[advances in chemistry] aquatic humic substances volume 219 (influence on fate and treatment of...

27 Effects of Humic Substances on Particle Formation, Growth, and Removal During Coagulation

Gary L . Amy, Michael R. Collins, C. James Kuo, and Zaid K. Chowdhury

Environmental Engineering Program, Department of Civil Engineering, University of Arizona, Tucson, AZ 85721

Roger C. Bales

Department of Hydrology and Water Resources, University of Arizona, Tucson, AZ 85721

This chapter describes research examining (1) the effect of natural organic matter (NOM) on the formation, growth, and removal of particles during coagulation using aluminum sulfate and (2) the effect of particles on NOM removal. Several sources of NOM and several types of mineral particles were studied under water treatment conditions. Particle formation, growth, and removal were found to be significantly affected by initial particle type and concentration, NOM type and concentration, pH, and aluminum sulfate dose.

SURFACE WATER CONTAINS PARTICLES ranging in size from submicrometer to supramicrometer dimensions, as well as aquatic natural organic matter (NOM), including humic substances. Important objectives of both conventional surface-water treatment and direct filtration are removal of both turbidity-causing particles and ΝΟΜ. Total particle number is reduced during coagulation / flocculation by inducing particle growth, but subsequent solids-l iquid separation (i.e., sedimentation and filtration) removes floe composed of aggregated particles and precipitated aluminum hydroxide.

0065-2393/89/0219-0443$06.00/0 © 1989 American Chemical Society

Dow

nloa

ded

by U

CSF

LIB

CK

M R

SCS

MG

MT

on

Sept

embe

r 4,

201

4 | h

ttp://

pubs

.acs

.org

P

ublic

atio

n D

ate:

Dec

embe

r 15

, 198

8 | d

oi: 1

0.10

21/b

a-19

88-0

219.

ch02

7

In Aquatic Humic Substances; Suffet, I., et al.; Advances in Chemistry; American Chemical Society: Washington, DC, 1988.

444 AQUATIC HUMIC SUBSTANCES

Experimental Design

Four independent series of experiments (A through D) were conducted with solutions containing a model particle and either aquatic dissolved organic matter (DOM) or soil fulvic acid. A l l experiments evaluated aluminum sulfate coagulation by using a conventional jar test apparatus, but mixing conditions (Table I) and "postmixing" sample processing were different. The principal objective of series A experiments was to evaluate particle removal by passing flocculated water through a laboratory sand filter functioning as a batch-mode direct-filtration apparatus. The dimensions and operating conditions of the laboratory sand filter are described elsewhere (J). In contrast, the major objective of series Β through D experiments was to study particle formation and growth. The filtrate was characterized in series A. In series B - D , the flocculated suspension was analyzed. For each series, a preliminary set of experiments was run to provide a basis for selecting aluminum doses.

The organic substances examined were a soil-derived fulvic acid (series A and D) and D O M isolated from two surface waters: the Grasse River in New York (series B) and the Edisto River in South Carolina (series C). D O M was defined as organic matter passing a 0.45-μπι membrane filter.

Two parameters used for particle characterization were the total particle number (TPN), which represents the summation of particle concentrations observed in all channels of the particle counter, and the volume average diameter (dv), which is a volume-weighted average. T P N was used to indicate overall particle removal, and dv indicated particle-size changes. Although values of T P N and dv for the initial conditions reflect only the model particle, these same parameters for the final water reflect aluminum-particle-humic aggregates in the filtered water (series A) or in the flocculated suspension (B through D).

Experimental Details The hydrophobic (humic) fraction of the DOM, which was used in the experiments, was isolated by adsorption onto resin (XAD-8, Rohm and Haas), according to the method of Thurman and Malcolm (2). The average molecular weight of the DOM was determined by ultrafiltration (3). Carboxylic acidity was measured by potentiometric titration of the hydrophobic fraction from pH 3.0 to 8.0 (4). Non-purgeable organic carbon (NPOC) was determined with an organic carbon analyzer (Dorhmann DC-80), and UV absorbance was measured with a UV-visible spectrophotometer (Perkin-Elmer Model 200).

Three different particles were studied: (1) kaolinite clay that contained significant amounts of submicrometer material, series A; (2) Min-U-Sil-5 (Pennsylvania Glass Sand Corporation), series Β and C; and (3) Min-U-Sil-15, series D. The Min-U-Sil materials represent crystaline silica, Si02. Particle enumeration and size characterization were done with an optical particle counter (Hiac 4100, Pacific Scientific Inc.), using a 2.5-150-μπι sensor (series A and B) and a l-60-μπι sensor (series C and D). The more sensitive sensor was acquired midway through the research.

Dow

nloa

ded

by U

CSF

LIB

CK

M R

SCS

MG

MT

on

Sept

embe

r 4,

201

4 | h

ttp://

pubs

.acs

.org

P

ublic

atio

n D

ate:

Dec

embe

r 15

, 198

8 | d

oi: 1

0.10

21/b

a-19

88-0

219.

ch02

7


27. AMY ET AL. Particle Formation, Growth, and Removal 445

Electrophoretic mobilities were measured on the coagulant suspensions with a microelectrophoresis apparatus (Rank Brothers Mark II). An equilibration time of 12 h was used for DOM-coating of particles.

Results

The Grasse River contained slightly more D O M than the Edisto River (Table II). Average NPOC-based molecular weights for the two rivers were comparable, although that for the fulvic acid was much higher. Carboxylic acidities differed by a factor of 2, but are comparable to values reported in the literature. The characteristics summarized in Table II for the natural waters reflect the original water, as well as actual experimental conditions.

The M i n - U - S i l - 5 (Si0 2 ) had average dv values of 4.2 and 3.4 μηα, using the 2.5-150 and l -60 -μπ ι sensors, respectively. Addition of the lower range sensor midway through the experiments provided more information on smaller-particle removal. The dv for M i n - U - S i l - 1 5 (Si0 2) was 6.3 μπι, using the 1-60-μπι sensor.

Electrophoretic mobilities of the model particles as a function of p H are shown in Figure 1. Without the presence of humic substances and D O M , mobilities for S i 0 2 and kaolinite ranged from positive to negative, with observed p H values at the isoelectric point (pH i e p ) of 2.6 and 4.6, respectively. The DOM-coated silica exhibited a more negative mobility than the uncoated silica over the p H range of about 4.0 to 7.0, while the mobility of the DOM-coated kaolinite remained negative over the p H range of about 3.0 to 6.0. The effects of D O M on mobility were more pronounced for kaolinite than silica. Also shown in Figure 1 are mobilities for Al(OH) 3(s) formed under conditions of homogeneous nucleation, indicating that the aluminum hydroxide precipitate formed can exhibit a positive or negative charge with a p H ^ of about 6.5.

The statistical significance of observed differences was evaluated by using the Duncan's multiple range (DMR) test, which has inherent advantages over the more commonly used f-test when a number of comparisons are made within a data set (5). The D M R test also permits a statistical comparison of pairs of parameter values. The data shown in Table I were subjected to a statistical analysis by the D M R test (6-8). Comparison of initial and final water characteristics shows an increase in dv, a decrease in T P N , and a decrease in N P O C for most experiments (Table I). The following specific data comparisons are based on statistically significant differences in parameter levels at the 95% confidence level.

In Series A, fulvic acid more adversely affected particle removal at the lower p H . Higher fulvic acid concentrations generally resulted in higher T P N levels in the filtrate at p H 5.5; at a higher p H of 8.5, no statistically significant differences were found for T P N levels in experiments with higher versus lower initial fulvic acid levels. This finding may be due to the greater

Dow

nloa

ded

by U

CSF

LIB

CK

M R

SCS

MG

MT

on

Sept

embe

r 4,

201

4 | h

ttp://

pubs

.acs

.org

P

ublic

atio

n D

ate:

Dec

embe

r 15

, 198

8 | d

oi: 1

0.10

21/b

a-19

88-0

219.

ch02

7


Tabl

e I.

Sum

mar

y of

Initi

al a

nd O

pera

tiona

l Con

ditio

ns f

or S

erie

s A

, B, C

, and

D E

xper

imen

ts

Init

ial

and

Ope

rati

ng

Con

diti

ons

Init

ial

Wat

er

Fin

al W

ater

E

xp.

Par

ticl

e O

rgan

ic

Mix

ing

Tur

b.

Al

TP

Nb

d v

NP

OC

T

PN

b d v

N

PO

C

ID

Typ

e S

ourc

e C

ond.

a pH

(N

TU)

(mg/

L)

(mh

-1) (μ

τη)

(mg/

L)

(mL

-1) (μ

τη)

(mg/

L)

A4

none

fu

lvic

I

5.5

0 2.

0 1,

100

—

1.9

90

4.9

1.0

A-2

ka

olin

fu

lvic

I

5.5

10

2.0

14,0

00

4.4

1.9

170

4.6

1.1

A-3

no

ne

fulv

ic

I 8.

5 0

4.0

1,10

0 —

1.

9 90

5.

1 0.

9 A

-4

kaol

in

fulv

ic

I 8.

5 10

4.

0 14

,000

4.

4 1.

9 80

5.

4 0.

8 A

-5

none

fu

lvic

I

5.5

0 2.

0 1,

100

—

7.7

200

4.8

3.4

A-6

ka

olin

fu

lvic

I

5.5

10

2.0

14,0

00

4.4

7.7

230

4.7

3.5

A-7

no

ne

fulv

ic

I 8.

5 0

4.0

1,10

0 —

7.

7 13

0 5.

0 3.

0 A

-8

kaol

in

fulv

ic

I 8.

5 10

4.

0 14

,000

4.

4 7.

7 11

0 5.

0 3.

3

B-l

none

G

rass

e II

7.

0 0

2.0

210

5.0

930

6.5

2.8

B-2

silic

a G

rass

e II

7.

0 5

2.0

140,

000

4.2

5.0

10,0

00

8.8

3.0

B-3

silic

a G

rass

e II

7.

0 10

2.

0 29

0,00

0 4.

2 5.

0 5,

100

1.2

2.7

B-4

none

G

rass

e II

8.

5 0

2.0

210

—

5.0

70

8.5

3.7

B-5

silic

a G

rass

e II

8.

5 10

2.

0 29

0,00

0 4.

2 5.

0 5,

700

10.7

3.

4 B-

6 si

lica

Gra

sse

II

5.5

5 2.

0 14

0,00

0 4.

2 5.

0 3,

500

18.5

2.

0 B-

7 si

lica

Gra

sse

II

5.5

5 4.

0 14

0,00

0 4.

2 5.

0 6,

200

14.9

1.

7

Dow

nloa

ded

by U

CSF

LIB

CK

M R

SCS

MG

MT

on

Sept

embe

r 4,

201

4 | h

ttp://

pubs

.acs

.org

P

ublic

atio

n D

ate:

Dec

embe

r 15

, 198

8 | d

oi: 1

0.10

21/b

a-19

88-0

219.

ch02

7



l>- CD ΙΟ 05 00 *"H OS oi c4 c4 c4 c4 c4 ^

IC CD CD (M 00 Φ CD ΙΟ 1Λ © CD © l£> CD

e l co co ©J <M

8888888 ic oo Φ ιο t> CO θί »-H r-T Η CO* «—*

CO CO xf* ^ CO CO CO

§ 8 S § ί

f-H © © 1-H < © © © © © «-H <Μ (M Η Η

oo οο ^

© © © © © © © ©i ©4 ©4 ©4 ©4 ©4 5̂

© in © © © in in

© © © in i n in in i> αό od in in

0 5 c D ^ i o c o © i q c o © i c ô c o c D i n c b ' - * » ^

• © © © < ^ © ©3 ς© «

© " f -π oo" oo"

' © ' © si

φ φ φ © © ^ ° ^

CO CO CO CD CD CD

8 © © CO CO © © i -H f-H

i-H ι—I i-H ©ί Csf OÎ

© < © © © © ©

i -H »-H f-H ^

© © © © ©

m io in io in

ο ο ο ο ο ο ο •t-f +-> -W ·«-» -W .*-» (Λ C/Î C/5 !Λ ίΛ

**Q Τ 3 ^«3

ω ω ω ω ω ω ω

g S S g 8 8 3 Ο ^ ΐ Ι ο 3̂ 3 α β *(Λ *C/5 C *5Λ *C/5 *WJ

f-H © q co ^ in CD i>

I I I i I I I u u u u u u u

Q Ο 4) ο

8 s 3 3 [Λ « V) ΙΛ 1Λ W

S ce ed Q .2 .2 C C

J i3 ^ ο ~ C

^ ΙΟ CD l> 00 ι ι ι ι ι

QQQQÛ

ο Ο .a ε

S ε Ο Q

o C ς »

5 S as . . s o i

6 « M

s i ! 7 >>

S -o

ο > g

a * ! Ε ι_ ^ * ο S a c 5 0 W5 «Λ

J 8 ^

.s "3 rt

«S s 1 « .S 6 " " S i

5Λ C Î

Je

s i

American Chemical Society. Library

1155 15th St., N.W. Washington, D.C. 20036

Dow

nloa

ded

by U

CSF

LIB

CK

M R

SCS

MG

MT

on

Sept

embe

r 4,

201

4 | h

ttp://

pubs

.acs

.org

P

ublic

atio

n D

ate:

Dec

embe

r 15

, 198

8 | d

oi: 1

0.10

21/b

a-19

88-0

219.

ch02

7



Table II. Characteristics of Dissolved Organic Matter and Soil-Derived Fulvie Acid

Parameter Grasse Edisto Fulvic Acid NPOC (mg/L) 4.96 4.35 1.94 or 7.71 UV absorbance (cm1, 254 nm, pH 7) 0.186 0.164 0.085 or 0.334 Hydrophobic NPOC (% of total NPOC) 63 60 73 Carboxylic acidity (meq/g) 6.4 8.5 13.5 Average molecular weight 3290 3570 9700 Bicarbonate alkalinity (mg/L as CaC03) 31 24 50

1

0.5 i

-2 i 1 1 I 1 I 1 ! 2 3 4 5 6 7 8 9

PH Figure 1. Electrophoretic mobilities of particles with and without organic coatings: • , kaolinite; O, Al(OH)3; Δ, Min-U-Sil-15; +, FA-coated

Min-U-Sil-15; and x , FA-coated kaolinite.

stabilizing effects of fulvic acid on kaolinite clay at the lower p H , as evidenced by the divergence of the mobility curves below p H 6 (Figure 1). At a p H of 5.5, the fulvic acid would have a lower charge density and the kaolinite clay would be near its p H i e p , thus enhancing adsorption of fulvic acid. The presence or absence of kaolinite did not influence T P N levels in the filtrate at the higher p H , although T P N trends were less clear at the lower p H . No significant differences were seen in άυ for series A; finished-water particle size was controlled to a large extent by the sand filtration.

Results of series Β and C (with the objective of particle growth) indicate that dv was greater in the presence than in the absence of silica, and was also greater when more silica was present. The effects of silica on T P N levels were less clear; with Edisto River D O M , increasing the initial silica con-

Dow

nloa

ded

by U

CSF

LIB

CK

M R

SCS

MG

MT

on

Sept

embe

r 4,

201

4 | h

ttp://

pubs

.acs

.org

P

ublic

atio

n D

ate:

Dec

embe

r 15

, 198

8 | d

oi: 1

0.10

21/b

a-19

88-0

219.

ch02

7



centration resulted in no statistically significant differences in final T P N . For Grasse River D O M , the final T P N concentration was generally highest; 5 N T U of initial silica was present. Also, with Grasse River D O M , higher p H and higher aluminum dose lead to increases in T P N and lower values of dv. Conversely, for Edisto River D O M , higher p H and higher aluminum dose resulted in lower T P N levels, with no significant change in dv.

In series D , the presence of silica, with all other factors equal, resulted in lower T P N levels and higher values of dc. The presence of fulvic acid led to larger particles and a smaller T P N when silica was absent; and to smaller particles, yet no systematic change in D O M when silica was present. This result suggests that an aluminum fulvate system (without particles) provides slightly better precipitation than a homogeneous aluminum hydroxide system. Snodgrass et al. (9) also observed that the average size of particles formed in an aluminum-fulvic acid solution was larger than those observed with aluminum alone. In an aluminum fulvate-silica system, more competing (pH-dependent) reactions occur; the extent of particle aggregation is not expected to be a simple function of aluminum dose, fulvic acid concentration, and so on.

The presence of kaolinite generally inhibited N P O C removal in series A. This result is apparently due to the competing reactions between adsorption and precipitation for particle destabilization and aggregation, as opposed to aluminum fulvate precipitation. In contrast, in series Β and C the presence or absence of silica had little effect on N P O C removal, a result probably reflecting less adsorption of N O M onto silica than onto kaolinite. The mobility data reflect a greater effect of N O M on kaolinite than on silica (Figure 1). This effect is consistent with observations that N O M adsorbs to aluminum (hydr)oxide surfaces (a portion of kaolinite) to a greater extent than to silica (10, 11). For series Β and C , better N P O C removals were observed at p H 5.5 than at 7.0; at p H 5.5, one would expect a lesser charge density associated with carboxylic groups of the N O M , thus affecting charge neutralization.

A further analysis of data was conducted to delineate a possible stoi-chiometry between N P O C removal and aluminum added. For the seven experiments conducted with Grasse River D O M , the ratio o fNPOC removed to aluminum added was 0.99 ± 0.28 mg/mg, while the corresponding ratio for the seven experiments conducted with the Edisto River D O M was 0.84 ± 0.16 mg/mg.

Discussion

Particle growth occurred in most of the series Β and C experiments, probably because mixing conditions were conducive to particle growth. Although different particle-counter sensors were used in the series Β and C experiments, qualitative comparisons between the two are still possible. Particles

Dow

nloa

ded

by U

CSF

LIB

CK

M R

SCS

MG

MT

on

Sept

embe

r 4,

201

4 | h

ttp://

pubs

.acs

.org

P

ublic

atio

n D

ate:

Dec

embe

r 15

, 198

8 | d

oi: 1

0.10

21/b

a-19

88-0

219.

ch02

7



formed with Edisto River D O M in the presence of silica were generally larger than those in Grasse River D O M ; T P N levels were correspondingly lower for Edisto River versus Grasse River. The Edisto River D O M was present at a lower concentration (as NPOC) than Grasse River D O M , and it had a slightly greater carboxylic acidity and molecular weight.

Significant increases in dv (particle growth) did not occur in either series A or D . Series D experiments used the largest particle and had a high-velocity gradient, approaching rapid-mix conditions. Further, substantial numbers of small aluminum hydroxide particles were produced in these experiments. Recall that in series A, the objective was particle removal and the flocculated water was filtered prior to analysis. Although T P N levels were significantly reduced, final dO values (3.6-4.7 μπι) were comparable to those in initial waters (d0 = 4.3 μπι). This result is not surprising, as clean sand filters are most effective in removing particles larger than a few micrometers (12). In all experiments, one would expect the initial formation of submicrometer precipitates of aluminum hydroxide, aluminum fulvate, or aluminum-particle aggregates, with the subsequent growth of these precipitates until eventually they pass through the detection limit (1.0 or 2.5 μπι) of the optical particle counter. In Series D experiments, it is likely that an observation of particle growth of the original silica particles was outweighed by the appearance of small aluminum hydroxide floe.

Results show qualitative consistency with the notion of differing removals or growth, due to competing dissolved-aluminum complexation and precipitation reactions that differ in rate and extent as a function of p H . In the absence of foreign particles or humic substances, homogeneous nuclea-tion and precipitation of aluminum hydroxide occur. In the presence of foreign particles, there is heterogeneous nucleation of aluminum hydroxide onto the initially present "seed" particles. Either homogeneous or heterogeneous nucleation can be the first step in particle formation and growth. When aluminum hydrolysis species predominate, particle destabilization occurs via adsorption of the hydrolysis species, creating conditions suitable for subsequent particle aggregation. Thus, two principal mechanisms involved in particle removal by chemical coagulation are sweep coagulation, in which particles are enmeshed in an aluminum hydroxide precipitate, and charge neutralization, whereby particles are destabilized by adsorbed, positively charged hydrolysis species.

When humic substances are present at lower p H levels (e.g., 5.5), cationic aluminum species react with negatively charged humic and fulvic acids to form aluminum-organic complexes and precipitates (13, 14). The observed greater N P O C removal in the current work at lower p H is consistent with adsorption versus coprecipitation; the results do not permit an unambiguous conclusion, however. Although these aluminum fulvate—humate precipitates are too small to settle effectively, they can be removed by filtration. At higher p H levels (e.g., 7.5), humic-substance ad-

Dow

nloa

ded

by U

CSF

LIB

CK

M R

SCS

MG

MT

on

Sept

embe

r 4,

201

4 | h

ttp://

pubs

.acs

.org

P

ublic

atio

n D

ate:

Dec

embe

r 15

, 198

8 | d

oi: 1

0.10

21/b

a-19

88-0

219.

ch02

7



sorption onto an aluminum hydroxide precipitate is more important. Humic substances also adsorb onto materials such as clays and metal oxides, and thereby change their surface identity (10, 11, 15) and stabilize the particles.

Humic substances can affect aluminum-particle interactions, and particles can affect aluminum-humic interactions. Because of the higher charge density associated with the fulvic acid, as opposed to the clay found in a typical surface water, the coagulant dosage required for charge neutralization will normally be controlled by the fulvic acid (16).

Summary and Conclusions

On the basis of T P N levels, the presence of fulvic acid resulted in a poorer quality of sand filtrate, and the most adverse effect was observed at lower p H conditions. In the presence of D O M , greater amounts of initial seed particles resulted in greater values of άυ in flocculated suspensions. The presence of fulvic acid and the absence of initial particles led to an increase in dv in flocculated suspensions, although the presence of both fulvic acid and initial particles led to a decrease in dv.

Particle formation, growth, and removal were significantly affected by initial particle and organic matter or humic substance concentrations, as well as p H and aluminum dose. The effect of particles on D O M removal differs for kaolinite versus silica.

Acknowledgment Partial financial support for this research was provided by the U.S. Environmental Protection Agency (EPA) under Grant R-812325-01-0. The contents do not necessarily reflect the views and policies of the EPA.

References 1. Collins, M . ; Amy, G.; Bryant, C. J. Environ. Eng. Div. (Am. Soc. Civ. Eng.)

1987, 112(2), 330-344. 2. Thurman, E.; Malcolm, R. Environ. Sci. Technol. 1981, 15, 463-466. 3. Amy, G.; Collins, M.; Kuo, C.; King, P. J. Am. Water Works Assoc. 1987, 79,

43-49. 4. Collins, M . ; Amy, G.; Steelink, C. Environ. Sci. Technol. 1986, 20, 1028-1032. 5. Dowdy, S.; Wearden, S. Statistics for Research; John Wiley: New York, 1983. 6. Collins, M. Ph.D. Dissertation, University of Arizona, 1985. 7. Kuo, C. Ph.D. Dissertation, University of Arizona, 1986. 8. Chowdhury, Z. Ph.D. Dissertation, University of Arizona, 1987. 9. Snodgrass, W.; Clark, M.; O'Melia, C. Water Res. 1984, 18, 479-488.

10. Davis, J. Geochim. Cosmochim. Acta 1982, 46, 2381-2391. 11. Davis, J.; Gloor, R. Environ. Sci. Technol. 1981, 15, 1223-1229. 12. O'Melia, C. J. Environ. Eng. Div. (Am. Soc. Civ. Eng.) 1985, 111, 874-891.

Dow

nloa

ded

by U

CSF

LIB

CK

M R

SCS

MG

MT

on

Sept

embe

r 4,

201

4 | h

ttp://

pubs

.acs

.org

P

ublic

atio

n D

ate:

Dec

embe

r 15

, 198

8 | d

oi: 1

0.10

21/b

a-19

88-0

219.

ch02

7



13. Wiesner, M . ; Lahoussine-Turcaud, V.; Fiessinger, F. Proc. AWWA Annu. Conf. American Water Works Association Annual Conference, American Water Works Association: Denver, CO, 1986; p 1685.

14. Hundt, T. Proc. AWWA Annu. Conf. American Water Works Association Annual Conference, American Water Works Association: Denver, CO, 1986; p 1199.

15. Gibbs, R. Environ. Sci. Technol. 1983, 17, 237-240. 16. Dempsey, B.; Ganho, R.; O'Melia, C. J. Am. Water Works Assoc. 1984, 76,

141-150.

RECEIVED for review July 24, 1987. ACCEPTED for publication December 7, 1987.

Dow

nloa

ded

by U

CSF

LIB

CK

M R

SCS

MG

MT

on

Sept

embe

r 4,

201

4 | h

ttp://

pubs

.acs

.org

P

ublic

atio

n D

ate:

Dec

embe

r 15

, 198

8 | d

oi: 1

0.10

21/b

a-19

88-0

219.

ch02

7


[advances in chemistry] aquatic humic substances volume 219 (influence on fate and treatment of...

Documents