiation
chloride (PeCl) as an internal standard (5).The feedstocks
containing BuCl were prepared by spiking a known amount
H,O,+ hv - 2'OH (1)

of BuC1-saturatedMilli-Q water into containers containing propagation

predetermined volumes of water at 20 f 1 "C. During the
experiments, samples of about 300 mL were taken from the 'OH + H,O,/HO,- - HO,'/O,- + H,O k, (2)
influent and effluent lines and analyzed as follows: 20 mL
of the sample was transferred into a 30-mL vial and was
spiked with 1mL of PeC1-saturated Milli-Q water. The vial
was then capped and thermostated at 78 "C in a water bath termination
for about 1 h. A 500-pL sample of the equilibrium gas
mixture was drawn from the head space by a gas-tight HO,' + HO,' - H,O, + 0, k, (4)
syringe and was injected into GC equipped with a flame
In addition, the disproportionation reaction of of HO2.i
ionization detector. The coefficient of variation (CV), Le.,
is expected to contribute to H 2 0 ~regeneration (161,as
02*--
the standard deviation expressed as the percent of the
shown below:
arithmetic mean was measured as 10% and 11%for BuCl
and PeCl concentrations, respectively. However, the CV HO,' + O,-* + H,O - H,O, + 0, + OH- k, (5)
for the ratio of the peak areas of BuCUPeCI was 0.6% in the
absence of humic material and 3%in the presence of humic
material. A statistical t-test indicated that the effect of
HO,' - O,*- + H+ pK,,,, = 4.8 (17, 18) (6)

humic acid on the peak areas of BuCl and PeCl was not
significant at 95% confidence interval.
Hydrogen peroxide concentrations in the water samples
collected from the influent and effluent lines of the reactor
were measured by the method of horseradish peroxidase-
catalyzed oxidation of N,N-diethyl-p-phenylenediamine
(PODIDPD) (7). The precision and the accuracy of this
method was checked by diluting the H202 solutions at
various predetermined proportions and by measuring the
concentrations using a prepared calibration curve. The
maximum relative bias estimated using this technique was
10%. The concentration of H202 in the stock solution was
measured by the potassium permanganate method (8).Both
methods were calibrated by the direct spectrophotometric
measurement at 254 nm. W intensity was measured by
using the ferrioxalate actinometry (9, 10). The incident
light intensity measured in this study varied between 2.46
x 10 ~'and6.79x 10-4EinsteinL-'min-'with6-161amps
operating simultaneously. The measured intensity over a
period of 1year showed no deterioration of the lamps, and
the random variation in the measured intensity was within
5% of the average value.
The dissolved or total organic carbon content (DOC/
TOC) of water samples was measured by a Dohrman DC-
80 carbon analyzer, according to the Standard Methods
( I 1 ) . The alkalinity of water was measured by titrating it
with a standardized sulfuric acid ( 1 1 ) . The p H was
measured by an Accumet pH meter (Fisher Scientific). A
Tn'-visible spectrophotometer (Milton Roy) was used to
Reaction 3 above as well as other possible termination
reactions involving reactions of 'OH with HOz' and 02.-
were found to have no significant contribution to the overall
reaction mechanism (4)according to a kinetic model that
is described later in this paper.
Reaction Mechanism in the Presence of Carbonate
Species. The carbonate species, HC03- and C03?-, are
expected to affect the photolysis of hydrogen peroxide in
aqueous solution through various pathways. Both species
are known as effective 'OH scavengers, and hence their
presence may cause a reduction in the oxidation rate of the
target contaminants. Furthermore, the reaction of'OH with
H C 0 3 - and COS2-generates the carbonate radicals (HC03'
and C03'- ) as oxidation transients ( 19). These two forms
of carbonate radicals exhibit similar reactivities toward other
species (20). It was suggested that these radicals may
disappear rapidly by attacking hydrogen peroxide, resulting
in the formation of hydroperoxyl radical (H02./02*-) (21).
Ultimately,H02*/02*- may react to regenerate Hz02 through
disproportionation reaction, or they may react with organic
substances if their reaction rate is relatively high. For pH
values above 8, the radical COS'- becomes dominant (pK,,e
of HC03' is 7.9; ref 22) and reacts with 02*- to terminate the
reaction. These reactions of the carbonate species are
summarized below:
'OH + HCO,-/COS'- - HCO,'/CO,'- + OH- k; ( 7 )
HC0,'- C0,'- + H+ pK,,, = 7.9 (8)

3008 - ENVIRONMENTAL SCIENCE & T E C H N O L O G Y , VOL. 29 N O 12,1995

HCO,'/CO,'- + H2O,/H0,- - nm, b is the optical path length, and [H20~1is the
concentration of H202. The rate of H202 decomposition
HO,'/O,'- + HCO,-/CO,'- k9 (9) can also be described as

c0;- + 0;- - co,,- + 0, k,, (10) rH",o,= (2.303@T6b10~)

[H2021 (17)
Oxidation mechanism of an organic compound, M, in where
this system includes the reaction of M with 'OH and HO2'/
1 - 10-tb[HzOzI
02'-, as follows:

M f 'OH - MOA kM (11)

'= 2.303~b[H,O,]
(18)

M + H0,'/02'- -... M*,xi kM* (12)

Note that q, which is a fraction, becomes close to 1 at
relatively small concentrations of H202. Furthermore, for
Note that here the organic compound is assumed to a CSTR, the mass balance for H202 through substitution of
react only with 'OH and H02'/02'-. However, certain eq 17 leads to
compounds can photolyze under UV light, e.g., certain
aromatics and olefins, and some compounds can also react
with H202 directly,e.g.,phenol. These reactions may further
enhance the oxidation of such compounds.
which implies a linear relationship between the ratio of the
Reaction Mechanism in the Presence of Natural
Organic Matter. Natural organic matter might affect the influent to effluent concentration of hydrogen peroxide,
H2021UV oxidation process through several mechanisms. Le., [H~021~/[H~021 and (lot) if r remains rather constant
First, the process might be inhibited due to absorption of within a narrow range of [HzOZI;furthermore, the slope of
UV light by the organic matter. Since the major source for the straight line becomes equal to 2.303Q~cby.
'OH formation is the photolytic decomposition of H202, Oxidation Reactions. According to the reaction scheme
the rate of which is controlled by the incident light intensity presented above, the oxidation rate of the model compound
available for H202, the fraction of the incident light intensity M (rM) can be expressed by the following equation:
available for H202 is of major concern when the UV-
rM = { k ~ r O H+
l kM*([HOZ'l f [O,'-l)}[Ml (20)
absorbing species such as humic material (HM)are present
in the solution. In addition to this light filtering effect, Additionally, the overall decomposition rate of hydrogen
humic material is also known as an effective'OH scavenger. peroxide (rH20z)will be
Finally, humic material might affect the chain reactions by
consumingtheradicals H02'/02*-.These possible reactions 'H,O, - {k,+ k,['OHl + k,([CO,'-l +
involving humic material might be incorporated into the
reaction scheme as follows: [HCO,'l)l [H2O21- k4,,([HO,'1 + [02'-1)2/2 (21)
HM + hv - excited HM - - - where for sufficiently small concentrations of H202
A,,, (cm-') = €,,*DOC (mg/L) (13) k , = 2.303@p~b10 (22)
'OH + HM - HMO, - - - and
ICHM, s-' (mg/L of DOC)-' (14)

HO,'/O,'- + HM - HM*,,--- In a CSTR, the application of the law of mass balance

kHM*,s-' (mg/L of DOC)-' (15) with respect to H202 and the model compound M respec-
tively leads to the the following expressions at steady state:
Here, A254 is the absorbance of light by humic material
and EHM is the absorptivity of humic material at 254 nm,
and DOC represents the concentration of the humic
material as dissolved organic carbon. The ~ H Mand k H M *
are the rate constants for reaction of humic material, in
terms of DOC, with'OH and HO2*/O2.-,respectively. These
constants are not available in the literature; however, the
extent of radical scavenging by humic material is believed where [MI, and [MI are the influent and effluent concen-
to remain constant during the reaction period (23). trations of M. and
Equations for the Kinetic Model. Decomposition of
HzOz in Pure Water. According to the definition of quantum
yield and the Beer-Lambert law, the overall decomposition
rate of H202 in pure water, rI+02,can be described as follows:
- @ I = @ I ( 1 - 10-mz0211
-
~ H , O ~ T a T 0 (16)
Oxidation Reactions in the Presence of Natural Organic
where Q ~ ithe
s overall quantum yield, lais the light intensity Matter. In the presence of humic material, the transmitted
absorbed by H202, I, is the incident light intensity, E is the light intensity available for absorption by H202 after
absorptivity of H202, which is 19.6 M-l cm-l (14) at 254 traveling a distance of x cm can be expressed for 254 nm

VOL. 29. NO. 12, 1995 /ENVIRONMENTAL SCIENCE & TECHNOLOGY 3009
 4
-
TABLE 1
- 3.5 10=(2.46 6.79).IUp Einsteidl m i n .
-
r=2.59 16 7 6 rnm
Reaction Rate Constants 1 3 22 rnM
[H10,jo=0.?4-2
c12.5
rate constant literature value (M-l s-l)
+
2.7 x 107 7.5 x 109 I O P H - ~ ~i*19)
: -
51.5
2
kz
k3 3.7 (24) e" 1
I

k4 8.3 x 105 ( 1 6 ) 0.5

k5a 9.7 x 107 (16) 0
k6 7.9 105 ( ~ - 1 ) 0 0.004 0.008 0.012
k-6 5 x 1O'O (76) 1-7, Einsteidl
k7 8.5 x 106 + 3.9 x io8 x 1 0 ( p ~ - 1 0 3 1 ( 7 9 )
kg 4.3 x 105 + 3 x 107 x I O ( P H - ~ ~ *(PR
) FIGURE 2. [H~Odo/[H202]- 1 versus lor.
ki o 6.5 x IO8 (25)
kM 3 x 109 (26) value predicted by the model; k; is the value of the
kM * not availableC
kHM not availableC experimental data. The MRE for the normalized concen-
kHM* not availablec trations of BuCl and H202 was found to be less than 12%
a ks I S calculated from pK, = 1.2 (27). ks IS calculated from pK, =
and 5%, respectively. These MRE values were then evalu-
4.8 ( 77, 18). Model sensitivity to these constants IS described i n this ated statistically for model verification. For that purpose,
paper pK,(H202) = 11.8; pK,(H2CO3) = 6.3; pKJHCO3-) = 10 3. a sensitivity analysis of the kinetic model was performed
by assuming an error of 110%in the measurement of each
as process parameter. The sign of error was selected so that
I, = 1 ~ ~ - 30j..i2;1
2 it may result in either lower or upper-bound conditions.
(28)
Then the measured concentrations of BuCl and HZO2were
Accordingly, the average photon intensity available for checked to see whether they remain within the lower and
absorption by H202 within the reactor can be estimated as upper bounds of the model predictions ( 4 ) .
- L'I~ cix J610e-2 cix
I,, = ___ - - -
- Results and Discussion
b b
Experiments in Pure Water. These experiments were
conducted by irradiating in the CSTR a solution of H202
that contained no additional organic or inorganic matter
where r]', which represents the fraction of light available at various light intensities and liquid residence times. The
for absorption by H202,is given by concentration of H202 in the influent and effluent was
measured. The results showed that increasing the light
intensity or the residence time significantly increased the
decomposition rate and, consequently, reduced the effluent
Hence, in the presence of humic material, eq 22 needs to concentration of H202. These results were plotted as
be modified as follows to describe ki: [H202],/[H202]versus lotin Figure 2. A linear relationship
k , = (2.3O3@,cb)(7'Ic,) (31) was observed as expected, according to eq 19. The optical
pathlength, b, was then determined as 7.0 cm from the
Furthermore, the radical reactions with humic material slope of the straight line (2.303@3eby)of 318 Einstein L - I
represented by reactions 28-30 need to be incorporated by substituting the literature values of @T = 1 and = 19.6
into the kinetic model. cm-I M-' at 254 nm and assuming that r] is 1.0. It should
The system of nonlinear ordinary equations (eqs 20- be noted that when corrections were made for 7 for higher
3 1)were solved numerically by using the MATLAB program
concentrations of H202, the slope yielded a slightly different
developed by Math Works Inc. The steady-state concen- b value of 6.8 cm, which was adopted for further modeling
trations of the species [HzO~I, [MI, POHI, ([HO;] + [ 0 2 * - 1 ) ,
purposes even though the model was not sensitive to such
and ([C03*-] + [HCO?'])were predicted under different
a small variation in the b value. The path length of 6.8 cm
specified conditions. For this, the process parameters t,I,,
is a reasonable value for our system, which consists of a 10
b, and [H2O2Ioand the water quality conditions [MI,, pH,
cm diameter quartz reactor surrounded with equally
and Ci (defined as the summation of [C032-l, [HCO3-1,and
distanced 6-16 U V lamps.
[H&03*]),concentration of humic material as DOC, and
the absorbance of the solution as A2j4 = E ~ I M -DOC were Effect of Carbonate Ions. Several experiments were
specified as inputs to the program. The reaction rate conducted in the CSTR by using solutions containing
constants presented in Table 1 were also used as input to various concentrations of BuC1, H202,and carbonate ions.
the program. In these experiments, the UV light intensity and the liquid
The reliability of the kinetic model was tested by residence time were kept constant to be able to interpret
statistical comparison of the effluent concentrations of BuCl the effect of carbonate ions on the oxidation rate of BuCl
and H202 predicted by the model to the experimental and the photolysis rate of H202. In Figure 3 , the data points
measurements. For each set of experiments the mean for normalized concentrations, Le., the ratio of the effluent
relative error (MRE) defined as to influent concentrations, are presented in squares for
I
BuCl and in circles for H202. Each pair of data represent
an experimental run under steady-state conditions for the
selected CT. It is obvious that carbonate ions have a
detrimental effect on the oxidation rate of BuC1, whereas
their effect on the decomposition rate of H202is relatively
was computed. Here, Nis the number of samples; Xi is the negligible. According to Figure 4, the effect of carbonate

3010 ENVIRONMENTAL SCIENCE &TECHNOLOGY / VOL. 29, NO. 12, 1995

 [H,O,l0=284 pM;

uI
[BUCI]~=~
pM;
0.8
r=1.16 min; pH=7.6;
pH=7.6; C,=4 mM;
-
.-rn
0.6
10=2.46*l o 4 Einsteidl-min r=7.16 min:
10=2.46*10"Einstein4-min
0.4 - 'k&O
t 0.2 ... .. . .. ..%*=io3 M.',.]

-
- . , k..'=104m
M.'s''
- 0
0 10 20 30 40 0 4 a 12
C,, mM Fluka Humic Acid as DOC, mgll

FIGURE 3. Model sensitivity to k ~ " . FIGURE 5. Effect of humic acid.

1 of carbonate ions on H202 photolysis on the other hand

[H,0,10=454 pM;
g 0.8
kl=6.3.10-4 s.' [B~Cl]~=1.2 indicates
pM; that the overall quantum yield (@TI remains
0 T =1.76 min; unchanged in the presence of carbonate ions. For example,
U p H=8.2;
0.6 the observed rate constant (kobs) and the primary rate
-.-B
0.4
10=2.46*104Einsteidl-min
constant ( k J were determined to be 12.7 x loT4s-l and 6.3
x s-l, respectively, for the experimental conditions
L

o.2
0
0
u 40
C,. mM
a0
FIGURE 4. Model verification: the effect of G.
ions on BuCl removal becomes even more pronounced
under conditions of lower concentration of BuCl. On the
other hand, the decomposition rate of H202 remains rather
constant for CT values up to 90 mM. Similar effects of
carbonate on the oxidation rate of 4-chloronitrobenzene
was reported earlier (28)without a mechanistic explanation.
The kinetic model was used to predict the concentrations
of BuCl and H202presented in Figure 3. The experimental
conditions (Le., z = 7.76 min, Io= 2.46 x Einstein L-I conducted in the presence of humic acid clearly showed
min-', b = 6.8 cm, [H20210= 270 pM, [MI, = 8 pM, pH =
7.6, and CT)as well as all the reaction rate constants were
fed to the program as input. However, since the reaction
rate constant kh,*, which is the rate constant for the reaction experimental conditions (Le., I,, [H2O2Io,[MI,, t,and pH)
of radicals H02*/02'- with BuCl, is not available in the
literature, some arbitrary values were assigned to the k ~ * humic acid as measured by DOC.
(0, lo3, and lo4 M-' s-l), and the concentrations of H202
and BuCl predicted based on these arbitrary values were
compared to the experimentally measured concentrations.
The effluent concentrations of H202 were found insensitive
to the value of kM*; however, the concentration of BuCl was
insensitive only when the k ~ was * less than about lo3 M-'
SKI. This indicates no appreciable reaction of H02*/02*-
with BuCl, which is reasonable since the rate constant of
HO2./O2.-with aliphatic compounds is generally low (16).
Consequently, this reaction was ignored for modeling
purposes for the rest of the study.
For verification of the kinetic model, the model equations obtained in the presence humic acid to model predictions
were solved for the experimental conditions presented in
Figure 4. The value for k ~ was * assumed to be zero, as
implied by the sensitivity analysis. The data points for BuCl
and Hz02 concentrations are presented in Figure 4 by the
squares and circles, respectively, and the solid lines
represent the predicted values by the model. It was
concluded that the experimental data were predicted
remarkably well by the kinetic model.
In view of the kinetic model, the effect of the carbonate Figure 6. This value for k H M is remarkably close to 1.7 x
ions on BuCl concentration profiles can be explained by
the reactions of HC03- and C032- ions with 'OH in
competition with BuCl. The lack of any significant effect
presented in Figure 4. Hence, the ratio kl / k o b s , which equals
@'p/@Pabstis 0.5 (note that kl = 2.303@.,~bl,and that kobs =
2.303@0b,~bl,).This calculation indicates that the quantum
yield observed in the presence of carbonate ions is about
1. This conclusion is understandable in view of the
proposed reaction mechanism in which the carbonate
radicals generated by reaction of 'OH with HC03-/C032-
(reaction 7) react further with H202 (reaction 9) and thus
regenerate the chain-propagating radicals, H02'/02*-. Ac-
cordingly, we conclude that the carbonate ions behave
differently than organic radical scavengers, e.g., acetic acid
or allyl alcohol, which can reduce the quantum yield to 0.5
by quenching the chain reactions of hydrogen peroxide
photolysis (14, 15).
Effect of Natural Organic Matter. The experiments
that the rate of H202 photolysis and the oxidation rate of
BuCl reduced significantly with increasing concentration
of humic acid (Figure 5 ) . In these experiments, all the
were maintained constant, varying only the quantity of the
To be able to test the kinetic model in the presence of
humic material, the rate constants for the reaction ofhumic
material with'OH ( ~ H M )and with H02'/02'- ( ~ H M * )needed
to be estimated. The kinetic model itself was used for this
purpose by first performing a sensitivity analysis with
respect to the rate constant kHM*. The results showed that
the model was insensitive to the value of ~ H M in * the range
of 0-lo4 s-l (mg/L)-', and therefore the reaction of humic
material with H02'/02*- was ignored for further modeling
purposes (4). The rate constant ~ H Mfor the humic acid
was estimated by force fitting the experimental data
by the least squares method, Le., by minimizing the
summation of the squared differences between the pre-
dicted and measured concentrations ( 4 ) . As shown in
Figure 6, within the range of ~ H values
M tested, that is 1 x
103-3 x lo4 s-l (mglL of DOC)-', the model was found to
be relatively insensitive; however, the best fit of the model
prediction to the experimental data was still obtained for
a ICHM value of 1.6 x lo4 s-] (mg/L of DOC)-', as shown in
IO4 s-l (mg/L of DOC)-', which was estimated by an
ozonation technique as an average value for the natural
organic matter in several Swiss lakes and rivers (29)and to

VOL. 29. NO. 12, 1995 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 301 1

-0
0 400 800 1200
[ H,O,la, PM
FIGURE 6. Estimation of ~HM.
2.3 x lo4 s-' (mg/L of DOC)-' estimated by Haag by the
ozonation technique for 11North American natural waters
(30).
For verification of the kinetic model, the results of the
experiments presented in Figure 5 were used. The equa-
tions of the kinetic model were solved numerically by using
the independently determined parameters of b = 6.8 cm,
FHM = 0.0867 cm-' (mg/L of DOC)-', and k H M = 1.6 x lo3
s-' (mg/L of DOC)-'. In Figure 5, the open squares and
circles represent the experimental data for BuCl and H202,
and the solid lines trace the values predicted by the kinetic
model. According to the statistical analysis, the mean
relative error (MRE) between the model predictions and
experimental data was found to be less than 4% for both
H202 and BuCl concentrations. Furthermore, the data
points remained within the the dashed lines, which trace
the model predictions for the lower- and upper-bound
conditions (based on f 1 0 % error).
The remarkable success of the model in predicting the
effect of humic acid (the model is equally effective in
predicting the effect of two other humic materials as well
( 4 ) )indicates that the assumptions made while developing
the model, Le., that humic material primarily functions as
a competitor to the target contaminants through the dual
effects of 'OH scavenging and light absorption, are ac-
ceptable. However, humic material might also promote
the radical chain reactions to generate H202, as reported
for sunlight-induced reactions in natural waters (31,32).
Hence, to test the possibility of generating any significant
amount of H202 by direct UV photolysis under our
experimental conditions, a solution of BuCl and humic acid
(DOC= 4.66 mg/L) was irradiated by UVlight as anintensity
of 2.46 x Einstein L-' min-' in the absence of any
H202for up to 17 min. The maximum concentration of
H202detected in the solution was 2pM. During that time,
the BuCl concentration was reduced by about lo%,
indicating the presence of 'OH in the reaction mixture.
However, compared to the experimental results obtained
in the presence of H202, these numbers are still too small,
implying that the role of the humic material as a promoter
of the chain reaction is only minor, and humic material
primarily functions as an effective inhibitor in this process.
The slight difference between the measured and the
predicted values of H202 in our experiments might be due
to this chain-promoting effect of the human substances.
This effect has not been incorporated into the kinetic model
because of the difficulty in defining the interaction of the
UV light with the ambiguous chemical structure of the
humic material.
Effect of [HzOz],,.The experimental results regarding
the effect of influent H202 concentration on the normalized
concentrations of H202 and BuCl are presented in Figure
3012 ENVIRONMENTAL SCIENCE €
TECHNOLOGY
i / VOL. 29, NO. 12, 1995
k,,=l 6.10's ' ( m a DOC) '
[H,0,10=284 pM,
(BUCI]~=X pM.
r=l 76 min.
E DOC=]X m u 1
Q
0 4 a 12
PH
FIGURE 7. Effect of pH.
6. The concentration of BuCl decreased dramatically with
increasing [H2O2Iobut then stabilized for [HL0210 greater
than about 500pM. This phenomenon could be explained
by two opposing effects of H202 on the steady-state
concentration of 'OH, Le., 'OH is generated through
photolysis of H202but is also consumed by H202. This
finding implies that there should be an optimal dosage of
the chemical oxidant H202 to provide the maximum removal
of the contaminant per unit of H202.
The normalized concentration for H202was independent
of [H202Iounder the experimental conditions. This ob-
servation implied that the kobs according to eq 27 remained
unchanged as [H202Io was varied. This observation is
understandable since the light intensity was relatively high
and the contribution of the primary photolysis to the overall
decomposition rate of H202 was dominant over the
contributions of the subsequent radical reactions. Hunt
and Taube (12) also reported that the observed quantum
yield was independent of H202concentration at high light
intensities.
Effect of pH. The effect of pH on concentrations of
BuCl and H202 in the presence of humic acid is presented
in Figure 7. The highest oxidation efficiency, Le., the lowest
BuCl concentration, was obtained at acidic pH values, and
it was independent of pH for pH values below 5 where
carbonic acid constitutes the major fraction of Cl. However,
the oxidation efficiency was reduced drastically with
increasing pH above pH 5. At this pH, the equilibrium
shifts toward the bicarbonate ion, which is expected to
successfully scavenge 'OH. Increasing the pH beyond 7
transforms the bicarbonate to the carbonate ion, which
has an even higher reactivity toward 'OH (the rate constant
for carbonate and bicarbonate ions are respectively 3.9 x
lo8 and 8.5 x lo6 M-' s-l) and as a result is more successful
in reducing the oxidation rate of BuC1. Furthermore, the
reaction of 'OH with HOz- is much faster than its reaction
with H202. For example, the value of k2 is 2.7 x 10- at pH
7 and 1.5 x lo8 M-ls-' at pH 10. Hence, at higher pH less
'OH is available for oxidation of n-chlorobutane. These
results indicate that acidification ofwater prior to treatment
by H202/UV may prove to be quite beneficial in practical
applications. The H202 concentration was basically in-
dependent of pH, except that [H202] decreased slightly at
the high end of the pH range, probably because of more
pronounced attack of carbonate radicals on H ~ O ~ I H O, J
as depicted by reaction 9.
The trends for both H202 and BuCl were predicted well
by the kinetic model. In Figure 7, the solid lines present
the vaues predicted based on the average values of the
measured process parameters, and the dashed lines trace
the model predictions for the lower- and upper-bound are susceptible to 'OH, especially aliphatic compounds.
conditions (based on f 1 0 % error). It should be noted that However, the validity of the model may need to be checked
these solutions contain humic acid, which might change further for compounds that might react appreciably with
its structure and behavior as a function of pH. However, H02'/02'- as well. The model is capable of accommodating
in the model, the rate constant ICHM and the light absorbance mixtures of organic compounds as long as the rate constant
of humic acid estimated and measured respectively at pH of each individual compound with 'OH is available. The
7.6 were applied throughout the pH range of 2-10. model needs to be be modified to accommodate organic
Furthermore, the reaction products of BuCl and humic acid compounds that might photolyze or react directlywith H202.
might differ with varying pH. In fact, the model seems to The successful verification of the model on various field
fit the data better for the pH values above 6. However, samples is going to be the subject of an upcoming
even under these uncertainties, the experimental data still publication.
remained within the lower and the upper bounds of the
predicted values, which indicates satisfactory performance Acknowledgments
of the model, especially for the purpose of process This work was supported entirely by the US. National
engineering design. Science Foundation, Environmental Engineering Program
directed by Dr. Edward H. Bryan. C.H.L. was a graduate
Summary and Conclusions student at Drexel University at the time of this study. This
This study was conducted to investigate the HzO~/UV workis based on his Ph.D. Dissertation, which received the
process for oxidation of a model organic compound (BuCl) 1994 Doctoral Thesis Award of the Association of Envi-
as well as for photolytic decomposition of H202 under ronmental Engineering Professors sponsored by Engineer-
various types of water quality and operational conditions. ing Science.
The experiments were conducted in a continuous-flow
stirred tank reactor (CSTR) in the presence of carbonate Abbreviations and Symbols
ions and humic acid at various pH values, residence times, CSTR continuous-flow stirred tank reactor
light intensities, and hydrogen peroxide dosages. It was BuCl n-chlorobutane
observed that the oxidation efficiencyfor the target pollutant *OH hydroxyl radical
BuC1, i.e., ([MI, - [MI)/[Ml,, was determined by the rate
of photolytic decomposition of H202 as well as the quality cv coefficient of variation
of the water to be treated. The higher the rate of H202 POD/DPD N,N-diethyl-p-phenylenediamine
photolysis, the higher was the efficiency of oxidation of the 5 residence time
pollutant; the rate of photolytic decomposition of H202 DOC/TOC dissolved/total organic carbon
increased with increasing incident light intensity and H202 @T overall quantum yield
dosage but decreased with increasing W absorbance (A254)
@!J primary quantum yield
of the aqueous solution. The oxidation efficiency of BuCl
was hindered with increasing pH, the total inorganic carbon light intensity absorbed by H202
concentration (CT),and the concentration of humic material IO incident light intensity
in water. The extent of the dependency of the oxidation E absorptivity of H202 at 254 nm
rate of the organic compound on the presence of humic b optical path length
acid is so strong that avoidance of this process for treatment absorbance of light by humic material
A254
of highly colored waters might be recommended. The
adverse effects of carbonate ions and high pH can be EHM absorptivity of humic material at 254 nm
alleviated by acidification of water prior to application of 11 a fraction defined by eq 3
the process. 1' fraction of light available for absorption by
A kinetic model based on a proposed reaction mech- H202
anism was tested by comparing the model predictions to MRE mean relative error
the experimental data for various operational conditions
and water quality. The major inputs to the model are the
CT +
[C032-l [HC03-] + [HzC03*]

operational parameters (Le., [H202Io,I,, and t)and the water

quality parameters described by pH, CT, DOC, and A2s4.
The model exhibited a great capability for predicting the
oxidation efficiency for the model compound and the rate
of decomposition of H202 by photolysis under all the
conditions tested. Carbonate/bicarbonate ions were found
to compete effectively with BuCl for consumption of 'OH,
causing a dramatic reduction in the oxidation rate of BuCl.
However, it was observed that carbonate ions did not affect
the overall quantum yield for hydrogen peroxide decom-
position, which was predicted to be 1.0 in the presence as
well as the absence of carbonate ions. On the other hand,
humic material not only competed with BuCl for *OHbut
also effectively blocked the light available for photolysis of
H202. The rate constant for the reaction of'OH with humic
acid was estimated to be 1.6 x lo4 s-l (mg/L of DOC)-'.
This model is expected to work reasonably well for
predicting the oxidation rate of organic compounds that
literature Cited
(1) Glaze, W. H.; Lay, L. Proceedings of the 9th International Ozone
Association; International Ozone Assoc.: New York, 1989; pp
688-708.
(2) Yao, C. C. D.; Haag, W. R.; Mill, T. Chemical Oxidation Technology
for the Nineties, 2nd International Symposium, Vanderbilt
University, Nashville, TN, 1992; Technomic: Lancaster, 1994.
(3) Wolfrum, E. J. Symposium on Environmental Applications of
Advanced Oxidation Techniques; EPRI & NSF: San Francisco,
1993.
(4) Liao, C. H. Ph.D. Dissertation, Drexel University, Philadelphia,
1993.
(5) Ioffe, B. V.; Vitenferg, A. G. Head-space Analysis and Related
Methods in Gns Chromarography;John Wiley & Sons, Inc.: New
York, 1984.
(6) Zellner, R.; Exner, M.; Herrman, H. J.Atmos. Chem. 1990, 10 (4),
41 1-415.
(7) Bader, H.; Sturzenegger, V.; Hoigne, J. Water Res. 1988, 22 (9),
1109-1115.
(8) Vogel, A. A Text Book of Quantitative Inorganic Analysis, 4th ed.;
Longman Scientific and Technical: Harlow, U.K. 1978.

VOL. 29, NO. 12, 1995 / E N V I R O N M E N T A L SCIENCE & TECHNOLOGY 3013

 19) Parker, C. A. Proc. R. SOC. London, Ser. A 1953, 220, 104-116.
!IO) Hatchard, C. G.; Parker. C. A. Proc. R. Soc. London, Ser. .A 1956,
235, 518-536.
( 11) StandardMethods for tlzeExamination of Waterand Wustewater.
14th ed; APHA, AL%TVA,LVPCF: Washington, DC, 1985.
112; Hunt, i. P.;Taube, N.J. Am. Chern. Soc. 1952, 74, 5999-6002.
r13; iVeeks, J. L.: Matheson, M. S. J . Arn. C h e m Soc. 1956, 78 (11,
1273-1279.
i 1.11 Bauendale, J. H.: Wilson. I. A. Trans. FaruduySoc. 1956,53,344-
-6.
Chen, I. C . J. Am. Chern. Soc. 1959. 81, 4141-
olman, D. H.:
4144.
I 16) Bielski, B. H. J.; Cabelli, D. E.; Arudi. R. L.J. Phys. Chem.Ref: Dntn
1985, 14, 1041-1100.
I 1 7 Hehar, D.; Czapski, C.; Rabani, 7.; Dorfman. I,. M.; Schwarz, 14.
A, J. Phys. Chem. 1970, 74, 3209-3211.
(18) Bielski, B. H. J.;Allen,A.0.1.Pliys. Chem. 1977, 81, 1048-1050.
(19) Buxton. G. V.; Greenstock, C. L.;Helman, LV. P.; Ross, A. R.1.
Phys. Chem. Ref: Data 1988, 17, 513-886.
i2Oi (.:hen, S. N.; Hoffman, bf. 2. Radiat. Res. 1975, 62, 18-27.
(21) Peyton, G. R.: Glaze, LY. H. Environ. Sci. Teechnol. 1988. 22 (71,
761 --767.
112) Eriksen. T. E.; Lind, I.: hlerenyi, G. Radinr. Phys. Chern. 1985,26.
19-199.
( 2 3 ) Hoigne. 1. Handbook of Ozone TechnologyandApplicutions; Rice,
R. G., Ketzer, A,, Eds.; Ann.4rhor Science: Ann Arbor. hlI. 1982:
i'ol. 1, Chapter 12.
121) tarhataziz; Koss, A. B. Selected Specific Rates of Reactions of
Iransients from LVater in Aqueous Solution. 111. HydroTl
3014 ENVIRONMENTAL SCIENCE & TECHNOLOGY VOL 29, NO 1 2 1995
Radical and Perhydroxyl Radical and Their Radical Ions. )Vat/.
Stand. Ref: Data Ser. ( U S . Natl. Bur. Stand.) 1977. 59.
!25) Neta, P.; Huie, R. E.; Ross, A. B. J . Phys. Clzern. Ref: Datu 1988,
17. 1027-1284.
(26) Haag. W. R.: Hoigne. 1. Chemosphere 1985, 14 (111121, 1659-
1671.
127) Sehested, K.; Rasmussen, 0. L.; Fricke, H. 1. Plzys. Chem. 1968,
72 ( I ] , 626-628.
Laat, J. D.; Duguet, J. P.; Bonnel, C.; Dore, hl.
Ozone Sci. Eng. 1990, 12, 73-94.
(291 Nowell, L. H.; Hoigne, I. Water Res. 1992, 26 (51, 599-605.
(30) Haag, W.R. Personal communication, 1992.
(311 Cooper, iV. 7.; Zika, R. G.; Petasne, R. G.; Fischer, A. M. In Aquatic
HurnicSitbstances: Influence on Fateand Treatment ofPollutunts;
Suffet, I. H.. MacCarthy. P., Eds.; American Chemical Society:
LVashington, DC, 1989; Chapter 22.
132) Hoigne, 1.; Faust, B. C.; Haag, LV. R.; Scully, F. E.; Zepp, R. C. In
Aquatic Humic Substances: Influence on Fate and Treatment of
Pollutants; Suffet, I. H., MacCarthy, P., Eds.: American Chemical
Society: Ll'ashington, DC, 1989; Chapter 23.
Received for review April 25, 1994. Revised manuscript re-
ceived June 28, 1995. Accepted July 31, 1995.@
ES940254I
Abstract published in AdvanceACSAbstracts, November 1 , 1995.