Document text
AD-A038 148 AMICON CORP LEXIN6T0N MASS F/6 13/2
DEVELOPMENT OF TECHNIQUES FOR DETECTION OF LOW MOLECULAR WEI6HT— ETC(U)
AU6 76 T W MIX* H SCHAREN DADA17-72-C-2169
END
DATE
FILMED
4-77
DEVELOPMENT OF TECHNIQUES FOR DETECTION OF LOW MOLECULAR WEIGHT
CONTAMINANTS IN PRODUCT WATER FROM WATER PURIFICATION
OR WATER RE-USE SYSTEMS
FINAL COMPREHENSIVE REPORT
Thomas W. Mix
Hans Scharen
August 1976
U. S. ARMY MEDICAL RESEARCH AND DEVELOPMENT COMMAND
Washington, D. C. 20315
Contract No. DADA17-72-C-2169
Amicon Corporation
25 Hartwell Avenue
Lexington, Massachusetts 02173
Approved for public release; distribution unlimited
The findings in this report are not to be construed as an
official Department of the Army position unless so
designated by other authorized documents
'■< r
MMte SMttso I If
Hrt S«CltM □
O
-2-
ssaww/gMiwm coo ei
" ML mt/m srlcur
m
FINAL COMPREHENSIVE REPORT
DEVELOPMENT OF TECHNIQUES FOR DETECTION OF
LOW MOLECULAR WEIGHT CONTAMINANTS IN PRODUCT
WATER FROM WATER PURIFICATION OR WATER RE-USE SYSTEMS
Thomas W. Mix
Hans Scharen
Supported by
U. S. ARMY MEDICAL RESEARCH AND DEVELOPMENT COMMAND
Washington, D. C. 20315
Contract No. DADA17-72-C-2169
Amicon Corporation
25 Hartwell Avenue
Lexington, Massachusetts 02173
D D C
APB » 1977
"distribution STATEMENT a
Approved fox public wleosaj
Distribution Unlimited
£
SECURITY CLASSIFICATION OF THIS PAGE (When Data Bntarad) _
~ ~ REPORT DOCUMENTATION PAGE befor^comple^gtorm
1. REPORT NUMBER ‘ |2. GOVT ACCESSION NO. 3. RKGJPIENT’S CATALOG NUMBER
REPORT NUMBER
DEVELOPMENT OF TECHNIQUES FOR DETECTION
OF LOW MOLECULAR WEIGHT CONTAMINANTS IN | 1 £°mPrehenSi ve FeP
PRODUCT WATER FROM WATER PURIFICATION Or\ V performing org. report number —
WATER RE-USE SYSTEMS * \
Thomas W.7Mix
Hans/Scharen
61
CONTRACT OR GRANT NUMBERf*)
DADA17-72-C-2169 6
9. PERFORMING ORGANIZATION NAME AND ADDRESS
10. program ELEMENT. PROJECT, TASK
AREA ft WORK UNIT NUMBERS
Amicon Corporation
25 Hartwell Avenue
Lexington, Massachusetts 02173 /"7
11. CONTROLLING OFFICE NAME AND ADDRESS
U. S. Army Medical Research and Develop^*] Augwwfc 15976
ment Command, Washington, D. C. 20315 1 is. number oTpages
88
14. MONITORING AGENCY NAME ft ADDRESS^// dlftarant from Controlling Ottlca) 15. SECURITY CLASS, (of thla raport)
\Sa. DECLASSIFIC ATI ON/ DOWNGRADING
SCHEDULE
16. DISTRIBUTION STATEMENT (ol thlt Report;
Approved for public release; distribution unlimited
The findings in this report are not to be construed as an
official Department of the Army position unless so designated by
other authorized documents.
17. DISTRIBUTION STATEMENT ( of tha aba tract antarad In Block 20, If dlffarant from Raport)
IS. SUPPLEMENTARY NOTES
19. KEY WOROS ( Contlnua on rararaa alda If nacaaaary and Idantlfy by block numbar)
Analysis
Ozone
TOC
Total Organic Carbon
ST
Reuse
Instrumentation
Wastewater
Permanganate
Carbon Adsorption
Colorimetry
ABSTRACT fCanftau* an rararaa aid* ff naeaaaary mod. Idantlfy by block numbar)
TV permanganate colorimeter instrument to monitor the quality of MUST product
water has been developed and successfully tested at the breadboard level.
During the program it was demonstrated that (1) aldehydes are the last of the
constituents of MUST RO permeate to be ozonated , (2) residual MUST aldehydes
(principally acetaldehyde) rapidly reduce permanganate solutions, and (3) the
degree of permanganate reduction may be easily quantitated colorimetrically
and is directly proportional to residual MUST water contamination. The colori-
meter and required hardware is judged low-cost, rugged, simple, and reliable.
WWW l"- ».
- -'" -n - '
-4-
SUMMARY
For three years Amicon Corporation has been under
contract to the U. S. Army Medical Research and
Development Command for the Development of Techniques
for Detection of Low Molecular Weight Contaminants in
Product Water from Water Purification or Water Reuse
Systems .
Principal emphasis during the latter portion of the
program has centered on the development of a permanganate
colorimeter instrument to monitor the quality of MUST
product water which has been processed through both
reverse osmosis and ozonation. The selection of the
permanganate colorimeter was based on a number of
factors :
1. The rate limiting step in the ozonation of the
MUST RO permeate appears to be the ozonation
of aldehydes (acetaldehyde, in particular) .
These aldehydes may therefore be expected to
accumulate and to represent an appreciable
fraction of the TOC in the product water;
2. Aldehydes are rapidly oxidized by basic
permanganate solutions (as are alcohols and
many other organic compounds) ;
3. Permanganate and manganate ions have high
extinction coefficients and are rapidly
detectable in low concentration spectro-
photometrically ; and
4. A colorimeter is compatible with the require-
ments for low cost, ruggedness, simplicity,
reliability, and field operation.
In addition, the technique offers the fail-safe backup
of a simple color comparator.
Initial testing of a breadboard colorimeter with a MUST
pilot unit tends to confirm the suitability of the
technique for MUST product quality monitoring but is
insufficient in scope to establish it with the requisite
certainty. Additional work with MUST pilot plant samples
and with known or suspected contaminants is therefore
recommended.
Additional work done under the program deals with the
following subjects;
1. A preliminary investigation of a spectrophoto-
meter technique to monitor the rate of decay
of ozone concentration as a measure of product
water purity.
J
l
-5-
2. Carbon adsorption experiments related to
monitoring of carbon column adsorption when
this was a processing step under considera-
tion for the MUST; and
3. A preliminary study of the monitoring of the
combination of reverse osmosis and ozonation
and/or carbon adsorption as tertiary treatment
steps for the output from a municipal treatment
plant.
-6-
TABLE OF CONTENTS
I . SUMMARY
II. OZONATION STUDIES
A. Ozonation Equipment
B. Experimental Procedures
C. Interpretation of Data
III. MONITORING TECHNIQUES
A. Development of Acetaldehyde Assay Using
Potassium Permanganate in Basic Solution
B. Preliminary Design Work on a Continuous
Flow Monitoring Technique
C. Abcor Pilot Run
D. Rate of Decay of Ozone
IV. CARBON ADSORPTION EXPERIMENTS
A. Experimental Procedures
B. Discussion of Results
V. SECONDARY EFFLUENT WASTE WATER STUDIES
A. Pilot Plant - Construction and Operation
B. Experimental Procedure
Page
4
10
10
10
13
22
22
29
41
66
73
73
75
81
81
81
VI.
REFERENCES
86
I
'» 1
I
: l
f ■ *
TABLE
TABLE
TABLE
TABLE
TABLE
TABLE
TABLE
TABLE
TABLE
TABLE
TABLE
-7-
LIST OF TABLES
Page
0 - 20 HOUR A OD OF KMn04 DECOMPOSITION AS
A FUNCTION OF {OH-} AT WAVELENGTHS OF
426 and 522 run
33
I
- RECORDER OUTPUT READINGS
52
II
- COMPARISON OF TOCSIN DATA
55
III
- CALCULATED A OD FOR 1 PPM
ACETALDEHYDE
61
IV
- INSTRUMENT RESPONSE AS A FUNCTION OF
THE LEVEL OF BASE CONCENTRATION AND
THE PERMANGANATE ACETALDEHYDE RATIO
63
V
- ABSORBANCE RANGE WITH A 1 CM PATH
LENGTH AND 4 MOLS Mn04~ PER MOL
CH3COH
65
VI
- PART A: DECOMPOSITION OF
DISTILLED WATER
OZONE IN
68
PART B: DECOMPOSITION OF
DISTILLED WATER
OZONE IN
68
VII
DECOMPOSITION OF OZONE IN
EFFLUENT RO PERMFATE
SECONDARY
74
VIII
- DECOMPOSITION OF OZONE IN
EFFLUENT RO PERMEATE
SECONDARY
77
IX
- BATCH TYPE CARBON ADSORPTION EXPERI-
MENTS
78
X - OZONATION OF RO PERMEATE OF SECONDARY
EFFLUENT FROM THE BROCKTON,
MASSACHUSETTS, MUNICIPAL WASTEWATER
^'EATMENT FACILITY
85
-8-
LIST OF FIGURES
Page
FIGURE 1 - OZONATION TEST SET-UP 11
FIGURE 2 - DETERMINATION OF QUINONE 12
FIGURE 3 - OZONATION STUDIES 14
FIGURE 4 - OZONATION OF FORMALDEHYDE @ 60°C 15
FIGURE 5 - TOC REDUCTION OF MUST PERMEATE (X-RAY
WASTE) AT pH 6.5 AND A 100 PPM
ACETALDEHYDE SOLUTION, A 100 PPM
ETHANOL SOLUTION, AND A 100 PPM
METHANOL SOLUTION 17
FIGURE 6 - DETERMINATION OF RESIDUAL ETHANOL WITH
DICHROMATE METHOD AFTER OZONATION 18
FIGURE 7 - OZONATION OF ACETALDEHYDE AT 60°C 20
FIGURE 8 - OZONATION OF ETHANOL 21
FIGURE 9 - ABSORPTION SPECTRUM OF KMn04 IN BASIC
SOLUTION REACTING WITH VARIOUS CONCEN-
TRATIONS OF CH3CHO 23
FIGURE 10 - REACTION BETWEEN KMn04 AND CH3COH IN
BASIC SOLUTION 26
FIGURE 11 - REACTION BETWEEN KMn04 AND CH3COH IN
BASIC SOLUTION 27
FIGURE 12 - STANDARD CURVE OF KMn04 VERSUS CH3CHO
IN BASIC SOLUTION 28
FIGURE 13 - KMn04 REACTION WITH CH3CHO IN BASIC SOLU-
TION FOR 10 MINUTES @ 21°C AS A FUNCTION
OF WAVELENGTH AND {OH~} CONCENTRATION 30
FIGURE 14 - KMn04 REACTION WITH CH3CHO IN BASIC SOLU-
TION FOR 10 MINUTES AT 21°C AS A FUNCTION
OF WAVELENGTH AND {OH~} CONCENTRATION 31
FIGURE 15 - KMn04 DECOMPOSITION IN BASIC SOLUTION AT
ROOM TEMPERATURE 32
FIGURE 16 - SCHEMATIC OF PERMANGANATE COLORIMETER
MONITOR 34
-9-
LIST OF FIGURES (continued)
Page
FIGURE 17 - MONITOR SCHEMATIC 36
FIGURE 18 - DUAL WAVELENGTH MODE OF OPERATION 37
FIGURE 19 - RECORDER TRACINGS OF ABCOR RUNS 42
FIGURE 20 - TCS DUAL WAVELENGTH SPECTROGRAM OF
{KMn04} 72 mg/1 AT ROOM TEMPERATURE
IN BASIC SOLUTION 62
FIGURE 21 - OZONE AUTODECOMPOSITION IN DISTILLED
WATER AT ROOM TEMPERATURE (^25 C) 69
FIGURE 22 - DECOMPOSITION OF OZONE IN SECONDARY
EFFLUENT 70
FIGURE 23 - RATE OF DECAY OF OZONE IN DISTILLED
WATER AND SECONDARY EFFLUENT RO
PERMEATE AT ROOM TEMPERATURE 71
FIGURE 24 - COLUMN ADSORPTION EXPERIMENTS 76
FIGURE 25 - FREUNDLICH ADSORPTION ISOTHERM FOR
ETHANOL 7 9
FIGURE
26
- FREUNDLICH ADSORPTION ISOTHERM FOR
ACETALDEHYDE
80
FIGURE
27
- PILOT PLANT FOR
REVERSE OSMOSIS
SECONDARY
TREATMENT
EFFLUENT
82
FIGURE
28
- RO TREATMENT OF
WASTE WATER
SECONDARY
EFFT.UENT
84
I
5
t
(
i ]
j j
i 1
I
me <
-10-
A. Ozonation Equipment
The apparatus used for the ozonation studies is
shown in Figure 1 and consists of a four-necked
reaction flask containing a high shear stirrer, a
gas dispersion tube, a sample withdrawal tube, and
a thermometer. An automatically-controlled
temperature bath was used to keep the temperature
within ±1°C of the reaction temperature. The ozone
was generated by a Welsbach Ozonator (Model T-816)
with an ozone production rate of 1.7% by weight or
16 g/hour using oxygen. In order to make the system
airtight, all the joints were fitted with Teflon
sleeves. A special shaft of stainless steel with a
Teflon bearing was machined for the high shear
dispersator disc. Stainless steel, glass and Teflon
were used as piping materia] since they are all
resistant to attack by ozone.
B. Experimental Procedures
Ethanol Ozonation
A 100 ppm solution of ethanol was added to the
reaction flask and ozonated for three hours. Samples
for TOC analysis and ethanol determination were with-
drawn periodically every 30 minutes; the last sample
being taken at 180 minutes. All TOC analysis was
performed at Fram Corporation, Pawtucket, Rhode
Island. For TOC analysis, a Beckman TOC Analyzer
was used. Ethanol concentrations were determined
by oxidizing with acidified dichromate and measuring
the disappearance of the characteristic dichromate
color at 549 nm with a spectrophotometer. The O.D.
values obtained were then compared to a previously
prepared standard reference curve from which the
actual concentration in mg/liter was read^) .
Ozonations of acetic acid, acetaldehyde, quinone,
oxalic acid, formaldehyde, methanol, glycine, and
MUST RO permeate were performed in a similar manner.
A 100 ppm solution of each compound was prepared.
They were ozonated for three hours each and samples
were withdrawn for TOC analysis. A portion of the
acetic acid sample was used for titration with sodium
hydroxide and a portion of the quinone sample was
used for the spectrophotometric determination of
quinone (Figure 2) . The ozonation conditions
typically used were: the ozone concentration, 38
mg/liter; gas flow, 0.5 liters/minute; and the
reaction temperature, 60°C.
i
i
TANK
DETERMINANT
Formaldehyde was selected because it is likely to be
formed in the ozonation of methanol.
Acetic acid was also selected because Professor
Wiberg of Yale University (4) pointed out that it is
frequently part of the oxidation pathway for ali-
phatic compounds (and aromatic compounds where ring
opening occurs) and is likely to be the most oxidation
resistant step in the pathway. Similarly, quinone was
selected because it is frequently the rate limiting
step in the oxidation of many aromatic compounds (4 T .
Glycine was selected because amino acids can be detected
in very small concentrations through use of fluorescent
dye indicators, such as Roche's Fluram(5). The rapid
oxidation of both glycine and quinone makes them un-
suitable as tracers. The normally oxidation-resistant j
quinone is probably susceptible to attack by ozone
at its double bonds. Acetic acid is also not useful
as a tracer but for a different reason: it is
essentially impervious to attack by ozone. Since
MUST reverse osmosis permeate is susceptible to
oxidation by ozone and does not show a sufficient
lowering of pH during ozonation to indicate appreciable
accumulation of acid, acetic acid is not formed to
any significant extent during MUST permeate ozonation
at neutral pHs. Because of acetic acid's inertness,
it can be tolerated by the body in fairly high con-
centration, making its removal from the MUST water
to be recycled of limited importance.
Oxalic acid, which is a likely intermediate in many
of the oxidation sequences, is relatively rapidly
ozonated, which is also consistent with the slight
change in pH as ozonation proceeds. Methanol and
formaldehyde also disappear relatively rapidly under
ozonation conditions. The broad scatter in the
formaldehyde data (Figure 4) is not understood but
may perhaps be attributable to some polymeric
material or particulate impurities in the formaldehyde
solution used.
OZONATION STUDIES
FIGURE 3
Acetaldehyde
Glvcine
Quinone
Time
— j— -
— |—
i
■ 1
— — |i —
— — —
r ♦- —
—
I
I j
I
L-
I
1
-16-
Both ethanol and acetaldehyde, however, have TOC
disappearance characteristics quite similar to
those of the MUST RO permeate (Figure 5) . The
similarity among the three TOC disappearance
curves suggests that acetaldehyde disappearance is
the rate limiting step for all three. This is
confirmed by studies of the rate of ethanol
disappearance using potassium dichromate which is
specific in the time allotted for reaction^. The
rate of disappearance of ethyl alcohol during
ozonation, as measured by dichromate disappearance,
shown in Figure 6, is much more rapid than the rate
of intermediary oxidation product of ethanol which
is relatively resistant to ozonation. There are
two possible pathways for ethanol oxidation: one is
through acetaldehyde to oxalic acid, which is
favored by basic conditions, and the other to acetic
acid, which is favored by acidic conditions. The
measured pH of the ethanol solution before and after
ozonation and the known dissociation constants for
acetic and oxalic acids indicate that after ozonation
these acids, if present, should be completely ionized
Their concentration may, therefore, be estimated from
the pH since electrical neutrality requires that the
anion concentration equal the difference between the
hydrogen ion concentration and the hydroxyl ion
concentration. This difference is small enough to
indicate that there are only negligible levels of
either acetic acid or oxalic acid present. Acetal-
dehyde does, therefore, appear to be the rate
limiting step in the ozonation of ethyl alcohol to
C02 and may, therefore, be an appropriate monitor
through which the effectiveness of the ozonation step
may be followed provided a sensitive enough method
of detection can be developed.
The rate of volatilization (gas stripping) of
acetaldehyde may be estimated from its known activity
coefficient at infinite dilution C7) and vapor
pressure^7) using the equation:
In (x^/xt)
GgP°
Vfl
(1)
where x = concentration of acetaldehyde in water
at time, t
x^ = initial acetaldehyde concentration
G = oxygen supply rate, g moles/min
V = liquid reservoir holdup, g moles
FIGURE 5
TOC REDUCTION OF MUST PERMEATE (X-RAY WASTE)
AT pH 6.5 AND A 100 PPM ACETALDEHYDE SOLUTION
A 100 PPM ETHANOL SOLUTION
AND A 100 PPM METHANOL SOLUTION
■k.
.*
I
41
c
I
« = activity coefficient for acetaldehyde
in water at infinite dilution and 60°C
P° = vapor pressure of acetaldehyde at 60°C,
atm
ir = atmospheric pressure, atm
t = time, minutes
For the acetaldehyde ozonation, G * 0.021 g/moles/
min., V = 55.5 g moles, = 2 4.75 P^_ = 4.05, and In
(x^/x ) = . 0073t . it
As may be seen from Figure 7, the disappearance of
acetaldehyde may be accounted for entirely by
stripping.
For ethyl alcohol, however, the volatilization (gas
stripping) rate is given by the equation:
In x./x. = ,0076t
l t
As may be seen in Figure 8, gas stripping makes only
a very minor contribution to the disappearance of
ethanol.
For formaldehyde, the very low activity coefficient
in aqueous solution^®' more than compensates for
the high vapor pressure of the pure material, so
that gas stripping contributes negligibly to its
loss during ozonation.
Potassium permanganate is a strong oxidizing agent
affecting many organic compounds in acid, neutral,
or basic solution. During preliminary investigations,
it was found that raw and treated MUST waste water
when added to a roughly 100 ppm basic potassium
permanganate solution effected a noticeable change
in the deeply purple color. Aldehydes are known to
be oxidized in basic potassium permanganate and when
roughly stoichiometric amounts of potassium
permanganate and acetaldehyde were allowed to react
in basic solution at room temperature, a noticeable
color change occurred in less than ten minutes.
The results of these preliminary experiments were
encouraging enough to warrant investigation into a
colorimeter monitor for ozonated MUST product water
based on permanganate oxidation of residual con-
taminants .
liiqiuid Volume = 1 liter
■genrFlew Rate - 0.5 1/ittin
loncebttation in Oxygen - 38 mg/1
1— + ! — ►— - * - ♦ -
“ .0077 t
disappearance predicted by stripping)
Ozone 1C
1 •
1 ' J
i l
• I
i
1-4
i i
!
i " i
} 1
1 :
l
1 1
i
. 1
i
. L ■ .
t~4r j
j
1
:
i : .
I i
1 _u.
! . !
1- 1 J
( 1
! ..... -
f
1 1
ppm Ethanol
.UtAtlon By<$a* Stripping)
OZOHATIOR OF ETHANOL
(Ethanol concentration determined
spectrophotomeitricaily after reaction
with di chromate)
-22-
III. MONITORING TECHNIQUES
A. Development of Acetaldehyde Assay Using Potassium
Permanganate in Basic Solution
A stock solution of 10,000 ppm acetaldehyde in dis-
tilled water was prepared. Appropriate dilutions
were made and samples of 5, 10, and 20 ppm were
prepared. A slight excess of basic potassium per-
manganate was added to the reaction flasks which
were allowed to stand at room temperature for 20
minutes. The total reaction volume of each flask
was 5 ml and the pH was between 12 and 12.5. A
blank was prepared in the same way using distilled
water. The spectral absorption curve of each solu-
tion was then determined using a Beckman 505 Scanning
Spectrophotometer. For all measurements, the
reference cell contained a solution of potassium
hydroxide with a pH of approximately 12. The solu-
tions were scanned from 200 nm to 600 nm. Figure
9 shows the absorption spectrum of the various
samples. From the spectra, it is obvious that
suitable wavelengths for measurements of permanganate
and manganate concentrations in a solution are 522
and 426 nm, consistent with literature findings (6).
The absorption at the former wavelength is due mainly
to permanganate and at the latter mainly due to
manganate.
If permanganate ion is reduced no further than
manganate ion, the ratio of the decrease in optical
density at 522 nm to the increase in optical density
at 426 nm may be shown to be:
(A OD)
(A OD)
522
426
11.54
From Figure 9 it is clear that the A OD ratio experi-
mentally observed is less. For 5 ppm acetaldehyde,
the experimental ratio is 1.28; for 10 ppm, it is
1.23; and for 20 ppm, it is 1.12. Since acetaldehyde
and its oxidation products should have negligible
absorption at 522 nm and 426 nm, and since reduc-
tion of permanganate to manganese dioxide is likely
to occur, the observed discrepancy is doubtless
attributable to the production of colloidal manganese
dioxide. Typically the size of colloidal manganese
dioxide particles produced will be of the order of
the wavelength of light'^ . The scattering effect
is likely to be greater at the shorter wavelength
-24-
so that a relatively small ratio of manganese
dioxide to manganate ion could account for the
discrepancy in A OD ratios. The formation of
manganese dioxide can be suppressed by increasing
the concentration of base and maintaining a
sufficient excess of permanganate ion (see later
discussion and Figures 13, 14, and 15).
Calibration curves, which show the disappearance
of permanganate at 522 nm or the appearance of
manganate at 426 nm for a given temperature as a
function of ppm of acetaldehyde, should be adequate
for estimation of acetaldehyde concentration in
solution provided adequate time is allowed for com-
pletion of the reaction. Thus, the next step was
to investigate the reaction at 522 nm and 426 nm
and to determine optimum time and temperature for
the reaction.
Stoichiometric requirements for permanganate oxida-
tion of aldehydes in alkaline solution are not clear
because oxidation to the corresponding carboxylic
acid competes with carbon-carbon bond scission^).
Oxidation to the corresponding carboxylic acid in
alkaline solution requires 2 moles of permanganate
per mole of aldehyde, as indicated in Equation 1,
while carbon-carbon bond cleavage can consume many
more moles of permanganate per mole of aldehyde (6),
CH3COH + 2 Mn04 + 2 OH = CH3COOH + 2 Mn04 + H20
The U. S. Army Medical Research and Development
Command has established as an interim standard, a
TOC level of 5 ppm as the maximum acceptable for
treated MUST waste water reuse. The acetaldehyde
concentration in the treated MUST waste water must,
therefore, be less than 10 ppm since this corresponds
to a TOC of 5 ppm, presuming no other compounds
present. As treatment processes are improved and
optimized, the acceptable TOC level will eventually
be reduced further. The range of the calibration
curve for an acetaldehyde assay was therefore
selected as from 1-10 ppm. Based upon reaction (1)
above, the stoichiometric amount of KMn04 to react
with 10 ppm of CH3COH is 72 ppm. A solution A was
prepared in the following way: 1 ml of (KMn04)
solution containing 7200 mg/liter plus 4 ml distilled
water plus 5 ml 1 MKOH solution were added to a
reagent flask. The final KMn04 concentration of
this solution is 720 mg/1. Since KMn04 undergoes
slight decomposition in basic solution, reagent A
has to be prepared freshly before each test. To
r
r
/■
1
-25-
t j
r
i
i
;
■
i
i
»
-
\
3 ml of distilled water (for blank) 0.3 ml of solu-
tion A was added. This gave a final KMn04 reaction
concentration of 65.4 mg/liter and a pH of 12.2.
The reaction was followed spectrophotometrically at
522 nm and OD readings were recorded at two-minute
intervals. This procedure was repeated with 1, 5,
and 10 ppm acetaldehyde solutions replacing the
distilled water. As mentioned above, at 522 nm
the disappearance of permanganate is measured. The
entire procedure was then repeated but this time
the measuring wavelength of the spectrophotometer
was set at 426 nm and the appearance of manganate
was determined. The results of these experiments
are shown in Figures 10 and 11. The reactions at
both wavelengths were followed over a period of one
hour. In Figures 10 and 11, the results were
plotted up to 20 minutes since the curves were
fairly straight from that point onwards after the
blank values were subtracted. It is also evident
from the two figures that the curves reach a near
maximum level at different times; when observed at
522 nm, it is about six minutes and at 426 nm, it
is about ten minutes. The longer time to reach
equilibrium at 426 nm is probably due to the kinetics
of formation and growth of colloidal manganese
dioxide particles. These particles are likely to
scatter light more at 426 nm than at 522 nm. As
short as possible a reaction time is essential so
that variations of acetaldehyde levels in the processed
effluent stream can be detected quickly.
Advantages of monitoring the 522 nm wavelength are
the shorter time to a constant reading and the
higher extinction coefficient for the permanganate
ion. Advantages of monitoring the 426 nm wavelength
are the possibly higher sensitivity of the colori-
meter in the low OD range and the additional A O.D.
resulting from manganese dioxide production, provided
the effect is reproducible.
Standard curves were obtained at both 522 nm and at
426 nm, with the 522 nm readings taken after six
minutes at room temperature for reaction, and those
at 426 nm taken after 20 minutes. These curves are
presented in Figure 12. Additional data were
obtained using a Cary 14 Recording Spectrophotometer
to study the effect of base concentration on the OD
readings and on permanganate decomposition. An
increase in base level should make manganese dioxide
less likely to form but should accelerate the slow
minut
I
-29-
autodecomposition of permanganate. Results are
presented in Figures 13, 14, and 15 and Table 0.
A higher base concentration does result in slightly
more reaction and in a higher A 00522/^ OD426 rati°
and is, therefore, helpful in suppressing manganese
dioxide formation. It has negligible effect on
permanganate decomposition kinetics for the time
scale of interest. A higher base concentration is
therefore desirable.
From Figure 14 the moles of permanganate which react
with a mole of acetaldehyde may be estimated to be
at least 2.8. The true number is somewhat higher
because the manganese dioxide scattering at 522 nm
raises the OD and therefore makes the permanganate
disappearance appear to be smaller than it is.
B. Preliminary Design Work on a Continuous Flow
Monitoring Technique
Permanganate oxidation of acetaldehyde appears to
meet the requirements for a continuous monitoring
technique. If necessary, ethanol or acetaldehyde
may be added continuously into the process stream
upstream of the ozonation reactor in an appropriate
concentration and acetaldehyde monitored continuously
in the effluent stream. Permanganate will also
oxidize ethanol, although more slowly; if ethanol is
added as a tracer, the amount of tracer utilized and
the permanganate oxidation conditions must be
selected so as to guarantee tripping of the system
alarm should the ozonation reactor be malfunctioning
so as not even to oxidize ethanol to acetaldehyde.
One schematic for a permanganate colorimeter monitor
is shown in Figure 16. KMn04 and KOH are stored in
separate tanks. Originally distilled water and P.O
permeate are also stored in tanks but once the
system is on line, RO permeate is fed to the monitor
via a direct line from the processed effluent stream
and distilled water is produced by an appropriate
unit on a continuous basis. The reagents are fed
into two Vortex mixers with volumes of about one ml
and tangental inlets into the mixing chambers.
From the Vortex mixers the reaction mixture enters
static mixers and coils of appropriate holdup to
give the desired residence time for reaction. The
length of the coils is chosen to allow adequate
reaction time; from Figure 10, a residence time of
about six minutes is desirable. From the coils, the
reaction mixture enters flow-through cells of a
double beam colorimeter where the difference in
optical density between the two streams is determined.
The reaction mixture of the sample stream consists
FIGURE 13: KMn04 REACTION WITH CH3CHO IN BASIC SOLUTION FOR
10 MINUTES @ 21°C AS A FUNCTION OF WAVELENGTH AND
{OH1 CONCENTRATION
{ KM11O4 } = 36 mg/1
{OH'} = 0.02 5 M
pH = 12.24
^ • — > 5 2 2 nra
— >4 26 nm
L 2
Acetaldehyde
3 4 E
Concentration, mg/1
t ■ ..... ■ i
)
{KMn04} = 36
mg/1
{ OH*} = 0.1 M
pH = 12.88
- >522 nm
- • —
>426 nm
2 3
Acetaldehyde Concentration, mg/1
-33-
^ ' 1 ~ •v»’r
TABLE 0
20 HOUR A OD OF KMn04 DECOMPOSITION AS A
FUNCTION OF { OHl
AT WAVELENGTHS OF
426 AND 522
{ OH"!
X 4 2 6 nm
X 5 2 2 nm
0.0 M
0.026
0.006
0.025 M
0.120
0.025
8 ft- 3/16 I. D. TUBING
-35-
of KMn04 , KOH and RO effluent whereas the reference
stream contains KMn04 , KOH, and distilled water.
The colorimeter compares the disappearance of
permanganate ion in ozonated MUST RO permeate, to
which potassium permanganate has been added, with
the permanganate ion concentration in a distilled
water sample, to which an identical quantity of
permanganate solution has been added.
A second schematic for a permanganate colorimeter
monitor is shown in Figure 17. MUST product water
to which permanganate and base have been added is
fed to the colorimeter cells in series. The mixture
is fed to the first cell immediately after mixing;
then flows through a coil of suitable holdup to
allow time for the permanganate oxidation to occur;
and then flows through the second colorimeter cell.
The residence time between the mixing of the
permanganate and base with the product water and
the flow of this solution through the first colori-
meter cell is kept sufficiently short that rela-
tively little reaction occurs during this time.
This first cell can then serve as a reference cell
and indicate the concentration of permanganate
prior to reaction while the second cell indicates
the permanganate concentration after reaction.
A third schematic for a permanganate colorimeter is
shown in Figure 18. A colorimeter with a single
cell is used and the absorbance at two wavelengths
is monitored. MUST Product water to which perman-
ganate and base have been added and which has flowed
through a delay coil which provides the requisite
residence time for reaction to occur, flows through
the cell. One of the monitoring wavelengths is 522 nm
to monitor the disappearance of permanganate as a
result of reaction. The second wavelength may be
chosen at an isosbestic point where the absorbance
of permanganate and manganate ions are equal; the
absorbance of the cell at this wavelength will then
be independent of the relative concentrations of
these two ions but will vary directly as the sum of
their concentrations. The absorbance at this
second wavelength then provides the requisite
information on the concentration of permanganate
added which is provided by the reference cell in
the two previously discussed colorimeter schematics.
Alternatively, the two wavelengths may be chosen to
be 522 nm and 426 nm, with the ratio of trans-
mittances at these wavelengths the monitoring output.
For low contaminant levels, the ratio OD^g will be
OD
522
FIGURE 17
MONITOR SCHEMATIC
The KMn04 and KOH solutions and the distilled water were nitrogen
pressurized at -14.2 psi. The pressure on the MUST water feed was
-13.0 psi .
5-mil Capillaries
Distilled
Water
KMn04
Solution
7.2 g/1
- . 046N
5M
KOH
Solution
15 -mil
Capillary
_ Ozonated
— * ^MUST
Water
(teed off feed to
TOC unit)
-Static Mixer
Inlet (reference) Cell
Colorimeter
Coil with -5 minutes
residence time for
reaction .
Outlet
Cell
Colorimeter monitoring transmission at 522 nm
FI CURE 18
DUAL WAVELENGTH MODE OF OPERATION
Lens
Supply Supply
f
low and for high contaminant ratios, it will be
high. One advantage of using these two wavelengths
is that for a given contaminant level, they will
produce a larger instrument reading than the other
approaches. A second advantage is that the ratio
will be relatively insensitive to fluctuations in
permanganate concentration at low contaminant levels
since these will have little effect on the numerator
^OD426^ '
The response of the monitor as the concentration of
impurities which are oxidizable by permanganate
varies may be analyzed as follows:
)522 = 2370 {Mn04“} + 387 {Mn042~}
D = 77 ^ Mn°4 ~ } + 1370 (Mn04 2~ }
Disosbestic * c (IM"0-") + <Mn<V
where
D522 = optical density at 522 nanometers
Disosbestic = optical density at an isobestic
point
D42g = optical density at 426 nanometers
{MnC4 } = molarity of permanganate ion
{Mn042 } = molarity of manganate ion
and the coefficients are the extinction coefficients
for the respective ions and wavelengths.
From the definition of optical density
I,._ = I e 522
522 °522
T426 X426
-D
e 426
-39-
where I = light intensity incident on the colorimeter
I = transmitted light intensity.
For the two cell mode of operation, the colorimeter
output is the log of the ratio of the transmitted
light intensities through the two cells:
'522
DW, 5 min
522
WW, 5 min
522
DW, 5 min
522
WW, 5 min
and S. =
522 r WW
~ 30 sec
'522, WW
- 5 min
522, WW
30 sec
522, WW
5 min
where the subscript DW refers to distilled water;
the subscript WW refers to waste water;
II
the subscripts 30 sec and 5 min refer to the
time allowed for permanganate reaction before
monitoring transmittance;
S refers to the colorimeter signal output
1 with the mode of operation depicted in
Figure 16;
S_ refers to the colorimeter signal output
with the mode of operation depicted in
Figure 17.
Similarly for the single cell, dual wavelength
mode of operation, the colorimeter output is the log
of the ratio of the transmitted light intensities
at the two wavelengths:
-40-
KMn04 has a very high extinction coefficient and
this reaqent has to be added very accurately.
Suitable micropumps are very expensive and of
questionable accuracy and reliability and pose
maintenance problems. For this reason, reagent dis-
pensing is accomplished by the use of capillary
metering restrictions with a pressure head to force
the fluid through the capillaries. Using Pciseuille's
Law, the dimensions of the capillaries to deliver a
predetermined volume with a known pressure head can
be calculated. To generate the necessary pressure,
a separate pump for each line could be used as shown
in Figure 10. A probably less-expensive method
(also fewer moving parts) is to use a single nitrogen
tank to pressurize the reagent dispensers and this
approach was selected for initial development. The
pressure necessary to force the liquid through the
capillaries and deliver the required volume was
kept relatively low, between 10 and 20 psi, so as to
minimize problems generated by the dissolving of
gas in the reagents. At higher pressures diaphragms
would have to be used to separate the two media and
prevent the dissolving of gas.
The diameters of the capillaries are very small and
proper precautions must be taken to prevent blocking.
An inexpensive method is to use fine filters to
remove small particulate impurities. The filter
material has to be inert to chemical attack by the
reagents, i.e., KMn04 and KOH. Very fine mesh
stainless steel clotch is used for this purpose.
Pressure gauges indicate the proper flow or reagents
through the system. The gauges are protected from
chemical corrosion by stainless steel diaphragms.
Eventually all tubing containing reagents as well as
the Vortex mixers should be made of stainless steel.
Slow decomposition of KMn04 occurs, especially after
the Vortex mixer where the solution is made basic by
the addition of KOH. A fine, dark-brown film of
manganese dioxide deposits gradually on the surface
of the materials in contact with the reaction
mixture. The stainless steel tubing and the flow-
through cell of the colorimeter therefore require
periodic cleaning, particularly since the decompo-
sition reaction is autocatalytic , i.e., is catalyzed
by manganese dioxide. A 1% solution of sodium or
potassium bisulphite can remove all Mn02 rapidly
by chemical reaction^'. A reservoir containing
K2S205 is connected to the monitor so the parts in
question may be flushed periodically whenever the
need arises as determined experimentally.
Abcor Pilot Run
The permanganate colorimeter monitor was set up at
the Abcor pilot plant late in August of 1974 to
operate in parallel with the Ionics TOC instrument
so as to determine its suitability for monitoring
the quality of MUST product water. The colorimeter
was set up to operate as shown in Figure 17; the
product water with permanganate and base added
flowed through the two instrument cells in series
with a delay loop in between. The instrument base-
line was periodically established using a distilled
water feed. Early instrument malfunction was traced
to improper electrical grounding of the ground wire
in the power line. This was corrected by separate
grounding of the instrument and recorder cases.
Unfortunately, the instrument became operational
only in time for the last pilot run. Although good
data were obtained over a significant portion of
this run, a number of instrument defects were
apparent which need correction before the instrument
can qualify for field use.
These defects and their correction will be discussed
later. The Abcor run started at about 1:00 P.M. on
Thursday, August 28, and was completed at about 8:00
A.M. the following morning. The bulk of the useful
Amicon data was obtained between about 10:30 P.M.
Thursday evening and 3:30 A.M. Friday morning. Xerox
copies of the recorder traces for these data are
presented in Figure 19. Tabulations of recorder
readings and "A OD" numbers derived from these
recorder traces are presented in Table I along with
TOC readings from the Ionics monitor and TOC data
obtained subsequently at Amicon usi ,g a Technicon/
Phase Separations Tocsin I Analyzer. Recorder
readings for the stream, W, refer to operation with
-52-
TABLE I
RECORDER OUTPUT READINGS
A OD = 1.0 = 10 ppm acetaldehyde
Monitor
Time
Stream
TOC
Reading
(A OD
1200
0230
W
7.2
.48
.092
0240
W
7.2
.475
.087
0240
DW
. 388
0254
W
0304
W
7.4
.488
0.100
0 325
W
7.4
.49
0.102
0334
W
9.3
.51
0.113
0343
W
7.3
.51
0.115
0343
DW
.395
0356
W
0400
W
7.0
.50
0.105
0405
W
7.2
.510
0.115
0410
w
8.3
.525
0.130
0415
w
7.7
.520
0.125
0420
w
7.0
.51
0.112
0425
w
7.6
.51
0.112
0430
w
7.8
.51
0.112
0435
w
7.9
.51
0.112
0440
w
7.8
.51
0.112
0445
w
8.0
.515
0.117
0450
w
8.1
.510
0.108
0455
w
8.0
.515
0.113
0500
w
8.4
.520
0.118
0505
w
8.3
.520
0.118
0510
w
8.6
.520
0.118
0515
w
8.5
.517
0.117
0520
w
8.8
.521
0.121
0525
w
8.4
0.318
0.118
0530
w
8.9
0.320
0.120
0535
w
8.8
0.520
0.120
0540
w
8.6
0545
w
9.0
0550
0555
0600
DW
0.400
0610
w
9.2
0.518
0.118
0615
w
8.7
0.511
0.111
0620
w
8.6
0.514
0.114
0625
w
8.7
0.510
0.110
0630
w
8.6
0.510
0.110
Time TOC
1530 6.1
1730 6.0,6
-53-
TABLE I (continued)
Monitor
Time
Stream
TOC
Reading
(A OD)
0635
W
8.8
0.511
0.111
0640
W
8.3
0.502
0.102
0645
W
8.4
0.502
0.102
0650
W
8.4
0.500
0.100
0655
W
8.2
0.520
0.120
0700
w
0705
w
8.2
0.510
0.110
0710
w
0715
w
7.9
0.511
0.111
0720
w
7.9
0725
w
7.5
0.500
0.100
w
7.3
0.492
0.092
1930 5.8
I
5
.
r
i
I
i
i
k
f
B
-54-
MUST product water (to which permanganate and base
have been added) flowing through the two cells in
series. Recorder readings for the stream, DW, are
obtained with distilled water (to which permanganate
and base have been added) flowing through the two
cells in series. These latter readings represent
the instrument baseline-- the output with no perman-
ganate oxidation occurring. The A OD column repre-
sents the difference between the readings with
the product water flowing through the distilled
water baseline. Since a 10 ppm acetaldehyde solu-
tion gives a A OD of about 1.0, the A OD of about
0.1 consistently obtained is equivalent to about
a one ppm concentration of acetaldehyde. In order
to obtain a uniform mixture entering the first cell
of the colorimeter, it was, howeve , necessary to
use a Kenics Static Mixer following the Vortex
Mixer. This introduced a time delay so that the
solution had about 30 seconds to react before
entering this first colorimeter cell. From Figure
11, the rate of acetaldehyde oxidation by permanganate
is such that with a solution of about one ppm
acetaldehyde, the oxidation would have gone about 25%
towards completion in 30 seconds. Thus, the readings
are really equivalent to an acetaldehyde concentra-
tion of about 1.5 ppm. Of course, if higher alde-
hydes are being oxidized only to the corresponding
acid, the permanganate readings correspond to a
higher ppm. One important conclusion from these
tests is that it does not appear necessary to use
a tracer with the permanganate colorimeter to
monitor the ozonation. Aldehydes appear to be pro-
duced and to accumulate naturally during ozonation.
From the rate of flow through the permanganate
monitor and the holdup in the tubing between the
tee into the TOC monitor supply and the second flow
cell in the colorimeter, a delay may be calculated
to determine what TOC monitor reading to compare
with the permanganate monitor reading. This delay
is equivalent to about 13 minutes; a permanganate
colorimeter reading should therefore be compared
with a TOC reading taken 13 minutes earlier. In
Table II data for the Ionics TOC monitor, the
permanganate colorimeter, and the Phase Separations
Tocsin I TOC Analyzer are compared for the three
samples in this time period for which Tocsin data
are available. Although there are data for the
permanganate monitor at close to the desired times
(10 minutes after the TOC readings) for comparative
evaluation, the data are not always available at
precisely corresponding times. From the trends of
the Ionics and permanganate monitor readings, how-
ever, the changes which are apt to occur in the
i
*
t
-55-
L
'
TABLE II
Run
Time
Ionics
TOC
Reading
Technicon
Phase Separations
Tocsin I Reading
Permanganate Colorimeter
Monitor Reading
1525
7.4
1530
6.1
1543
0.115
1725
8.4
1730
8.9
6.2
1735
0.120
1915
7.9
1925
7.5
0.100
1930
5.8
short time differences from direct correspondence
are small. From Table II it is apparent that there
is not enough variation in any of the data to es-
tablish whether a correspondence between TOC and the
permanganate colorimeter readings exists or not.
It is at least clear that such a correspondence is
not ruled out for MUST product water by these data.
The instrument does appear to have the required
sensitivity, particularly since the noise, once micro
bubbles are eliminated, appears to be less than .01
OD which corresponds to about 0.1 ppm of acetaldehyde
and from Table II assuming the Phase Separations
Tocsin I measurements to be more accurate, to be
equivalent to about 0.6 ppm of TOC.
The response characteristics of the instrument may
be deduced from the recorder data by analyzing the
response when the feed is shifted from the MUST
product water to distilled water and then back again
to the MUST product water. A typical recorder trace
for this sequence together with an explanation for
the changes observed is shown in Figure 19. A
typical response is sketched below:
D v0
a
f\
I
5
>
-57-
a = recorder reading with wastewater feed to colori-
meter. Gas pressure on KMn04 and KOH feed
streams = 14.3 psig; pump pressure on waste-
water feed stream = 13.0 psi. KMn04 feed stream
concentration and capillary diameters and
lengths are chosen to give a potassium perman-
ganate concentration of 72 mg/liter - 0.455
millimolar with the same pressure upstream of
all capillaries. With the pressure on the waste-
water being less than that on the permanganate
stream upstream of the capillaries, the
permanganate concentration before any oxidation/
reduction reaction will be
L
*
14.3
13.0
x
. 455
0.500 millimolar.
b = feed shifted from wastewater at 13.0 psi to dis-
tilled water at 14.3 psi,
c = the diluted distilled water appears in the inlet
cell of the colorimeter. This generates a drop
in A (OD) of about 0.10. The calculated drop,
assuming no reaction in either the distilled
water or wastewater before flow through the
inlet cell:
A (OD) = e A{Mn04 }
where e = extinction coefficient for permanganate
ion at 522 nm.
A (OD) = 2370 A { Mn04 ~ }
= 2370 x .0455 x lo"3
= 0.108
The observed A (OD) of 0.100 is quite close to
this, but the agreement is good in part because
of two cancelling factors: (1) the wastewater
is partially oxidized and the permanganate
partially reduced when the wastewater permangan-
ate stream flows through the inlet colorimeter
cell; and (2) the distilled water contains some
organics which are apt to be oxidized prior to
flow through the inlet colorimeter cell.
i
d = diluted reacted waste appears in the outlet
cell. Because of axial mixing, the rise is
lower than would be expected with distilled
water in the inlet cell and reacted waste in
the outlet cell. The solution flowing through
the outlet cell is a mixture of reacted waste
and distilled water.
e = distilled water-permanganate solution appears
in the outlet cell. This produces the distilled
water baseline reference.
f = the feed is shifted from distilled water to
wastewater .
g = This peak results from an increased concentra-
tion of permanganate ion in the inlet cell due
to the decreased pressure on the wastewater
stream. As the more concentrated wastewater
stream enters the inlet cell, the rise in A OD
is less than expected from the straight dilu-
tion effect due to reaction of permanganate
with the wastewater prior to entering the inlet
colorimeter cell.
h = concentrated distilled water in the outlet cell,
concentrated wastewater in the inlet cell.
i = concentrated wastewater in the inlet and outlet
cells .
Since the recorder paper speed is inches/minute,
the time from shifting the feed to its appearance
in full concentration in the outlet cell is seen to
be about 13 minutes, while it first begins to affect
the outlet cell reading in about 7 minutes. Back-
mixing due to laminar flow pipeline mixing there-
fore gives a time spread of about 6 minutes from
first appearance of the feed in the outlet cell to
its passing through the outlet cell in full conentra-
tion. These times can be significantly reduced by
shortening lines, but the delay line between the
cells is of course required to ensure time for reac-
tion. This delay line is coiled which generates a
stable secondary flow which reduces axial disper-
sion^). Tighter coiling of a larger diameter tube
can be utilized to increase the Dean number and
thereby further minimize the axial dispersion in
this delay line. Elevating the fluid temperature
in this delay line will increase the reaction rate
and decrease the required residence time in the
delay line but will also increase the rate of decom-
position of permanganate under the alkaline condi-
tions in this delay loop.
t
i
r
✓
-59-
1. Gas Evolution and Entrapment
A number of design modifications to the perman-
ganate colorimeter monitor appear desirable in
light of the operational experience obtained
with it thus far. The principal design change
called for is one to reduce gas evolution and
entrapment in the colorimeter cells, primarily
in the cell following the delay loop. The gas
probably resulted from a combination of:
permanganate oxidation; the slow decomposition
of permanganate under alkaline conditions (6) .
dissolution of nitrogen used for pressurization
of the distilled water, permanganate and base
streams; and dissolved ozone and air evolution
from the ozonated MUST stream. The pressure
drop through the capillary flow restrictions
will clearly lead to some gas bubble formation
if the liquid is saturated with nitrogen at
the same temperature upstream of the capillaries.
The delay loop between colorimeter flow cells
appears to be a source of bubbles, probably both
due to nucleation and growth of bubbles from a
supersaturated solution, due to permanganate
oxidation and due to the slow breakdown of
permanganate under strongly alkaline conditions ( ^ #
Buildup of manganese dioxide in the delay loop
or in other parts of the system will, of course,
aggravate the gas evolution problem because of
the autocatalytic nature of the decomposition
reaction. The flow cell configurations, with
leads into and out of the top of the cells, com-
pounds the problem because these cells are
effective bubble traps. A cell with flow inlet
at its base and flow outlet at its top should
do much to alleviate the entrapment of bubbles
in the cell.
2. Mode of Operation
Use of the single cell, two wavelength mode of
colorimeter operation discussed above would
appear to introduce substantial improvements:
a. the noise level ir. the output signal should
be significantly reduced since most of the
factors generating noise will affect both
wavelength cancelling out. Problems due to
microbubbles, manganese dioxide deposition
on the windows or scattering of the light
beam and fluctuations in flow rates should
be markedly diminished.
r
.
I
f t
I
I i
r
l I
»
-60-
b. The operation is simpler and involves less
equipment than the other possible modes of
operation. There is no need for rapid
mixing since the flow enters the colori-
meter only after a suitable delay time for
oxidation to occur.
c. Operation with 522 nm and 426 nm as the
two wavelengths monitored produces a higher
signal than with operation at one of these
wavelengths and an isosbestic point or with
wither of the single frequency modes of
operation. Experimentally, about 4 moles
of permanganate are found to react with
each mole of acetaldehyde when permanganate
is in excess and the concentration is
sufficiently basic. Oxidation of acetaldehyde
to acetic acid requires only two moles of
permanganate per mole of acetaldehyde, so
some carbon-carbon bond scission is occurring
as well. The calculated changes in perman-
ganate colorimeter OD with single cell opera-
tion at 522 nm and 426 nm and reaction with
1 ppm of acetaldehyde is compared in Table
III with the A OD obtained monitoring either
at 522 nm or at 426 nm with either a dual
cell, single-wavelength mode of operation
or a single cell, dual wavelength mode of
operation where the second wavelength is at
an isosbestic point.
The colorimeter was subsequently modified at
Amicon to convert it to the dual wavelength
mode of operation, with 522 nm and 460 nm (an
isosbestic point) chosen as the two wavelengths.
A flow-through cell with positive flushing was
introduced. These modifications essentially
eliminated the gas bubble entrapment problem
and reduced baseline noise to a small fraction
of 0.01 OD as may be seen from the recorder
tracing reproduced in Figure 20. The data
show that, even after the Kenics Mixer, small
concentration fluctuations due to incomplete
mixing are present. These are eliminated by
flow through the delay coil.
a. Instrument response
Experiments were carried out to determine
the instrument response as a function of
the level of base concentration and the
permanganate acetaldehyde ratio. Results
are presented in Table IV. The calculated
TABLE III
CALCULATED A OD FOR 1 PPM ACETALDEHYDE
Time (min)
-63-
S
I
TABLE IV
Initial {Mn04“} = 36 mg/1 = 0.228 x 10_3M
:2.5 ppm CH3COH if 4 mols Mn04 react with 1 mol CH3COH
Calculated A OD
0
0.42
1.00
2.12
2.5
4.24
■0.523
0.126
0.298
0.632
0.747
Experimental A OD
ppm
_D522
A OD
D522
0.1M {OH~} -Dj. 22
0.2M {OH-}
CH3COH
+D426
+D426
+D426
-0.522 -0.527
0.127 0.131
0.582
0.812
0.540
0.775
-64-
and experimental agreement is good when
only enough acetaldehyde is present to
react with one-sixth of the permanganate,
but is not as good when the acetaldehyde
concentration is high enough to react with
80% of the permanganate. In Table V the
range of optical densities are presented
which would be observed with permanganate
concentration sufficient to react with three
different levels of acetaldehyde and a one
cm path length cell. Optical densities
above one become increasingly difficult to
measure because of stray light, interference
with the low levels of transmitted light.
As the table indicates, if permanganate
utilization is kept below 40%, and the
correspondence between ozonated MUST TOC
and permanganate is that observed in the
limited experiments to date where_one ppm
of TOC reacts with about 0.2 x 10 4 moles
of permanganate, then from 0 to 10 ppm of
TOC can be monitored by noise, which in the
dual wavelength mode of operation appears to
be less than .002 OD by baseline drift which
can be limited by frequent calibration and
by permanganate decomposition which con be
factored out. Conservatively, with baseline
calibration every few hours, a sensitivity
of .01 OD equivalent to 0.2 ppm of TOC
should be readily attainable.
b. Instrument calibration
In the dual wavelength mode of operation,
after flushing with bisulfite and rinsing
with distilled water, the instrument base-
line may be calibrated by introducing a
neutral filter of known absorbance.
3. Flow Capillaries
The use of capillaries of selected diameters
and lengths to control the flow ratios of the
reactants appears sound, but the diameter of
the capillary used to feed the permanganate
should be enlarged. Its length should probably
also be increased so as to maintain the same
hydraulic resistance and it should either be
folded back and forth or coiled to keep dimen-
sions small. A larger diameter capillary will
show less tendency to plug due to permanganate
decomposition and will be easier to clean by
flow or bisulfite through the system.
-65-
TABLE V
ABSORBANCE RANGE WITH A 1 CM PATH LENGTH AND 4 MOLS Mn04“ PER MOL CH3COH
Range of acetaldehyde
concentrations to be
monitored
Lower
Upper
°522
°426
Limit
Limit
ppm-
CH3COH
0
2.5
. 540 -
.072
.018 -
.296
0
5.0
1.080 -
.144
.036 -
.592
0
10.0
2.16 -
.288
.072 -
1.184
ppm CH3COH ^ ^
ppm TOC 5
A OP £ 0 . 3
ppm TOC ^ 6
If Mn04 utilization is kept to 40% to prevent Mn02 formation, a high
enough Mn04 concentration to react with 5 ppm CH3COH will permit
determination of 0. 4x5x5= 10 ppm TOC with a 1 cm path length and
a sensitivity of about 0.2 ppm TOC.
-66-
Precise control of the permanganate concentra-
tion in the monitoring stream is a prerequisite
for reliable monitoring. This requirement is
considerably alleviated by operation in the
dual wavelength mode with 426 nm and 522 nm,
the two wavelengths monitored, since the
absorbance at 426 nm results almost entirely
from reaction with contaminant and is essentially
proportional to contaminant level (for contami-
nants of a given reactivity level) .
To control the permanganate/sample ratio more
accurately, it may be desirable to pump the
sample into a container which is pressurized
by the same gas pres; are used to drive the
permanganate through its flow capillary. Limit
switches would be used to control the liquid
level in the container, and the container volume
would be kept small to minimize the time lag it
introduces .
4. Pressure Measurement
Replacement of the present Bourdon tube-type
pressure gauges by electromechanical (piezo-
electric or semiconductor) gauges will eliminate
dead spot regions where permanganate can slowly
decompose and release manganese dioxide into the
system. Measurement of pressure after the
metallic filters may prove unnecessary, parti-
cularly if a periodic bisulfite cleaning flush
schedule is set up.
5. Instrument Simpl if ication and Cost Reduction
Considerable simplification and ruggedization
of the permanganate colorimeter is desirable
and should be feasible once the desired
characteristics have been pinned down.
D. Rate of Decay of Ozone
An alternate technique for monitoring the ozonation
reactor is to monitor the rate of decay of ozone
concentration in the effluent from the reactor.
This can be done spectrophotometrically since ozone
has a molar absorptivity of 2.500 to 3000 (10) at 360 my
using a long (10 cm) path cell to obtain the desired
sensitivity. The rate of decay of ozone concentra-
tion may be represented approximately by the equa-
tion :
r
1
i
i
-67-
(1)
“dO 3
dt
14.6
k2 (03 ) (TOC) + kde RT
The first term on the right represents the rate
of oxidation of TOC and is a reasonable empirical
approximation for typical waste streams, particu-
larly for intermediate ozonation times^' 1^.
The rate constant, k2 , is of course a function,
both of the organic material present, the pH, and
catalyst (e.g., UV) concentration. The second
term on the right represents the rate of auto-
decomposition of dissolved ozone as given by Li
et ald3) , The rate constant, k , is also a func-
tion of pH, catalyst concentration , and ionic
strength. For product water which has been treated
with both reverse osmosis and ozonation, however,
k^ should be essentially constant since the pH
will be close to neutrality and since the impurity
levels are quite low.
For very low impurity concentrations, the rate of
decay of ozone concentration will be determined
solely by the rate of autodecomposition; as impurity
levels are increased, the rate of decay of ozone
concentration will increase due to the first term
on the right-hand side of equation (1), provided
the impurities react with ozone. Measurement
of the rate of decay of ozone concentration will
then provide a measure of whether material which
is oxidizable by ozone remains in the system.
To provide a preliminary indication of the utility
of this concept, experiments were conducted with
both distilled water and secondary effluent RO
permeate, monitoring the rate of decay of ozone
concentration in a long path (10 cm) cell at 254 nm
and at room temperature (=25 C) as a function of
ozonation time. Results are presented in Table VI
and Figures 21, 22, and 23. As may be seen, the
ozone rate of decomposition in distilled water
shows the three-halves dependency on ozone concen-
tration found by Li (13) ancj 2ero time
extrapolated intercept agrees with the solubility
data of Rawson'^' . The relatively low OD readings
at early times are probably attributable to Beer's
Law deviations at high concentrations due to stray
light interference in the Coleman 139 Spectrophoto-
meter. The two different intercepts are probably
partially attributable to a difference in
temperature, but the data for the sample ozonated
one hour (circular data points, Figure 1) are
believed more accurate.
;
i,
:
I
I
[
I
l
t:
r
-68-
TABLE VI - PART A
DECOMPOSITION OF OZONE IN DISTILLED WATEP
Absorbance monitored at 254 nm in a 10 cm path length cell
500 ml of water ozonated 30 minutes at room temperature with
oxygen containing 38 mg/1 ozone
TIME
OD
(OD)
7
0.23
10
0.25
2.0
20
0.22
2.13
40
0.21
2.18
55
0.18
2.35
120
0.C76
3.63
130
0.070
3.78
145
0.060
4.08
175
0.040
5.00
205
0.035
5.35
265
0.025
6.33
TABLE VI - PART B
DECOMPOSITION OF OZONE IN DISTILLED WATER
500 ml of water ozonated one hour at room temperature
('25 C) with oxygen containing = 38 mg/1 o zone
Absorbance monitored at 254 nm
TIME
OD
(OD)
5
0.242
2.03
15
0.232
2.07
30
0.195
2 .27
45
0.153
2.56
60
0.128
2.80
75
0.105
3.09
90
0.088
3.38
105
0.075
3.65
120
0.062
4.01
135
0.054
4.31
153
0.047
4.60
167
0.042
4.88
190
0.036
5.27
OZONE AUTODECOMPOSITION IN DISTILLED WATER AT ROOM TEMPERATURE
RATE OF DECAY OF OZONE IN DISTILLED WATER
AND SECONDARY EFFLUENT RO PERMEATE AT
ROOM TEMPERATURE
0, Treatment
a 2ndy effl
^ Secondary
Effluent
Saturated
30 min. in
reaction
vessel
X Secondary
Ef f luent
t (minutes )
With the secondary effluent, the rate of decay of
ozone concentration is clearly higher than with
distilled water, particularly for material ozonated
for less than one hour. The effluent contacted
with ozone for a few minutes to saturate it and that
ozonated for 30 minutes both show rates of decay
dominated by the rate of oxidation of TOC (first
right-hand term in Equation 1) . The zero time
intercepts for these data are again in approximate
agreement with the Rawson solubility data(^) . For
secondary effluent ozonated at room temperature
for one hour, the rate of decay of ozone (Figures
2 and 3) is only slightly greater than in distilled
water and the further purification achievable with
room temperature ozonation in a reasonable time
period is therefore small.
To evaluate the true potential of this monitoring
technique, low holdup flow cells should be utilized
and measurements at higher temperatures, where
reaction rates are faster, should be investigated.
The technique appears to hold promise.
CARBON ADSORPTION EXPERIMENTS
Three primary processing steps presently contemplated
for the purification of MUST water are RO treatment,
ozonation and adsorption by activated carbon. In
order to investigate the adsorption characteristics
of activated carbon relative to MUST RO permeate,
adsorption isotherms were studied for both ethanol
and acetaldehyde. Ethanol was selected as a known
principal impurity in the MUST RO permeate. Acetalde-
hyde was selected as a principal contaminant in ozonated
MUST RO permeate.
A. Experimental Procedures
A 790 ppm solution of ethanol was prepared by
adding 1 ml of ethanol to 1,000 ml of distilled
water. This solution was added to a reaction
flask. One hundred grams of previously-washed
Pittsburgh Charcoal 20/50 mesh was then added to
the reaction flask. This solution was then agitated
overnight on a rotary mill. The following day the
charcoal was removed by using a Buchner Funnel with
suction and Whatman #1 Filter Paper. A sample of
the filtrate was then analyzed and the ethanol
concentration determined as usual by the acidified
dichromate method. This entire experiment was
carried out at room temperature. The same experi-
ment was repeated at 60 C. The solution was stirred
overnight in a reaction vessel at 60 C and filtered
hot before being analyzed for ethanol concentration .
A third experiment was carried out at room
temperature with a 800 ppm solution of acetaldehyde.
The acetaldehyde concentration was determined by
reacting it with basic potassium permanganate as
described previously in this report. The results
are represented in Table VII.
Two more carbon adsorption experiments were performed
and both times a Pyrex glass column was used since
this setup is more in line with the actual processing
step than batch-type experiments. In the first
column experiment 114 g of activated carbon were
added to a 1" I.D. x 2' long Pyrex column with a
fritted glass base. The carbon filled approximately
10" of the column. The column was filled with
distilled water and then four liters of a 79 ppm
ethanol solution were passed through the column at
room temperature at a flow rate of 20 ml/minute.
Small samples of about 20 ml were collected every
250 ml, analyzed for ethanol using an acidified
potassium dichromate colorimeter test. A second
-74-
TABLE VII
DECOMPOSITION OF OZONE IN SECONDARY EFFLUENT RO PERMEATE
700 ml of secondary effluent (Brockton) RO permeate
was ozonated for one hour at room temperature with
oxygen containing 33 mg/1 of ozone, and the
absorbance then monitored at 254 nm in a
10 cm path length cell
Time
OD
1/ (OD)
0
^0.25
15
0.235
2.06
30
0 . 225
2.11
50
0.192
2.28
60
0.165
2 .47
75
0.137
2.71
93
0.107
3.06
105
0.091
3.31
120
0.075
3.65
135
0.063
3 .98
150
0.053
4 .35
180
0.042
4 .88
210
0.039
5.06
240
0 .039
5.06
285
0.039
5 .06
-75-
identical experiment was performed with slightly
less activated carbon, i.e., only 73 grams of acti-
vated carbon were used. The results are presented
in Figure 24.
Discussion of Results
From the effluent curves, the total alcohol adsorp-
tion on the activated carbon may be estimated and
a distribution coefficient for alcohol between the
carbon and the solution calculated and compared with
those derived from batch experiments. These data
are presented in Table VIII.
The adsorption experiments indicate (Table IX and
Figure 25 and 26) that the batch equilibrium data
fit Freundlich adsorption isotherms. The slight
effect of temperature on ethanol distribution
indicates a low heat of adsorption which is consistent
with the relatively weak adsorption of the alcohol.
Acetaldehyde is somewhat more strongly adsorbed
than the ethyl alcohol, which is consistent with
the higher hydrogen bonding potential of the alcohol
and its consequent greater affinity for water. The
data indicate both ethyl alcohol and acetaldehyde
can be adsorbed from MUST RO permeate using activated
carbon columns but the low distribution coefficients
indicate that the carbon would have to be regenerated
frequently. With a distribution coefficient of ,
the weight of carbon is at a minimum W/<*. Thus, to
process 3500 gallons of water with an alcohol dis-
tribution coefficient of 30, approximately 18 ft.3
of carbon would be required. With a volume of 18 ft.3
of carbon to process 3500 gallons of water, the
carbon would have to be regenerated on a daily basis.
From the standpoint of monitoring, neither alcohol
nor acetaldehyde is strongly adsorbed and either
might be a suitable tracer materia) depending on
the location of the activated carbon and the
impurities which were to be removed by it. For
example, if it were located after the ozonation step
and were intended to remove substantially all residual
impurities, then acetaldehyde might again prove to
be a suitable tracer compound, both because it is
likely to be the principal residual impurity and
because it is not strongly adsorbed.
One
for
liter of permeate ozonated at room temperature
30 minutes with oxygen containing 38 mg/1 ozone
Absorbance monitored
length cell
. at 254 nm in a 10 cm path
using a Coleman 139 Spectrophotometer
Time
(minu tes )
OD ' pH
5
>0.20
20
0.175
25
0.150
30
0.135
47
0 .092
80
0 .055
87
0.048
^24 hours
0.0146 4.8
One
38
liter of permeate
mg/1 ozone for a
contacted with gas containing
few minutes to saturate it
Absorbance
monitored as above
Time
(mi nutes )
OD
5
0.020
20
0.0166
30
0.0126
4 0
0.092
50
0.068
6 0
0.048
70
0.036
80
0.02C
90
0.020
BATCH TYPE CARBON ADSORPTION EXPERIMENTS
gms acetaldehyde absorbed/gms carbon
gms acetaldehyde in solution/gms water
solution
%.*>r
-81-
V. SECONDARY EFFLUENT WASTE WATER STUDIES
Amicon has initiated studies for the monitoring of
reverse osmosis, ozonation, and activated carbon
treatment steps in various combinations for the
processing of secondary effluent of municipal waste
water to upgrade it to tertiary quality. Amicon
arranged to receive secondary effluent waste water
from the municipal waste treatment plant in Brockton,
Massachusetts.
A. Pilot Plant - Construction and Operation
A pilot plant for the treatment of secondary
effluent waste water was constructed as shown in
Figure 27. The ten-gallon, stainless steel reser-
voir is filled with secondary effluent waste water.
With the appropriate valves open and shut, the
untreated effluent is transferred from the reser-
voir to the accumulator by pressurizing the reser-
voir with the aid of the nitrogen tank. Once the
accumulator is filled, the reservoir is isolated
from the system by closing the values leading to
it. The whole system is then pres: arized to
operating pressure via the nitrogen tank. The
tubing used to build the plant consists of 1/2"
stainless steel tubing except for the cooling coil
which is 5/8" stainless steel acting as a simple
heat exchanger. The pump used in this system is a
sealless magnet drive pump bought from the Kontro
Co. , Inc. Since the recirculating pump has a
capacity of 20 gpm, which is considerably in excess
of that required by the RO unit, most of the waste
water is recirculated through the bypass loop, i.e.,
the cooling coil. The cooling coil is situated in
a large vessel which is filled with ice before each
run since an appreciable rise in temperature occurs
during the course of an experiment. The RO unit
was received from Universal Oil Products Company.
It is a spiral-wound, 620 mini-module with 1.0 sq.
ft. membrane area, producing 10.5 gfd with 96.6%
sodium chloride rejection at 600 psi, 25 C and
5,000 ppm NaCl solution. The recommended flow rate
through the module is 2.0 gpm.
B . Experimental Procedure
To five gallons of waste water enough sodium chloride
was added to make a 5,000 ppm solution. This solu-
tion was added to the reservoir and then transferred
to the accumulator and pressurized at 600 psi. Ice
was added to the cooling coil container and the pump
switched on. The flow through the RO unit was ad-
justed to 2 gpm and samples of approximately 20 ml
f
r
were collected for TOC analysis at two-liter inter-
vals. A total volume of eight liters was collected.
The results of this experiment are represented in
Figure 28.
'
1
In a subsequent experiment, water from the Brockton
waste treatment plant was again processed through
the VOP RO module as before, and the RO permeate
was then divided into three parts. The first was
ozonated for 30 minutes at room temperature with
oxygen containing 38 mg/1 of ozone using the
apparatus shown in Figure 1; the second was ozonated
under the same conditions for 60 minutes; and the
third was not subjected to further treatment. All
three samples were then reacted with alkaline per-
manganate to determine whether a detectable change
in OD would occur and all three samples were
analyzed for TOC. Results are presented in Table
X.
i A
-86-
✓
VI. REFERENCES
1. Annual Report on Development of a Monitor for
Recycle of Waste Water, December 1, 1972 - August
15, 1973, Contract No. DADA 17-72-C-2169 , Amicon
Corporation, Lexington, Massachusetts.
2. Final Report on MUST Waste Water Treatment System,
Contract No. DADA 17-71-C-1090 , AiResearch
Manufacturing Company of Arizona.
3. First Quarterly Report, Fiscal Year 1974, Contract
No. DADA 17-73-C-3013 , University of Cincinnati,
Department of Environmental Health, Kettering
Laboratory, Cincinnati, Ohio.
4. Wiberg, Professor Kenneth 3., Yale University,
personal communication.
5. Lingg, Dr. Robert, Environmental Protection Agency,
Cincinnati, Ohio, personal communication.
6. Oxidation in Organic Chemistry, Part A, Chapters 1
and 2, edited by Kenneth Wiberg.
7. Perry's Chemical Engineer's Handbook.
8. International Critical Tables, Edition , Volume
3 , p . 290 .
9. Koutsky, J. A. and Adler, R. J. "Minimization of
Axial Dispersion by Use of Secondary Flow in
Helical Tubes," Can. J. Chem. Eng., 42, 239 (1964).
10. Lagton, R. F. "Analytical Methods for Ozone in
Water and Waste Water Applications." Chapter II
in Ozone in Water and Wastewater Treatment, F. L.
Evans III , Editor , Ann Arbor Science Publishers
Inc. , 1972 .
11. Hewes, C. G. III. "Renovation of Secondary Waste
Water Effluents by Ozonation of Dissolved Organic
Compounds," pH.D. Thesis, Texas A&M, Department of
Chem. Eng . , 1971 .
12. Gollan, A. "Evaluation of Membrane Separation
Processes. Carbon Adsorption and Ozonation for
Treatment of MUST Hospitfl Wastes" 25 March 1975
Walden Research Progress Report on USAMRDC Contract
No. DAMD 17-74-C-4066 .
13.
Li, K. Y. , Wen, C. P., Weeks, J. L., and Kuo, C.
Absorption and Decomposition of Ozone in Aqueous
Solutions, paper presented at the 68th National
Meeting of the AIChE, Los Angeles, California,
November 1975.
14. Seidell/Linke . "Solubilities of Inorganic and
Metal Organic Compounds, 4th Edition, Volume 2,
American Chemical Society, 1965, pp. 1239-40.
-88-
DISTRIBUTION LIST
4 copies HQDA (SGRD-RP)
Washington, D. C. 20314
12 copies Defense Documentation Center (DDC)
ATTN: DDC-TCA
Cameron Station
Alexandria, Virginia 22314
1 copy Superintendent
Academy of Health Sciences, U.S. Army
ATTN: AHS-COM
Fort Sam Houston, Texas 78234
1 copy Dean
School of Medicine
Uniformed Services University of the
Health Sciences
Office of the Secretary of Defense
6917 Arlington Road
Bethesda, Maryland 20014
25 copies Environmental Protection Department
ATTN: SGRD-UBG
US Army Medical Bioengineering
Research and Development Laboratory
Fort Detrick, Frederick,
Maryland 21701