Degradation of ethylene glycol using Fenton's reagent and UV

Article abstract of DOI:10.1016/S0045-6535(00)00597-X

Oxidation of ethylene glycol in aqueous solutions was found to occur with the addition of Fenton's reagent with further conversion observed upon UV irradiation. The pH range studied was 2.5-9.0 with initial H₂O₂ concentrations rangin

Full text of DOI:10.1016/S0045-6535(00)00597-X

Chemosphere 45 ꢀ2001) 101±108
www.elsevier.com/locate/chemosphere
Degradation of ethylene glycol using Fenton's reagent and UV
a,
b
B. Dietrick McGinnis , V. Dean Adams , E. Joe Middlebrooks
c
*
a
McGinnis and Associates, 6920 Forsythia, Reno, NV 89506, USA
Department ofCivil Engineering, MS 258, University ofNevada, Reno, NV 89557, USA
b
c
3
60 Blackhawk Lane, Lafayette, CO 80026, USA
Received 27 March 2000; received in revised form 8 September 2000; accepted 25 September 2000
Abstract
Oxidation of ethylene glycol in aqueous solutions was found to occur with the addition of Fenton's reagent with
further conversion observed upon UV irradiation. The pH range studied was 2.5±9.0 with initial H concentrations
ranging from 100 to 1000 mg/l. Application of this method to airport storm-water could potentially result in reduction
of chemical oxygen demand by conversion of ethylene glycol to oxalic and formic acids. Although the amount of H
added follows the amount of ethylene glycol degraded, smaller H doses were associated with increases in the ratio of
ethylene glycol removed per unit H added indicating the potential of pulsed doses or constant H feed systems.
Ethylene glycol removal was enhanced by exposure to UV light after treatment with Fenton's reagent, with rates de-
2
O
2
2
O
2
2
O
2
2
O
2
2 2
O
pendent on initial H
formic acids, have been shown to be mineralized in other HO generating systems presenting the potential for ethylene
2
O
2
concentration. In addition to ethylene glycol, the principle products of this reaction, oxalic and
Å
Å
glycol mineralization in this system with increased HO production. Ó 2001 Elsevier Science Ltd. All rights reserved.
Keywords: Hydrogen peroxide; Ultraviolet photodegradation; Hydroxyl radical; Deicing solutions
1
. Introduction
Because ethylene glycol is relatively non-toxic, the LD50
for ethylene glycol has been determined to be approxi-
mately 11 g/kg for domestic animals ꢀEvans and David,
1974). The environmental impact is principally due to
the relatively high COD and the quantities of ethylene
glycol used for deicing.
Ethylene glycol is used as an airplane and runway
deicer and poses a disposal problem for many airports.
With a typical deicing formulation having a chemical
oxygen demand ꢀCOD) of 1.39 mg/mg ꢀUnion Carbide
Chemicals, 1992), the introduction of ethylene glycol
from deicing into wastewater can drastically increase
treatment needs. A typical deicing event consisting of
Since ethylene glycol is not chemically altered during
deicing, recycling of deicing solutions has been presented
as a treatment alternative. The recycling alternative has
been limited by both economics as well as the quality of
recovered materials ꢀMericas and Wagoner, 1994). Used
deicing solutions are typically disposed of in publicly
owned treatment works ꢀPOTWs), impoundments or re-
leased directly into the environment. Airport runo€ con-
taining ethylene glycol can impact the quality of surface
water due to its high COD, create hazards to wildlife and
domestic animals when impounded or add excess COD to
wastewater treatment plants during cold weather when
the facilities are sensitive to COD overloading.
1
has a 5-day biological oxygen demand ꢀBOD ) equiva-
000 gallons of deicing ¯uid ꢀ20±80 % ethylene glycol)
5
lent to the daily domestic wastewater generated by ap-
proximately 5000 people ꢀMericas and Wagoner, 1994).
*
Corresponding author. Tel.:+1-775-972-1725; fax: +1-775-
9
72-8062.
E-mail addresses: dietrick@plumasnet.com ꢀB. Dietrick
McGinnis), vdadam@scs.unr.edu ꢀV. Dean Adams).
0
045-6535/01/$ - see front matter Ó 2001 Elsevier Science Ltd. All rights reserved.
PII: S 0 0 4 5 - 6 5 3 5 ꢀ 0 0 ) 0 0 5 9 7 - X

102
B. Dietrick McGinnis et al. / Chemosphere 45 ꢀ2001) 101±108
Although ethylene glycol is relatively biodegradable,
photodegradation processes possibly using natural sun-
light. This system is of particular interest due to the
potential for use with variable treatment ¯ows and low
temperatures associated with deicing during winter
there are two problems with biological treatment sys-
tems on the scale required for the treatment of airport
wastewater. First, typical airport wastewater containing
Å
approximately 5% ethylene glycol has a CBOD
3
5
of
0,000 mg/l ꢀMericas and Wagoner, 1994) and can
storm events. Although descriptions of HO generating
systems for use with a variety of compounds are avail-
able, treatment of airport storm-water is a novel appli-
cation.
overwhelm POTWs during winter storm events. Second,
due to the high concentrations and large quantities of
ethylene glycol required for deicing, large impound-
ments or lagoons are required for adequate COD re-
duction. In addition to the construction and
maintenance cost of these impoundments, they create
both direct health hazards and sources of ground and
surface water contamination.
2. Materials and methods
2.1. Reagents and bu€ers
Chemical removal by reaction with hydroxyl radicals
Å
Acetalaldehyde, ethanol, ethylene glycol, formalde-
ꢀ
HO ) created by the use of Fenton's reagent may be
2 2
hyde ꢀ37%), glacial acetic acid, H O ꢀ30%), methanol,
able to reduce the COD of airport wastewater. Similar
compounds have been shown to degrade using Fenton's
oxalic acid and the inorganic reagents were purchased
from Fisher Scienti®c. Formic acid ꢀ99%), glycoalde-
hyde ꢀ98%), glycolic acid, glyoxal ꢀ40%) and glyoxilic
acid ꢀ98%) were purchased from Aldrich.
3
reagent. Polyethylene glycol is degraded in O /UV sys-
tems ꢀLanglais et al., 1991) and undergoes photodegra-
dation with the addition of iron carboxylate and nickel
and iron dithiocarbamates ꢀLemaire, 1993). Wang et al.
All tests were conducted in a solution with 200 mg/l
NaHCO
using dilute HCl or NaOH in addition to various con-
centrations of H and ethylene glycol. Although the
composition of water contaminated with ethylene glycol
can vary greatly, including NaHCO and CaCl was
3 2
and 50 mg/l CaCl added and the pH adjusted
Å
ꢀ
1992) proposed HO produced by UV absorbed by
O
H
2
2
as a mechanism for degradation of ethylene glycol.
2 2
O
The reaction of several oxidized two-carbon species in
photo Fenton systems was described by Klementova
and Wagnerova ꢀ1994). The reactive systems involved
organic acids, aldehydes and alcohols. The systems were
similar to those used by Allen et al. ꢀ1996), that pro-
3
2
intended to provide a more realistic water chemistry for
the tests. For the degradation evaluations, concentra-
tions of 100, 500 and 1000 mg/l H
2
O
2
, 200, 500 and 1000
2‡
Å
duced OH capable of unspeci®c oxidation reactions
with a collection of organic compounds. Oxalic, glyox-
mg/l ethylene glycol, 10, 50 and 100 mg/l FeSO
and a pH range of 3.0±6.4 were evaluated.
4
as Fe
,
Å
ilic and glycolic acids have been shown to react with HO
ꢀKarpel and Dore', 1997). In the systems described by
Karpel and Dore' ꢀ1997), glycolic type compounds were
The e€ect of post-treatment by UV was also evalu-
ated. The two Fenton's reagent systems used, both
2
‡
contained 1000 mg/l ethylene glycol and 100 mg/l Fe
before addition of H . Two di€erent H additions
Å
oxidized when they reacted with HO and were ulti-
mately mineralized.
2
O
2
2 2
O
were used, 50 mg/l which was calculated to be a cost
e€ective addition, and 1000 mg/l, which was calculated
to be near optimal for ethylene glycol removal, but not
as cost e€ective.
Å
Applications of HO generating systems can be cost
inhibitive. One method to overcome this is to enhance
Å
HO generating systems or couple them to cheaper sec-
ondary methods. This can include increasing the e-
ciency of the system through photo-enhancement
Å
2.2. Experimental methods
ꢀPignatello et al, 1999) or using HO generating systems
as a pretreatment for bioremediation ꢀMiller et al.,
2‡
Various concentrations of ethylene glycol, Fe and
2 2
H O as well as various pH conditions were evaluated.
Å
1
996). These applications of HO generating systems
may rely on production of other radical processes that
can be created in applications of Fenton and photo
Fenton chemistry to increase eciency ꢀSafarzadeh-
Amiri et al., 1997; Watts et al., 1999; McGinnis et al.,
Bulk solutions at designated pH and ethylene glycol
concentrations were generated and added to 44 ml vials.
2
‡
2 2
Fe and H O were then added and the vials sealed to
be analyzed twenty minutes to four hours later. The
reaction with respect to ethylene glycol was found to be
relatively fast with no additional degradation of ethylene
glycol occurring after about 10 min. Only initial and
®nal concentrations of ethylene glycol were determined.
A Trojan model 702 UV light with a 650 ml stainless
steel jacket was used as a UV source. The generated UV
light had a maximum output at 265 nm. The Trojan
Model 702 was connected via a variable speed peristalic
2
000).
Hydroxyl radical generating systems such as the
Fenton reaction have the potential for treating airport
winter storm runo€ by chemical oxidation of ethylene
glycol. Ethylene glycol waste treated with Fenton's re-
agent may contain photo active constituents such as
2‡
carboxyl radicals ꢀMcGinnis et al., 2000) as well as Fe
that could be utilized in more passive and less expensive

B. Dietrick McGinnis et al. / Chemosphere 45 ꢀ2001) 101±108
ni®cance of the studied parameters ꢀH
103
2
O
2
addition, pH,
2
‡
Fe addition and initial ethylene glycol concentration)
and their interaction with each other.
2
.3. Analytical methods
2 2
Oxalic acid, phenol and H O were determined using
a Hewlett Packard 1050 HPLC with an Absorbosphere
C18 4.8 mmÂ250 mm column and 1:1 acetonitrile:water
mobile phase pumped at 1 ml/min. The UV/VIS detector
was set at 254 nm and a 100 ll sample volume used. For
the determination of glycolic acid, glyoxilic acid and
formic acid, a Hewlett Packard 1050 HPLC with an
Aminexâ HPX-87H Ion Exclusion Column, 300
mmÂ7.8 mm was used. The mobile phase was 0.005 M
sulfuric acid pumped at 0.5 ml/min. The UV/VIS de-
tector was set at 210 nm and a 100 ll sample volume
injected. Ethylene glycol, glycoaldehyde and acetic acid
were determined using a Hewlett Packard Model 5890
GC/FID with a Supelco Nukol 15 mÂ0.53 mm column,
a carrier of 15 ml/min helium and a 10 ll sample volume
directly injected. The temperature program for the GC
started at 67°C for 1 min then increased to 140°C at 8°C/
min. pH was determined using an Orion Model 230A
pH Meter with a Fisher Glass Combination pH Elec-
trode ꢀAPHA, 1995).
Fig. 1. UV reactor. Sample constantly pumped from the stirred
bubbler through te¯on tubing to the UV source then returned
to the bubbler.
pump to a 850 ml modi®ed bubbler with te¯on tubing
ꢀ
Fig. 1). A sampling port was installed at the UV source
jacket exit. A 2.5 cm coarse fritted disk at the bottom of
the bubbler attached to a manifold supplying air was
used to aerate the contents of the bubbler. The solution
in the bubbler was kept constantly stirred by a magnetic
stirrer. One liter of solution was circulated at 1 l/min
between the UV source and the bubbler resulting in 21 s
of aeration within the bubbler and 39 sec of UV expo-
sure to the light source per minute. The UV radiation
3. Results
3
dose was determined to be 200 mW-s/cm following the
method of Harris et al. ꢀ1987) for ferrioxalate acti-
nometery.
Ethylene glycol loss was noted in several of the sys-
tems studied ꢀFig. 2). The reaction was relatively rapid,
2 2
with 90% of the H O depleted and no signi®cant change
Rates were considered signi®cant at the 95% con®-
dence interval. Stated errors were also based on a 95%
con®dence interval. The quadratic surface model was
developed using SAS Release 6.08 for Widows 3.10
in ethylene glycol concentration observed after 12 min.
As indicated by the P values, losses in ethylene glycol
2 2
were dependent on the amount of iron and H O added
and initial ethylene glycol concentration ꢀTable 1). The
signi®cance of pH was most likely due to its strong
ꢀ
SAS Institute). This model is used to estimate the sig-
2 2
Fig. 2. Ethylene glycol and H O loss with oxalic acid and formic acid production in Fenton's reagent systems at pH 3 with initial
2
‡
concentrations of 100 mg/l Fe , 1000 mg/l H
system.
2 2
O , and 1000 mg/l ethylene glycol. A small amount of acetic acid was also detected in the

104
B. Dietrick McGinnis et al. / Chemosphere 45 ꢀ2001) 101±108
Table 1
2
‡
2 2
F-ratios and P values for amount of H O and Fe added, initial ethylene glycol concentration, pH and regression components for
a
both the removal and cost factor models
Parameter estimates
Removal
Cost factor
F-ratio
P value
F-ratio
P value
H
Fe
2
2
O
2
13.68
47.10
25.94
12.89
96.5
0.000
0.000
0.000
0.000
0.000
0.005
0.000
0.000
19.02
6.46
2.81
4.17
30.6
8.66
2.88
12.5
0.000
0.000
0.023
0.002
0.000
0.000
0.015
0.000
‡
Initial ethylene glycol
pH
Linear
Quadratic
Crossproduct
Total regression
4.13
9.63
32.9
a
All P values were less than 0.05 indicating that each factor was signi®cant at the 95% con®dence interval. Initial concentrations of 100,
2‡
2 2 4
500 and 1000 mg/l H O , 200, 500 and 1000 mg/l ethylene glycol, 10, 50 and 100 mg/l FeSO as Fe , and a pH range of 3.0±6.4 were
evaluated.
correlation with iron concentration ꢀhigh concentrations
of iron could not be tested at high pH due to solubility
constraints). Each of the components of the quadratic
surface model ꢀlinear, quadratic and crossproduct) was
signi®cant ꢀTable 1) and was used to construct a model
to predict ethylene glycol degradation ꢀTable 2, Fig. 3).
In addition to evaluating the system for removal, a
cost factor was derived. The cost factor was a function
Ethylene glycol loss=ꢀH
2
O
2
‡ 1=3FeSO
4
†
ꢀ1†
To determine the most ecient Fenton's reagent system,
removal was substituted with the cost factor in the
quadratic surface response model. The results were
similar with all the measured variables being signi®cant
at the 95% con®dence interval as indicated by the F-
ratios and P values less than 0.05 for both the removal
and cost factors in Table 1. The F-ratios, P values, and
parameter estimates for both the removal and cost fac-
tors are also presented side by side in Tables 1 and 2. All
of the regression components and most of the parameter
estimates were signi®cant ꢀTables 1 and 2).
2‡
of the removal and relative cost for Fe ꢀas FeSO
. A limited survey of industrial suppliers of FeSO
and H determined that FeSO is approximately one-
third the cost of H by weight ꢀcorrected for con-
4
) and
H
2
O
2
4
2
O
2
4
2
O
2
centrations). A unitless cost factor was calculated using
Eq. ꢀ1). This cost factor had the advantage of being
For both the removal and cost factor model, the
stationary point was a saddle point. Uncoded parameter
values for the model indicate increases in removal with
increasing amounts of all the added parameters ꢀTable
3). The pH data were in¯uenced by iron addition and
not e€ectively controlled in the experiment. Subse-
applicable to FeSO
ratio of the cost of FeSO
4
and H
2
O
and H is 1:3, respectively.
2
at any cost, as long as the
4
2 2
O
The cost factor is applied as an indicator of relative
expense accounting for unit cost di€erence between
4 2 2
FeSO and H O .
Table 2
Parameter estimates and P values of the regression components for the removal and cost factor models
Parameter
Removal
Cost factor
Parameter estimate
3.618
Parameter estimate
P value
P value
Intercept
586
0.000
0.159
0.035
0.004
0.000
0.254
0.222
0.161
0.001
0.000
0.003
0.211
0.003
0.029
0.000
0.005
0.000
0.035
0.028
0.005
0.000
0.995
0.000
0.505
0.106
0.048
0.049
0.002
0.249
0.011
À3
H
Fe
2
2
O
2
0.128
)2.25
0.337
)258
)2.98 Â 10
‡
)0.0104
1.81 Â 10À3
)1.21
Initial ethylene glycol
pH
À5
À4
À3
À4
À3
À4
À6
H
2
2
O
2
ÃH
Fe ÃH
Fe ÃFe
2
O
2
)6.85 Â 10
1.46 Â 10
‡
À9
2
O
2‡
2
5.08 Â 10
3.27 Â 10
2
‡
À6
8.19 Â 10
)9.16 Â 10
À7
Initial ethylene glycolÃH
2
2
O
2
1.43 Â 10
)1.86 Â 10
‡
À6
Initial ethylene glycolÃFe
2.38 Â 10
)4.97 Â 10
2
À6
Initial ethylene glycol
)2.14 Â 10
)1.02 Â 10
À4
pHÃH
2
O
2
)0.0141
0.447
1.52 Â 10
2
‡
À3
pHÃFe
3.27 Â 10
À5
pHÃInitial ethylene glycol
pHÃpH
)0.0261
26.1
)9.29 Â 10
0.109

B. Dietrick McGinnis et al. / Chemosphere 45 ꢀ2001) 101±108
105
2
Fig. 3. Ethylene glycol loss in mg/l predicted by the model in Table 2 compared to analytically determined losses. r ˆ 0:881 and
m ˆ 0:85 Æ 0:08.
quently, the pH data are only considered for the equa-
tion parameters in Table 2.
Table 4
Estimated responses from uncoded parameter values for the
a
cost factor
Increasing values for cost factors indicate increased
eciency. The cost factors in this model increased with
2
‡
Estimated
cost factor
H
2
O
2
Fe
Added
Initial ethylene
glycol
lower H
2
O
2
addition, higher iron addition and higher
Added
initial ethylene glycol concentrations ꢀTable 4). As with
the removal data analysis ꢀTable 3), pH is a factor of
iron concentration and is not presented in Table 4.
Samples treated with Fenton's reagent were also ex-
posed to UV light as a post-treatment ꢀFig. 4). The
concentration of ethylene glycol continued to decrease
with exposure to UV light in both systems studied. The
1
.84
500
492
484
478
472
467
462
459
456
453
451
50.0
49.8
50.5
51.6
53.1
54.7
56.4
58.2
60.1
62.1
64
642
658
673
688
701
713
725
736
747
757
767
2.12
2.41
2.72
3
3
3
4
4
4
5
.04
.37
.72
.09
.48
.88
.31
2 2
Fenton's treatment with higher initial H O concentra-
tion resulted in higher ®rst-order k values for ethylene
glycol removal ꢀTable 5). The system with only 50 mg/l
H
2
O
2
added contained no detectable H
2 2
O when ex-
a
Stationary point ꢀwhich is a saddle point) is 0.721. All values
except the unitless estimated cost factor are in mg/l.
posed to UV light. The system with 1000 mg/l H
added contained a residual concentration of H
2
O
2
2 2
O of
approximately 100 mg/l when exposed to UV light
ꢀ
Fig. 2).
Intermediate products were monitored in the system
with 50 mg/l H added ꢀFig. 5). Formic acid con-
Table 3
Estimated responses from uncoded parameter values for eth-
a
ylene glycol removal
2 2
O
2
‡
centrations generally increased while oxalic acid con-
centrations quickly decreased with exposure to UV light.
Estimated
removal
H
2
O
2
Fe
Added
Initial ethylene
glycol
Added
1
1
1
2
2
2
2
2
2
3
3
60
75
90
07
23
41
59
78
97
17
38
525
541
557
572
587
601
616
630
643
656
667
50
54
58
63
67
71
76
80
85
89
94
642
659
677
694
712
729
747
764
781
797
812
4
. Discussion
Ethylene glycol loss in these systems followed the
results of Wang et al. ꢀ1992) and McGinnis et al. ꢀ2000)
Å
for degradation of ethylene glycol by HO produced in
UV/H
2
O
2
and photo Fenton systems, respectively. Al-
Å
though these systems di€er in method of HO produc-
tion, each facilitates radical mediated oxidation as the
mechanism for ethylene glycol loss. The formation of
oxalic acid and formic acid ꢀFig. 2) follows pathways for
ethylene glycol degradation, sequential oxidation,
a
Stationary point ꢀwhich is a saddle point) is 20.66. All values
are in mg/l except the estimated cost factor ꢀEq. ꢀ1) which is
unitless.

106
B. Dietrick McGinnis et al. / Chemosphere 45 ꢀ2001) 101±108
2
‡
Fig. 4. Ethylene Glycol concentration in two systems initially treated with Fenton's reagent then exposed to UV light. 100 mg/l Fe
added, initial concentration of ethylene glycol ꢀEG) 1000 mg/l. Time zero concentration is the amount of ethylene glycol remaining
2 2
after the initial treatment with Fenton's reagent ꢀ100 or 1000 mg/l H O initially).
outlined by McGinnis et al. ꢀ2000) and for glycolic,
oxalic and glyoxylic acid determined by Karpel and
Dore' ꢀ1997) ꢀFig. 6).
Calculation of total organic carbon ꢀTOC) from re-
sidual ethylene glycol concentration and the resulting
concentrations of formic and oxalic acids indicate less
than a 25% TOC loss. Despite the limited TOC loss, the
conversion of ethylene glycol to oxalic acid and formic
acid can result in a substantial loss of TCOD. This was
noted in systems studied previously ꢀMcGinnis et al.,
Table 5
Degradation rate constants ꢀk) for ethylene glycol with expo-
sure to UV light from two systems initially treated with Fen-
a
ton's reagent
2
System/Reac-
tion order
r
k
P-value
2 2
50 mg/l H O added
2
tional production of HO . Oxalic acid and formic acid
000) with additional TOC losses in systems with addi-
Å
Zero
First
0.810
0.788
2.62 Æ 1.76
0.015
0.018
0.004 Æ 0.002
Å
1000 mg/l H
2
O
2
added
have both been shown to mineralize in HO and similar
Zero
First
0.830
0.981
8.30 Æ 5.32
0.012
0.000
systems ꢀMurphy et al., 1989; Karpel and Dore', 1997).
The results of the quadratic response surface model
indicated that the loss of ethylene glycol can be pre-
0.040 Æ 0.009
a
2‡
00 mg/l Fe added, initial concentration of ethylene glycol
1
000 mg/l. k values are con®ned to zero- and ®rst-order models
1
for comparison and are in mg/l-min and min , respectively.
dicted as a function of initial ethylene glycol concen-
2‡
À1
tration, Fe and H
2
O
2
added and pH ꢀFig. 3). The pH
Fig. 5. Formic and oxalic acid in 50 mg/l H
UV exposure concentration.
2 2
O test exposed to UV light. Time zero concentrations are post-Fenton's reagent and pre-

B. Dietrick McGinnis et al. / Chemosphere 45 ꢀ2001) 101±108
107
2 2
Fig. 6. Possible intermediate products of UV/H O ethylene glycol degradation.
component in this model and the cost factor model is
related in both cases directly to the Fe concentrations
higher pH values are associated with lower dissolved
glycol loss ꢀFig. 4, Table 5). Variation of the rate con-
stants depending on the amount of H added as part
of the Fenton's reagent was observed. During irradia-
tion with UV light, the solution with 50 mg/l H ad-
ded demonstrated a signi®cantly lower ethylene glycol
degradation rate than the solution with 1000 mg/l H
added. The rate in the low H system was lower than
rates reported in comparable photo Fenton systems
ꢀMcGinnis et al., 2000), but the high H had a rate
exceeding that determined in photo Fenton systems
ꢀMcGinnis et al., 2000). The high H system ethylene
glycol degradation rate was not signi®cantly di€erent at
the 95% con®dence level than rates for UV/H sys-
tems ꢀMcGinnis, 1998). This is important for two rea-
sons: ꢀ1) The comparable UV/H system had 1000
mg/l H at the beginning of UV treatment while the
system used for this experiment only had 100 mg/l H
ꢀremaining after treatment with Fenton's reagent with
1000 mg/l H ) upon exposure to UV. ꢀ2) The Fenton's
2‡
2 2
O
±
2
‡
initial Fe concentrations. In addition, pH was not ef-
fectively controlled in these systems. Although pH is a
signi®cant factor ꢀTable 1), this result may be an artifact
of iron speciation and/or solubility. Increased pH values
are related to decreases in soluble iron concentration. In
addition, these results do not specify an optimal pH, but
no evidence is presented to contradict an optimal pH
range of pH 2±5 which were the pH conditions utilized
by Murphy et al. ꢀ1989), Lin and Lo ꢀ1997) and Safar-
zadeh-Amiri et al. ꢀ1997).
2 2
O
2
O
2
2
O
2
2
O
2
2
O
2
2
O
2
The optimal conditions for this system may include
ranges of pH and H
study. The data do indicate that the ranges, pH values
above 9.0 and below 2.5 and H addition greater than
000 mg/l ꢀFig. 2) are not always practical for ®eld use.
2
O
2
addition not included in this
2 2
O
2
O
2
2
O
2
2 2
O
1
Due to this condition, ridge analysis of the ethylene
glycol loss model to ®nd the stationary and/or saddle
point was not performed, the optimal result lies outside
the range of the study.
2 2
O
reagent system had the bene®t of initial treatment re-
sulting in signi®cant loss of ethylene glycol with only the
addition of a relatively inexpensive additional compo-
2
‡
Similar results were noted in the cost factor model
nent, Fe
.
with the exception of decreased H
ecient removal. This indicated that higher concentra-
tions of H in the reactors result in less ecient oxi-
2
O
2
resulted in more
Loss of oxalic acid and increased concentrations of
formic acid in the system exposed to UV light ꢀFig. 5)
are similar to results of ethylene glycol treatment in
photo Fenton systems ꢀMcGinnis et al., 2000). This
suggested that oxidation of oxalic acid may be in¯u-
enced by the formation of ferrioxalate that degrade with
UV as described by Safarzadeh-Amiri et al. ꢀ1997). This
also indicated that the continued loss of ethylene glycol
in the system may be due to either the production of
carboxylate radicals or from the photo Fenton reaction
2
O
2
dation of ethylene glycol although the total amount
oxidized increased. This decreased eciency is most
likely due to scavenging of radicals by carbonates and
bicarbonates ꢀChan and Larson, 1991) or H
2 2
O itself
ꢀ
actions have been reported as second-order ꢀKarpel and
Karpel and Dore', 1997). Rates for both of these re-
Å
Dore', 1997), increasing with higher HO concentrations
2
‡
3‡
resulting from higher concentrations of H
Intermittent addition of H may increase the e-
ciency of the removal of ethylene glycol. By limiting the
2
O
2
.
utilizing the Fe converted to Fe by the initial addi-
tion of H . The di€erence in the ethylene glycol loss in
the 50 and 1000 mg/l H may then be due to the lack
of Fe oxidized to Fe by H in the 50 mg/l H
system before exposure to UV light, resulting in ethylene
2
O
2
2 2
O
2
3‡
O
2
2‡
concentration of H
2
O
2
at any one time, the amount of
2
O
2
2 2
O
Å
HO lost to scavenging can be decreased due to the
relative concentration of ethylene glycol and the con-
centration dependence of the reaction kinetics. Due to
3‡
glycol degradation limited by Fe for the photo Fenton
reaction.
The quick loss of oxalic acid, 80% loss in 5 min in the
the relative cost of H
would be expected to be H
2
O
2
, cost-ecient applications
limited.
2
O
2
2 2
50 mg/l H O Fenton system exposed to UV light ꢀFig.
3
‡
Samples treated with UV light after the addition of
Fenton's reagent showed signi®cant rates of ethylene
5), does indicate adequate amounts of Fe present to
form UV sensitive ferric oxalate. Which system, photo

108
B. Dietrick McGinnis et al. / Chemosphere 45 ꢀ2001) 101±108
Fenton or ferric oxalate, is the dominant consumer for
ultraviolet reactors. Journal of Environmental Engineering
13, 612±627.
Karpel, N., Dore', M., 1997. Mechanisme d'action des radicaux
3‡
1
Fe is not indicated despite the relatively slow ethylene
glycol degradation rate ꢀFig. 4). Production of the re-
quired radicals can be a function of ferric oxalate de-
gradation ꢀEq. ꢀ2)) as well as photo Fenton chemistry.
Å
OH sur les acides glycolique, glyoxylique, acetique et
oxalique en solution aquesuse: incidence sur al consamma-
tion de peroxyde d'hydrogene dans les systemes H
and ozone/H O systems. Water Research 31, 1383±1397.
2 2
O /uv
3
3
À
‡ hv ) Fe2‡ ‡ 2C
2À
‡ COÅ2À ꢀ2†
2
2
FeꢀC
2
O
4
†
2
O
‡ CO
2
4
Klementova, S., Wagnerova, D.M., 1994. Photocatalytic e€ect
of FeꢀIII) on oxidation of two-carbon substrates related to
natural waters. Collective Czech Chemistry Communica-
tions 59, 1066±1076.
The potential for treatment utilizing ferrioxalate has
several advantages. Treating airport storm ˆ water with
Fenton's reagent to oxalate as opposed to mineralizing
the ethylene glycol may be less expensive. Ferrioxalate
absorbs light up to 500 nm ꢀSafarzadeh-Amiri et al.,
Langlais, B., Reckhow, D.A., Brink, D.R., 1991. Ozone in
Water Treatment: Application and Engineering. Lewis
Publishing, Chelsea, MI.
1
997) making a secondary passive treatment in open
Lemaire, J., 1993. Mechanisms of photooxidation of polyole-
®ns; prediction of lifetime in weathering conditions. In:
Clough, R.L., Billingham, N.C., Gillen, K.T. ꢀEds.), Poly-
mer Durability, Degradation, Stabilization and Lifetime
Prediction. American Chemical Society, Washington, DC,
pp. 1±35.
lagoons possible. Subsequently, primary treatment with
Fenton's reagent followed by a secondary passive
treatment in open lagoons has the potential to be a low-
tech, inexpensive, although undeveloped, treatment for
airport storm-water.
Lin, S., Lo, C., 1997. Fenton process for treatment of desizing
wastewater. Water Research 31, 2050±2056.
McGinnis, D., 1998. Degradation of ethylene glycol by dye-
sensitized and hydroxyl radical oxidation. Ph.D. Disserta-
tion, University of Nevada, Reno.
5
. Conclusions
Fenton's reagent e€ectively reduced the concentra-
McGinnis, B.D., Middlebrooks, E.J., Adams, V.D., 2000.
Degradation of ethylene glycol in photo Fenton systems.
Water Research 34, 2346±2354.
tion of ethylene glycol in these systems. Application of
this method to airport storm-water can result in reduc-
tions in the COD by conversion of ethylene glycol to
oxalic acid and formic acid. The most ecient applica-
Mericas, D., Wagoner, B., 1994. Balancing safety and the
environment; Managing aircraft deicing ¯uid's impact on
airport storm-water. Water Environment Technology 12,
2 2
tion of this system may be sequential addition of H O
3
8±43.
for treatment with Fenton's reagent with post-treatment
with UV light or utilization of ferrioxalate to substitute
natural sunlight. The principle products of this reaction,
oxalic and formic acids as well as ethylene glycol have
Miller, C.M., Valentine, R.L., Roehl, M.E., Alvarez, P.J., 1996.
Chemical and microbiological assessment of pendimethalin-
contaminated soil after treatment with Fenton's reagent.
Water Research 30 ꢀ11), 2579±2586.
Å
been shown to mineralize in HO and similar systems
ꢀMurphy et al., 1989; Karpel and Dore', 1997; McGinnis
et al., 2000) presenting the potential of ethylene glycol
Murphy, A.P., Boegli, W.J., Price, M.K., Moody, C.D., 1989.
A Fenton-like reaction to neutralize formaldehyde waste
solutions. Environmental Science and Technology 23, 166±
169.
Å
mineralization in this system with increased HO pro-
duction.
Pignatello, J.J., Liu, B., Huston, P., 1999. Evidence for an
additional oxidant in the photoassisted Fenton reaction.
Environmental Science and Technology 33 ꢀ11), 1832±1839.
Safarzadeh-Amiri, A., Bolton, J.B., Cater, S., 1997. Ferrioxa-
late-mediated photodegradation of organic pollutants in
contaminated water. Water Research 31, 787±798.
References
â
Allen, J.M., Lucas, S., Allen, S.K., 1996. Formation of
Å
Union Carbide Chemicals, 1992. UCAR Aircraft Deicing
hydroxyl radical ꢀHO ) in illuminated surface waters con-
taminated with acidic mine drainage. Environmental Tox-
icology and Chemistry 15, 107±113.
Fluids Composition and Trace Contaminants. Technical
Information Bulletin. Union Carbide Chemicals and Plas-
tics Company, Inc. Technical Center, South Charleston,
WV.
APHA., 1995. Standard Methods for the Examination of Water
and Wastewater, 19th ed. APHA-AWWA-WPCF, Wash-
ington, DC.
Wang, F., Cassidy, F., Lum, B., 1992. Incineration alternatives
for combustible waste: ultraviolet/hydrogen peroxide pro-
cess. Final report to rocky ¯ats plant, FY92 tasks under
TTP SF 221207. Lawrence Livermore National Laboratory,
Livermore, CA 94550.
Chan, W.F., Larson, R.A., 1991. Mechanisms and products of
ozonolysis of aniline in aqueous solution containing nitrite
ion. Water Research 25, 1455±1539.
Evans, W.H., David, E.J., 1974. Biodegradation of mono-, di-
and triethylene glycols in river waters under controlled
laboratory conditions. Water Research 8, 97±100.
Harris, G.D., Adams, V.D., Moore, W.M., Sorenson, D.L.,
Watts, R.J., Bottenberg, B.C., Hess, T.F., Jensen, M.D., Teel,
A.L., 1999. Role of reductants in the enhanced desorption
and transformation of chloroaliphatic compounds by mod-
i®ed Fenton's reactions. Environmental Science and Tech-
nology 33 ꢀ19), 3432±3437.
1987. Potassium ferrioxalate as chemical actinometer in