El

Ra

b

c

a

ARR1AA

KLHORR

1

imIhtd

Te[

s(p

1h

Journal of Photochemistry and Photobiology A: Chemistry 274 (2014) 13– 19

Contents lists available at ScienceDirect

Journal of Photochemistry and Photobiology A:Chemistry

journa l h om epa ge: www.elsev ier .com/ locate / jphotochem

ffect of humidity on the production of ozone and other radicals byow-pressure mercury lamps

yo Onoa,∗, Yusuke Nakagawab, Yusuke Tokumitsua, Hiroyuki Matsumotoc, Tetsuji Odab

Department of Advanced Energy, The University of Tokyo, 5-1-5 Kashiwanoha, Kashiwa, Chiba 227-8568, JapanDepartment of Electrical Engineering, The University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-8656, JapanCorporate Advanced Technology Center, Iwasaki Electric Co. Ltd., 1-20 Fujimicho, Gyoda, Saitama 361-0021, Japan

r t i c l e i n f o

rticle history:eceived 13 June 2013eceived in revised form6 September 2013ccepted 27 September 2013vailable online 10 October 2013

eywords:ow-pressure mercury lamp

a b s t r a c t

An ultraviolet (UV) process using a low-pressure mercury lamp is affected by ambient humidity. It is dueto strong influence of humidity on the production of ozone and other radicals by the UV light. In thispaper, a photochemical reaction model under the irradiation of a low-pressure mercury lamp is devel-oped, and the effect of humidity on the production of ozone and other radicals [O, O(1D), O2(a1�g),O2(b1˙+

g ), OH, HO2, H, and H2O2] by a low-pressure mercury lamp is discussed using the reactionmodel. The validity of the reaction model is confirmed by comparing the ozone densities calculatedusing the model with experimentally measured ozone densities, and they showed good agreement. Thereaction model shows that the ozone density decreases with increasing humidity for three reasons:

umidityzone densityadicalseaction model

(i) attenuation of 185 nm light due to absorption by H2O, leading to a decreased O atom production byO2 + h�(185 nm) −→ O + O which is required to produce ozone by O + O2 + M −→ O3 + M; (ii) ozone destruc-tion by O3 + h�(254 nm) −→ O(1D) + O2(a1�g), where the resulting O(1D) partly reacts with H2O beforeconverting back to O3 after quenching to O; and (iii) an ozone destruction cycle OH + O3 −→ HO2 + O2 andHO2 + O3 −→ OH + 2O2. The effect of humidity on the densities of other radicals is also discussed using thereaction model.

. Introduction

An ultraviolet (UV) process using a low-pressure mercury lamps widely used for surface treatment, sterilization, and water treat-

ent. The UV process often utilizes ozone produced by the UV light.n the “UV/O3 process”, it is empirically known that an increase inumidity reduces the ozone density, which can lower the effect ofhe UV/O3 process. This is partially caused by the catalytic ozoneestruction cycle involving OH and HO2 [1,2]:

OH + O3 −→ HO2 + O2,

HO2 + O3 −→ OH + 2O2.

he production of OH and HO2 by a low-pressure mercury lamp wasxperimentally confirmed using light absorption measurements1,3].

In contrast, some UV processes are enhanced by humidification,uch as those that use radicals produced in a humid environmente.g., OH and HO2). The decomposition of volatile organic com-ounds (VOCs) using a low-pressure mercury lamp is enhanced

∗ Corresponding author. Tel.: +81 3 5841 6663; fax: +81 3 5841 6663.E-mail address: ryo-ono@k.u-tokyo.ac.jp (R. Ono).

010-6030/$ – see front matter © 2013 Elsevier B.V. All rights reserved.ttp://dx.doi.org/10.1016/j.jphotochem.2013.09.012

© 2013 Elsevier B.V. All rights reserved.

by humidity because OH radicals efficiently decompose VOCs [4,5].Our group used a low-pressure mercury lamp for the surface treat-ment of dye-sensitized solar cells and found that the effect of usingthe lamp is improved by humidification [6]. Furthermore, someUV processes are not significantly influenced by humidity [7,8].To understand the effects of humidity on the UV processes, thephotochemical reactions occurring under the irradiation of a low-pressure mercury lamp should be investigated.

In this paper, a photochemical reaction model under the irradi-ation of a low-pressure mercury lamp is developed, and the effectof humidity on the densities of ozone and other radicals [O, O(1D),O2(a1�g), O2(b1˙+

g ), OH, HO2, H, and H2O2] produced by the lampis examined using the reaction model. After the reaction model isdescribed in the next section, the validity of the reaction model isconfirmed by comparing the ozone densities calculated using themodel with experimentally measured ozone densities. Then, theeffect of humidity on the production of ozone and other radicalsis discussed using the reaction model. Particularly, the decrease inozone density with increasing humidity is discussed in detail.

2. Reaction model

The reaction model under the irradiation of a low-pressure mer-cury lamp is described in this section. The low-pressure mercury

14 R. Ono et al. / Journal of Photochemistry and Photobiology A: Chemistry 274 (2014) 13– 19

Table 1Rate coefficients (k [cm3/s]) at 298 K and cross-sections (� [cm2]) of the reactions considered in the reaction model. The ratios written in R7, R10, R13–R16, and R22 arebranching ratios.

Reactions k and � Refs.

(R1) O2 + h�(185 nm) −→ O + O a�1 = 1.2 × 10−20 –(R2) O + O2 + M −→ O3 + M k2 = 6.0 × 10−34[O2] [10]

= 5.6 × 10−34[N2](R3) O3 + h�(254 nm) −→ products �3 = 1.16 × 10−17 [11](R4) H2O + h�(185 nm) −→ H + OH �4 = 7.14 × 10−20 [10](R5) O + O3 −→ O2 + O2 k5 = 8.0 × 10−15 [10](R6) O(1D) + N2 −→ O + N2 k6 = 2.6 × 10−11 [12](R7) O(1D) + O2 −→ O + O2(X, a, b) k7 = 4.0 × 10−11 (15:5:80) [10](R8) O(1D) + H2O −→ OH + OH k8 = 2.2 × 10−10 [10](R9) O(1D) + O3 −→ O2 + O2, O2 + O + O k9 = 2.4 × 10−10 (1:1) [10](R10) O2(a) + O2 −→ O2 + O2 k10 = 1.6 × 10−18 [10](R11) O2(a) + O3 −→ O2 + O2 + O k11 = 3.8 × 10−15 [10](R12) O2(a) + H2O −→ O2 + H2O k12 = 5 ×10−18 [10](R13) O2(a) + O2(a) −→ O2(X, b) + O2 k13 = 2.8 × 10−17 (1:1) [13](R14) O2(b) + N2 −→ O2(X, a) + O2 k14 = 2.1 × 10−15 (1:9) [10,13](R15) O2(b) + H2O −→ O2(X, a) + H2O k15 = 4.6 × 10−12 (1:9) [10,13](R16) O2(b) + O3 −→ O2(X, a) + O3, O + 2O2 k16 = 2.2 × 10−11 (3:27:70) [10,13](R17) OH + O −→ O2 + H k17 = 3.5 × 10−11 [10](R18) H + O2 + M −→ HO2 + M k18 = 5.5 × 10−32[N2] [10](R19) HO2 + O −→ OH + O2 k19 = 5.8 × 10−11 [10](R20) OH + HO2 −→ H2O + O2 k20 = 1.1 × 10−10 [10](R21) H + O3 −→ OH + O2 k21 = 2.9 × 10−11 [14](R22) H + HO2 −→ OH + OH, H2 + O2, H2O + O k22 = 8.0 × 10−11 (90:7:3) [10](R23) H + OH + M −→ H2O + M k23 = 6.9 × 10−31[N2] [15]

= 4.4 × 10−30[H2O](R24) OH + OH −→ H2O + O k24 = 1.5 × 10−12 [10](R25) OH + OH + M −→ H2O2 + M bk25 = 5.2 × 10−12 [10](R26) OH + H2O2 −→ H2O + HO2 k26 = 1.7 × 10−12 [10](R27) OH + O3 −→ HO2 + O2 k27 = 7.3 × 10−14 [10](R28) HO2 + HO2 + M −→ H2O2 + O2 + M b,ck0

28 = 2.9 × 10−12 [10](R29) HO2 + O3 −→ OH + 2O2 k29 = 2.0 × 10−15 [10](R30) O + H2O2 −→ OH + HO2 k30 = 1.7 × 10−15 [10](R31) H2O2 + h�(185 nm) −→ OH + OH �31 = 8 ×10−19 [10](R32) H2O2 + h�(254 nm) −→ OH + OH �32 = 7.0 × 10−20 [10]

l1

O

O

O

OO

O

w

H

pstt

rO[tst

A 36-cm long low-pressure mercury lamp is placed in the cylin-drical cell, as shown in Fig. 1. The photo-reaction cell is made fromaluminum, and black non-glossy paint is applied to the inner wallof the cell to reduce reflections. The radius of the lamp, r0, is 0.9 cm

a See text.b Calculated using low- and high-pressure limits and broadening factor.c k28 = k0

28{1 + 1.4 × 10−21 exp(2200/T)[H2O]}.

amp emits radiation at wavelengths of 185 nm and 254 nm. The85 nm radiation dissociates O2 into O(3P):

2 + h�(185 nm) −→ O(3P) + O(3P). (R1)

(3P) reacts with O2 to produce O3:

(3P) + O2 + M −→ O3 + M. (R2)

3 is dissociated by the 254 nm radiation to produce O(1D), O(3P),2(a1�g), and O2(v):

3 + h�(254 nm)�3−→O(1D) + O2(a1�g), (R3a)

1−�3−→ O(3P) + O2(X3˙−g , v), (R3b)

here �3 = 0.9 [9].In addition, the 185 nm radiation dissociates H2O:

2O + h�(185 nm) −→ H + OH. (R4)

Thus, the UV radiation from the low-pressure mercury lamproduces O(3P), O(1D), O2(a), O3, OH, and H. These radicals react ashown in reactions R5–R32, which are listed in Table 1 along withheir rate coefficients at 298 K. The absorption cross-sections andhe rate coefficients of reactions R1–R4 are also shown in Table 1.

The cross-section of R1, �1, is uncertain. It was previouslyeported that �1 of the low-pressure mercury lamp depends on the2 column, with values ranging from 0.6 × 10−20 to 1.2 × 10−20 cm2

16]. In another study [17], the range of values for �1 were reportedo be 1.4 × 10−20 to 1.8 × 10−20 cm2. In our experiment, �1 is mea-ured using our lamp, as shown in the next section, and is estimatedo be 1.2 × 10−20 cm2. This �1 value is used in our model.

3. Verification of reaction model

For the verification of the reaction model developed in the previ-ous section, the ozone densities calculated using the reaction modelare compared with experimentally measured ozone densities. Acylindrical reaction cell equipped with a low-pressure mercurylamp shown in Fig. 1 is used for the verification. The ozone den-sity at the gas outlet of the reaction cell is measured and comparedwith that calculated using the reaction model. The experiment andsimulation are detailed below.

3.1. Experiment

Fig. 1. Photo-reaction cell equipped with a low-pressure mercury lamp.

R. Ono et al. / Journal of Photochemistry and Photobiology A: Chemistry 274 (2014) 13– 19 15

0 20 40 60 80 100

Relative humidity [%]

1015

1016

1017

Ozo

ne d

ensi

ty [c

m-3

] ExperimentSimulation (model 2)Simulation (model 1)

(a) R = 4 cm, 1.5 L/min, air, 25.4 oC

0 20 40 60 80 100

Relative humidity [%]

1015

1016

1017

Ozo

ne d

ensi

ty [c

m-3

] ExperimentSimulation (model 2)Simulation (model 1)

(b) R = 4 cm, 3.0 L/min, air, 24.7 oC

0 20 40 60 80 100

Relative humidity [%]

1015

1016

1017

Ozo

ne d

nsity

[cm

-3] Experiment

Simulation (model 2)Simulation (model 1)

(c) R = 8 cm, 6.0 L/min, air, 22.3 oC

0 20 40 60 80 100

Relative humidity [%]

1015

1016

1017

Ozo

ne d

ensi

ty [c

m-3

] ExperimentSimulation (model 2)Simulation (model 1)

(d) R = 8 cm, 12.0 L/min, air, 20.2 oC

0 20 40 60 80 100

Relative humidity [%]

1015

1016

1017

Ozo

ne d

ensi

ty [c

m-3

] ExperimentSimulation (model 2)Simulation (model 1)

(e) R = 4 cm, 3.0 L/min, O2(50%)/N 2, 23.9 oC

exper

actuTtrtu2uamoHhfaaaot

lof

Fig. 2. Comparison of the ozone densities obtained from simulation and

nd the length of the cell, L, is 30 cm. The UV lamp emits radiationontinuously. Humid air or humid O2(50%)/N2 is allowed to flowhrough the reaction cell. Two conditions of O2 = 20% and 50% aresed to verify the reaction model for different O2 concentrations.emperature and humidity are measured at the gas outlet. Afterhe lamp is switched on, the ozone density increases with time andeaches equilibrium after around 20 min. After equilibrium condi-ions are reached, the ozone density at the gas outlet is measuredsing an UV absorption method. A 2-cm quartz absorption cell and54-nm radiation from another low-pressure mercury lamp aresed for the UV absorption. The 254-nm light is absorbed by H2O2s well as ozone (see R32). However, the ozone density measure-ent is not disturbed by H2O2 because the absorption cross section

f ozone at 254 nm is two orders of magnitude larger than that of2O2 (Table 1) and the density of ozone is orders of magnitudeigher than that of H2O2 (see Fig. 4).To verify the reaction model

or different conditions, the gas flow rate is changed between 1.5nd 12 L/min and two sizes of photo-reaction cells are used: R = 4 cmnd 8 cm where R is the inner radius of the cell. The gas flow ratesnd the cell sizes are chosen to fix the ozone density into a rangef about 2.5 × 1015 to 2.5 × 1016 cm−3 (100–1000 ppm), which isypical values for actual UV/O3 processes.

The radiation power of the lamp is measured as follows. Theamp is placed outside the photo-reaction cell, and the illuminancef 185 nm and 254 nm radiation is measured at different distancesrom the lamp surface at the midst of the lamp. The 185 nm and

iments under different humidity, O2 concentration, and flow conditions.

254 nm light is measured using the illuminometers H9535-172(Hamamatsu) and UV-M02 (Orc), respectively. H9535-172 is cal-ibrated for 172 nm light but is also sensitive to 185 nm. A spectralresponse provided on its manual is used for calibration. Then, theilluminance is numerically simulated, and the radiation power isdetermined by fitting the simulated values to the measured results.In the simulation of the illuminance, the photon flux at the mea-surement point is volume integrated over the lamp volume. Thecos � dependence of the 254-nm illuminometer is not good, and thedetails of this dependency provided by the manufacturer is consid-ered in the simulation. The radiation power is determined to be1.9 W for 185 nm and 12.2 W for 254 nm.

In addition, the cross section of R1, �1, is determined by thefitting of measured and simulated 185-nm illuminance. It is deter-mined to be 1.2 × 10−20 cm2. In the calculation of �1, the absorptionof 185 nm by H2O (see R4) in the presence of humidity, where the185-nm illuminance was measured, is also considered.

3.2. Simulation

The radical densities in the photo-reaction cell are simulatedusing the reaction model. The aim of the simulation is to verify

the reaction model and not to develop an exhaustive simulation.Therefore, assumptions are made to simplify the modeling.

The cylindrical axis (r, z) shown in Fig. 1 is used in the simula-tion. The inside of the photo-reaction cell is divided into uniform

1 nd Photobiology A: Chemistry 274 (2014) 13– 19

cOsrrrto1ptfl

Itetfliicm

F

wRrH

ubttsflwtwtfwsoa

w

oa

4

4

tOcFs

0 20 40 60 80 100

Relative humidity [%]

0

1e+16

2e+16

3e+16

Ozo

ne d

ensi

ty [c

m-3

]

All reactions

w/o R4, R8, R27, and R29

+ R4

+ R27 and R29

+ R8

+ R27 and R29

1

2

3

4

5

6

6 R. Ono et al. / Journal of Photochemistry a

omputational mesh. The densities of short-lived radicals [O, O(1D),2(a), O2(b), OH, HO2, and H] at a point (r, z) are calculated by

olving the rate equations obtained from Table 1 under equilib-ium conditions (e.g., d[O]r,z/dt = 0). Diffusion is neglected for thoseadicals. The numbers of the mesh are sufficiently large: 100 for-direction and 300 for z-direction, respectively. It was confirmedhat the calculated radical densities are not affected by the numberf the mesh for larger than 100 × 300 meshes. The photon flux at85 nm and 254 nm at a point (r, z) is calculated by integrating thehoton flux over the lamp volume. The light absorption due to reac-ions R1, R3, and R4 is considered in the calculation of the photonux.

For long-lived species (O3 and H2O2), two models are assumed.n “model 1”, it is assumed that O3 and H2O2 diffuse sufficiently andhat their densities are uniform in the photo-reaction cell. Underquilibrium conditions, the number of ozone molecules in the reac-ion cell does not change with time. The number of ozone moleculesowing out from the reaction cell per unit time is F[O3]0, where F

s the volumetric flow rate of the gas and [O3]0 is the ozone densityn the reaction cell which is uniform in the cell. Under equilibriumonditions, the value of F[O3]0 is equal to the number of ozoneolecules produced in the reaction cell per unit time:

[O3]0 =∫

V

(d[O3]

dt

)r,z

dV, (1)

here (d[O3]/dt)r,z is increase in ozone density at (r, z) by reactions1–R32 per unit time and the integral is volume integration in theeaction cell. The O3 density is calculated by solving Eq. (1). The2O2 density is also calculated using the same way.

In model 1, it was assumed that the O3 and H2O2 densities areniform in the photo-reaction cell. However, their densities muste slightly or much lower at the upstream side (near the inlet ofhe photo-reaction cell) and higher at the downstream side (nearhe outlet) because they are accumulated as the gas flows down-tream. Accurate calculation of their density distribution requiresuid dynamics model, which is beyond the scope of the presentork. Instead, in “model 2”, it is simply assumed that their densi-

ies have gradients in the z-direction and uniform in the r-directionith boundary conditions of [O3] = 0 and [H2O2] = 0 at the inlet of

he reaction cell (z = 0). The ozone density at z, [O3]z , is calculated asollows. In an infinitely short time of �t, the gas flows by �z = vz�t,here vz is the average flow velocity of the gas, and the ozone den-

ity increases by (d[O3]z/dz)�z. At the same time, the increase inzone density in �t by reactions R1–R32 at (r, z) can also be writtens (d[O3]/dt)r, z�t. Therefore, the following equation holds:∫S

d[O3]z

dz�zdS =

∫S

(d[O3]

dt

)r,z

�tdS,

∴ vzd[O3]z

dz

∫S

dS =∫

S

(d[O3]

dt

)r,z

dS,

(2)

here the integral,∫

SdS =

∫ R

r02�r dr, is surface integral at z. The

zone density is calculated by solving Eq. (2). The H2O2 density islso calculated using the same way.

. Results and discussion

.1. Comparison between simulation and experiment

Fig. 2 compares the simulated and measured ozone densi-ies under different flow and humidity conditions. In Fig. 2(e),

2(50%)/N2 gas is used instead of air. The O3 and H2O2 densities arealculated using both models 1 and 2, and their results are shown.ig. 2 shows that the simulation underestimates the ozone den-ity by 30–70% than the measured values for all conditions, but it

Fig. 3. Simulation of ozone density versus relative humidity using model 2. Curve1 is calculated using all reactions R1–R32. Curves 2–6 are calculated by excludingsome reactions. Here R = 4 cm, T = 25 ◦C, and 1.5 L/min.

simulates well the effect of humidity on ozone density. Consideringthat the rate coefficients listed in Table 1 have generally an uncer-tainty of 10–30% [10] and some simplifying assumptions are madein the simulation, the errors of 30–70% are allowable. Thus, it canbe concluded that the reaction model is suitable.

The reaction model has been verified. Next, the decrease inozone density with increasing humidity is discussed using the reac-tion model. Then, the effect of humidity on the densities of otherradicals is examined using the reaction model. These information isimportant for an efficient use of the UV process with a low-pressuremercury lamp in a humid environment.

4.2. Reduction of O3 density with increasing humidity

Fig. 2 shows that the ozone density is reduced with increasinghumidity. There are three reasons for this reduction.

1. Absorption of 185 nm radiation by H2O: R4.2. Consumption of O(1D) by H2O: R8.3. Catalytic ozone destruction by OH and HO2: R27 and R29.

Curve 1 in Fig. 3 shows a simulated ozone density calculated usingall reactions R1–R32 and curve 2 is calculated excluding the follow-ing reactions: R4, R8, R27, and R29. It can be observed that curve 2 isalmost independent of humidity, indicating that the ozone reduc-tion by humidity is in fact caused by these three processes. Theseprocesses are discussed below in detail.

4.2.1. Absorption of 185 nm radiation by H2OThe 185 nm radiation from the lamp is absorbed by O2 and

H2O according to R1 and R4. The ratio of the absorption byO2 is �1[O2]/(�1[O2] + �4[H2O]), which decreases with increasinghumidity. As a result, the production of O(3P) by R1 is reduced withincreasing humidity, leading to decreased ozone production by R2.Curve 3 in Fig. 3 is a simulation where R4 is added to the condi-tions of curve 2. The difference between curves 2 and 3 indicatesthe ozone reduction caused by the absorption of 185 nm light byH2O (R4).

4.2.2. Consumption of O(1D) by H2O

Ozone is photodissociated by R3 to produce O(3P) and O(1D).

O(3P) converts back to O3 by R2, and O(1D) also reacts to form O3by R2 after quenching to O(3P) by N2 and O2 as R6 and R7. Because ofthis regeneration cycle, ozone is not practically decomposed by R3.

R. Ono et al. / Journal of Photochemistry and Photobiology A: Chemistry 274 (2014) 13– 19 17

0

5e+15

1e+16

0

2e+13

4e+13

6e+13

0

1e+12

2e+12

Den

sity

[cm

-3]

0 20 40 60 80 100

Relative humidity [%]

0

2e+10

4e+10

O3

O2(a)

H2O2

HO2

OH

O

(r, z) = ( R, L/2)

0

5e+15

1e+16

0

1e+14

2e+14

3e+14

0

2e+12

4e+12

Den

sity

[cm

-3]

0 20 40 60 80 100

Relative humidity [%]

0

2e+11

4e+11

O3

O2(a)

H2O2

HO2

OH

O

(r, z) = ( r0, L/2)

F ) = (R,F

Haibtir

4

t4car

4

iRpcatatrdo1dt

clts

ig. 4. Simulation of the radical densities for different humidity conditions at (r, z = 1.5 L/min.

owever, in humid conditions, a part of O(1D) reacts with H2O (R8)nd does not get converted back to ozone. Because of this branch-ng from the ozone regeneration cycle, ozone is partly decomposedy R3. Curve 5 in Fig. 3 is a simulation obtained by adding R8 tohe conditions of curve 2. The difference between curves 2 and 5ndicates the ozone reduction caused by R8 that breaks the ozoneegeneration cycle.

.2.3. Catalytic ozone destruction by OH and HO2In humid conditions, ozone is decomposed by the ozone destruc-

ion cycle involving OH and HO2, as shown in R27 and R29. Curves and 6 in Fig. 3 are the simulations where the ozone destructionycle, R27 and R29, have been added to the conditions of curves 3nd 5, respectively. In both cases, the ozone density is remarkablyeduced by the ozone destruction cycle.

.3. Other radical densities

The effect of humidity on the densities of other radicals is exam-ned. The radical densities are calculated in the reaction cell of

= 4 cm with T = 25 ◦C and F = 1.5 L/min conditions as an exam-le. Fig. 4 shows the calculated densities under different humidityonditions near the inner wall of the photo-reaction cell (r = R)nd near the lamp surface (r = r0) at the center cross-section ofhe photo-reaction cell (z = L/2). The densities of O(1D), O2(b),nd H are not plotted in Fig. 4 because they are too small ofhe orders of 105 ∼ 106 cm−3, 106 ∼ 107 cm−3, and 107 ∼ 109 cm−3,espectively. Near the inner wall of the reaction cell (r = R), theensities of O3, O2(a), H2O2, HO2, OH, and O are of the ordersf 1015 cm−3, 1013 cm−3, 1013 cm−3, 1012 cm−3, 1011 cm−3, and010 cm−3, respectively. Near the lamp surface (r = r0), the radicalensities increase because the photon flux from the lamp is largerhan that at r = R.

Fig. 4 shows radical densities and effect of humidity for typical

onditions for actual UV/O3 process: several cm distance from theamp surface with O3 density of the order of 1015 cm−3. It showshat ozone has the highest density among the radicals. The den-ity of atomic oxygen is 4–5 orders of magnitude lower than the

L/2) and (r0, L/2). They are calculated using model 2. Here R = 4 cm, T = 25 ◦C, and

ozone density, but its dependence on humidity is approximatelythe same as the ozone density. Radicals originating from watervapor, OH, HO2, and H2O2, are produced in humid environment.But their densities do not necessarily increase with humidity. Thedensities of HO2 and H2O2 reach maxima at relative humidity of5–20%. Further increase in humidity decreases their densities. Onthe other hand, the OH density increases in nearly proportional tohumidity. The density of O2(a) is almost independent of humidityat r = R, while decreases with increasing humidity at r = r0.

The densities of radicals and their dependence on humidity canbe obtained from Fig. 4, but only for the limited condition. Next,a simple model is deduced to estimate those radical densities formore general conditions. If the ozone density and illuminance of185 nm and 254 nm light are known, approximate radical densitiescan be analytically estimated. For example, the density of O(1D) canbe approximated by

[O(1D)] ∼= �3�3I254[O3]k6[N2] + k7[O2]

= ˚O(D)I254[O3], (3)

where I254 is the 254-nm illuminance and˚O(D) = �3�3/(k6[N2] + k7[O2]) is a constant. This approximation isvalid because O(1D) is mostly produced by R3 and predominantlyreacts via R6 and R7. If [O3] and I254 are known, [O(1D)] can beanalytically calculated using Eq. (3).

Similarly, for experimental conditions similar to ours, the densi-ties of O, O2(a), O2(b), OH, and HO2 can be approximated as follows:

[O] ∼= 2�1I185[O2] + �3I254[O3]k2[O2]

, (4)

[O2(a)] ∼= �3�3I254

k11, (5)

[O2(b)] ∼= 0.8k7[O2]˚ I254[O3], (6)

k15[H2O] O(D)

[OH] ∼= 2k8˚O(D)[H2O]I254 + k29[HO2]k20[HO2]/[O3] + k27

, (7)

18 R. Ono et al. / Journal of Photochemistry and Photobiology A: Chemistry 274 (2014) 13– 19

0 20 40 60 80 100

Relative humidity [%]

1e+05

1e+06

1e+07

1e+08

1e+09

1e+10

1e+11

1e+12

1e+13

1e+14

1e+15

Den

sitie

s [c

m-3

]

Full modelSimple model

0 20 40 60 80 100

Relative humidity [%]

Full modelSimple model

O2(a)

HO2

OH

O

O2(b)

O(1D)

r = r0

O2(a)

HO2

OH

O

O2(b)

O(1D)

r = R

Fig. 5. Radical densities calculated using the numerical simulation with all reactions and those calculated using the analytical model for different humidity conditions at (r,z ulati

[

wcsuaocm

oaiamartcbr(

5

camiwtiabtOa

[

[

[

) = (R, L/2) and (r0, L/2). Here R = 4 cm, T = 25 ◦C, and F = 1.5 L/min. The numerical sim

HO2] ∼= k27[OH][O3]k20[OH] + k29[O3]

, (8)

here I185 is the 185-nm illuminance. [OH] and [HO2] can be cal-ulated by solving the simultaneous equations (7) and (8). Fig. 5hows the radical densities calculated using the numerical sim-lation with all reactions R1–R32 and those calculated using thenalytical model of Eqs. (3)–(8) with values of [O3], I185, and I254btained from the simulation. The difference between the analyti-al and full reaction models is within a factor of two. The analyticalodel gives a good approximation of the radical densities.For the analytical model, the main production and reaction paths

f the radicals can be discussed. For example, the main productionnd reaction paths of O2(a) are R3 and R11, respectively, as shownn Eq. (5); i.e., O2(a) is produced by ozone photodissociation (R3)nd quenched by reaction with ozone (R11). Similarly, O atoms areainly produced by the photodissociation of O2 and O3 (R1 and R3)

nd consumed by reaction with O2 (R2) to produce ozone. The OHadicals are mainly produced by R8 and the catalytic ozone destruc-ion cycle (R29) and consumed by reaction with HO2 (R20) and theatalytic ozone destruction cycle (R27). HO2 is mainly producedy the catalytic ozone destruction cycle (R27) and consumed byeaction with OH (R20) and the catalytic ozone destruction cycleR29).

. Conclusions

The effect of humidity on the densities of ozone and other radi-als produced by a low-pressure mercury lamp was examined using

reaction model developed in the present work. After the reactionodel was developed, the validity of the reaction model was ver-

fied by showing that the simulated ozone densities agreed wellith the measured ones for different conditions. It was shown that

here are three reasons for the decrease in the ozone density withncreasing humidity: (i) decrease in O(3P) production (R1) due tottenuation of 185 nm radiation by H2O (R4); (ii) O3 destruction

y 254 nm radiation arising from R3 and R8; and (iii) O3 destruc-ion cycle by OH and HO2 (R27 and R29). The densities of O, O(1D),2(a1�g), O2(b1˙+

g ), OH, HO2, H, and H2O2 were also calculatednd the effect of humidity on those densities were discussed. Their

[

[

on used is model 2.

main production and reaction paths were discussed based on thereaction model.

Acknowledgments

This work is partially supported by A-STEP of Japan Science andTechnology Agency, and by the Grant-in-Aid for Science Researchby the Ministry of Education, Culture, Sport, Science and Technol-ogy.

References

[1] J.P. Burrows, R.A. Cox, R.G. Derwent, Modulated photolysis of the ozone–watervapour system: kinetics of the reaction of OH with HO2, J. Photochem. 16 (1981)147–168.

[2] M. Salvermoser, D.E. Murnick, U. Kogelschatz, Influence of water vapor onphotochemical ozone generation with efficient 172 nm xenon excimer lamps,Ozone Sci. Eng. 30 (2008) 228–237.

[3] K. Yoshihara, Y. Takatori, K. Miyazaki, Y. Kajii, Ultraviolet light-induced water-droplet formation from wet ambient air, Proc. Jpn. Acad. Ser. B 83 (2007)320–325.

[4] Y.S. Shen, Y. Ku, Treatment of gas-phase volatile organic compounds (VOCs) bythe UV/O3 process, Chemosphere 38 (1999) 1855–1866.

[5] J.H. Wang, M.B. Ray, Application of ultraviolet photooxidation to removeorganic pollutants in the gas phase, Sep. Pur. Technol. 19 (2000) 11–20.

[6] S. Zen, D. Saito, R. Ono, T. Oda, Low-temperature sintered dye-sensitized solarcell using surface treatment of TiO2 photoelectrode with ultraviolet light, Chem.Lett. 42 (2013) 624–626.

[7] J.R. Vig, UV/ozone cleaning of surfaces, J. Vac. Sci. Technol. A 3 (1985)1027–1034.

[8] L.F. Macmanus, M.J. Walzak, N.S. Mcintyre, Study of ultraviolet light and ozonesurface modification of polypropylene, J. Polym. Sci. A 37 (1999) 2489–2501.

[9] J.C. Brock, R.T. Watson, Ozone photolysis: determination of the O(3P) quantumyield at 266 nm, Chem. Phys. Lett. 71 (1980) 371–375.

10] R. Atkinson, D.L. Baulch, R.A. Cox, J.N. Crowley, R.F. Hampson, R.G. Hynes, M.E.Jenkin, M.J. Rossi, J. Troe, Evaluated kinetic and photochemical data for atmo-spheric chemistry: volume I – gas phase reactions of Ox , HOx , NOx and SOx

species, Atmos. Chem. Phys. 4 (2004) 1461–1738.11] L.T. Molina, M.J. Molina, Absolute absorption cross sections of ozone in the 185-

to 350-nm wavelength range, J. Geophys. Res. 91 (1986) 14501–14508.12] R. Atkinson, D.L. Baulch, R.A. Cox, R.F. Hampson Jr., J.A. Kerr, J. Troe, Evaluated

kinetic and photochemical data for atmospheric chemistry supplement IV, J.Phys. Chem. Ref. Data 21 (1992) 1125–1568.

13] D.S. Stafford, M.J. Kushner, Ø2(1�) production in He/O2 mixtures in flowinglow pressure plasmas, J. Appl. Phys. 96 (2004) 2451–2465.

14] J.A. Manion, R.E. Huie, R.D. Levin, D.R. Burgess Jr., V.L. Orkin, W. Tsang, W.S.McGivern, J.W. Hudgens, V.D. Knyazev, D.B. Atkinson, E. Chai, A.M. Tereza, C.-Y. Lin, T.C. Allison, W.G. Mallard, F. Westley, J.T. Herron, R.F. Hampson, D.H.

nd Pho

[

[

R. Ono et al. / Journal of Photochemistry a

Frizzell, NIST Chemical Kinetics Database, NIST Standard Reference Database17, Version 7.0 (Web Version), Release 1.4.3, Data Version 2008.12, NationalInstitute of Standards and Technology, Gaithersburg, MD 20899-8320. Avail-

able from: http://kinetics.nist.gov/

15] D.L. Baulch, C.T. Bowman, C.J. Cobos, R.A. Cox, Th. Just, J.A. Kerr, M.J. Pilling,D. Stocker, J. Troe, W. Tsang, R.W. Walker, J. Warnatz, Evaluated kinetic datafor combustion modeling: supplement II, J. Phys. Chem. Ref. Data 34 (2005)757–1397.

[

tobiology A: Chemistry 274 (2014) 13– 19 19

16] D.J. Creasey, D.E. Heard, J.D. Lee, Absorption cross-section measurements ofwater vapour and oxygen at 185 nm. Implications for the calibration of fieldinstruments to measure OH, HO2 and RO2 radicals, Geophys. Res. Lett. 27 (2000)

1651–1654.

17] D. Kartal, M.D. Andrés-Hernández, L. Reichert, H. Schlager, J.P. Burrows,Technical note: characterization of a DUALER instrument for the airborne mea-surement of peroxy radicals during AMMA 2006, Atmos. Chem. Phys. 10 (2010)3047–3062.