Predicting

liquid bleach decomposition

A hypochlorite ion degradation model helps utilities
using liquid bleach to develop control strategies
to minimize formation of chlorate ion.

Gilbert Gordon, Luke C. Adam,

T
Bernard P. Bubnis, Ching Kuo,
Robert S. Cushing,
and Richard H. Sakaji

h e treatm en t of drin kin g w ater w ith eith er

sodiu m h ypoch lorite (NaOCl) or calciu m h ypoch lo-
rite [Ca(OCl) 2 ] is a sou rce of ch lorate ion (ClO 3 –) in
finished water. Currently, ClO3– is not regulated by the
US Environmental Protection Agency (USEPA), pend-
in g th e resu lts of h ealth effects stu dies, bu t th e state
of Californ ia h as establish ed a 200-µ g/ L action level
th at is im posed on w ater u tilities. It is an ticipated
th at th e USEPA will regu late ClO3 – in drin kin g water
in th e fu tu re.1
ClO 3 – is a problem for
Stu d ie s h ave le d to an u n d e rstan d in g o f h o w ch lo rate io n (ClO 3 ) – u t ilit ie s t h a t st o r e a n d
fo rm atio n in th e p H 6–13 re gio n fro m d e co m p o sin g ble ach is apply liqu id bleach . Lab-
affe cte d by p H, te m p e ratu re , N aOCl co n ce n tratio n , ch lo rid e io n oratory an d field stu dies
co n ce n tratio n (by m e an s o f an io n ic stre n gth e ffe ct), an d provide direct evidence of
tran sitio n m e tal io n im p u ritie s. Th e h yp o ch lo rite io n d e grad atio n ClO3 – formation resulting
m o d e l d e ve lo p e d by Go rd o n an d A d am p re d icts fre e available from th e decom position
ch lo rin e d e co m p o sitio n , ClO 3 – fo rm atio n , an d o x yge n ge n e ratio n . of hypochlorite ion (OCl–)
Th u s, a u tility ap p lyin g liqu id ble ach can u se th is n e w ly in co m m e r cia l liq u id
d e ve lo p e d m o d e l to p re d ict th e tim e -d e p e n d e n t ch an ge s th at ble a ch . 2 –4 Da t a fr o m a
o ccu r in sto re d liqu id ble ach an d to d e ve lo p ch e m ical survey of raw water, stock
sp e cificatio n s an d co n tro l strate gie s to m in im ize th e fo rm atio n o f
Th e Gordon –Adam bleach de-
ClO 3 – d u rin g th e sto rage o f so d iu m h yp o ch lo rite . Th e fie ld stu d ie s com position m odel is available on
re p o rte d h e re p ro vid e ve rificatio n o f th e m o d e l sh o w in g a stro n g d isk fr o m t h e AW W A Re se a r ch
co rre latio n be tw e e n th e m o d e l p re d ictio n s an d th e m e asu re d Fou n dation with th e report “Min i-
m izin g Ch lorate Ion Form ation in
d ata tak e n in th e fie ld . Drin kin g Water Wh en Hypoch lorite
Ion is th e Ch lorin atin g Agen t.”

Copyright (C) 1997 American Water Works Association

142 VOLUME 8 9 , ISSUE 4 JOURNAL AWWA
 Formation of chlorate ion is a problem for water utilities that store
and apply liquid bleach.
liqu id bleach , an d fin ish ed water sam ples from m ore
th an 100 u tilities applyin g liqu id bleach for ch lori-
nation clearly indicate that ClO3 – is entering the pub-
lic water su pply at levels equ al to or greater th an at
system s applyin g ch lorin e dioxide.3
There are three possible sources of ClO3 – in drink-
in g water wh en liqu id bleach is u sed as th e ch lorin -
atin g agen t. On e sou rce is from th e raw water itself.
Data from field surveys indicate that approximately 20
percent of responding utilities had ClO3 – of unknown
origin at con cen tration s >10 µ g/ L in th e raw water.
A secon d poten tial sou rce of ClO 3 – is th e decom -
position of residu al h ypoch lorou s acid (HOCl) du rin g
th e disin fection process. It h as been clearly sh own ,
h owever, th at in th e dark, decom position of HOCl
at th e m g/ L level between pH 5 an d pH 8 is relatively
slow.2 On th e oth er h an d, in th e presen ce of UV ligh t
(su n ligh t), HOCl decom poses with a ph otolysis h alf-
life 5 of 12 m in at pH 8 an d in creases as th e ratio of
OCl– to HOCl in creases. Ph otocatalyzed decom posi-
tion can be avoided by operatin g at low pH or by
sh ieldin g from su n ligh t. Given th ese facts, th e for-
m ation of m easu rable ClO 3 – attribu table to HOCl
decom position u n der drin kin g water con dition s an d
in th e absen ce of su n ligh t is u n likely.
Th e th ird possible sou rce of ClO 3 – is th e decom -
position of OCl– in liqu id bleach or Ca(OCl) 2 du rin g
Copyright (C) 1997 American Water Works Association
APRIL 1 9 9 7
st o r a ge p r io r t o a p p lica t io n . Th e
detailed m ech an ism of OCl– decom -
position over th e pH 6–13 ran ge h as
been reported.2,3
Th e rate of OCl– decom position
above pH 10 is at least two orders of
m agn itu de slower th an th e decom -
position of HOCl at pH 7 u n der sim -
ilar experim en tal con dition s. How-
ever, OCl– stocks are frequently stored
at con cen tration s n ear 12–15 trade
p e r ce n t Na OCl m e a su r e d a s Cl 2 ,
which is equivalent to 1.5 X 10 5 mg/L
free available ch lorin e (FAC). (Trade
percen t = g/ L Cl2 ÷ 10, or percen t of
available Cl by weigh t X specific grav-
ity.) In con trast, th e FAC residu al in
fin ish ed drin kin g water is arou n d 1
m g/ L. Th u s, even th ou gh th e rate of
HOCl decomposition and subsequent
ClO3– formation is two orders of mag-
n itu de slower above pH 10 th an at
pH 7, th e 10 5 -con cen tration differ-
en ce in th e applied FAC solu tion s
m akes th e decom position of con cen -
trated liquid bleach the most probable
sou rce of ClO 3 – in fin ish ed water.
OCl – chemistry
Commercially produced OCl– typ-
ically contains between 0.001 and 0.1
M hydroxide ion (OH–). Thus, the liq-
uid bleach typically has a pH between
11 an d 13. In basic solu tion , OCl–
decomposition to form ClO3– has been shown to be a sec-
ond-order process3 with the following stoichiometry:
3OCl– → ClO 3 – + 2Cl– (1)
Th e decom position of OCl– in volves ch lorite ion
(ClO 2 –) as an in term ediate in th e followin g gen er-
ally accepted m ech an ism :3,4
OCl– + OCl– → ClO 2 – + Cl– (2)
OCl– + ClO 2 – → ClO 3 – + Cl– (3)
Th e reaction of ClO 2 – to form ClO 3 – is th e faster step,
m ean in g th at th e rate-lim itin g step is described in Eq
2—th e produ ction of ClO2 –. Th e ClO2 – con cen tration
in liqu id bleach is at least 200 tim es lower th an th e
OCl– con cen tration . Th u s, after application of th e
bleach for disin fection , a dose of 1 m g/ L FAC w ill
con tain <0.005 m g/ L ClO 2 –.
Th e form ation of oxygen from decom posin g OCl–
is a very slow side reaction in solu tion s of pu re OCl–
an d is con sidered to be on ly a m in or decom position
path way (<10 percen t) 3 :
OCl– + OCl– → O 2 + 2Cl– (4)
G. GORDON ET AL 143
 It h a s a lso be e n sh o w n
TAB LE 1 Sample initial data settings and program results for the predictive
model*
th at in creasin g th e tem pera-
ture of liquid bleach increases
Residual Below its rate of decom position . 2,3
Time NaOCl Chlorate Oxygen Chlorate 0.2 mg/ L Th u s, at in creased tem pera-
days wt percent g/ L L gas/ L OCl – mg/ L ClO3 – tu res (e.g., in th e su m m er
0 1 5 .8 9 0 0 .0 0 3 0 .0 0 0 0 .0 0 OK m on th s in u n ven tilated stor-
1 1 5 .6 3 7 1 .1 4 5 0 .0 3 7 0 .0 1 OK age sh eds), th e decom posi-
7 1 4 .2 7 1 7 .2 9 8 0 .2 3 5 0 .0 4 OK
14 1 2 .9 5 2 1 3 .2 4 4 0 .4 2 6 0 .0 8 OK
tion of liqu id bleach will be
28 1 0 .9 3 1 2 2 .3 5 2 0 .7 1 9 0 .1 7 OK m ore rapid.
35 1 0 .1 3 9 2 5 .9 1 7 0 .8 3 4 0 .2 1 High Th e r o le o f t r a n sit io n
42 9 .4 5 5 2 9 .0 0 1 0 .9 3 3 0 .2 5 High
56 8 .3 3 0 3 4 .0 6 9 1 .0 9 6 0 .3 3 High
m etal ion catalysis in liqu id
bleach is com plex. Alth ou gh
* Co nditio ns : c o nc e ntratio n o f NaOCl— 1 5 .8 9 0 wt pe rc e nt FAC, s pe c ific gravity— 1 .2 3 9 g/ mL, te mpe rature —
o
2 5 .0 C, pH— 1 3 .0 0 0 , c hlo rate c o nc e ntratio n— 0 .0 0 3 g/ L, c hlo ride c o nc e ntratio n— 2 .7 7 7 mo l/ L, applie d FAC
the formation of ClO3– is gen-
–
do s e — 1 .0 mg/ L, ClO maximum c o ntaminant le ve l go al— 0 .2 mg/ L, half-life — 6 1 .7 days
3 erally n ot a m ajor con cern ,
O2 formation is a minor path-
w ay for decom position cat-
alyzed by tran sition m etal
ion s. Nickel ion appears to
e ffe ct ive ly ca t a lyze FAC
TAB LE 2 Experimental and predicted half-lives of NaOCl decomposition
at 25 C o decom position an d produ ces
O 2 eith er alon e or in com bi-
Experimental Predicted n ation with oth er tran sition
NaOCl Half-lives Half-lives Percent m etal ion s. Th e im plication
Bleach Type M days days Difference*
fo r u t ilit ie s is t h a t FAC
Co ntinuo us 2 .8 0 1 5 6 .4 5 7 .5 2 .0 stren gth w ill be redu ced in
Co ntinuo us 2 .4 6 1 8 7 .2 8 1 .4 6 .7 th e presen ce of som e tran si-
Co ntinuo us 1 .5 3 7 225 247 9 .8
Co ntinuo us 0 .7 1 4 886 974 9 .9 tion m etal ion s. Th u s, tran si-
Batc h 2 .0 6 5 116 125 7 .8 tion metal ion concentrations
Batc h 1 .7 4 7 181 192 6 .1
Batc h 1 .2 5 0 357 372 4 .2
sh o u ld be m in im ize d in
Labo rato ry 2 .1 4 6 218 227 4 .1 bleach stock. Th e m axim u m
Labo rato ry 2 .1 0 8 225 248 1 0 .2 con cen tration of tran sition
1/ 2 –
* Pe rc e nt diffe re nc e as c alc ulate d by the e quatio n [(e xpe rime ntal 1/ 2pre dic te d )/ e xpe rime ntal
1/ 2 ] X 100. m etal ion s th at will n ot sig-
–
Me as ure d Cl c o nc e ntratio ns we re us e d in the pre dic tio n. Ave rage pe rc e nt diffe re nc e was 6 .7 . n ifican tly affect bleach de-
composition is ~0.1 mg/L Ni2+
a n d ~ 1 m g/ L Cu 2 + . W h e n
p r e se n t a lo n e , fe r r ic io n
Stored liquid bleach with a pH between 12 and 13 (Fe 3+) and Mn 2+ are not considered to be effective cat-
is the most stable. The pH of decomposing OCl– should alysts for bleach decom position .3
n ot ch an ge with tim e above pH 11, as in ferred from Although ClO3– formation is not accelerated in the
th e stoich iom etric equ ation for th e prim ary decom - presence of catalytic amounts of transition metal ions,
position pathway6 above pH 11 (no acid dependence). th e redu ction of FAC via th e oxygen path way is a
On th e oth er h an d, as th e pH of th e bleach stock con cern . Man u factu rers of liqu id bleach u n derstan d
solu tion is lowered, th e pH of th e OCl– stock solu tion th at tran sition m etal ion s redu ce th e FAC con cen tra-
will continue to decrease over time because of a com- tion , an d m an y h ave begu n to elim in ate tran sition
petin g acid-produ cin g decom position path way 2 with m etal ion sou rces in th e m an u factu rin g process by
th e followin g stoich iom etry: filterin g th e fin ish ed produ ct. Filterin g is th e su rest
way to redu ce tran sition m etal ion con cen tration s. If
2HOCl + OCl– → ClO 3 – + 2H+ + 2Cl– (5) th e bleach su pplier does n ot filter, 3 u tilities sh ou ld
specify that transition metal ions must be < 0.10 mg/L.
At th e u tility, th is cou ld be a seriou s problem
becau se th e h igh NaOCl con cen tration in th e stored Treatment of chemical data versus
liqu id bleach w ill lead to a large ClO 3 – residu al in empirical data
the storage tank and eventually in the finished water. Predictive m ath em atical m odels for ch em ical sys-
Dilu tion of stock bleach can be an effective strat- tem s can be based on eith er a ch em ical u n derstan d-
egy for m in im izin g OCl– decom position an d con - ing of the system or an empirical analysis of the avail-
com itan t ClO 3 – form ation . To m in im ize th e rate of able data.
ClO 3 – form ation , h ow ever, th e pH also m u st be Th e m ore accu rate ch em ical m odel requ ires de-
adju sted to an d m ain tain ed at ⱖ pH 11. Typically, tailed kin etic data an d a fu ll m ech an istic u n der-
com m ercial bleach con tain s en ou gh excess cau stic stan din g of th e im portan t variables an d m olecu lar
to m ain tain a pH ⱖ 12 after 50 percen t dilu tion . in teraction s th at affect th e ch em ical system . From

Copyright (C) 1997 American Water Works Association

144 VOLUME 8 9 , ISSUE 4 JOURNAL AWWA
th is u n derstan din g an d detail, a com plete ch em ical
FIG U R E 1 Measured and predicted NaOCl
description of a dyn am ic system can be developed. decomposition and ClO3 – formation
In con trast, em pirical m odels provide approxi- using the predictive model
m ate m ath em atical description s of ch em ical system s
an d are n ot based on a fu ll u n derstan din g of th e Predicted Values Experimental Values
ch em ical detail. An em pirical m odel wou ld rath er fit OCI
–
OCI
–
– –
a fu n ction , su ch as th e rate con stan t–ion ic stren gth CIO3 CIO3
relation sh ip, w ith an in fin ite series equ ation . Th is O2
2.5
best-fit equ ation provides a m ath em atical solu tion
to a predeterm in ed degree of certain ty. Th u s, th is
model does not provide an accurate chemical descrip- 2.0
tion of th e process, an d th e resu ltin g m ath em atical

represen tation h as n o ch em ical im plication , is n ot
totally predictive, an d m ay n ot be applicable in situ -
ation s in wh ich th e con dition s differ from th ose u sed
to develop th e em pirical regression . Application of
an em pirical m odel lackin g ch em ical detail m ay be
risky, because the assumptions of the model itself can
lead to con fu sion abou t th e adverse effects on an d
im plication s to th e ch em ical process. Th e degree of
error and uncertainty associated with empirical mod-
elin g is h igh , an d th e possibility of in correct extrap-
olation s can lead to poor tim e-depen den t prediction s
th at h ave th e poten tial to be catastroph ic if or wh en
th e m odel fails.
Specific ch em ical equ ation s th at in corporate tem -
peratu re an d ion ic stren gth were u sed in develop-
m en t of th e Gordon –Adam bleach decom position
model. The kinetic information was used to formulate
fu n dam en tal ch em ical an d th erm odyn am ic equ a-
tion s describin g th e bleach decom position in th e pH
11–14 range. The model is based on a complete chem-
ical reaction set, experim en tal reaction rates, an d a
detailed rate law th at takes in to accou n t th e reaction
stoich iom etry.3
Gordon–Adam bleach decomposition model
Th e Gordon –Adam com pu ter-based m odel w as
d e ve lo p e d to p r e d ict ClO 3 – fo r m a tio n a n d OCl –
decom position in stored liqu id bleach . Th e ch em ical
model has been incorporated into a computer program
th at requ ires a u tility to in pu t on ly th e bleach an d
ClO 3 – con cen tration s (preferably recen tly m easu red
valu es) alon g with th e storage tem peratu res an d pH.
If available, addition al in pu t factors (su ch as Cl–) can
be en tered by th e u ser or, if u n available, will be esti-
m ated by th e program .
Th e bleach decom position m odel h elps u tilities
using liquid bleach answer some important questions:
• How lon g does th e stored bleach m ain tain its
stren gth ?
• How rapidly will ClO 3 – bu ild u p?
• Will oxygen form ation be a problem ?
• Wh at is th e projected m axim u m tim e to store
th e bleach before u se?
• Wh at are th e econ om ies of scale in pu rch asin g
versu s storin g liqu id bleach ?
• Wh at sh ou ld be th e m in im u m ClO 3 – con cen -
tration in th e desired produ ct?
Field data in dicate th at th e program can be u sed
to predict bleach con cen tration to better th an ±10
Concentration—mol/L
1.5
1.0
0.5
0.0
0 25 50 75 100
Time—days
percen t. Table 1 sh ows th e kin d of in form ation gen -
erated by th e program ; th e prin tou t in clu des in pu t
bleach in form ation , predicted bleach decom position ,
ClO3– formation, oxygen formation, the residual ClO3–
concentration in the distribution system, and the like-
lih ood of th e predicted distribu tion con cen tration
exceedin g th e u ser-defin ed m axim u m con tam in an t
level goal.
The computer program was developed to be used
with a spreadsheet program.* In practice, a user will
proceed through a series of program prompts that are
automatically displayed on the viewing monitor. No
programming or advanced computer skills are required.
If an entered value is outside the acceptable range for
a particular parameter (e.g., the pH must be between
11 and 14), the program will inform the user that a
problem exists and will require a new input value.
Verif icat ion of half -lif e predict ions. Table 2
sh ow s experim en tal an d m odel prediction s for th e
h alf-lives of variou s con cen tration s of th ree bleach
types (con tin u ou s process, batch process, an d labo-
ratory-produ ced). Th e differen ces betw een th e ex-
perim en tal an d predicted h alf-lives average less th an
7 percen t.
Figure 1 illustrates FAC decomposition at 25 o C for
bleach man u factu red by th e con tin u ou s-bleach proc-
ess. Th e valu es of experim en tal data poin ts (m arked
as squ ares an d trian gles) fall on th e predicted OCl–
decom position an d ClO 3 – form ation cu rves.
Predictions under conditions of temperature
variation. Tables 3 an d 4 sh ow m easu red an d pre-
*Lotu s 1–2–3

Copyright (C) 1997 American Water Works Association

APRIL 1 9 9 7 G. GORDON ET AL 145
 in g tran sportation . A con -
TAB LE 3 ClO3 – measurements* from liquid bleach holding study
stan t tem peratu re of 18 o C
Measured Predicted
w as applied for th e sam ple
FAC Temperature ClO3 – ClO3 – tran sit day. Du rin g storage,
oC
Stock trade percent Day g/ L g/ L tem peratu re was m ain tain ed
A 15 13 28 2 .9 9 3 .1 1
at 8 o C. On th e day of m ea-
10 13 28 1 .1 7 1 .4 2 su rem en t, tem peratu re w as
5 13 28 0 .5 0 0 .5 5 m ain tain ed at 19 o C.
15 27 28 1 2 .7 1 1 0 .8 6
10 27 28 4 .5 6 4 .0 8 Full-scale utility evalu-
5 27 28 1 .0 7 1 .0 2 ation. Many utilities are plan-
B 15 13 30 3 .8 1 3 .9 9
10 13 30 1 .4 5 1 .5 8
n in g m ajor con version s to
5 13 30 0 .4 9 0 .5 1 switch from gaseou s ch lorin e
15 27 30 1 7 .4 6 1 5 .0 7 to liqu id bleach as th eir pri-
10 27 30 6 .1 9 5 .5 1
5 27 30 1 .0 8 1 .1 6
m ary disin fectan t. Th is deci-
C 15 13 30 3 .0 4 3 .1 4 sion is bein g m ade in part
10 13 30 0 .9 9 1 .1 4 because of a movement away
5 13 30 0 .2 6 0 .2 8
15 27 30 1 7 .0 4 1 3 .8 5 from the stricter requirements
10 27 30 5 .6 0 5 .0 0 for gaseou s ch lorin e storage
5 27 30 0 .9 9 0 .9 3
a n d h a n d lin g. Du r in g a
–
* All ClO me as ure me nts we re made by io n c hro mato graphy ac c o rding to USEPA me tho d 3 0 0 .
3 period of u pgradin g ch emical
and storage-handling systems,
th e East Bay Mu n icipal Util-
TAB LE 4 FAC measurements* from liquid bleach holding study ity District u n dertook a six-
m on th stu dy to test con trol
Measured Predicted
FAC Temperature FAC FAC st r a t e gie s fo r m in im izin g
Stock trade percent oC Day g/ L g/ L ClO 3 – fo r m a t io n a t t h e
A 15 13 28 1 3 2 .6 1 3 1 .1
Lafayette Water Treatm en t
10 13 28 9 0 .1 9 0 .7 Plan t. Th e plan t is an in lin e
5 13 28 5 3 .5 5 3 .0 direct filtration facility u sin g
15 27 28 1 0 8 .1 1 0 9 .8
10 27 28 8 3 .5 8 3 .4
surface water and has a nom-
5 27 28 5 0 .1 5 1 .8 in al capacity of 40 m gd (0.15
B 15 13 30 1 4 6 .4 1 4 4 .6 X 10 6 m 3 / d).
10 13 30 1 0 2 .0 1 0 1 .8
5 13 30 5 2 .7 5 1 .2 As expected, stu dies u n -
15 27 30 1 1 0 .1 1 1 4 .1 d e r t a k e n o n st o r e d liq u id
10 27 30 8 9 .6 9 1 .0
5 27 30 4 9 .5 4 9 .4
ble a ch sh o w e d th a t a s th e
C 15 13 30 1 4 9 .8 1 4 3 .7 o r igin a l co n ce n tr a tio n d e -
10 13 30 1 0 2 .5 1 0 1 .6 creased, th e rate of ClO3 – for-
5 13 30 5 2 .1 5 1 .3
15 27 30 1 1 1 .3 1 1 4 .2
m ation also decreased. For
10 27 30 7 6 .3 9 1 .0 e x a m p le , a ft e r 3 0 d a ys, a
5 27 30 5 0 .0 4 9 .5 dilu ted liqu id bleach stock
* All FAC me as ure me nts we re made us ing po te ntio me tric titratio n. h avin g a con cen tration of 10
trade percent had a measured
ClO 3 – con cen tration of 9,000
m g/ L. By com parison , a fu ll-
dicted ClO 3 – an d FAC con cen tration s for a liqu id stren gth bleach stock (15 trade percen t) con tain ed
bleach h oldin g stu dy con du cted by a con su ltin g 14,000 m g/ L ClO 3 – after 30 days.
firm .* Th ree bleach su ppliers provided bleach sam - Prelim in ary estim ates on th e m axim u m ClO 3 –
ples. Th e th ree bleach es w ere dilu ted by 33 an d 66 con cen tration s allowable in th e liqu id bleach stor-
percen t, respectively, an d in cu bated in glass con - age tan ks were m ade based on th e Californ ia action
tain ers in th e dark at 13 an d 27 o C for 28–30 days. level (0.2 mg/L in the finished water). The maximum
Periodically, th e sam ples were sh ipped by overn igh t allowable ClO3– concentration at an applied FAC dose
delivery to a tech n ical facility† for m easu rem en t of of 2 m g/ L was calcu lated to be 15,000 m g/ L for a 15
ClO 3 – an d FAC. trade percen t bleach solu tion . Th e ClO 3 – in th e stock
ClO 3 – m easu rem en t data in Table 3 h ave a calcu - 10 trade percen t solu tion m easu red after 40 days
lated average deviation from predicted ClO 3 – con - exceeded th e m axim u m allowable ClO 3 – con cen tra-
cen tration of ±10 percen t. FAC m easu rem en t data tion (10,000 m g/ L) for an applied FAC dose of 2
in Table 4 h ave a calcu lated average deviation from mg/L. The measured ClO3 – concentration was 11,250
th e predicted FAC con cen tration of ±2.5 percen t. m g/ L. Neverth eless, becau se th e actu al applied FAC
Th e predicted an d m easu red valu es for ClO 3 – an d
FAC con cen tration s are in good agreem en t despite *Malcolm Pirn ie In c., Mah wah , N.J.
assu m ption s m ade abou t tem peratu re ch an ges du r- †Notavek, Oxford, Oh io

Copyright (C) 1997 American Water Works Association

146 VOLUME 8 9 , ISSUE 4 JOURNAL AWWA
dose was less than 2 mg/L, the amount of ClO3– enter-
in g th e system did n ot exceed th e Californ ia action FIG U R E 2 Regression of predicted and measured
ClO3 – concentrations in storage tank 1
level of 0.2 m g/ L.
over a 30-day period
Figu res 2 an d 3 sh ow th e correlation betw een
predicted and measured ClO3 – concentrations for two Perfect correlation (m = 1) Measured CIO3
–

batch es of liqu id bleach over th e 30-day storage Calculated regression (m = 0.76)

period. In Figu re 2, th e m odel u n derestim ated th e 20

ClO 3 – con cen tration by an average of 24 percen t, 18

u sin g th e differen ce in th e slope of th e regression

Measured CIO3 —mg/L ✕ 1,000

16
lin e for m easu red ClO 3 – con cen tration an d a perfect
correlation . Th is deviation valu e does n ot accou n t 14
for possible errors in the analysis and in measuring the 12
actu al tem peratu re an d dilu tion of th e delivered liq-
10

–
u id bleach solu tion .
Figu re 3 sh ow s th at th e m odel overestim ates 8
th e ClO 3 – con cen tration in a secon d liqu id bleach 6
storage tan k by approxim ately 13 percen t. Non eth e-
4
less, th e agreem en t between th e predicted an d m ea-
su red ClO 3 – con cen tration s in th e storage tan k is 2
im proved, i.e., th e data sh ow less scatter alon g th e 0
regression lin e. Th is im provem en t cou ld be du e in 0 2 4 6 8 10 12 14 16 18 20
part to im proved an alysis of th e h igh ly con cen - –
Predicted CIO3 —mg/L ✕ 1,000
trated liqu id bleach , in clu din g dilu tion s an d tem - The solid line is the regression line of the measured
peratu re con trol. data points.
In gen eral, th e data sh ow th at in th e field, dif-
feren ces betw een predicted an d m easu red ClO 3 –
con cen tration s w ere greater. Th e in crease in th e
differen ces betw een th e predicted an d m easu red
FIG U R E 3 Regression of predicted and measured
valu es is believed to be th e resu lt of u n certain ties
ClO3 – concentrations in storage tank 2
du rin g dilu tion of th e con cen trated bleach . Th is is over a 30-day period
n ot su rprisin g, becau se th e laboratory develop-
m en t of th e m odel w as perform ed u n der tigh tly Perfect correlation (m = 1) Measured CIO3
–

con trolled con dition s, w h ereas in th e field proce- Calculated regression (m = 1.13)
20
du res su ch as dilu tin g stock solu tion s u sin g an
exterior sigh t glass provided on ly rou gh estim ates 18
Measured CIO3 —mg/L ✕ 1,000

( ± 1 0 –2 5 p e r ce n t ) . Co n sid e r in g t h e n u m e r o u s 16
poten tial sou rces of error, th e field data in dicate
14
th at th e m odel can be u sed to predict ClO 3 – con -
cen tration s in stored liqu id bleach at a level th at is 12
acceptable to u tilities operatin g u n der fu ll-scale 10
–

con dition s.
The predicted model and operational observations 8

made during the full-scale evaluation revealed several 6

operation al con sideration s th at deserve m en tion for 4
fu ll-scale con trol of ClO 3 –.
• Oxygen can form du rin g th e decom position of 2

stored liqu id bleach . Allowan ces m u st be m ade for 0

relieving gas bubbles on the suction side of the chem-
ical feed pu m ps an d at an y h igh poin t of th e pipin g
system .
• Polyvin yl ch loride (PVC) ball valves w ith ou t
som e sort of pressu re relief capability sh ou ld n ot be
u sed in th e bleach delivery or tran sfer lin es. On two
separate occasion s, PVC valves split u n expectedly
after less than six months of use in the piping system.
It was h ypoth esized th at th e gen eration of oxygen
in th e isolated ch am ber of th e ball placed a stress on
the molded valve seam, forcing the two valve parts to
split. At least on e u tility district h as recom m en ded
th at ru bber-lin ed diaph ragm valves be u sed in ch em -
ical-h an dlin g facilities.
0 2 4 6 8 10 12 14 16 18 20
–
Predicted CIO3 —mg/L ✕ 1,000
• Stain less-steel ball valves sh ou ld n ot be u sed
because the surface of the ball becomes pitted and can
cau se th e valve to freeze an d becom e n on fu n ction al.
• Wh en sigh t glasses are u sed for dilu tin g, th e
difference in specific gravity between water and bleach
m u st be taken in to accou n t to preven t overfillin g th e
storage tan k.
• Becau se leaks frequ en tly occu r at th readed
join ts, on ly slip-on type join ts are recom m en ded.

Copyright (C) 1997 American Water Works Association

APRIL 1 9 9 7 G. GORDON ET AL 147
 Ca ( O Cl) 2 d e co m p o sit io n
TAB LE 5 MWDSC holding study of aqueous Ca(OCl) 2 * prior to dissolu tion :
25 oC 38 oC
Measured Predicted Measured Predicted
3 Ca(OCl) 2 → 2 CaCl2 (6)
ClO3 – ClO3 – Difference† ClO3 – ClO3 – Difference‡ + Ca(ClO 3 ) 2
Day mg/ L mg/ L percent mg/ L mg/ L percent

0 320
1 330 323 2 .1 340 333 2 .1 Ca(OCl) 2 → 2 CaCl2 + O 2 (7)
3 340 327 3 .8 400 360 1 0 .0
7 370§ 336 9 .2 420 411 2 .1
9 340 341 0 .3 430 437 1 .6 At room tem peratu re for
14 350 352 0 .6 510 500 2 .0 so lid Ca ( OCl) 2 , t h e ClO 3 –
21 380 367 3 .4 590 587 0 .5
29 380 385 1 .3 660 683 3 .5
path w ay is proposed to ac-
35 380 398 4 .7 710 754 6 .2 cou n t for rou gh ly 70 percen t
42 380 414 8 .9 780 835 7 .1 decom position , w h ereas O 2
49 390 429 1 0 .0 890 913 2 .6
56 400 444 1 1 .0 940 990 5 .3 accou n ts for approxim ately
65 420 464 1 0 .5 980 1 ,0 8 7 1 0 .9 30 percent. Experience shows
77 510 490 3 .9 1 ,1 7 0 1 ,2 1 1 3 .5
th at at th e tim e of m an u fac-
* Parame te rs us e d inc lude 2 pe rc e nt FAC, pH = 1 1 .8 9 , Cl– = 0 .2 8 6 M. ture, typically the weight per-
† Me an = 5 .4 cen t Ca(ClO 3 ) 2 is 1.4 percen t
‡ Me an = 4 .4
– o
§ Statis tic ally, it appe ars that the me as ure d ClO3 c o nc e ntratio n fo r day 7 at 2 5 C is to o high.
a n d t h e w e igh t p e r ce n t
Re mo val o f this data po int re duc e s the ave rage pe rc e nt diffe re nc e to 5 .0 . Ca(OCl) 2 ) is between 70 an d
75 percen t. Th u s, w h en on e
m o le o f Ca ( O Cl) 2 d e co m -
p o se s, 0 .2 3 –0 .2 5 m o le s o f
Ca(ClO 3 ) 2 w ill form . In th e
TAB LE 6 Measured and predicted Ca(ClO3 ) 2 concentrations* as a function absence of more detailed data
of holding time for solid Ca(OCl) 2 to th e con trary, th e w eigh t
p e r ce n t Ca ( ClO 3 ) 2 ca n b e
Measured Predicted
Time Ca(OCl) 2 Ca(ClO2 ) 2 Ca(ClO3 ) 2 Mass [Ca(ClO3 ) 2 ] t mg/ L ClO3 –/
e st im a t e d u sin g a cu r r e n t
days mol/ 100 g mol/ 100 g mol/ 100 g Balance mol/ 100 g mg/ L FAC titration of w eigh t percen t
Ca(OCl) 2 ).
0 0 .4 9 3 0 .0 0 0 0 7 0 .0 0 6 5 0 .5 1 3 0 .0 2
9 0 .4 9 1 0 .0 0 0 0 7 0 .0 0 6 6 0 .5 1 1 0 .0 0 7 0 0 .0 2 If [Ca(ClO 3 ) 2 ] 0 is taken as
18 0 .4 9 0 0 .0 0 0 1 0 .0 0 6 5 0 .5 1 0 0 .0 0 7 2 0 .0 2 th e in itial assay of calciu m
32 0 .4 8 0 0 .0 0 0 2 0 .0 0 6 7 0 .5 0 1 0 .0 0 9 5 0 .0 2
46 0 .4 9 1 0 .0 0 0 5 0 .0 0 7 3 0 .5 1 4 0 .0 0 7 0 0 .0 2
ch lorate ( t = 0; e.g., date of
74 0 .4 8 2 0 .0 0 0 3 0 .0 0 7 1 0 .5 0 4 0 .0 0 9 0 0 .0 2 m an u factu re or as m easu red
116 0 .4 7 6 0 .0 0 0 4 0 .0 0 6 1 0 .4 9 5 0 .0 1 0 4 0 .0 2 on th e day of receipt) an d
* Co nc e ntratio n units are in mo le s o f e ac h s pe c ie s pe r 1 0 0 grams o f s ample .
[Ca(OCl) 2 ] 0 is th e assay of
Ca(OCl) 2 at t = 0, th en ass-
u m in g 70 percen t form ation
of ClO 3 –, th e calciu m ch lo-
–
Solid Ca(OCl) 2 . OCl is also available to u tilities rate after storage for som e tim e ( t) can be estim ated
as Ca(OCl) 2 . Becau se Ca(OCl) 2 is m ore expen sive by th e equ ation
than liquid bleach, many utilities do not use it regard-
less of th e fact th at Ca(OCl) 2 —a solid—can be stored [wt percen t Ca(ClO 3 ) 2 ] t = [wt percen t Ca(ClO 3 ) 2 ] 0
and used over longer periods of time with less decom- + ([wt percen t Ca(OCl) 2 )] 0 (8)
position . Utilities u sin g Ca(OCl) 2 prepare a h oldin g – [wt percen t Ca(OCl) 2 )] t)
tan k of con cen trated FAC. In solu tion , Ca(OCl) 2 X 0.23
decom poses to form ClO 3 – by th e sam e path ways as
liqu id bleach . Th e gen eral applicability of th e equ ation w as
Th e Metropolitan Water District of Sou th ern Cal- dem on strated by stu dyin g seven batch es of recen tly
iforn ia (MWDSC) u n dertook a h oldin g stu dy by dis- m an u factu red Ca(OCl) 2 stored at room tem peratu re
solving solid Ca(OCl)2 in treated Colorado River water. in glass and plastic containers. The container lids were
Sam ples (approxim ately 2 percen t FAC) were placed tigh tly sealed an d open ed on ly to rem ove sam ple
in tem peratu re-con trolled water bath s at 25 o C an d aliqu ots for m easu rem en t. Periodically, th e sam ples
38 o C. Usin g ion ch rom atograph y, MWDSC periodi- were measured. Some results are presented in Table 6.
cally an alyzed sam ples for ClO 3 –. Resu lts are sh own Table 6 demonstrates that the decomposition of solid
in Table 5. Predicted ClO 3 – con cen tration s u sin g th e Ca(OCl) 2 is a very slow process. ClO3– is formed at a
model are in good agreement with the measured val- considerably slower rate in solid Ca(OCl)2 than in solu-
u es (± 5 percen t). tion . In itially in th e solid Ca(OCl) 2 th e in term ediate
With specific h elp* an d th e existin g ch em ical lit-
e r a t u r e , t h e fo llo w in g m o d e l is p r o p o se d fo r *Man fred Noack, Olin Corp., Ch esh ire, Con n .

Copyright (C) 1997 American Water Works Association

148 VOLUME 8 9 , ISSUE 4 JOURNAL AWWA
Ca(ClO2)2 builds up to a relatively low steady-state con-
centration. This is a clear indication that the Ca(OCl)2 is
undergoing a very slow decomposition (1/2 of years) by a
pathway similar to that which occurs in liquid bleach. This
means that when Ca(OCl)2 is used to prepare a holding
tank of FAC, some ClO3– will come directly from decom-
position of the solid Ca(OCl)2 that occurred during stor-
age as shown by the data at t = 0 in Table 6. Following dis-
solution, additional ClO3– will be formed in the holding
tank as a result of FAC decomposition occurring in solu-
tion. For individual solid samples of Ca(OCl) 2 , as the
con cen tration of th e solid OCl– decreases, th e mea-
sured and predicted ClO3– concentrations increase. These
values are within an average of ± 0.0005 mol/100g. Table
6 also points out that the mg/L ClO3– contributed per
mg/L FAC added using the solid Ca(OCl) 2 product is
0.02 mg/L or less after 116 days of storage.
Conclusions
The temperature variation, dilution, and full-scale
studies confirmed that the Gordon–Adam OCl– com-
puter-based decomposition model is a good predictor
of ClO3– formation in stored liquid bleach. The com-
puter program can be used to implement effective stages
to minimize ClO3– formation. As the model predicts and
pilot-scale studies have shown, storage and handling
facilities must be properly designed to account for oxy-
gen formation resulting from the decomposition of OCl–
that is a result of transition metal ion impurities. The
accuracy of the model for predicting ClO3– in field stud-
ies is excellent despite the potential for a high degree of
variability caused by experimental conditions.
Th ree basic strategies are available to u tilities for
m in im izin g ClO 3 – form ation in liqu id bleach : (1)
dilu tin g th e con cen trated liqu id bleach im m ediately
after delivery, (2) u sin g lower storage tem peratu res,
an d (3) avoidin g su n ligh t du rin g storage. Dilu tion of
con cen trated liqu id bleach solu tion s in creases th e
OCl– h alf-life an d decreases th e rate of OCl– decom -
position . Du rin g dilu tion , th e pH sh ou ld be m ain -
tained at or adjusted to between pH 12 to 13. Decreas-
in g th e tem peratu re at wh ich th e liqu id bleach stock
is stored decreases th e rate of decom position . Sh ield-
in g storage tan ks from su n ligh t is also h elpfu l.
Th e role of tran sition m etal ion s is com plex. De-
com position attribu table to tran sition m etal ion s is
depen den t on wh eth er th e ion s are in con cen trated
liqu id bleach , are presen t at m illigram -per-litre levels
in solu tion s con tain in g m g/ L FAC, are presen t in
com bin ation with oth er tran sition m etal ion s, or are
in th e presen ce of UV ligh t. Th erefore, it is im por-
tan t th at Ni2+ or Cu 2+ con cen tration s be < 0.1 m g/ L
in con cen trated liqu id bleach solu tion s. Th e pH of
liquid bleach at delivery should be between 12 and 13.
Both O 2 form ation an d ClO 3 – form ation are cat-
alyzed by UV ligh t (su n ligh t). As th e in ten sity of UV
ligh t in creases, th e form ation of O2 is favored. Th ere-
fore, m in im izin g exposu re to su n ligh t is im portan t
du rin g both th e storage an d application of FAC.
The amount of ClO3– that is acceptable in delivered
liquid bleach depends on many factors. Obviously, fresh
Copyright (C) 1997 American Water Works Association
APRIL 1 9 9 7
or n ewly m ade liqu id bleach will con tain less ClO 3 –
than bleach that is stored without concern for temper-
ature or other variables that affect OCl– decomposition
and ClO3– formation. If a utility is using a single tank to
store liquid bleach, a residual ClO3– concentration prob-
ably is building in the tank. Thus, storage tanks should
be periodically flushed and cleaned.
Acknowledgment
Th e au th ors th an k th e AWWA Research Fou n da-
tion; Miami University, Oxford, Ohio; Malcolm Pirnie,
Newport News, Va.; th e East Bay Mu n icipal Utility
District, Oaklan d, Calif.; th e Metropolitan Water Dis-
trict of Sou th ern Californ ia, La Vern e, Calif.; an d
Man fred Noack, Olin Corp., Ch esh ire, Con n .
References
1. USEPA Nation al Prim ary Drin kin g Water Regu -
lation s: Disin fectan ts an d Disin fection By-prod-
u cts. Proposed Ru le. 40 CFR 141 an d 142 (Ju ly
29, 1994).
2. ADAM , L. ET AL. Hypoch lorou s Acid Decom posi-
t io n in t h e p H 5 t o 8 Re gio n . In org. Ch em . ,
31:3534 (1992).
3. GORDON, G.; ADAM , L.; & BUBNIS, B. Min im izin g
Chlorate Ion Formation in Drinking Water When
Hypoch lorite Ion is th e Ch lorin atin g Agen t.
AWWA, Den ver (1995).
4. GORDON, G; ADAM , L; & BUBNIS, B. Min im izin g
Ch lorate Ion Form ation . Jour. AWWA , 87:6:97
(Ju n e 1995).
5. NOWELL, L & HOIGNÉ, J. Ph otolysis of Aqu eou s
Chlorine at Sunlight and Ultraviolet Wavelengths—
I. Degradation Rates. Water Res., 26:593 (1992).
6. LISTER, M.W. Uncatalyzed and Catalyzed Decom-
position of Sodiu m Hypoch lorite. Canadian Jour.
Chem., 34:465 an d 34:479 (1956).
About the authors: Gilbert Gor-
don* is the Volwiler Research Professor
in the Department of Chemistry, Miami
University, Hughes Laboratories, Ox-
ford, OH 45056. He holds a PhD from
Michigan State University (East Lans-
ing), has completed postdoctoral work
at the University of Chicago, and has
spent more than 30 years researching the reactions of chlorine,
chlorine dioxide, and ozone. Luke C. Adam is an analytical
inorganic chemist, Safety-Kleen Inc., 3700 La Grange Rd.,
Smithfield, KY 40068. Bernard P. Bubnis is president of
Novatek, a Division of EBB Inc., 5172 College Corner Pike, POB
608, Oxford, OH 45056. Ching Kuo is a chemist with the Met-
ropolitan Water District of Southern California, 700 Moreno
Ave., La Verne, CA 91750. Robert S. Cushing is a project engi-
neer at Malcolm Pirnie Inc., 11832 Rock Landing Dr., Suite
400, Newport News, VA 23606. Richard H. Sakaji is a senior
sanitary engineer with the California Department of Health
Services, Division of Drinking Water and Environmental Man-
agement, 2151 Berkeley Way, Room 449, Berkeley, CA 94704.
*To wh om correspon den ce sh ou ld be addressed.
G. GORDON ET AL 149