SURVEY OF WATER DESALTING INVESTIGATIONS, IN PARTICULAR THE ELECTRODIALYTIC METHOD.

• Document Type: CREST
• Collection: General CIA Records
• Document Number (FOIA) /ESDN (CREST): CIA-RDP82-00041R000100160001-0
• Release Decision: RIPPUB
• Original Classification: K
• Document Page Count: 107
• Document Creation Date: December 15, 2016
• Document Release Date: August 19, 2002
• Sequence Number: 1
• Publication Date: October 1, 1952
• Content Type: REPORT

STAT
Approved For Release 2002108128 : CIA-RDP82-00041R000100160001-0
Approved For Release 2002108128 : CIA-RDP82-00041R000100160001-0
 Approved For. Release 2002108128 CIA-RDP82-00041R000100160001-0
PORT T.A. .2
OF THE
TECHNICAL DEPARTMENT T.N.O.
OF THE CENTRAL NATIONAL COIINCIL FOR APPLIED SCIENTIFIC
RESEARCH IN THE NETHERLANDS
i Konin9skade
The Hague
ArL RIGHTS RESERVED.
Tel. 777830`
Appwu d.Fc~ elease~;2QQ 8.28 :4( ROP8 00041 R000100160001-0
Survey of water desalting investigations, in
particular the electrodialytic method.
 Approved For Release 2002108128 : CIA-F '82-00041R000100160001-0
GENERAL TECHNICAL DEPARTMENT T.N.O.
The Hague
Report T.A. No 270
TITLE
AUTHORS
DATE
PASSED BY
' Survey of water desalting in-
vestigations, in particular the
electrodialytic method.
Dra Y. Boer-Nieveld and D. Pauli.
October 1952?
C. van Hoek, Dr E. We8elin and
Dr Ir S.G. C7iechers.
PARTICIPANTS Drs
F. Bergsma
w. Doornenbal
J.F. Tilstra
P. h'estdorp
Approved For Release 2002108128 : CIA-RDP82-00041R000100160001-0
 Approved For Release 2002108128 : CIA-RDP00041R000100160001-0
OBJECT.
To find an ecoromicall
Y justifiable method of desalting water for
domestic, industrial and agricultural purposes.
METHOD.
Literature research was carried out into methods of desalting sea
water and brackish water in
general and into electricialYtic removal
of salts from brackish and similar waters in particular.
An investigation was made into the factors determining energy con-
sumption for electrodialYsis.
Experimental investigations were made into the
results obtainable
b electrodial tic
by y and electrolytic purification of water from about
1650 to about 00
5 mg sodium chloride
per luxe.
C027CLUSI Oli S .
1. It appears to be possible to desalt 1 m3 water from 1650 to
500 mg
sodium chloride per litre at a cost of F1. 0.37 per m while further
reduction of this figure may be attained.
2. There are sound economic reasons for intensive continuation of
research into electrodialytic desalting of brackish and similar
waters.
3. As regards the most economic method
for the desalting of seawater
no proper comparison is yet possible between vapour compression
distillation and electrolytic or
electrodialytic purification
methods.
Approved For Release 2002108128 : CIA-Rl P82-00041R000100160001-0
 Approved For Release 2002108128 : CIA-RDPF~00041R000100160001-0
C O N T E N T S
I. STARTING POINT TO THE INVESTIGATIONS.
II. GENERAL METHODS OF DESALTING WATER.
A. Desalting methods for sea water.
B. The desalting of brackish waters.
III. ELECTRODIALYTIC REMOVAL OF SALTS FROM BRACKISH
AND Sfl ILAR WATERS.
A. Introduction.
B. The principal patents relating to electxo-
dialytic desalting of water.
C. Brief survey of the literature on the electxo-
18
18
24
27
29
dialytic water desalting. 30
D. Review of the theoretical computations by gten in
the electrodial tic desalting of water in a three-
compartment cell.
35
E. The experiments of Hoffmann.
42
F. The electrode processes. 43
IV. ENERGY CONSUMPTION OF ELECTRODIAJ.,ySIS. 47
A. Introduction.
47
B. Electrode
potentials. 4$
C. Voltage drops caused by Ohmic resistance.
53
D. Voltage drops at the membranes.
56
E. Summary. 55
F. The quantity of charge passed through in
electrodialytic desalting.
0. Calculation of energy consumption and the desalting
effect for the electrodialytic desalting of water
from experimental data.
56
57
Approved For Release 2002108128 : CIA-RDP82-00041R000100160001-0
 Approved For Release 2002108128: CIA-RDP82-00041R000100160001-0
V. CHARGE EFFICIENCY.
A. Introduction.
B. Definitions of the e various efficiencies.
C. Derivation of
a general expression for current density
of an ion in a system of a number of ions in the case
of charged membranes.
D. A simple calculation of the current density efficiency
of anion present in a system o
of a number of ions in
the case of non-selective membranes.
E. Calculation of the charge efficiency of the chloride
ion in the cathodic and the anodic membranes from
experimental data of elec
tro dia lytic desalting ex-
periments of water.
F. Charge efficiencies and current efficiencies
found
in literature.
VI. THE EXPERIMENTAL RESEARCH INTO t'(ATER DESALTING.
A. Description of the apparatus.
B. Experiments
arith non- or only slightly selective
membranes.
C. Research into the usefulness of selective membranes
for electrodialytic desalting of water.
D.
Tentative experiments on chloride removal at the anode
in a two-compartment cell with non-reversible electrodes.
E. Several desalting experiments on potassium chloride
solutions in a two-compartment cell with reversible
silver-silverchloride-electrodes.
F. Disct;ssion of the results.
63
63
63
66
70
74
76
79
79
81
88
93
94
97
BRIEF SUL~tS6ItY. 105
REFERENCES. 107
Tables I - XVII incl.
Graphs 1 - 18 incl.
Figures 1 - 6 incl.
Appendices 1 - 10 incl.
Approved For Release 2002108128 : CIA-RDP82-00041R000100160001-0
 Approved For Release 2002108128 IA-RDP82-00041 R000100160001-0
List of symbols and units
In the text of this report in general no explanation of the symbols
usec is given. Their meaning can be looked up in this last. In a few
cases it was unavoidable to use the same symbols for different items. At
such places a supplementary explanation is given in the text and the
same applies to the units. Vbherever in the text of this report no special
units are given all the quantities are expressed in the units listed
here.
Symbol
dE
a~:
Unit
2.
Coefficient of diffusion cm sec.
Terminal voltage volts
Field strength volts cm
F Faraday's constant
I Current strength
M Molecular weight
NE Average energy consumption in
aesalting rrom zo to o m equiv.
Area
Desalting velocity according
to Aten (31)
Charge
Resistance
aasconstant
Time
Equivalent conductance of cations
Equivalent conductance of anions
Volume
Energy consumption
Average energy consumption in
desalting from c to c m equiv.
p k
C1' 1.
k%Yh m3
2.
cm
g equiv~ per
per sec.
Coulombs
ohms
2
cm
joules per degree C
per mole
hours
cm2per ohm per
gram equiv.
2
cm per ohm per
gram equiv.
litres
3
kVh m
kWh~m3
Approved For Release 2002108128 :CIA-RDP82-00041 R000100160001-0
 Approved For Release 200210812CIA-RDP82-00041 R000100160001-0
- 11 -
Concentration within the membrane
pores
3
Concentration in the free solution g equiv. cm
2
calculated per cm
Current density,
membrane area
100 amp.hr
i Current density in the solution amperes/cm
of concentrations in the
m Ratio
anolyte cf. . Aten 31
Transference number
n
n Idem in the membrane
m
n Idem in the cathodic membrane
an
n Idem in the anodic membrane
am
P Ratio of concentrations in the
catholyte cf. Aten 31
Desalting effect
q
-1 -1
Electrolytic mobility of cations cm2volt sec
-1 -1
Electrolytic mobility of anions cm2volt sec.
q (c ' ck Average desalting effect during
mg NaCl per 100 ap.hr
p equiv.
e
from c to c m q
d
lt
ing
esa
p, k
Cl- 1
r Donnan concentration ratio
(of Teorell 90
Time sec.
T
Energy consumption
~
g equiv./cm
g NaCl per
watt sec per cm~
Coordinate of length. cm
3
Concentration in electric units eoulombe.per cm
Thickness of membrane
Plancks variable
(of Teorell 90p 461
i
Current density efficiency
I
Current efficiency
-1 -1
Specific conductance cm
K
P
T
Specific resistance f cm
Absolute temperature degrees Kelvin
Average charge efficiency for
the time interval Opt
3
fixed membrane charges, g equiv. per cm
W X Density of the
expressed as concentration
Approved For Release 2002108128 : CIA-RDP82-00041R000100160001-0
 Approved For Release 2002108128 IA-RDP82-00041 R000100160001-0
-13-
- +
The indices + and -, e.g. c 9 c, i, indicate that the symbols
apply to the cations or to the anions. The indices ', " and ", show
C' + c" + . The indices 1 and 2 indicate the two sides of
without these indices relate to the whole the total concentration c
that the indexed symbols relate to different types of ions; symbols
the membrane. Thus, c is the concentration of a certain cation in the
1 ,
free solution on one side + c that on the other side of the membrane
Approved For Release 2002108128 :~A-RDP82-00041 R000100160001-0
 Approved For Release 2002/08/28:
I. STARTING POINT OF TAE IhWESTIGATIONS.
Because of the difficulties occuring in obtaining water supplies
for domestic, industrial and agricultural purposes, considerable
. attention is devoted to the desalting of water 1,8,15924,99.
.
This applies in the first place to countries where the supply of
drinking water is already an important problem or will be within a:
measurable time 1, 2, 3, 4). In this respect for instance there,might
be mentioned the Netherlands, the United States and islands such as
Aruba and Guraao on these :slands practically all drinking water has
to be prepared in a more or less complex and expansive man ner, for
where .
instance by distilling sea water, or by obtaining supplies from else-
Approved For Release 2002/08/28:
Furthermore, the supply of water for irrigation purposes is of
great agricultural importance. In the first place we have in mind
countries in which it is desired to start reclaiming deserts: _Israel,
Irak Persia.
The Netherlands' future supplies of drinking water are endangered
by a steady increase?in the salinity of underground water caused by
the continuous penetration of sea water displacement of the so-called,.
U Galt water
salt water boundarya eastwards and by the discharging of wast a
rish in ions by industry, while - owing to the continuous growth in
population, the higher standard of living and increasing industrialis-
ation the water supplies formed by precipitation are failing to meet.
requirements. ?
Fig. 1 (5) gives an idea ofthe fresh water layer floating,on
sea water, the boundary of which is constantly shifting inland.
In a summary of present and possible future water supplies in
the Netherlands 6 it is stated that the amount,?of water which, as
estimated will be required by the year A.D. 2000 is 890,000 m3, day,
whilst the amount of water available is:
3
from the dunes
.
113,500 m
day
from other underground sources
227,000 m3day
total water available approx.
400,000 m3day
The deficiency will have to be made up by surface water and
desalted brackish and/or sea water.
 Salinity in the. North Holland polder and polder basi
n area is
already 500-1000 mg chloride per litre9 whilst the criteria for fresh,
brackish and salt water according to Krul and Liefrinck 6 are:
.
fresh water
up to 100 mg chloride per litre
brackish water 100 1000 h
salt water over 1000 a I, ~l n
The salinity of agricultural water must not exceed 00
3 mg chloride
e
per litre. Th water of the. North Sea Canal near IJmuiden is salty,
while its salinity gradually decreases towards Amsterdam.
That of the
New Claterwad
y and of the New Maas between Rotterdam and Hook of Holland
varies according to the tides and the. run off of the river Rhine.
The increasing salinity of polder
basin and river water in the
Netherlands is becoming serious and occurs in the West o
_ primarily of
the country 7
Owing to the great problems arising from the penetration of salt
in older basin and river water in this country, the Research Institute
for Public-
Health Engineering T.N.O. approached the Plastics Research
Institute T. ;0
N on the question whether it was. possible to formulate
process for large-scale desalting sea water or brackish water by
ionexchanger treatment.
In reply to
this question the Plastics Research Institute T.N.O.
issued a report K.I.
49/151, dated September 14th, 1949 from which.
it appeared that in the production of large quantities of drinking
water
from saline river water by ion exchange in most cac-~
?s, the cost
of regenerating the exchangers, even if effected
electrolytically,
would make the
price of the water rather high.
'Arising from the conclusions in this report and owing to the
fact that the General Technical Department T.N.O. was . undertaking
extensive research into the electrodia1ytic desalting of various
liquids, this Department was requested on January 20th, 1950 by the
Committee for
Hydrological Research T.N.O. through the Research In-
stitute for Public Health Engineering T,N.O, to examine he
the
'
litres Rf electrodial tic desalting of brackish
Y g and sea water.
In -t
he countries referred to on page 15 there is of course
also great interest in processes enabling efficient?largeoscale re=
duction of the salt content of 18,000 mg C1 1 seawater
1,000 mg Cl 1 brackish water to about 300 mg C1 1.
 Desalting methods for sea water.
Various methods of desalting sea water for domestic use are
known. The most important methods are mentioned in Table I with
a note of the cost found in the literature 19 2f 8).
The average rate for water for industrial purposes in the
3
United States is fl. 0.10 to fl. 0.12 per m
, and for smaller
quantities for instance for domestic use f1.,0.25 Per m3. Wa-
ter for agricultural purposes hasf of coursef to be much cheaper.
It will be clear that the desalting problem for any parti-
cular region is not solved until a method is found which can be
economically justified. Therefore the cost per m3 reclaimed Wa-
ter must be calculated separately fore c d foreac
each process an each
location.
1. Distillation processes 2,
a. Distillation under atmospheric pressure.
The production of drinking water from sea water by
means of distillation is applied fairly generallYf during
the Second World War the United States in this way daily
produced water for 1,000,000 people.
By distilling 1.25 kg of sea water one obtains 1
kg
of drinking water and 0.25 kg of brine.
The possibility of reaching high efficiency of heat
recovery by employing vapour compression has been known
1
The patent literature was reviewed by the Patents Department
T.IS.O. Appendix 1 lists patents in the United States, Great
Britain, Germany, France, Switzerland, the Netherlands up to
1951 in the class 85 b 1 processes for the purification of
water for industrial use and of drinking water f distillation
processes excepted.
2
Appendix 2 lists the numbers of patents up to 1950 in the
class 12 a 3 a distillation processes relating to desalting
of water.
for some time, but was not generally applied until the Second World
War. The considerable saving obtained in operating costs by applying
vapour compression as compared with the outmoded' distillation methods
is clearly shown in Table II.
At present it is Possible to produce 175 to 200 kg of water
with the aid of 1 kg of fuel by means of this pxooess_ 1, 8).
Leicester 8 calculated that the maximum water-to-fuel ratio
270 0 1.
Accordingto Latham 10. it thus appears very unlikely that
cost of distillation can be reduced to:less than half the
cost if the cost of fuel remains about the same.
Systems which in one way or another make use of solar energy
are of course not restricted to this ratio of 270.; 1. In the
3
, per day'ie,prepared through
Virgin-islands for instance, 3.8 m
distillation with solar energy see also 9
Power and depreciation are the predominant.itoins in the total
production cost of a vapour compression still Graph 1 Dior this
reason distillation is attractive in places where fuel can be ob
tained at very low cost oil wellsf ; natural gas deposits . The
biggest sea water distillation installation is that built by the
.
Kuwait Oil Co. in the Persian gulf capacity 2600 m3/day). This
installation is associated with oil development and presumably
uses refinery wastes, for fuel (2).
The distillation of sea water is also applied extensively in
Aruba and Curacao (ii).
f
The size of the vapour compression distilling plant, compares
very favourably withthat of the conventional evaporator. The
'
overall size of atypical commercial unit, designed for an output
.
of 25f400 kg water /day is 2,7 x 1,7 x 1.8 in weight 5600 kg).
Therefore the compression still can also be used instead of
stored water on diesel and Petrol' engine -propelled ships, which
,
cargo or fuel with a corresponding
will release valuable space for increase the'ship's cruising radius.
in d.
As regards the use of atomic energy for distillation Dubrige.
 12 says that thirty to fifty Years will elapse before uranium can
become a major source of
possibly ~ power and that this power will
certainly cost much more than power from coal. It will be advisable
to watch developments in this respect.
b. High-pressure distillation.
In 1950 Von Platen 13 developed a process for separating
dissolved salts from their solvent with a very small amount of
energy, which is especially suited for the production of drinking
water from sea water.
.If sea water. is' subjected to distillation at a pressure far
above the critical value for water9 say 300 to 360 kgcm2, a
modest-sized heat exchanger will do in transmitting practically
all 'the heat of condensation to the sea . water feed, thus giving
a low fuel ratio.
This process can also be used for preparing drinking water on,
board ocean-going ships. No cost evaluation is given. Von Platen
shows thermodynamically
that it does make a considerable difference
whether one works at or well above critical pressure.
2. Desalting _b freezing out.
With freezing out the quantity of "heat" used is less than with
distillation, on the other hand, however, "heat-calories" are cheaper
than "cold-calories, and therefore these processes should again be
compared for each location.
There are three methods of freezing out sea water of. which Stein-
bach gives a review without, however, stating the cost 14 Salt
.
concentrations of 0.14 per cent can be attained.
Aultman 1 points out that freezing out sea water would probably
Desalting b ion-exchange.
The 't
Th mixed bed's demineralisation-process was originally suggested
by Akeroyd and Ktessmann 15 for desalting sea water in case of
emergency, for instance in wartime. "Mixed
bed" demineralisation units
were used in fairly large quantities during the latter part of World
1
War II b the R.A.F. Furthermore y , they were occasionally used_in sea-
going vessels.
However, an expense which is permissible in wartime is not usually
accepted in peace time. The unit cost of the process is directly pro-
portional to the salinity of the water being demineralised and is rather
high in the case of sea
water. In South California the chemicals cost
only is estimated to be Fl. 25.-/m3).
As it requires twenty to thirty times the amount of demineralised
water produced just w
~ to ash the regenerating acid an alkali fromthe
demineralising material, this process is
unattractive from
an economical
point.of view. According to Aultman 1 and Moore 99 the economic
limit of the present process is reached when the raw water contains
2000-2500 mg/1 total salines less than one-tenth the salt content
sP water. .. According to Showell cost o io exchange
a (16) the of n -is Un-
economical compared to distillation for concentrations above 1500-2000
mg/1 solids.
1
4. Electrolytic desalting .
This paragraph deals with the systems which use electric power
for desalting sea water.
a. Electrolysis of sea water utilizing one 'diaphragm.
Electrolysis of sea water using one diaphragm is in use for
production of caustic soda and or chlorine, sterilization of sea
waters but not for desalting purposes. It will be quite obvious
that the ode and 'corrosio
an an the diaphragm are very liable to n
cco d' to A r According tInnoue 17 in the production of caustic soda a.
moulded anode made of a mixture of lead dust and powdered graphite
in 7.5 - 15:1 ratio appeared to resist corrosion the 'best. A tYPica
process for producing caustic soda from sea water owes 'its effect-
iveness to the use of asbestos diaPhrans (18). Nishida and co-
1
Appendices 3 4 and 5 respectively list patents up to 1950 in
the classes 12 h y electro-osmosis , 12 h 3 (diaphragms for
electrolytic purposes in general), and in 12.d 1 d (clarifying
and separating liquids by electrical action and 12 h 1 elec-
trolytic processes and equipment in general).
 workers 19 use diaphragms of ion exchange resin i,e, melamine
resin to prevent the moving of hydroxyl ions produced around the
cathode towards the anodic compartment.
Electrolysis of sea water is also employed for its bacteriolo-
gical purification as such electrolytic chlorination can kill all
living organisms and is more satisfactory in this respect than dosing
with hypochlorite or chlorite solutions (21). A Gloucester fish plant
employs a simple electrolytic unit to generate chlorine in harboir
water. After dilution 120 mg/1 of free chlorine to 15 mg 1 the
latter is then used to wash all incoming fish, floors, walls and
equipment t20). Prime benefits area reduction in bacteria counts
on.the fish and subsequent'improvement in keeping quality of the
fillets, decrease in objectionable "fishy" odours about the plant,
?
and elimination of, slime from the various working surfaces.
Carbon or graphite-amalgam electrodes are used = 1.9;
2. 2
0 = 45 dm ? I = V = . d = + 1 ma cm ? energy consum tion
9 75'~ 9 75, _ 7 , p 3
1
kWh wh ch must be replaced about once a year. Cost of replace-
ment is approximately F1. 247. per electrode. At a capacity of
the entire unit of 11 m3 water of 120 mg free chlorine per litre
per hour the power cost, at Fl. 0.075 per kVth, is Fl. 0.225 per
hour. A portable chlorine generatingunit has been designed for
processing fresh water or water with a low salt content, especially
suitable for use in fruit and vegetable canneries.
?A computation of power for these electrolytic processes with
one diaphragm is found only in Aultman's.artiole 1 .. If the elec-
trolytic process developed by Briggs 22, 23 which is used for
boiler water treatment, were to be used for treating sea water,
the cost for power alone would be F1. 0.90 m3 at Fl.. 0.02 per kWh.
Furthermore this method would require a water waste of four times
the recovery.
b. Electrodial sis.
E1ec
trodialysis has been known for some time and9 as appears
from the literature in this field, has been fairly extensively
investigated, at least as regards its application on a laboratory
scale.
E
lectrodialysis should be looked upon as a combination of
dialysis and electrolysis: in ctrolysis? ~.n electrodialysis the diffusion of
ions through the membrane is
influenced in part by the electric
field'; colloid
, al solutions for instance can be freed of dissolved
salts more
quickly with the aid of alectrodialysis than of ordinary
dialysis.
E1ectrodialysis was first applied on an industrial scale about
1925 in thee
leotrodxalytic production of "distilled" water see
sectio s I
n II.D and C), while in 1951 industrial electrodialYtic
desalting of milk-whey materialised.
A method for the technical demineralisation of sea ureter by
electrode
alysis with new synthetic membranes the so-called
Permionic membranes,
has been announced by the Tonics Incorporated
3
24, 25, 26). 1.5 m3 sea water yields 1 m purified water and 0,
5
m3 brine.
At a comparatively lore flow rate of the sea water the energy
'
consumption can be as low as 5.2 kWh zn3. As to the cost of the
equipment: the ~~
apparatus has no moving parts other than the water
stream; water pumping costs are negligible; the process is con-
tinuous and uses no heat or chemicals. Therefore at a kCm rate of
Fl. 0.04 the total cost is estimated at F1. 0,30 to F1. 0.35 per
3
m , which is one half to'one third of that required by the most
economical process now commercially available: the vapour com-
pression destillation. .
Since the membranes are quickly affected in electrodialYsis
of water containing
chloride ions they will have to be replaced
regularly. This cost of replacement may be considerable and has
very probably been left out of account in calculating the so-called
"total " cost and therefore this total cost is probably nothing
but the operating cost.
 From the summary given by Aultman of present known methods of
sea water desalting he concludes that for the time being the e cost
will be
at least twelve to more than five hundred times as much as
the average for existing supplies. plies. According to
g p~ ng author from an
engineering standpoint there is - within the foreseeable future and
in the light of current technical knowledge - no question of sea
water being considered as a source of domestic, industrial or agri-
cultural water along either coast of the U.S.A. or of local water
supplies in general being superseded bYthe ocean. He would rather,
make every effort to improve and develop existing local supplies:
treating of sewage and industrial wastes, better rainfall collection.
At places where supplies of potable water are not available and
water is vitally needed now, a choice should however be made of the
reviewed, desalting methods of either sea water or brackish water.
Research into these methods should therefote be enoodraged. Which
method is most economical depends upon the salinity of the water
to be treated and the form of energy available.
B. The desalting of brackish waters.
Water is brackish when it has been contaminated with moderate salt
concentrations (1/10 to 1/5 that of sea water by intrusion of sea
water or oil brines into fresh well water or by percolation of fresh
water through rocks and soils containing soluble minerals. According
to Krul and Liefrinck 6 the criteria for fresh brackish and salt
water are:
fresh crater up to 100 mg C1 /1
brackish rater 100 - 1000 mg C1 71
salt water over 1000 mg C1 /1
Brackish waters are now widely found in the Western States of the
U.S.A., Bermuda, The Hawaiian Islands, the Bahamas, Cuba in certain
regions of Europe and in many desert areas of Asia Africa and the
Middle East.
In principle all the methods mentioned in A can be used for de-
salting brackish water.
Distillation.
Because of the lower concentration of solids, process costs
will be somewhat lower than those for seawater section II.A.1 .
Ion-exchange.
Owing to the lower salinity of the starting liquid, ion ex-
change is more favourable '
in this case. Some examples on this sub-
ject now follow.
From the
results of extensive. research into ion exchange for
water treatment by Showell 16 it is clear that for reducing a
solids content of water of 250 mg/1 demineralisation is cheaper
.
than vapour compression distillation. Typical values given for
operation cost of a 1000 1 /mm. capacity
compression still, ordinary
still and demineralisation plant are F1. 1.48 1.81 and 0.44 per m3
respectively. . In comparison with these prices chemical softening
is so far less expensive 16.
According to Juda the cost of the Ionics partial demineri-
alisation process for desalting brackish water from + 1000 mg Cl 1
_ ~
to + 300 mg C1 1 is F1. 0.14 m3 1. In this process regeneration
5 is carried out
with sulphuric acid and lime the prices of which
are F1. 10.-
and Fl. 5.- per 100 kg respectively.
Aultman 1 quotes a cost1of Fl. 0.26 for the 5 m3 production
of water equal in to disci
quality distilled water from raw water con-
taining about 370 mg/1 total dissolved solids.
At the central power station at Villers_Saint Paul FF.
90,000 m3 water per hour from the Oise is demineralised by cation-
anion-exchange (27).
At present research into the possibility of desalting water
in Holland by means of ion exchangers still carried out b
by the
Plastics Research Institute T.N.O.
3. Electrodial sis.
As electrode
alytic desalting of water is the subject of this
1
Ito further details of this cost are given and it is not
known
whether it is the total cost or cost of chemicals onl.
Y
 report, the electrodialytic desalting experiments already made with
brackish waters,will be discussed in detail in a separate section
see section III).
4. Electrolysis.
In a two-compartment cell, using one diaphragm, Briggs 22, 23
produces soft.water at lower power cost and with as lower percentage
of waste water than has formerly been possible with the three-
com-partment cell used by the Siemens and Halske A.G.
The energy consumption for decreasing the solids content of
III. ELECTRODIALYTIC RELaOVhI, OF SALTS FROM BRACKISH
AND STEuILAR ~lA_TERS.
At the cathode 1Ye,and hydrogen are formed, ; the hydroxyl
ions then move in the direction of the anode. At the anode acid
and oxygen are formed. The odic
an hydrogen ions are responsible
A. Introduction.
The principle of desalting by means of e1ectrodi
alysis is
that cations and anions
are removed from the dialysate via two
or more membranes under the influence of an electrical field.
for a decline in the pH of the anolyta and move' towards the Ca-
thode . The phenomena occurring in the presence of halogens are
extensively dealt with in section III F.
Hence when two not fully selective membranes a i
, y ar used, in
the absence of stirring the pH o
g p of the dialysate will show a decline.
from high at the cathodic membrane to low at the odic
th an membrane.
Intensive stirring will cause the hydrogen ions and the hydroxyl
ions to combine into water
molecules so that the pH will remain
constant9 if as many hydrogen ions as hydroxyl ions enter into the di
alysate This is not the case, however. If membranes are used
which are equally permeable for. both the aforementioned ions more
ions than hydroxyl io wi
hydrogen ions will enter the dialysate, since
the mobility o
~ f the hydrogen ion exceeds that of the hydroxyl ion.
This is evident from Table III which shows the mobilities of dif-
ferent ions at 250C. In the absence of special measures the dialy-
sateistherefore
diluted somewhat and its pH falls. Aocorthng to
Billitex (29) the dilution effect, when water is desalted, is ve
: ~ rr
slight, viz., 0.03,~ m.
As variation in the pH is p undesirable, particularly when col-
loidal solutions are desalted denaturation efforts have been made
to obviate this by the use of selective membranes, and much re-
search has been performed in this field.
At the same time it is obvious that when less-selective mem-
branes are used large proportion according to Billiter: 80
~ N
of the current txansPort is supplied bY by the hydroxyl ions and b
., . Y
 the hydrogen ions passing towards the anode or the cathode.
Not until
1 did r
939 b~anegold 30 clearly establish the correlation between the
use of selective
membranes and the efficiency of the desalting process.
By means of calculations he proves that all the current transport is
supplied b the ior~s to be ~
Y removed from the dia' ~.ysate if
. membranes are
used in which the c
, barge efficiency definition see section V B of
the anion or
cation to be concerned is 1004 see Fig. 2b). In this
~
case both the and the catholYto may be concentrated electro-
lytic solutions
, as a result of which the energy consumption will be
substantially reduced. It into be expected that the energy cost of
electrodialytic desalting wills become lower if the charge efficiencies
of the
salt. ions to be removed are as high as possible in the mem-
brane(s) in question.
..
If in the aforementioned case the membranes are interchanged
,
then the electrolyte will accumulate in the middle cell see Fig.2c .
The result of the use of a number of membranes is shown diagram-
matically in Fig. 3 The electrolyte becomes concentrated in alternate
cells.
Aten 31 has calculated that, if all compartments of the three-
compartment electrodial sis cell are filled with the same solution
desalting of the di
alysate is possible by using either selective mem-
branes or two identical membranes. In the latter case the degree of
,.
acidity or
alkalinity of the ?.nolyte or the catholYte should be of
a specific value. The
maximum "current efficiency's attainable see
section V F is then only p 20%, whereas in the other case
this maximum
value depends
upon the selectivity of the membrane and may be higher.
For further details and the history of electrodialytic desalting
we may refer to the publications of Stamber er Stauffer (32), 33
and of Pr '
ausnitz and Reitstdtter 34
In the next
pages we will give a survey of some important elec-
trodialytical water desalting experiments known from literature. Next
the above mentioned theoretical computations by Aten will be reviewed
,
whereafter a brief survey of thEelectrode processes occurring at the
electrodes during electrodialytic desalting of.water containing
chloride ions will be
given.
-29-
B. The prinoi al patents relating to electrodialytic
desalting of water
The use of electrodialysis for water purification is a special
feature
of the patents of the Elektro Osmose Cesellschaft'and of the
Siemens & Halske A.G. (35). However, direct conclusions regarding
the
problems related to the desalting of sea water and water with 1000
mg
cl 1 are not given. On the other hand they do contain some data.
which might provide a starting point.
An example. i
s given of the de-
salting of water from 3 m C1 1 to 10 mg C
5 g 1 1 in the..normal -three-
compartment cell, with an energy consumpption of 13.2 kWh per in3. In
addition there are other data which. relate to the elimination of bi-
carbonate etc. Briefly this procedure comes to the addition of the
spent rinsing liquids to the water to be desalted, whereupon owing
to their alkalinity they will reci itate the bicarbonate in the form
of carbonate.
An important feature of all these atents is t
p he aim of limiting
the distance between the electrodes. Attention is also paid to the pH
regulation. Fo
For this purpose systems are indicated which involve mixing
of the cathodic and anodic rinsing liquids.
A Dutch patent, (of 36 in the name of J. Billiter, is also of
importance in this investigation. He suggests to reduce the back-flow
of water of a low salt content towards compartments with water of a
high salt content by maintaining a hydrostatic pressure difference
and by means of a correct choice of diaphragms. In the case of porous
.
clay diaphragms,
where the osmotic transport
of water through the
membranes is about one hundred times as much as the H + OH --) H20
see . 27) it i clear quantity that this transport can be quite
considerable.
~7hen desalting liquids. of a .high salt content. is.concerned and
untreated water is used as 'rinsing liquid.a substantial difference in
.
concentration?vra,ll arise .between: dalysate and rinsing liquid..
The patents.referred to here.are also mentioned in appendices,
3, 4 and :5. , .
 -30-
Consequently: .
a. a higher hydrostatic pressure difference is necessary
to prevent undesirable water
transport.
ID. a higher current density is necessary suppress
undesirable dialYtic salt transport.
In many cases he
maintenance of a very large pressure difference
is needed, for instance by working with nearly empty y y electrode com-
partments The electrodes are then placed right
against the diaphragm.
Asbestos
and ceramics are eu ested as gg a suitable materials for the
diaphrams , since they are chlorine-resistant, The results of some
of Billiter's experiments are
listed in Table V. The Siemens d. Halske
A.G. later `on also suggests the application of a pressure difference
D. R . P.498 ,048 cf
(35)).
Brief survey of the literature on the electrodial tic
water desaltin .
A survey of the demineralis
anon of water by electrodial sis '
y is
given by hug Y g 37 and by Prausnitz and Rei
tst~tter (34). Like in
the case of sea water. germicidal
properties are attributed to electro-
lytical chlorination e.. .
g (31), (34)), whereas according to Prausnitz
and? Reitst,Stter 34 .water c
. an be more thoroughly purified by means
of electrodial sis
y than by distillation. A review is given of the
investigations of Siemens
, Billiter, Zhukov and Juda
as published
in the literature..
The investigations of Siemens and Halske A.
. G.
'The electrode
alytic elimination of salts
from water in the
three-compartment cell has been developed on a laboratory scale
.
and subsequently on a technic
Y al scale mainly by the Siemens
Y and
Halske A.G. of Illi ?
g 38 and Gerth
(39)). Apparatuses were
constructed with a capacity of 20-5000
Y 1 /day for the preparation
of various qualities of softened water. The water to be desalted
flowed through consecutive middle compartments, rinsing being
done with untreated water. For the 0
preparation f highly purified
-31-
water the last cells had to be rinsed with distilled water. -
. Various membranes were used. Vegetable fibre ~ '
(e.g. 'Kuttertuch
proved satisfactory as material for the cathodic membrane and animal
fibre for to ,
.e anodic membrane e.g. speciall tre ea leather, y at Vul-
kanfiber" ,?
or wool with chrome gelatins). Experiments
have also been
performed with microporous rubber. However, various difficulties a-
rose, owing to corrosion of the anodic memb
rane as soon as the water
contained chloride ions e.g, (31)),
The apparatuses of Siemens h ve'
a also been described by many other
investigators, viz. in Germany (40-44), Y 4 44 , France (45, 46,47)), Britain
5, ~ 47 ~ 48, 49, 50) and in the Netherlands where (5f), Aten 31) made. a
special study of the electrodialytic purification of water from the
river Vecht with the aid of Siemens'
apparatus.. Ceneially speaking
these articles contain no data other than those already referred to
the publications of Illi 8
g 3 and Gerth (39). Bartow 50carried
out research into various conditions and observed that time and
energy are lost when switching on, as it takes one hour before suitable
water is obtained.
The energy consumption depends upon the salt content of t
? he
water and the degree of purity desired. According to 'Beh
rmann - 49
the process is no longer
economical if the total solids content is
higher than 1000 mg/l. 'l
ater for breweries which has to
be only part_
` ly purified, requires 10 kWh m 3 (42), whereas otherwise in the litera-
ture values are found varying between 15 and 50 kWh m Sarrot
45
arrives at an energy consumption of 20- 3
gY 25 kl~h m for a.desalting from
250-300 mg solids 1 to 0 m l and r
/ remarks t
g that this figure can be
considerably reduced if a total solids content of 30=40 mg/1 is
permissible,
The flow rate of the dialysate can be varied only within y thin-certain
limits Bartow
(41). 48 reports a rate of 24, Ate
4, noneof3.61h.
Vlith Bartow the flow rate of the rinsing water was between the limits
of 12
and 49 1/h, whilst in Aten's case it was i1 h in the first
eight and 0.12 1/h in the last two cells.
The amount of rinsing
water to be used varies considerably for the
y different investigators,
viz. from - ?~
1~ times the production to 3 - 4 times
the production.
 _32_
The current density applied b Ate 2
Y n was 1 ma cm .
Patin
(46) fills the apparatus with four litres of water
and
p
one hour d ofaer
etermines the water and energy consumption,
the acidity of
anodic and the alkalinity of the cathodic liquid
, the conductivit
the pH dro ed Y9
pp to 6). ana the temperature increased by ?
2 _ 3C.
2. Bi .
llitt
er s method.
B.
_ illiter regards the desalting process during the elect
rodialysis
'in a three-com
partment cell with:
neutral membranes as
a substitution
of the salt ions
by hydrogen ions
and hydrox1 ion
3' s . According to hi
.:....the fewer foreign cations are Contained in the anol to b
i y eside hydrogen
ona, and the fewe r foreign anions
,. are contained in the catholyte beside
hydroxyl ..ions the
, better the -- "current yield t see section V.F will be.
It attains a maximal value when
the anol to is
Y a pure acid and the ca
-
tholyte a pure alkali. However
, in that case the formation of acid in
the dialysate is
likewise maximal.
These claims are
confirmed by the results of desalting experiments
of various natural as well as artificially salted vraters
salt , ranging in
content from 0 to
5 .5000 m 1
g (29). In these ex
perlments the middle
compartment and the cathodic compartment
Contained the non-treated water
and the anodic co '
mpartment a diluted
acid, viz. 0.001
n sulphuric acid.
The cathodic diaphragms were usually made of closely woven cotton
filter cloths
, except when the water
contained relatively high Con-
centrations of alkali salts and experiments extended over
the a long period
n asbestos dia hra '
p gms being used instead, The anodic diaphragms
conaisted of porous clay lay cells with
a wall thickness of
5 mm which
possessed a small negative charge.
The potentials
applied were
440, 2.0 110, ~ , 64 and 12 vol
the c is ? During
nurse of the e g
lectrodia] sis
Y , which was continued day and ni ht
without interru g
ption, the acid concentration of the ano
lyte graduall
increased, more r y
apidly when the untreated water contained sul hates..
In. man cases the ac ~ i~~' A
d co
ncentration rose to 0.1 n. However he c
not obse ~ , could
rve any effect on the "current efficiency" within this range
of acidity, provided there was
no defect in the di
it aPhragm. He found
a current efficiency" of 16-18% for t
he complete des
. siting of eaters
-33-
in the middle compartment, with salt contents of 160, 341 and 642 mg 1.
:
For details of his experiments and the apparatus we refer to the
article of Billiter itself 29
The principal conclusions of his investigations are:
a. the liquid transported by electro-osmosis through the
anodic diaphragm is pure or almost pure water.
b. natural water can be almost completely desalted even
when a relativ'dy concentrated acid solution is used
as anolyte.
C. no desalting of the water in the middle compartment
takes place if the anolyte and catholYta are maintained
neutral.
d. the sign or size of the static potential of the diaphragm
charge has no noticeable effect on the output of the cells,
IT) other words; a greater variety of anodic membrane
material!).
. According to Billiter: therefore, the flushing of the two outer
compartments, as has been done by the Siemens and Halske A.G. is
unnecessary providing perfect diaphragms are employed. Even omitting
the flushing of one of the cell compartments means a decided eimPli-
fication in cell design. The ever&Y consumption for complete desalting
equals 0.0023 V..s. kV7h m3 V = average voltage of the cells in volts;
s
S = salt content of the untreated water ip mg/i). By application of
this equation to our desalting range 1000 mg Cl- l - 300 mg Cl r
1
it would be approx. V x 3.8 kWh m3.
For the treatment of waters with a high salt content the operating
costs of his cells can be appreciably reduced if the untreated water
is first given a chemical treatment with zeolites before being fed
into the cells. Thus, for example the salt content of the water can
be reduced from approx. 30,000 m 1 down to m 1 before feeding
400-500 into the electrodialysis cells.
3. The investigations of Zhukov.
Z uko
h v 52 purifies water from the river Neva by electxodialYsis
 in a three-compartment cell as used b Billiter, a vr'
~ Y , ih two
unequal negatively charged diaphragms. The latter are
characterized
by the following porosity 9 mean pore diametre, transference
number
of the chloride ion in 0.01 n potassium chloride, electrokinetic
potential:
anodic (gro
g, 12000) Q.,.B9 4.08 .c 0.504 nd 15.1 m v
cathodic c ?
lay,800 0.35> 0.04 > 0.341 and 7.3 m v
2
At a current density of 1-5 ma cm and a flow rate of 3 1/h
the energy consumption to obtain water with a dry residue of
10-12 m 1 and 3
g an an ignition residue of 6-8 mg/]. is 12-14 k~'~'h m .
The salt content or the non-treated water is not mentioned. After
pre-filtering through sand 5 cm thick it was possible to in-
crease the flow rate to 5 1/h., as a result of which the energy
consumption fell to 8 k\Jh m3.
The anodic diaphragm is regenerated by a 1% solution of caustic
0
soda at 50 C, preferably after passing 350-400 1 of water. With via-
ter pre-filtered through sand and active coal the lifetime of the
anodic diaphragm is two to three times longer. The cathodic diaphragm
requires only scraping in of an 0.5 mm thin peptized layer after pro-
longed use.
In the same apparatus he has determined the "current effi-
ciency" for the calcium ion and the sulphate ion in a 0.01 n cal-
cium sulphate solution see section V.F .
Zhuko'v 52 stresses the fact that when selecting the rinsing
rate of the catholYto one should allow for
the fact that at higher
lye concentrations there may arise not only an increase in energy
consu~ption but also changes in the transference numbers of the
ions in the cathodic
membrane. In one experiment the transference
number of the chloride ion dropped from the initial value of 0.343
in 0.01 n KC1 to 0.27.
4. The use of Permionic membranes.
The new Permionic membranes (24), used for the desalting of
sea water can also be used in
the purification of many common
brackish graters with a salt co
. ntent of 3600 mg/]. and lower. Juda
(24) states
that in these cases the energy costs will amount to
Fl. 0.01 perm3, for 'a kWh-price of 1 Dutch cent. Energy consumption
is therefore approx. 1 kWh m3. However he did not take into ake rote account
the renewing costs of the membranes sea remark section II A.4.b .
D. Review of the theoretical computations b Ate _ n in the electro-
_
dial tic desalting
Y of water in a three-compartment cell
Aten 31 considers the situation in a three-compartment electro-
.
dialysis cell as used by the Siemens & Halske A.G.
all compartments
being continuously fed with raw water.
Neglecting.
the diffusion and the transport of water through the
membranes he arrives at the following conclusions ~ ng by means of .cal-
..
culation:
1. If the mobilities of the anions and the oatio s '
n in both diaphragms
are equal or proportional to the mobilities of these ions in the
free solution, desalting is possible only if the anidoo
and the
cathodic rinsing liquids possess the correct degree of acidity
and alkalinity.
2. Desalting of water with neutral raw water fl
ovring into all three
compartments can be
achieved only by the 'correct use of selective
membranes.
1. De
salon with non-selective membranes.
If two identical membranes are situated between the three
compartments filled with salt MZ ' containing water, the middle
Compartment
can nevertheless be desalted if by cexefullY regulating
the rinsing rates one ensures that the anolYte'and the catholyto
possess the right degree of acidity or alkalinity and the middle
Compartment remai
ns .neutral.
 Neglecting the number of hydrogen ions and hydroxyl ions in 'era-
ter the Ilstationary state" 1 :
may be represented as follows
Catholyte (3)
R (co)
When the assumption is made that u' = v' u" =
? 5 u1 and
v" = 3 u', and in addition the factors m and
p are introduced2 of
which m = c' c and = c3 c3 the
1 1 P , following equations can be
.
derived:
The reduction of the amount of the salt MZ in the stationary state
is given by:
5- geqjcm.sec
P is negative, i.e. the salinity in the middle compartment decreases,
if p is 1, in other words if the catholYto is alkaline.
The
quantity ,.
indicates the amount of salt which during the passage of 1 Faraday
.
is removed from the middle comp.rtment. The maximum value is 0.2
when p is co and so the maximum "current efficiency" is 20%
(of section V F).
1
in parenthesis the ion concentrations are given in geq per cm .
In section IV F we the des
qualify along effect by the quantity
Y
q: the number of ammes of sodium chloride removed from the middle.
compartment per 100 ampere-hours. In connection with the above
equation q cannot
exceed the value of 0.2 x 218.4 _ 43.6.
From the equation for P it is evident that the
absolute concen-
trations of acid and alkali in the 'electrode compartments and the
salt concentration in the middle compartment exert no
influence, ;but
that desalting depends exclusively on the relative alkalinity
acidity of the catholyte and theanolYto, viz.
catholyte and c0 the feed concentration then R
o and S are given: by
av e to be given a certain flow rate. If.A represents
the flow x
in cm3 per sec. per.cm2 diaphragm of the anol e
yt , B that of ,the:..
In order to obtain the desired ratio _
. the rinsing liquids
alkalinity
h
the equations:
1' 2
co (4P_ 2 P-1
The only value to be chosen at random is the acidity of the anolyte
or the
alkalinity of the
~ catholYte. For different values of p, the
+ 9
values for m, q, c , c and the 11ourrent efficiency" c
an be found
3
in the following table:
P
m
efficiency?
g Nai7~lpp
amp.hr.
l c1'
j co
co
'
I co
c_"_
co
F.c
S. d ?
F.c
R. a o
2
3
4
5
6 [
1.5
1.86
2.12
2.33
2.50
12.5
15.4
16.7
17.4
17.8
27.6
33.6 ~
36.6 ~
37.8 i
3
( 0.83
0.78
0.77
0.72 i
~
0.75
0.67
0.63
0.60
0.58
0.42
0.67
0.83
0.95
1.05
0.75
1.34
1.87
2.40
2.90
1.2 ~
0.60
0.43
0.35
0.30
0.66
0.30
0.19
0.14
0?ti
maximum possible value 20% ` maxi
um m possible value 43.6
 -38-
-39-
Hence for the electrodial 2. Desaltin with the use of selective membranes.
yta.c desalting of water contai
i
16
0
n
ng
5
m sodium ;chloride per litre _ _ 1000 m C17- When the three compartments, filled with salt b~Z containing wa-
_g 1 with the use of a mem-
brane of 500 cm2 area and of 2 ter, are divided by two diaphragms one of which -e.g. the cathodic-
a current density of ma cm -
3 the fol-
is selective and the current has passed through for some time the salt
lowing combinations and flow rates of,the rinsing liquids for the
"stationary state can be calculated for GGl;ucl,ulaoi~i; will drop in the membrane from c ge 9,/cm `in the cathodic
p-values of 2, 4 and 6: 3
3
to c2geq cm. in the dialysate compartment. ?
Anolirt
e
Catholyte
q
composition
flow rate
lh
composition
flow rate
g IJaCl 10Q
amp.hr ~ athodic compartment Diaphragm.. DialYsate comPartn;ent
I ~
lh
-.--
,
IJaCl 0.025 n +
HC1 0.013 n
NaCl 0.023 n +
NaOH 0.023 n
1.2 ~
27.6 j
.NaCl 0.022 n +
0
8
IdaCl 0.019 n +
I
--
" ,
"
HC1 0.025 n
idaCl 0.021 n +
HC1 0.032 n
.
0.55
NaOH 0.056 n
IdaCl 0.01 n +
NaOH 0.08 n
0.35
0.2
36.6
39.0
-~z
- v
i
B
In the calculations the di
ffusion and water transport
through the
membranes have
not been taken into account.
Yet a certain amount of
acid from '
m the anodic compartment and
alkali from the cathodic compart-
ment will diffuse towards
the dialysate compartment, depending upon
the nature of the membrane or the diaphragm and the acid
and alkali
concentrations. This will determine whether it is more economical to
work with
larger or smaller values of
p and m. When p increases the
acid and lye concentrations increase more than q (of the tables). In
the case of slight diffusion it is better to work with a
large value
of in the case of considerable diffusion with a small value of
p.
If in practice the dialysate compartment becomes acid or alkaline
one can slightly reduce the degree of acidity or alkalinity, or in-
crease the flow rates. In an case '
any it will be necessary to investigate
by means of, experiments, whether the stationary conditions mentioned
above and the q values of 28, and
37 39 can indeed be realised.
When A represents a random cross section of the dialysate compartment
parallel to the diaphragm, the decrease of the salt concentration in
the space between A en B amounts to.
P=-2
R c
T ,3 2-
.n ?n ,u
m
2
geq cm .sec.
n = transference number in the membrane
m
In this equation P is maximal if n = 0 m i.e. if in the experimental
conditions a 100% cation-permeable membrane is used.
t In that case
PF -
then equals n .
d max.
The current density must exceed the minimum value
-d =2RT
min
n -n
m
P = Oi 4 . 0):
 - 40 -
This `minimum current density is maximal when c = 0, viz
-d. =2Rt .
min
Since _ ' _ increases rapidly by an increase of n , it is advisable
m
n -
_
nm -
to select 'a diaphragm _ with a small n value. At a certain value of n
m m
the necessary current density is lower at greater thickness of the
+ -
diaphragm and 'a lower mobility of the ti -ion. For small values of n
m
say
0.1) desalting is not greatly influenced by the concentration,
whether the current density be great or small.
If a number of electrodialy,sis cells is placed in series the
concentration in the first cell will drop from c3 to c2 , in the second
from 02 to c1, etc, the concentration of the rinsing water is c
2 3
throughout). A calculation reveals that -as to the current yield-
an
apparatus works most economically if the current strength is adjusted
to a value rendering zero concentration in the last cell . In this case
all cells contribute equally well to the Purification.
.` If S represents the amount of water to be desalted passing through
the middle compartment o e o e e
of of 6 1 ctrodialytic cells' expressed in
em3 per second and per 2 cm diaphragm, he finds the following expression
or he
current f t ;density required '
c S
c
_ 4? _ .3
S .F I 2
,, y..
u~3,
wl
. amp. cm
,.m 1
When as little energy as possible is used for a given quantity of
electrolYto,, the. following ratio sho Ci.d be minimal:
the loss of electric energy per cm2 diaphra~n area,
the electric resistance o?f the diaPhr expressed in
~-,
ohms per cm2. 1
Apparently Aten neglects the resistance of the solutions in
the respective compartments.
-41-
In that case for one diaphragm, say the cathodic diaphragm, the optimum
desalting
velocity is.
2 R? "3 "2
P = - 2 ? n- u e om2.sec
gq
opt. F j m
and the corresponding current density:
-
a
3
R
_ -4R
rt . _
. Hence it is possible to effect desalting with the use of only
_
one selective diaPhran.
Of course it is more economical to use as well a selecting oa-
thodic as a selective anodic diaPhran. When two equally thick dia-
..
hra s are used and the concentrations P ~ in the cathodic.. and anodic.
compartments are the same, viz. c3 , the decrease of the salt con
.
centration is represented b :
by the following equation.
C -c _ 1
2 R'C 2 -m d 2
F ? . uc . n cm - u . n c n - n goq cm .sec.
m am F am am
and q
n -n
cm am
?218.4 g NaC1/100 amp.hr
As v for a cathodic and u for an anodid diaphragm have to be,
small n - n will be approximately twice as large as n - .
n
cm am m
The term for d. is roughly the same as with ons diaphr~
TAln ~ . .
When desalting is?performed.with a greater current densitYsthen:
d.
P = d - d aA am min
. ---'and q = ----
min F d
218.4
As compared with' the desalting using only one selective dia-
phragm n cm - riamis approximately twice as large, .but d , remains.
1 n
-
and nam are the 2 transference numbers of the anions in the
cm
cathodic
membrane and anodic membrane rasp.
m
. u .:::---- amp./cm
opt.
~ -
II,
 nearly the samea With the use of one cation- and on
. e anion-permeable
diaphragm
the degree of desalting therefore will be ne
arly doubled.
E. The ex eriments of.. Hoffmann. '
In view of the character of our ex eriment
p al research the ex-
periments of Hoffmann are briefly reviewed here.
Hoffmann 53 extensively studied the electrode
Y alytic desalting
of a 0.88 n sodium sul hat.e solution s d
P , as ura.ng electrodialysis of
uotassium chloride solutions the anodic di
aphragm was severely corroded.
Use was made of identical diaphragms of the "Makot ch'
u type which
after two or three day's
use were found to have acquired a very limited
permeability to water. .
The dialysate 1$00 cc was Vidro
g usly stirred, the circulating
rinsing liquids (totalling 4000 cc distilled Water in the is
anod and
c
3250 e in the cathodic com artment grew acid resp.
p alkaline during
desalting; the volumes of the three liquids were kept to the mark by
replenishing; the distance between the two diaphragms was 18 mm,
that between the electrodes 40 mm; the area of the diaphragms was
2- 2
10 dm 9 that of the electrodes 8 dm .
Electro
dialyses was carried out with current densities of
4, 8,
12r 16 and 24 ma cm2 an d at temperatures o
. of 200' 400 and 600C? In
each test a total of 64 amp. hr was passed through,
the duration of
the experiments therefore varying with the current.
. density and amoun-
ting to ?16, 89 5 1/3, 4 and 2 2/3 hours.
The course of the sodium sulphate removal in terms of tine
current density and temperature and the course of the acidity of the
dialysate are evident from the graphs 2, 3 and 4.
The conclusions of his experiments read as follows:
.
a. desalting with two identical diaphrams is possible.
b'
b. the current yield increases
by increasing the temperature
and reducing the current density: the effect of the tem-
erature is largest when current densities are small.
C. the increase of the acidity in the middle compartment is
minimal at a a high temperature and a low current density.
Introduction.
If the cathodic co 'onta~--
mpartmen.t _.,.a~ii~ raw water during electro-
dial sis sodium hydroxide
Y and, hydrogen will be formed at the cathode.
If the anodic compartment contains untreated crater the chloride ions
discharged at the anode will partly combine into chlorine molecules
9
resulting in the formation of chlorine
gas, which '
~ch is less soluble
in water so that part of it can escape.
Part of the chlorine in molecular solution is hydrolxsed
into hydrogen and chloride ions and into hypochloric acid molecules
in accordance with the equation:
012 +H20 `- + -
...; H + Cl + HC10
In section r 2 the
l~ypochloric acid
concentrations in the
anolyto are calculated for varying conditions, whilst section F 3
deals with the reactions which may result fro the simultaneous
presence of chlorine
hypochloric acid and chloride ions in the
anolyte.
2. The hypochiorite concentration in the
ano1 te.
It has been found that during the electro
g dialysis of liquids
containing chloride ions great difficulties
arise owing to corro-
sion of the membranes (31, 34, 36, 38, . The c
9 9 9 9 53 ause of thiR
trouble must be attruwst" robabl to the p Y presence of the
hypochloric acid.
In Table IV the byPochloric acid concentrations in the anolyte
are calculated for various conditions. The computations are in the
following form:'
The equilibrium constant for the above equation is represented
as follows:
K _ _ tx+l [ ci'l [acio)
t:ci2]
The chlorine in molecular solution is in equilibrium with the
chlorine gas which is contained in the gas bubbles of the anofyte.
 - 4.4-
If the partial pressure (p) ressure rpof this gas is less than 0.6 atm.y then
Henry's law holds1 and C12= H.p. Because of the low pressure and
concentrations one may use these quantities instead of the activities.
The concentration of the hypochloric acid in the anolyte now becomes:
H. .K.
[H] [cfJ [Hi) {ci]
If in this equation one substitutes the values o ?C
of Land K at 15?C as
-3
found by Vrhitney and Vivian 54, 55), viz. 9.35 x 10 mol 100 g H 0
2
6
-
per atm. and 2.3 x 10 mol 100 g H0, it becomes:
2,
[ irci~j =
From this equation as well as from Table IV it follows that the
hypochloric acid concentration is reduced by a factor 10:
a. by reducing the partial chlorine gas pressure by a factor 10,
b. by reducing the pH by 1,
C. by increasing the chloride ion concentration by a factor 10.
As it appeared not to be difficult to maintain the partial chIc-
-4
rine pressure in the gas phase outside the cell at 10 atm. [HC1O]
was calculated for a chloride on concentration of 3 x 10 ` n Dutch
- ,
brackish water), p = 10_4atm. and pH = 2. In these circumstances the
concentration of the hypochloric acid in the rinsing liquid was found
to be 0.72 x 10-6 n; for pH = 7 it was 0.072 n. It is true that the
values found by Whitney and Vivian 54, 55 for the constants K and
H, deviate approximately 25 ~ to 100' from those of Jakovkin (56).
However, this does not alter the value of [HCl0] , the hYdrolised
fraction.
Hence the concentration of the hypochloric acid in the anolyte
may be
aY limited e.g. by raising the concentration of the hydrogen ions
and/or of the chloride ions in the anolyte. However, if the anodic
membrane used is not fully selective, an excessive concentration of
hydrogen ions in the anolyte will have to be avoided owing to the
-45-
charge efficiency of the anion to be remo
ved,
3. Reactions between ochioric
i
acid chl
l
Y
or
ne mo
ecules.
.l_._...
d
hl
id
i
an
or
e
ons in a ueous solution
c
The
presence of hypochloric acid in the anolYto, formed by the
reaction C12 + H20 ~--- HC1 +
.-...~ HC10, may give rise. to various other
reactions. However
. , in the literature there is no agreement on the
mechanism of the different reactions of aqueous solutions of chlora,c
and hypochloric acid whilst the chemical nature of the chlorine-
oxygen-compounds is likewise a point of dispute.
It i
s generally assumed that as a result of the simultaneous
presence of chlorine, hydrochloric acid ochloric acid
~ h3'p , and, sodium
chloride the following reactions may occur:
a. According to Nernst and Sand the following equi]ibrium
is established:
Then the chloric acid c
oncentratjon is ,calculated with the aid of
the equilibrium constant found by Sand 58
for -5
a partial chlorine pressure of 10 atm, and for 0.02 n hydrochloric
-1
00
101_
acid. ? a chloric acid concentration of 1 03
t 2
x
C is found
However, much criticism has been levelled at Sand e.g. 5
~ o
,
1:oreov
er it is very doubtful whether this aq uilibrim of Sand still.:
applies to the low oonceptrat. ions occurring when water is desalted..
1
See section V B.
2 ?. _
the following considerations are a taken_ from a T.N.O. internal
report by Dr C.L. d'e'Vries.
[C12] is calculated with the values of K and H as found by
?thitney and Vivian (54, 55)r viz. 2.442z 1a6(mol/106 g HpO-
d
3
sue
7.75 x i0-
(mo1/100 g H20 per'etm.
 b. The decomposition of hypoch1oriq.acid according to
2 HC10 ~-- 2 HC1 + 0
~ 2
plays a very minor role 60, 61).
C. ochloric ac'
Hyp id acts upon chlorides in accordance with the equation
which reaction is strongly dependeht on the concentration of the
hypochlorite (62).
The ions formed can now react hydroxyl in accordance with
oa + acio ~ cio
and if the concentration of C10f becomes considerable the reaction
C10 +2HC10 -.*2HT+2Cl +C10
3
takes place with measurable velocity. On the other hand at low con-
centrations C10 is formed slowly.
3
d. If the hypochloric acid is regarded as one o the intermediate of pro-
ducts of the
reaction. between photo-activated chlorine molecules and
water, it may react as an oxygen acceptor and give rise to the form-
ati.on of chloric acid, perchloric acid and hydrogen peroxide' (61)-the
formation of oxygen being unimportant (63)- e.g.
5 HC10 --+ H010 + 4 HC1 + 0
3 2
However
, lllmand and his co-workers 'deny the formation of perchloric
acid 64
l~hen chlorine water is placed in ultraviolet light, conversion
takes lace into chlo c
p ri acid which can also be considered as the
end product of the photolysis of chlorine water and GYPochloric acid
solutions 60 and of solutions of chlorine dioxide in grater (65).
Summarizing we may therefore expect that if the concentration of
the
bypuohlorio acid in the anolYto can be kept sufficiently low, and
the electrodi
alysis is not carried out-in sunlight, practically no other
chlorine oxygen compounds will be generated.
IV. ENERGY COfdSULiPTION OF EI,ECTRODIALYSIS.
A. Introduction.
exist in heat generation and in formation of by-products e.. b
, g Y
formation of hydrogen chlorine caustic soda c
, , , hypohloric acid.
The energy consumption i
in the electrodialysis cell is gfiven
by the equation:
factors of which it is composed. Energy losses during electrodialysis
Energy consumption should be kept as low as possible. For this
purpose it is necessary to become acquainted with the various
when V is the 3
volume of the dialysate in cm . If E and I are
E.I.t
V
The terminal
voltage E can be regarded as the sum of the
electrode-potentials
Eel, the voltage drops caused by the Ohmic.
resistance of liquids E, and the voltage drops
1 across the mem-
branes (E).
m
The factors influencing Eel E, and Em will be discussed in
i
the sections IV B, C and D. In section IV E a summary will be
given of the factors which might lower the terminal voltage, while
in section IV F
the quantity of charge passed through I x t will
be considered.
The determination of the desalting effect q see section IV
F and of energy consumption for desalting of
water, taken from
experimental figures, is discussed in section IV C.
 . Introduction. ,.
The occurrence o "
f a certain potential when an electrode is
immersed in a liquid and no passage of current
takes place can be
taken as w c
aknonfat.
During the passage of current a considerable overvoltage fre-
quently occurs, i.e. a higher voltage must be applied to the elec-
trodes than the theoretical equilibrium potential.
The magnitude of the sum of the electrode potentials Eelis
not a very important factor with a high terminal voltage, but with
low terminal voltages such as are applied in desalting water the
energy loss caused by this potential may be a substantial percentage
of the total energy consumption.
The overvoltage is dependent on the current strength9 the corn-
position of the electrolyte, the nature of ' t
he electrode surface and
on other factors the influence of many of them not yet being fully
known. Although overvol
tage has already been studied for some 45
Years, there is still much confusion, the results of the. various
researchers showing large differences.
The following distinctions are made: hydrogen overvoltage, 3ygen
overvoltage, chlorine overvoltage, roncgrtration and polarisation
overvolta a, etc. The expressions cathodic overvoltage or anodic over-
voltage are also in use.
The magnitude of the overvoltage,is connected with the rate of
reactions taking place at the electrodes viz:
For more details on this subJject, tables, dia8ramsa etc. see
Literatuurstudie betreffende het o treden van overs i
p pann ng,
in het bijzonder bij e1cctrodia1se" by Ir B.C. Lippens
y ,
T.A.-report 264 published by the General Technical Department
T.N.O. .
The total voltage drop at the electrodes Eelis composed of an anodic
voltage drop (E and a cathodic one (E). Hence.
a c
A low reaction rate causes a high overvoltage, and only in the
case of an infinitely high reaction rate the overvoltage is nil.
There area m great any theories and hypotheses about the nature.
of the reactions which occur the
at electrodes upon current passage..
The closest study has been dy ha n made of hydrogen overvoltage. A.o. this
is connected with the fact that measurements can be made fairly easily
with a reversible hydrogen electrode. In the case of oxygen no such
electrode is known and furthermore many difficulties arise owing to
the corrosion of the electrode material. Therefore fewer data about
oxygen overvoltaga are found in literature.
Much ,',ork on this subject is done by Tafel (66).
2. The influence of various factors on overvolta e.
a. Current density.
According to Knobel, Caplan and Eiseman 67 in the case of
hydrogen overvoltaga on increasing the current density there is
always a top limit of 1.3 volts, provided there are no secondary
reactions.
Azzam and Bockriss 68 however, found that at very high cur-
rent densities the o e vo
v r ltage often rises considerably, while
there is sometimes a kind of hysteresis if the density is first.
increased and then diminished again. The results are difficult tb
reproduce, however.
b. Temperature.
Increasing the temperature lowers the overvoltage. At low cur-
rent densities the influence of the temperature is deterr4ned main-
ly by the temperature coefficients of the diffusion, reaction rate
,
etc. At high current densities, when the entire electrode is covered
with a gas film, the predominant factor is the ease of formation of
n
the gam bubbles, the temperature coefficient of the overvoltage
therefore seems to be independent from the electrode material employed.
 c. Pressure.
There is no agreement as to the influence of the pressure.
Goodwinn and dTilson 69 found that overvoltage decreased ti'iith
increasing pressure pressures of 5-100 cm of water). Bockrzss
and Parsons 70 s 9 however found that -allowing for the influence
of the pressure on the equilibrium potential of hydrogen- no change
in overvoltage with varying pressure could be established with
certainty.
d Nature of the electrode material.
As there is no agreement as to the way in which overvoltage
ought to be or can be determined, it is hardly surprising that
literature quotes very varying values for the overvoltage with a
given electrode. Differences as high as 1 v may occur, between the
lowest and highest values stated for overvoltaga for an electrode
of a certain metal and for the same current density. see Appendix
III of the T.A. Report mentioned before).
Oxygen overvoltage is still less reproducible and here again
the observed values do not agree.
Appendix IV of the T.A.Report 264 contains a number of values
for oxygen overvoltaga in a 1 mol solution of potassium hydroxide.
In general metals with a low melting point (Hg, Au> Pb, Zn, Cd
exhibit a high overvoltage and those with a high melting point
bio Vr
Pd a low overvoltage. Classified according to atomic
number the metals display a certain periodicity see Appendix vi
s
T.A. Report 264.
e. Finish of the electrode surface.
Research into the overvoltage of hydrogen on metal electrodes
with a rough surface metal-ceramics and powders by Koezmin 71
iron , Moertazajew 72 cobalt and llaitak 73 copper in-
dicates a possibility of reducing this overvoltage by the use of
metal powder as cathode material.
A recent investigation 74 has shown that hydrogen over-
voltage can be lowered by electrolytically precipitating a film of
divided o so
finely metal n a lid cathode. At a current density of
100 ma cm2 in this way a lowering of the overvolta a of 2 my.
g 45 was
obtained with nickel as the cathode material for the metals latinum
iron, copper, silver, bismuth and tin these values were respectively
221,
552, 405, , 406, 100 and 270 mv. According to the authors of the
latter investigation it is proved in a number of cases that such re-
duction is not only the result of surface enlargement
through roughen-
ing, but mainly o increasing
f the rate of discharge of the hydrogen.
ions.
f. Impurities.
Only recently the conclusion has been reached that the presence
of very small
quantities of impurities in the electrode material has
an important influence on overvoltage. Using platinized platinum
Bockriss and his co-workers 75 76 found that 10-10 mol. arsenic
per litre caused a considerable increase in overvoltage. The kind of
base metal determines the magnitude of the influence of the impurity
upon the overvoltage.
According to previous research there is a certain connection
between the extent of the overvoltage and the time during which cur-
rent is passed through. According to Hickling and Salt 77, 78 this
fact has only been established experimentally ands at the ,time of
.,
their publications,
no theoretical. explanation of the phenomenon
could be given.
Generally speaking the curve indicating the relation between
the extent of overvoltage and the logarithm of the time is a straight
line with a certain sl,o a relative to the coordinate
P axes. The kind
of the electrode material, impurities and other factors influence the
trend of the curve.
Even after careful purification of the electrode material the
overvoltage was found to rise with time. A certain number of minutes
after switching on the electric current, the overvoltage reaches a
constant value
(30-90 mini). It may be that in this period of time
part of the electrode material is dissolved 0
7 , 75, 77 t /m 84).
 - 53 -
h. Polarisation.
According to' Goode i
r n and i{nobel 79overvoltage drops if
alternating current is superimposed on the direct current. The re-
ductior. in overvoltage probablY is caused by ? partial depolarisation
of the electrode o ~ resulting
f r ,he time that n the voltage is lower
than the decomposition c' }
v ..tags. The extent to Which overvoltage is
reduced is determined by the ratio of the voltages of both electric
currents.
In general the addition of depoyarisers reduces the overvoltage.
Hicklin and Salt 80
g found that oxygen bubbled along an electrode
considerably reduced hydrogen overvoltage. This> however> depends
upon current density: dbove a specific value there is a sudden in-
crease in overvolta e until it
g has reached "normal" value. The higher
the oxygen concentration, the higher this "critical" current density
seems to be see Graph 5).
Chromates, wolframates9 titanates, etc. likewise reduce the
overvoltage. These substances stay unaltered 85.
i. Ultrasonic vibrations.
Accordin to Cur (86) and Pio e
8 Cup r nt llz (87) ultrasonic vibrations
of about 1200 kc sec seem to reduce overvoltage.
3. Reduction of overvolta a in electrodial sis. .
There are drawbacks to the use of de olarisers in electro '
p dialysis
because the ions of the substances under co '
nszderation molybdates,
chromates
, etc. will pass through the membrane into the liquid being
treated.
As to the use of oxygen on the cathode very little is
y known
positively.
Analogously it can be expected that hydrogen will reduce
the `anodic oxygen overvolt ~e. This is a mere speculation
~, hrnvever,
further research will have to show how far this is
true.
With regard to the
use of ultrasonic vibrations results are still
too re imi
p 1 nary to permit a fair Judgement of its usefulness.
?16th regard to
the factors mentioned in section IV.B. 2 it can be
stated now'that in electrodial sis the sum of the electrode potentials
may be reduced by selecting an electrode material which involves the
lowest possible overvoltage and/or by roughening the electrode
surface as described on page 50 ,
Frequently
a.o. in desalting water the electrode material is
determined by other factors -e.g.
y resistivity to corrosion- and
therefore in such cases reduction of the overvoltaga is only pos-
sible by surface roughening.
As in a1ec
trodialysis magnetite is an important electrode
material a few data are given in Tables VI and VII 8
8, 89 For
further information on overvoltaga at magnetito electrodes
we refer
again to report T.A. 264.
C. Voltage drops caused b Ohmic resistance.
1. Introduction.
The voltage drop caused by the 0hi
p y w~ c resistance in the cell
depends on this resistance and the current strength in accordance
with Ohm's law:
I.R
The resistance in the cell(R)is the sum of the resistances in anodic
compartment R cathodic compartment Rcand dialYsate compartment
a
(Rd). The following equation applies to each of these resistances.
therefore if all cross sections of the liquid have the same area:
The sections IV.C 2 and 3 are devoted to the relation between the
voltage drop caused by Ohmic resistance (Es), the specific con-
ductances K and the current density d
= de
J
2
sum of the products of the anodic cathodic and
dialysate compartments.
pth of the layer of liquid in question.
 -54-
2. Specific conductances,
The voltage drop caused by Ohmic resistance in an eleetrodialysis
cell is thus proportional to the sum of the products of the specific
resistances of the liquids and the depths of the layers of liquid in
each of the compartments. In the rinsing liquids the specific re-
sistance is dependent upon the electrolyte
concentration and decreases.
with increasing concentration. At the same time however, back-diffusion
will increase, i.e., diffusion towards the dialysate compartment. This
in turn can be reduced by increasing the rate of refreshing of the
electrode rinsing liquids. In order to obtain a minimal energy con-
sumption a definite combination of rinsing liquid composition and re-
freshing must be found for each g individual case of electrodia-
lysis. . .
The depth of the liquid layer in the various compartments depends
upon the dimensions of the cell. In the case of very slight depths the
pumping power needed to circulate the rinsing liquids at the required
rate amounts to considerable values, so that reduction of the depth
of the various compartments has a limit below which further decrease
does not pay.
There must be no local variation in the depths of the compartments
in other words the distance of the
membranes must. be sharply fixed. If
the membranes bend inwards, the resistance at the place of the shortest
distance '
will be lower than at the sides. The current density will in-
crease locally, in consequence whereof heat development will also in-
crease. This may give rise to all kinds of unpleas't
,,. ~. cc.:nplications
e.g. denaturation of colloids). In case the memb: arrhos:id touch
there will be practically no desalting. "hen the memb:nes ben: out-
wards the?resistance in the cell will increase. should it be dea..red
to keep the current strength at a fixed value, a higher errinal vol-
tage will have to be applied, resulting i
in a higla i energy consumption.
In this case the greatest passage of current will he found at he edges
where the membranes are trapped * between the frames or other , n
ranting
accessories. If a membrane bends so far that it touches ~.ti a,lectrode,
the normal circulation of the rinsing liquid is disturbed and there
-55-
will be a local increase in acidity or alkalinity.
y In consequence
of this more acid
resp. lye will be introduced into the dialYsate..
Therefore the supporting of the membranes is very essential in
thin compartments.
To prevent an increased resistance of the.eleotr
ode rinsing,
liquids and of the dialYsate, degassing is wanted.
The gases in the
rinsing li uidsorigi
q nate from the electrodes and they :should be re-
moved as effectively as possible. The as content. of a rinsing
8 .liquid
will be
lowered by increasing the circulation rate, - provided
the
gases are fully eliminated from the rinsing liquid at the degassing
vessel.
If possible foaming must be avoided dea too. ~~th Slightlyviscous
liquids
and liquids containing electrolytes only, few difficulties
are encountered in this respect. It is different if the rinsing
liquids are polluted
, for instance by proteins in such cases elec--
trodial sis has to be stopped.
3. Current density.
Another important factor determining the value of E. is the.
2 i
current density. The desalting effect per m area can 8 p a an be aocelareted
by increasing the current density, which causes the energy consump-
tion to rise
howaver. The amount of,ener consumed by the 8Y y formation
of chemicals e.g. oxygen, ch1on
hydrogen, ne caustic soda.. and.
~ hy-
drochloric acid in electrodialysis of salt water does not increase
in the same proportion and thus the
~ greater part of the energy.
introduced is released in the form of
.heat..
These simple considerations can only be
applied if the ratios
between the quantities of anions
and cations passing through the
membranes are independent of current density, which with various.
membranes is actually the case.
However, the great adv
antage; of: increasing the current density
is the decrease in time which means a
. ~ larger desalting capacity er
~.
_
. n1
of membrane area.
 the membranes.
With the use of thin membranes the voltage drop at the membranes
usually 'rather low so that it has no influence on terminal voltage.
This is different when by blocking or chemical conversions -for in-
stance by the introduction of certain chemical groups whereby the mem-
brane acquires a certain electric character: cross-linking or other-
wise- resistance in the membrane increases. Care should therefore be
taken to avoid fouling etc.
E. Summary.
low overvo)tage attained by selection of suitable electrode
material and/or roughening of the electrode surface.
2. short distances between membranes and electrodes.
3. low current density.
4. high electrolyte concentration of the rinsing liquids.
5degassing of dialYsate and electrode rinsing liquids.
6. Prevention of membrane fouling.
the use of thin membranes.
7
F. The Quantity of charge passed through in electro
_~
The following factors may reduce terminal voltage in electro-
dialysis:
dialYtic desaltin .
For electrodialytic desalting the most important item is of course
the energy consumption per gramme of the salt to be removed. Beside the
factors which lead to minimum terminal voltage see section IV.E , there
will also be several factors which influence the product I x t.
'on of migration of other ions, the rediffus~. f ions already re-
moved and the water transport through the membrane are a.o. the causes
that, during desalting of water in a three-compartment cell, all the
charge passed through the dialysate is not carried over, by chloride
ions moving towards the anode and sodium ions moving towards the Ca-
thode. Therefore efforts have to be made to ensure that in passing a
curtain number of coulombs through the da.'al a, this liquid is d
Ysat
salted to a maximum degree.
In general our experimental coworkers qualify the desalting effect
in various electrodialYtical exPeriments_bYq, i.e, the number of gram-
~
mes of salt removed from a compartment per 100 ampere.-hours, Therefore
in the water desalting investigations q becomes the number of grammes
of sodium `chloride removed from a compartment per1O0 ampere-hours.
The value of q of the salt to be removed electrodialYticallY from
the middle compartment is determined mainly by the values of the oharge
efficiencies of the cations and anions of the salt for the anodic and
cathodic membranes . Under certain conditions hi
a gh q-value will be
obtainable with the use of an anodic or cathodic membrane which at these
conditions exhibits a high charge efficiency for the anion or.the cation
to be removed.
In chapter V it is attempted to. deduce theoretically
influencing these charge efficiencies. The determination of
perimental data on the desalting of water will.be discussed
C. Calculation of energy oonSum tion and desalting effectfor the
electrodial tic desalting
of water from experimental data..
1. Definitions.
In order to be able to express the results of water desalting
experiments in terms which can be compared it is necessary to cal-
culate:
NE = the number of kVih needed for desalting 1 m3 water
over a desalting range of 28 to 8 meq. C1- 1.
.
=t1enumber of grammes of sodium chloride removed
per 100 amP.hr in desalting over the same range.
At the same time the followingsymbols are introduced
,
w(c
, ck) = the energy consumption in kh .mo of > a certain .
P
experiment for a desalting range from to
q c ,c = the nurc'ber of grammes of sodium chloride removed
n k
per 100 an).hr in desalting frc.~, a to ck meq. Cl- 1.
P
section V.B for the description of these efficiencies.
 If c 28 and ck 8 meq, C1^ 1s then in the same experiment
_P
V1 cPr ckwill be smaller
, than NEAs however q represents an average
value e.g. q 27, 20 can under certain conditions be higher than q.
This might be the case for q if -at the beginning of desalting the
salt concentrations of the water in the dialysate, anodic and ca-
thodic compartments
are the same. as desalting of the middle com-
rtme
pa nt progreases the counteraction of the diffusion will be
greater.
Calculation of.N.
.. In, every desalting experiment it is possible to measure at the
times T , T , ..-:.?., T in hours, the voltages E E1, . .
0 1, n
o'
E, in volts, the current strengths I, I1 , . .. ' I in amperes
n 0 n
and the Pchlori.de ion concentrations c_, c? . , c in -neq/1 .
energy consumption for the entire experiment in kWh/ m is thus:
10 3 Tn
IT ,I E(T).I(T).dT
3
with V representing the volume of the dialysate in m, and ;the
index x n the number of time intervals. ? .,.
. .
,
- P meq Cl 1
%(cp~ck) ~. y m,I: E(T) I(T) dT (1 ( k (n).
Therefore on.the basis of experimental data it is desired to
obtain as simply as possible the most. then accurate value practic-
able for the above integrals, i.e. for the. integral:
0
the functional relatt'.on f(T) n0
not being known.
If f(T) is determined i.e. if E and I are an ar measured at suf-
ficient close (equal) intervals linear interpolation is the proper
w and f(T) is adequately approximated by this broken line.
hen the above mentioned condition is fulfilled the above in-
tegral can simply be found from the experimental data.
In a three-compartment cell the 'number of grammes of sodium-
chloride removed from the dialysate per 100 amP.hr.,'4c `, ,is
.
. P
equal to the total number of
gram equivalents chloride xembved,
multiplied by the quotients 58.5/35.5 and 100 I x T. The total num-
ber of gram.equivalents C1^ removed is equal to the difference
_ d
between 10_3V.c d and 103V.c if the volume of the dialYsate V
n U
is expresse4 in litres, and therefore holds
(v_c)d _ (v old
58.5
qc, 10
S IT --?.
355 I 100 5
In a two-compartment cell the elimination of chloride ions takes
d d
place in the anolyte, and in that case V and c in these equations
become the volume in litres and concentration meq. Cl l of the
anol to viz. Va a
y , andc.
Removal of the chloride ions takes place at the anode where
conversion into chlorine and hypochloric acid takes place. The
chlorine
partly escapes as gas and can be determined experimentally.
C
i2henceforth means the number of milli equivalents of
chlorine gas determined.
The remainder of the chlorine remains in solution and can be
quantitatively determined there together with the various chlorine-
oxygen compounds.
ct.C
a 1 henceforth means the number of milli-equivalents of
.,
active chlorine in the anolYto and possibly due to diffusion in
the dial s to
y a as well), i.e. dissolved chlorine gas and the other
chlorine-oxygen
compounds together.
 in a t hree-compartment cell:
It ~ra.ll nowbe clear that
act.C1+ [(V.c.)k V.c a + V.c k -
p
and in a two-compartment cell:
V.c - V.c a _ Cl2+ act.Cl + [(v.c)k - V.c C
. p
P
V. c
so that c c can be determined as well directly as indirectly in
4. ~ k
P
a three-and a two-compartment cell.
:4. The error in NEi_
chloride ion concentrations.
The inaccuracy in determining the chloride ion concentrations
cP and ck leads to an error in C1 c ,ckand q c c .The value of this
p p k
.
error will be relatively higher as the accuracy in measurement and the
difference c are less. The influence of the experimental error
c
p k ar
in the determination of the chloride concentrations c and c will be
.
p k
clear. from the following examples in which an experimental error of
0.3,E has been assumed.
and
If c - 28 k 8 meq. Cl 1' the degree of accuracy is 0,54~, as
, c
P
shown by the calculation below.
- 61
28
+ 0.3% =
28
? 0.004
8
? 0.3% =
8
? 0.024
20
? 0.708 =
20 ?
0.54p
k
In the same way the values in the following table are calculated
fox various accuracies in measurement and differences in concentration.'
1
This is also the err in Vic,ck, cPc and ,( cpc .
P k k
Error in cp-ck per cent in the case of
various values of c and c ;p k
I cP ck
C
E
Accuracy titration (per cent
0.6
0.3
0.2
0.05
700
1000 (
1.1
0.6
0.4
0.1
7000
11.4
5.6
3.8
0.9
100
400
4.2
2.1
1.4
0.4
1000
33.2
16.6
11.1
2.8
35.5
335.5
11.1
5.6
3.7
0?9
These errors are caused by the actually measured concentra-
tions c and ck. In order to obtain anything like reliable results
p
the measuring points tk'ckcan only be used if the accuracy in
measurement of k is 0.05 and the points are no closer together
c w .
`
than 35.5 mg. In that case about eighteen or nineteen points are
obtained in the range from 1000 to 300 mg ci 1. In all other cases
the bast thing is to plot "as well as possible" i.e. in the sense
of the method of the smallest quadrates a smooth curve along the
measured points c(t) and use the concentrations cP and ck determined
from this curve for further calculations.
is desired therefore in certain cases to determine
If it
~l c'ckand q cPc this can only be done with any degree of
p'k'
 accuracy if the range c c is not taken too small. The limit
for this depends upon the accuracy of determining the chloride
ion concentrations.
The error in the "indirectly" measured q's depends chiefly
on the accuracy of determining the chlorine and active chlorine
concentrations.
V. CHARGE EFFICIEIFCY.
after which current is passed through the solution. Now what is the
mobilities. On either side of the membrane an 'electrode is placed,
..
cations and anions in different concentrations 'and with different
A membrane is placed between two solutions containing various
A. Introduction.
charge efficiency for each of these ions under the influence of the
electric field the water transport through the membrane and the
diffusion together .
In section V.C a general equation is found for the current
density efficiency of these ions, followed in section V,D by a
simpler computation, made on the basis of a very simple conception
of the membrane.
Since in the material dealt with in section V.C. and V.D.,
the terms current density efficiency, charge efficiency and current
efficiency will be used these concepts are first described in detail in
section V.B.
In section V.E we describe the war in which the charge efficien-
cy of the chloride ion for the cathodic and the anodic membrane is
g ytis
calculated from the experimental data relatin to electro dia 1
desalting of water.
FinallYs in section V.F a number of charge efficiencies are
given which were found by Zhukov 52 andBi11iter 29 in their
experimental investigations of electrodialysis.
B. Definitions of the various efficiencies.
The following efficiencies are each defined in respect of the
1f-ions in solution, possibly together with other ions. It should
be born in mind however, that the definitions can. be applied for
any ion.
Current density efficiency t the direction x is under-
stood to be the fractionof the total current carried bythe
R -ions through an infinitesimal surface element df situated at
 a random point P of the electrolyte the
orientation ofd is given
by the unit vector x df di'. x ?
In stating current densit efficiency therefore i
Y y it is always neces-
sary to give
the orientation of the unit area considered in the point
concerned.
Reference is sometimes made to ntheu current density efficiency
in a point P of
the electrolyte which then means the current density
-4
efficiency in the direction i the orientation of the unit area at
that point coincides with the total current density vector: =
y v ctor? . x i i.
Current efficiency. ?
ecurrentefficiency for an arbitrary surface S
situated
entirely in the electrolyte is that art oft
p he total current I
through
the surface, which is carried by the R -ions.
In she stationary state
closed surface..
1
'I 'S~..L-
3f
(s)
The current efficiency in respect of the R -i
ons of a membrane
is that part of the total current through the membrane which is
transported by the R -ions.
3. Charge efficient and coulomb efficiency.
The charge efficiency for an arbitrary surface. 5 situated
entirely in the electrolyte solution is that part of the total charge
Q flowing during a time interval t through th at surface, which is
transported by the R -ions. This efficiency is sometimes also called
dis !
putable? the coulomb efficiency.
S
S =
R QS
?The charge coulomb efficiency so defined is t erefo
y h xe a kind
of average efficiency for the time interval 0,t The charge effi
ciency of the R -rions at a given moment t then is:.
aR (s;t) zR(s;t) . ai
aQ s; t r s t . at
In the stationary state this fraction is independent of time and
the charge; efficiency therefore becomes a constant viz:
I S? t ' I
R ~
- S
I (S; t R
It is only in the stationary state therefore that for a random.sur-
face S situated entirP
ely in the electrolyte, the current. efficiencies
and charge efficiencies are numerically the same.
The charge coulomb efficiency of the R-'
y ~.ons for a membrane,
for a specific Ar^cess , during the time interval Opt , in respect.
of the R--ions means. that part of the charge which has. flowed through
the membrane in the time interval t, which is carried over by the R-
ions. This therefore is again an average efficiency. It is only in
the stationary state that this efficiency becomes a constant
while at the same time
 -66-
It follows x fom the above d
definitions that in the, stationary
state in a homogeneous field current density effidiencies9 current
efficiencies and charge efficiencies are numerically ec,ua1, and
that in such case the following computations of the current density
efficiencies for the various ions can also be used for calculating
the charge efficiencies.
. Derivation of a general expression for current density of an ion
in a system of a number of ions in the case of charged membranes.
1; General equation for current density of an ion.
Teorell 90 finds an expression for the current density of
the anions and cations under the conditions mentioned in section V.A.
making the following assumptions:
a. all cations and anions are monovalent.
b. the water transport through the membrane is negligible.
C. distribution of the so-called fixed charges" is the same
over the entire pore and over the entire membrane.
d. Planck's method (92), developed for treating the electrolyte
diffusion is correct, i.e. there is no linear mixing of the
.ions. in the membrane as assumed by Henderson 93.
e. the state is stationary.
Teorell uses e'
(92, 94 th fundamental equation of the 'ionic
flux i.e. the number of coulombs transferred per sec. by a certain
ion species M at an arbitrary place within the membrane through a
unit area perpendicular to the direction of diffusion = current
density of that ion in the membrane pores = . Hence for a mo-
novalent cation is
.+' , R'C dc' d
i _ -u . a a' dx ? .
a dx
----+--v-~''
-
~,-
-
osmotic electric
1
1
. tern term
2
See e.g. 91).
The concentrations in the membrane are here indicated by a's ,
in the free solution by c's.
3
-67-
After introducing Planck's variable see note 2 -Table VIII
F RT
and page 69 for e E equations are obtained for the current
densities of each monovalent cation
on and anion which for the stationary
state after integration -
from x = 0 to x- give the general equation
of Teorell: .
According to Teorell it appears from this equation that if a
membrane is placed
between two solutions with different monovalent
ions and if current passes through the current density of each ion
is proportional to its
mobility in the membrane and to the difference
'
of its electrochemical activities at the membrane surfaces2
All cations have the s
ame proportionality constant: while this
is likewise the case with the proportionality constant for the anions.
In general the proportionality constants of cations and anions eaoh
a product of three terms, will not be equal and will have different
values for?different membranes. Their ?enominator includes the thick-
ness of the membrane.
This equation of Teorell'sisgenerallY applicable as it.applies
both to current free systems free diffusion : I = o and to
an systems.
where current passes through t 0 . Numerical evaluation of this
general equation in a specific case has to be '
performed in different.
stages cf a of the is
p g 469 art le in. question .
In Table VIII
also several leas general equations are recorded.
Graph 6 is a completely worked
out example at passage of current
and equal tot
al concentrations c = c .Between solution.) 90 HCl +
1 2
3 _
10 NaBr and solution 2 10 HC1 + 90 Naar 3 a membrane W.X.= 200
is 1 ced?
p a .the Donnan potentials at the membrane surfaces 7(' and
1
2are + 22 v
m and - 22 my resp. The current densities of the ions
,
This equation is iven in full detail in Table VII.
g Abridged
computations are given in Appendix 6.
These activities are
a2 and a1 respectively as the value of
varies from 0 to through the membrane..
Arbitrary concentration units.
 and, the total current density are lotted as a function of various
p applied potentials :the potential difference equals the sum of these
Don
nan potentials and the dlffuslonpotential in the membrane.
Correction factor for water transport through the
membrane.
Water trspo
an rt through the membrane is not always negligible
pana strictly speaking the equation is
. not yet general enough. If
one assumes a constant flow of liquid
perpendicular to the membrane
with a rate of cm sec the equation: i = -u'. A' cf.
~ Appendix 6).
becomens:
So far, however, attempts to obtain a general solution for ~n frthis
equation have been unsuccesfull in respect of charged membranes.
The ratio influx outflux for each ion.
.The current density of each ion can be represented by
1 =-f .u . a -a
2 1
in which f is a proportionality ? constant. It appears to be composed
of an influx .
in the direction 2 -+ 1 and an outfl
ux
in the direction 1 --~ 2.
influx
The ratio for a cation is therefore
outflux
r represents the constant Donnan ratio:
T influx
he expression f
particularly appropriate
or the ratio outflux is
for radio active tracer experiments.
If the resulting current
approaches zero it is possible to
speak of an individual ion distribution equilibrium, in which in-
.
flux = outflux. In this case we find
It appears from this equation that has also the
meaning of the
ratio of equilibrium concentrations.
4. Ratio between current densities of two cations
and of two anion b
As f has the same value fox both ions the ratio between the
current densities of two cations becomes:
Hence it appears that the ratio of the current densities of
two cations in the same system and under the conditions mentioned
on page 66 is independent of the charge of the membrane and the
thickness of
the membrane. It is proportional to the ratio
of the
mooilities within the membrane and to the ratio between the dif-
ferences of their electrochemical activities at the membrane sur-
faces.
 The current density efficiency of each ion.
i
The current density efficiency of each ion is obtained by
dividing the second term of the general equation by - the total cur-
rent density in the pores
In this + - -
equation i = i i ,
Hence in the denominator consists of a sum of terms9 each of
which is proportional to the mobility of an ion in the membrane and
to the difference of its electrochemical activities at the membrane
surfaces, and inversely proportional to the thickness of the mem-
brane. In addition the numerator also consists of such a term.
D. A simple calculation of the current density efficiency of an ion
resent in a system of a number of ions in the case of non-
selective membranes.
1. General remarks.
To make calculations as'simple as possible various basic as-
sudo -
mp ns partly incorrect- are made. The assumptions a band c
of Teorell page 66 are adopted also in this case. In addition the
following assumptions are made:
a. Fixed membrane charges are either absent or their influence
is completely negligible.
b. The external electric field in a membrane pore is homogeneous
and longitudinal to the pore.
C. All the quantities concerned are variable only longitudinal
to the pore.
The current density of a cation in membrane pore is represented
in which j) = - , u
. Similar equations can be derived for the other c do s
a n and
anions. 'B Y adding these. equations and by bearing in mind that
dE
-' , the real field strength at each point, can be computed. After
dE
substituting in the equation mentioned above and then dividing
by i we obtain a differential equation for the current. density ef-
ficiency of the cation in question. After integration o
,of this dif-.
ad
ferential a uation i
q , ntxoduction of the boundary conditions
l
ti
c' (O) = C. and c' (c ) = A e direct
re
a
onsh p can be established
green current density concentrations membrane thick-
.
............ .
ness and current density.
These integrations have been effected for a two- and a three-
ion-system, calculations for which are given in Appendix 7.
The two-ion-system.
h cur-rent density efficiency of e.. the cation
in Y g atwo-
ion-s stem is :
Y given by the equation.
i + 3?
D
'l+= n + --- -
i 2 1
in which D represents the "mixed diffusion" coefficient of the
salt; ,
and are the concentrations o
2 1 on either aide of the membrane
coulombs per cm3.
.It appears from this result that the current density efficiency
is obtained by increasing or d
g ecreasinthe?electrical trans-
ferenc.e number by a diffusion term. The latter increases in im-
portance as the difference in concentration g is 2 greater
1
and the thickness of the membrane is less
see Graph 7). If
2
the diffusion effect is eliminated
1 and the current density
efficiency
equals the electrical transference number.
Numerical evaluation of this equation involves the substitution
 of the actual current density in the membrane pores i by the current
density as calculated per cm2 membrane area d . These two quantities
are connected by the "effective membrane area" 0e' which is that part
of the
membrane which corresponds with the openings of the pores.
The interrelation is '
0
e
According to experimental data on cellophane membranes, made by the
.
firm K
alle and Co. quality OJ, calibre 70 the value 0e = 0,12 has
been adopted. The
thickness of this sample was =
0.015 cm.
Taking into account the factors, necessary for conversion of
chemic o
alcncentration values into the proper units and substituting
the numerical values for n and D we obtain in the case of HC1:
n ' - = 0.179 + 22.8
Cl
c2 and Cl in geq. 1, d in ma cm2.
This equation is represented diagrammatically in Graph 8, where
the current density efficiency
is plotted as a function of the con-
centration difference for various current densities.
It appears that when c1 = c2 the th current density efficiency amounts
to 18 per cent.
In the region
c c2 O, diffusion brings about a decrease in
1 -
the current density efficiency which is greater as d is smaller, in
the region c1 - c2 0 there is
a corresponding increase. For an in-
finitely great current density diffusion relatively plays no further
part and the slope of the curve therefore approaches zero. If the thick-
ness of themembrane for a given current density is made f times as great
,
this
has the same effect as if the current density were taken f times
as great with the same thickness.. In this way the array of curves of
,.
Graph 8 can also be regarded as curves for various membrane c
thickness at
at constant current density. ,
3. The three-ion-system.
Vie have in mind a mixture of a salt MR with an acid HR as for
-73-
instance in the anodic membrane of an electrodialysis cell. Section II
of A uendix con ins i
p, 7 , ta the derivation of an expression for the
current density efficiency of the common R--ion. The result was:
in which Z =
This is an implicit exponential function of the current density
i
efficiency 17 . With the aid of this equation it is .row possible
R
numerically to show dia ramaticall the dependence of the current density
efficiency upon each other quantity.
Several general remarks follow from this derivation. It appeared
for instance, that
2 -
d
= 0, therefore
d'
d x
It is interesting to note that the latter conclusion with regard
to the R --ions applies general)Y , to the total concentration and there-
fore to a system with n ions as well.
Y The total concentration has there-
fore v ..
always a linear trend through the ores. On the other hand the trend
of concentratio
n of individual ions throughout the pores is definitely
riot linear as for instance in Graph 9). Furthermore obviously i
and
' are found to be absolute equivalents,
viz as product ix .This was
also the case in the result of the two-ion-system and continues to apply
therefore to a three-ion-
system.
Section III of Appendix 7 contains the calculation of the current
' _ .
`
~
de
tff
s n iy e ion-system.
 - 74 -
4. Conclusions from calculations.
The principal conclusions
arrived at from the simple calculat-
ions made in the three preceding paragraphs are:
1. Both in two- and three-ion-systems increasing the current density
has the same effect on the current density efficiency of an ion
as increasing the thickness of the membrane.
If diffusion
and electrodzalysis co-operate . it is advisable to
have the thinnest possible membrane and the lowest possible
current density in order to obtain the maximum current density
efficiency. If they counteract, such as for instance happens
in the desalting of water, a thick me br e
~ m anand a high current
density are advisable.
2. In a s yctem of thre + + -
y e tons, A , B and R the trend of the
concentration of R -ions through the pores is linear.
However, this all applies only to a very simple conceptions
of membranes !
E. Calculation of .the charge efficiencYof the chloride ion in the
cathodic and the anodic membranes from experimental data of
electro dia 1 tic desalting experiments of water.
In the case of electrodialytic desalting of water in a three-
compartment cell q is at a maximum if the charge efficiency of the
chloride ion in the anodic membrane and of
the sodium ion in the
cathodic membrane is at a maximum. As a membrane with a high charge
efficiency for cations will have a low charge efficiency for anions
,
?it is sufficient to consider only the charge efficiency of the chloride
ion in the cathodic membrane and in the anodic
membrane, which shaild
then be at, a minimum or at a maximum respectively.
When in,a certain experiment the dial sate is desalted
y from c
P
to ck meq. C171 , the energy consumption W cP, ck,and the de-
salting effect q c ckare calculated according to the equations in
p
section IV.G. If the concentration and the volume of the catholyte
C C C C
alters during this test from c and V to c and V
an and the concentra-
tionand the volume of the dial sate at the same time changes from
-75-
d d d d
c and V to c and V the number N of chloride ions which have
p P k 1.
passed through the cathodic membrane is:
c c
N = V.'c P - (V. c me
k q
The quantity which has passed through the anodic membrane is equal to
d d
this difference, increased by the term V.c - V.c ' i.e.
P k
c d -
r _ d
1V.c + l.c 1(V.C)c
+ V.c meq Cl
k
p
This latter can also be calculated directly Cf.,Page 59) and then
' J
amounts to;
a -
N = [ci2) + act. C1 - {(V.C)a V.c meq Cl
p k
The number of coulombs transported by these chloride ions is
6 00 .
obtained by multiplying the value of 1. with . The total charge
looo
which has passed through the membrane equals T
x T x 3600. Therefore
the c
average barge efficiencies cf.- page 60 of the chloride ions in
the cathodic and anodic membranes during the above, mentioned experiments
are obtained by multiplying the values of N with the factor
96500/1000 2.68
IxTx3600 - 100 I.T
In this reportthey are indicated by cpc and c ck
cm k am p
direct or indirect ,in contrast to the charge efficiencies cm and
direct or indirect), which relate to the des tig from
~ al n 28to8
~
meq. Cl- 1 . '
.
To sum up therefore:
~om(cp,ck) I:~ {(V.c)p - (V.c)kI ? 10 Z .......(1)
c ,c indirect =
am p k
v.c)C (V.c)C I . - j(v.c)? + (v.c)' ] J. io2
 As these charge efficiencies are averages, it is not impossible
that in some cases for instance (c ,c will be greater than (p
Co am p k am
When a two-compartment cell is used, fQ will mean the average
m
charge efficiency of the chloride ion in the membrane, when desalting
of the anolyto from 1000 to 300 mg C1 1 is considered; if the range
from c to ck is being considered, the membrane. efficiency will be in-
dicated by cOmcpck. It then appears from the above definitions that
,
= equation 1 = equation (2).
The errors in the various +s and c ,c 's caused by the
p k
determination of the chloride ion concentrations are given in the table
on page 61.
F. Charge efficiencies and current efficiencies found in literature.
Zhukov 52 and Billiter 29 who both studied the electrodialysis
process very thoroughly, including the removal of salts from water,
calculated several "current efficiencies", R's, from the equation:
Aten 31 also calculated his theoretical "current efficiencies" in
this way Cf. section III.D .
1 _
R - the so-called "current efficiency of an ion or of a salt.
P the number of gramme equivalents of the ion or salt
.
concerned removed from the dialYsate.
C the number of coulombs passed through the dialysate.
The "current efficiencies" found by these investigators, however,
.
are by no means the same as the current_ or charge efficiencies 'defined
by us Cf. section V.B . These efficiencies
always
s relate to an area
and therefore the description used by them for an efficiency related
to a volume is rather
misleading. Their "current efficiencies far more
resemble the
q Introduced in this report cf, section IV.G 9 the number
of rammes of salt removed n
G from the dialysato per hundred amp.hr.
If their experimental determined P's represent the number of gramme
equivalents of salt removed from the dialysate it is possible to cal-
culate
q from their "current efficiencies"
In the electrodialysis apparatus in which Zhukov studied purificat-
ion of the w r
ater of the r~eva X52 , the "current efficiency" of the
electrodialYsis was detected
gravimetrically on a 0.01 n solution of
calcium sulphate. 'rith a cathol to alkalinity. of 0.00 n "current
Y 5 ,the efficiency" o
of the calcium ion increased with decreasing acidity of the
anol to reaching 36p at 0.001 n. For the sulphate ions a value of
28-29 ,E was attained. This lower value for the anion was to be expected
as Zhukov used a pair of negative membranes, viz. two tubular membranes
of ceramic a
material. Ideverthe_ess the observed value is still well
above 20%.
Billiter 29 stated that for his equipment a ceramic anodic and
an asbestos cathodic membrane the "current efficiency"
of the salts
to be removed was, at its best, 20% but normally did not exceed 12 to
15%. As average molecular weight of these salts he assumed 50
.
Stenler and Sirak 96 investigated in electrodialysis a series
of membranes to '
find their ncurrent efficiency" In a 4 n so3.ution of
sodium sulphate. For negatively charged pairs of membranes, such as
asbestos-ceramic, they state 10 to 18%, for bakelite-gelatine membranes-
acetylcellulose 23%.
The pair of negatively charged ceramic membranes selected
by ?hukov,
1
= molecular weight.
 as compared with membranes not electrochemically active, therefore
it
give a higher current efficiencyrr.
The conclusion of, our research in literature is, that although
n
been found to which the name r, current efficiencies
some, quantities have
was given, the literature contains no data on efficiencies as defined
-
by us in section V.B. Nor could we calculate these with the assistance
of factual material found in the literature on the subject, as the
data required were not complete.
w_.__
-
VI. THE EXPERDAENTAL RESEARCH INTO V;ATER DESALTING.
The experimental research mar be divided into five sections, viz.:
A. Construction of the apparatus.
B. Investigation of the possibility of desalting with the use
of non- or only slightly selective membra as and acid an
liquids, followed by research into the
alkaline rinsing
causes of corrosion of the anodic membrane and the pos-
sibility of preventing this.
C. Desalting research with the aid of selective membranes.
D. Research into chlorine removal at the anode.
E. Investigations of the possibility of desalting water by
means of reversible electrodes.
These sections will be dealt with one by one. The apparatus used
in section VI.E will be described together with the corresponding
research pg 94 . Finally this chapter will be concluded with a
review of the results of these experiments and the possibilities of
water desalting on an industrial scale.
A. Description of the apparatus.
1. of the Arrangement f th
In the experimental work described in the sections VI.B and
VI.C a rectangular electrodialYsis cell was used 20 x 40 ~n con-
sisting of an anode 1 of magnetite thickness 10 mm and a nickel
cathode 2 thickness 5 mm).
Proceeding in the direction of the cathode according to the
diagrammatic cross-section in Figure 4a we find between the anode
and the cathode resp.:
a rubber sheet 3, cut out as , shown in Figure 4b, forming
the anodic rinsing compartment,
the anodic membrane 4 with a workingarea of about 580 ,
cm` .
..
the dialYsate cell in the form of a rubber sheet cut out
(5
as shown in Figure 4c
,
2
the cathodic membrane (6): working area approximately 580 cm ,
a rubber sheet 7 cut out in accordance with Figure 4d and
forming the cathodic rinsing compartment.
-
 In the case of non-rigid membranes, such as cellophane mem-
branes, supports in the form of saran or nylon gauze are fitted
against that side of the membrane which faces the electrode in
order to maintain the thickness of the three compartments at con-
stant values all over the membrane area. By arranging for a slight-
ly higher pressure in the middle compartment than in the outer
compartments, the membranes are pressed against the taut supports.
The whole
is compressed and sealed by two pressure plates
8 and,9 , attached to the outsides of the electrodes. For the
desalting experiments in the two-compartment cell with non-
re-versible electrodes section VI.D the same cell was used though
the dialYsate cell 5 and one membrane 4 or 6 were omitted, so
that a cathodic and anan anodic compartment remained, separated by
a membrane. For the membranes cellophane was used, treated or un-
treated, supplied by the Visking Corporation, Chicago thickness
dry: 0.12 mm, wet: 0.22 mm).
2. Circulation of the liquids.
The anodic rinsing liquid is fed through hole 10 to the bot-
tom of the anodic compartment, which it leaves again via hole 11.
The cathodic rinsing liquid is transported via holes 12 and
13, whilst the dialysate is fed to the bottom of the dialysis cell
via hole 14 and leaves this via hole 15.
During circulation of the liquids they are pumped back into
the respective compartments via a suPP1Y vessel after leaving the
cell. If one or more liquids pass the respective compartments only
once this is indicated in the report b
~ by the term dosing.
3. Thickness of the compartments.
During the course of the investigations the thickness of the
middle compartment, originally 3 mm, was reduced to 1.4 and sub-
' to 1.2 mm. The thickness of the rinsing compartments
was 1.4 mm.
For carrying out a number of experiments with neutral rinsing
liquids the thickness of the electrode compartment was. doubled
and divided in two by a diaphragm made of saran cloth. In this case
the neutral rinsing liquids ass through
pass the spaces between the
diaphragms and the membranes at a fair speed
approximately 4-8 1/b)
before leaving the cell along the electrode. In
this maner.com-
Pbetely neutral liquids flow along the membranes,
g whilst the corn-
pounds formed by the electrode processes enter the rinsing liquids
only afterwards.
4. Current density.
In the experiments described in
the sections VI.B D and E
a current density of 2
y approximately 2.6 ma cm was used. The cur-
rent densities used in section VI.C
are stated there.
B. Experiments with non- or only
- slightly selective membranes..
Two series of electrodial ses
y under considerably varying con-
dition s were
carried out, viz., one series in which untreated cel-
lophane was used, and one series in which the cellophane membranes
were treated with bakelite or some other lacquer.
The degree of acidity or alkalinity Y of the rinsing liquids can
be adjusted:
1. by adding acid and/or q
alkali,to the rinsing liquids.
..
2. by using the formation of acid and alkali at the
electrodes. This may occur i
aY in two ways:
.
a. by allowing the rinsing liquids g ids to circulate asa result
of which
an accumulation of hydrogen ions and
hydroxyl ions
takes place in the rinsing liquids.
g quids.
b? by arranging for their dosing
g rate to be so low that the con-
centration of the
hydrogen ions and the hydroxyl ions in the
rinsing liquids is
g sufficient to effect desalting of the
dialysate.
Neutral rinsing liquids are obtained in the manner indicated in
section VLA.3.
 The lowest energy consumption for desalting from 1000 to 300 mg
C1- i reached during desalting with non- or only slightly selective
membranes, and thicknesses of the anodic, dialysate and cathodic-
3
compartments of 1.5, 3 and. 1.5 mm respectively, was 10-11 kWh m
whilst for q the maximum was approximately 36 g NaC1/100 amp.hr
in this case o 30-20 amounted to a value of 48 Cf. Table IX
However, in the case of the electrodialYses carried out with Un-
treated cellophane membranes at dosing rates of the anodic and ca-
thodic liquid of 0.25 and 0.15 1/h respectively the anodic membranes
got strongly corroded after three to four hours. For this reason the
causes of this corrosion and the factor which might be able. to reduce
s
or even prevent it were investigated see section VI.B.2 . In this
investigation it was found that the q-value was increased by 35-50%
if in similar circumstances the electrodialYses were performed with
.
bakelitised instead of with untreated membranes. In view of this
the membranes were treated in various manners in order to effect an
increase of q see section VI,B..3 , and therefore a lower energy
consumption, for the above mentioned. energy consumption is too high
for industrial application of the process.
The various subjects referred to will be dealt with in greater
~
detail in the following paragraphs.
Electrodial ses with cello have membranes and var in
acidity and alkalinity of the rinsing liquids.
Some of the experiments are referred to in Table IX in which
the different circumstances under which electrodialysis was performed
are recorded.
One of the paints noticed about these electrodialyses was that
an excessive acid and/or lye concentration. of the anolyte..and/or the
catholYto -which would result in a lower terminal voltage- retards
chloride removal from the dialysate. In Graph 10, for instance the
time and the number of ampere-hours passed through in the four ex-
periments 1- incl. are 4 plotted against the chloride concentration
of the dialysate. From this graph the favourable influence of re-
duction of the sulphuric acid concentration is evident. Apart from
er 100 amp.hr a maximum of 218.4 Ggrammes of sodium chloride can
theoretically be removed.
this the four experiments were carried out under identical condition s
which -needless to say- applies to all series of experiments compared
in Graphs 10-18 incl.
Corrosion of the anodic membrane.
a. Detection of the corrosion.
Originally a simple method was used for detecting corrosion
of the anodic cellophane membrane, viz. colouring with methylene
blue.
The blue alkaline dye turns dar blue.
_ k those laces where
chemical
corrosion has resulted in the formation of oxycellulose
andhdoc'
y r ellulo?e, the cellophane itself turning o
my pale blue.
The intensity of the blue colour is an indication of the degree
of corrosion
(97).
Quite soon i
. t was decided to colour the membranes with
Turnbull's blue, a colour more characteristic for
the carboxyl
groups formed
sulphate and potassium ferric anide i
y n succession, which causes
the places where so-called oxycellulose has formed to turn dark
blue whilst the other places remain uncoloured . Figure (98) 5, a
picture of a piece of corroded membrane dyed with Turnbul 's
l blue
has been included in
the report. This clearly shows the pattern
Y
of the membrane support.
The colouring of cellophane membranes treated with bakelite-
cresurol- aralac uer
~ P q ,etc,, was not practicable moreover this
might cause blueing in
places other than those corroded. In the
determination of the degree of corrosion of cellophane membranes
treated in some way or other' and that o
? of membranes not made of
cellophane
for instance Permionic membranes only the mechanical
strength and the surface of these membranes were examined more
closely.
b. The cause and reducl;ion of corrosion.
By.means of a series of electrode
alyses, performed with non-
treated cellophane membranes under different circumstances the
causes of corrosion and the factors influencing it were inves-.
 tigated. Suffice it to recount the results and some technical details.
1. Corrosion of the anodic membrane starts ane tarts in the. anodic compartment
by the hypochloric acid which has formed at the anode.
2. The concentratior. of the hochloric acid i
3'A in the anolyte and the
consequent corrosion of the membrane may be reduced.
a. by adding sodium sulphite to the anolyte
b. by increasing the dosing rate of the anolYto
,
C. by raising t
y he acidity of the anolyte
d. by aerating the anolYto
,
e. in certain cases by increasing the current density.
- 3. Corrosion of the anodic membrane can be reduced by
bakelitising it.
4. When an acid anodic rinsing liquid is used the life of the
anodic membrane stays limited despite aeration or
bakelitisation.
C. Method of aerating the anolyte.
In section III it was explained already how the concentration
of the ochloric acid might
~'P be reduced by r
aped removal of the
chlorine gas for instance by aeration) and . Y an by increasing the acidity
of the anolyte.
. Two methods of aeration were used viz. bubbling air through the
anodic compartment and aerating the anolyte outside the cell.
Ylhen air i s bubbled through the anodic
. ,. ... compartment the maximum
amount of air is only small owing to the limited volume of the arioi te.
Experiments proved that u
nder the circumstances chosen cf. experiments
5 and 6, Table IX at least fifty litres of air had to be passed through
hourly in order to
Y prevent corrosion during one electrodialYsis ex-
periment . Industrial
. use demands such a long life that larger quanta -
ties of air are required which cannot be bubbled throu h the cell
without major difficulties.
Aeration outside the cell produces better results as it enables
larger quantities of air to be used
without any difficulty..The anodic
rinsing liquid drips pinto the top of a p tower filled with glass beads
or Raschig rings 'the air being fed '
~ g into the bottom of the tower at
a rate of hundred litres per hour. This ensures thorough desorption
of the chlorine. This metho& of
aeration, :unlike.the before mentioned
is effective only if a circulating anolyte is used. The 'circulation
rate should. not be too low as this would cause
the concentration of
active chlorine contained in the
anolyte to rise too;much and cor-
rosion still to occur.
d. Bakelitisin the membranes.
The membranes were dipped in various concentrated _ etl ?lalcoholio
and iso
propylalcoholic solutions of a condensation product
of phenol
and formaldehyde bakelite lacquer; dry ,matter:' ~
48 Thereupon,
~ in
order to evaporate the solvent the were dried for a few hours and.
subsequently bakelitised by heating
. y ng at 100?C. The.bakxn and-drying
times producing
the best results were determined e
xperimentally.-In
doing so chlorine resistivity aid mechanical properties had to
be
taken into account. 10o-so
lutxons, both in ethyl- and in isopropyl-
alcohol alcohol yielded the most stable
membranes. Since hovrever, mmoistening
left much to bedesired in the first case only the second solvent
was considered suitable, Drying for two
hours. at 60?C and baking at
100?C for seven hours: '
yielded the best. results.
Im Graph 11 two exam les e
p ar gxven.of electrodxalyses (exp. No.
.
7, 8 carried out with b
akelitised membranes under ci
. tcumstances
referred to in Table IX., The value of was approximately
q 36 and
for q 30-20 a value of
48 was obtained,. whilst. the electrodxalYsxs
experiment 9 with non-treated cellophane
membranes, but otherwise
performed under the same conditions, produced a q-value of 27.
In the experiments and 8
7 the value of the ever cons
umption
amounted to E approximately 13 kYlh m 3
,
3. Enhancing the selectivity
of the cellophane membranes by ?
treatin them with different
lacquers.
A series o '
f electrodxalyses with different membrane combinations
was performed, and for each des
altxng the q-values as well as some
charge efficiencies.
of the membranes. were calou1ated. The following
membranes were used;
 non-treated cellophane membranes,
cellophane membranes impregnated with bakelite lacquer, cresurol
lacquer or para lacquer;
cellophane membranes impregnated with A.F.-lacquer made by the
Fibre Research Institute T.N.O.,
cellophane membranes treated with hypochiorite indicated by cell.
these experiments the rinsing liquids 0.03 n sodium chloride
solutions were dosed at fairly high, rates (4-6 1/h, Cf. Table x). By
application of the method described in section VI.A.3 it was ensured
pH?of the rinsing liquids did not deviate much from 7.0, so
,
:that the membranes were txied in almost neutral solution. Data relating
some of these experiments are summarized in Table X.
Lacquer treatment of the anodic membrane.
Graph, 12 clearly indicates the favourable influence of treating
the anodic cellophane membrane with bakelite lacquex Cf. exp. No
12 and 13? and see also exp. No. 7, 8 and 9 in Graph ii). The effect
of treatment with cresurol- and para-lacquer is equally favourable
to q, viz. it roughly doubles it, whilst treatment with A.F.-lacquer
does not materially improve the result Cf. exp. 12 and 16). The
increase in the average terminal voltage is - 0.1, 0.6, 3.4 and 1O.8 v
for membranes treated resP. with A.F.-lacquer, bakelite-, cresurol-
and para-lacquer.
Since the membranes treated with cresurol lacquer are less
supple than bakelitised membranes and in addition they absorb less
.
water, the most favourable results in many respects have so far
been achieved by bakelitisin8 the anodic membrane.
b. Treatment of the cathodic membrane with lacquer.
If we compare the curves of desalting experiments 17 and 18 in
Graph 13, the favourable effect on q of a careful treatment of the
cathodic membrane with hypochlorite and boric acid, which treatment
result in slight oxydation, will be clearly observed. The effect of
impregnating the cathodic membrane with A.F.-lacquer experiment
No 19 on q is not very great.
4. The charge efficiencies in the membranes.
In a number of experiments the
charge efficiencies of the chloride-
ion in the anodic and in the cathodic membrane were calculated in ac-
cordance with the method described in section V.E. They are given in.
Table XI.
-87 -
From From this series of experiments in which rinsing was done with
neutral liquids, the detrimental effect of this rinsing. on the.de-
.de-
.
salting obvious, when g , n the membranes were but slightly selective.
In order clearly to demonstrate this influence Graph 14 was
plotted.
For experiments 11 and 17 the number of ampere-hours passed through
is lotted
p against the sodium chloride concentrations of the dialysate?
In both cases the same membrane combination ,was used. The values fora....
q 30-20 were 48 and 24 g NaCl 100 amP.hr. resP. The only difference
was the use of acid and alkaline rinsing liquids in experiment No.11
and of neutral ones in experiment No. 17.
In experiments 9, 10 and 12, in which two non-treated cellophane
membranes were used, the rinsing liquids were dosed at identical
rates. In experiment 12 however, the rinsing method with neutral
rinsing liquids was adopted. In this case the desalting process
took almost four times as long Cf. Graph 14).
The results of the experiments with non- or only sli htl -
selective membranes.
The principal.results of the experiments axe:
An energy consumption of less than approx. 10 kth m3 for the
desalting with the
use of non-selective membranes. from 1000 to 300
m C1 1 was not obtained. The
g q.-value is then roughly 36 g NaCl 100
amp.hr s and 4 30-20 approx. 48 g NaC1 100 amp.hr.
Even when the membrane is bakelitised or the anolYto is aerated
,
the life of the anodic membrane stays limited.
Excessive acid. and/or lye concentrations of the anolYto or ca-
tholyte retard the removal of chloride from the dialysate.`
The use of neutral rinsing liquids strongly retards the desalting
effect in experiments with non- or only slightly selective membranes.
theoretical maximum of q . 218.4 g NaC1 100 amP.hr.
 When non- or only slightly selective membranes are used q
declines sharply as the sodium +he sodi?m chloride content of the dialYsate
decreases.
As a result of different treatments of the membranes the q-
values are increased by 35-50%. From the point of view of increased
. ,.
terminal voltage the bakelitised membranes gave the best results.
Summarizing we can say that, though by various methods an im-
provement in energy consumption was achieved it remained unattractive
for industrial use. Since already very thin cells were used9 further
experiments were directed especially towards an increase in the
selectivity of the membranes.
C. Research into the usefulness of selective membranes for
electrodialytic desalting of water.
In the course of the investigation several Permionic membranes
were t our disposal, viz: placed membranes of the types CR 51, ARX 44
and ARX 102, consisting of a layer of cation or anion exchanger on a
diaphragm made, for example, of nylon gauze or paper. The Fibre Re-
search Institute T.N.O. made moreover a number of excellent selective
1
membranes on a cello have basis
), including the types A 17b, A 1
P 9 g 9,
A21', A22', and A40.
In the following sections we will describe research into several
of these
membranes, as to their physxco-chemical and mechanical pro-
erties and their selectivity on the of current. t
p y passage A first we
only got samples which were too small for the electrodialYsis ap-
paratus normally used. Special small electrodialYsis cells A 1 and
A 2 were built for the experiments with these samples. In these cells
the middle compartments were 6 and 31 mm thick respectively y and carbon
electrodes were used. Where the experiments described were performed
in the small cells this will be stated. The
experiments without any
such indication were carried out with larger membranes (subsequently
obtained in the three-compartment cell we normally used, in which
the thickness of the middle compartment was further reduced in size
1 Suggestions on the preparation of these membranes were made by.
Drs H.G.Roebersen.
Physico-chemical iro erties.
Table XII contains some data on the physxco-chemical properties
of These membranes i.e, with es i , th regard to their properties in a "state
of rest".
Membrane ARX proved u s i b1e for 'c
44 , nuto technical use on account
of its high resistance
2. Mechanical properties.
If membranes
are to be suitable for technical use they must
satisfy certain mechanical requirements, and consequently this
aspect was also considered.
Generally speaking, the mechanical properties of the Permionic
membranes are not good. They are fairly rigid plates
, measuring
20 x 20 cm which have to be kept under wet conditions to prevent
cracking. Mem'6rane ARX 44 had cracked after being kept some time
under wet conditions and in this rePsect it is unsuitable for in-
dustrial use. Furthermore these membranes,,
unlike cellophane mem-
branes have the great drawback for use in thin i e ec od'
n h n 1 tr ~alysis
cells that they are no longer absolutely flat after some time.
The T.N.O.- emb
m raves were much better and quite usable as
regards their mechanical properties.
3. Selectivity on passage of current.
In cells A 1 and A 2 the influence of the current density and
of intensive agitation agitation and/or circulation of'the
dialysate on q vras investigated. The results of these investigations
are shown in the Graphs 15 and 16, and in the Tables XIII and XIV.
At greater current densities appeared to q decline, while this fall
began only later if care was taken to ensure intensive movement of
the dialysate.
The explanation of this association is easy to find. With
.
highly selective membranes, close to the membrane thin films of
a greatly deviating sodium chloride concentration are formed.De-
salting will, take place more rapidly if these films are mixed more
 rapidly with the other liquid. The greater the current density used,
..
the quicker such films are formed again, i.e.a the more intensive the
agitation required. The decline caused by this agitation is greater
wits the use of the membrane combination ARX 102 - CR 51 than with the
use of the. combination A 19 - CR 51. This is perhaps caused by the
rougher surface of the membrane ARX 102, as a result of which the
disturbed.
close to the membrane surface is perhaps less easily
film
concentrations close
The presence of films of greatly differing
to the membranes Cf. Bethe-Toropoff is proved by the measurement
of the pH's close to themembranes. In a certain informative test
2 current density: . 4 macm , ; membranes A 19 and cell ox.), the following
pH's were measured at a certain moment:
anodic cathodic cathode
anode
membrane membrane
nle of the results that can be obtained with the use
As an exam,
of selective membranes, we give now a description of experiment No.49.
The membranes used were T N.0.-membranes A 21' and A 22', the
current 2 and the thickness of the three compartments
t density 2.6 ma cm
1 mm. The circulation rate of the dialYsate was'13 1/h, the dosing
rate of the rinsing liquids 5 1/h see Appendix 8).
The energy consumption for desalting from 30 to 9 meq.Cl 1
was 3.3 kWh m3, 1 which by interpolation gives an energy consumption
for desalting to 8 meq. C171 of 3 klh m . The q(30,9)-value
from 28
.hr i.e. 82% of the theoretical
attained was 1$0 g IdaC1100 amp N
maximum. Desalting took 55 minutes, the terminal voltage rose from
4.5 to 5.5 v, and averaged 4.8 v.
1
The interpolation or extrapolation of desalting ranges is not
strictly allowed. If, however, ranges are desalted which like
.
the above mentioned ranne?differ very slightly from the ae-
salting range normally used 28 to d meq C1 /l the difference has
been shown -by experience with these selective membranes- to be
too slight to cause major errors in the results found.
By means of a number of experiments it was ascertained how in
this experiment the average terminal
voltage of 4.8 v. was constituted.
a. The cell felled with a 0.03 n sodium chloride solution without
membranes and without middle compartment gave a terminal voltage
of 3.7 v with the current density used.
b. From a relation established experimentally between tenninal
voltage and current.density it follows that the voltage drop
at the electrodes may reasonably be taken 3 v.
Hence the voltage drop of 4.8 v in a single three-compartment cell
can be divided as follows:
voltage drop at electrodes
3.0 v
voltage drop in rinsing compartments
0.7 v
voltage drop in middle compartment
and two membranes .
1.1 v
terminal voltage
4.8 v
4. Influence of the acidity of the rinsing liquids.
The electrodia tic experiments T
ly 50 and 51 carried out with T.b.O.-
membranes clearly show the favourable effect cf the use of neutral
rinsing liquids, in contrast to those carried out with slightly or
non-selective membranes. The surveys of these electrodialYtis ex-
periments are given at the end of this report Appendices 9 and 10).
The normal rinsing method, in vrhich the liquids flow directly
along the ;nembranes and the electrodes, was used. In comparison with
experiment 50 the dosing rates of the anodic and cathodic rinsing
liquids were considerably higher in experiment 51 0.3 and 1 1/b in
No. 50 and 7 and 8 1/h in No. 51 res . and therefore the acidity
and alkalinity of these liquids were only very low. The hydrogen ion
and hydroxyl ion concentrations of the drawn-off rinsing liquids were
3 and 4 meq l in No. 51, while they ~ were ,8 and 30 meq1 in No. 50.
The influence of these less acid and alkaline rinsing liquids on the
values of q c ,c was great; these values were viz. 99 and 173
g
p k
NaCl 100 amp.hr resp.
 In experiment 51 the energy consumption interpolated 1) to the
normal desalting range, was 3.1 kWh m3. The average terminal voltage
in experiment 51 was 0.1 v higher than in experiment 50.
Pnt 5.
In experiment 50 W(30,8) and 4'(30,8) were calculated.
am can
TheYwere 50% and 5resPective1Y Cf. the values given in Table XI).
~
Extent of diffusion.
At lower current densities the counteraction of diffusion is
generally speaking,
greater. This diffusion is however relatively
smaller as the membranes are more selective. This is also shown in
Table XV; the experiments represented were carried out in the small
cells Al and A2.
The quantity of sodium chloride was determined which after a
certain course of time had diffused to the middle compartment9 when
it was filled with distilled water, whereas the chloride concentra-
tions of the liquids in the rinsing compartments were kept at approx.
0.031 n. Various combinations of membranes were used.
The results show that diffusion through selective membranes per
unit of time is about one-tenth that through cellophane.
6. Results.
The results of research into the nto th usefulness of selective mem-
branes for electrodialytic water purification can be summarised as
follows;
In a single three-compartment cell with a short distance between
the electrodes and with the use o se ect'
of 1 ~.ve membranes an energy con-
sumption of about 3 kdh m3 can be attained for the lowering of the
chloride concentration from 28 to 8 meq 1.
Over 60% ` o
~ f the loss of energy in the said apparatus occurs at
the electrodes.
!Tith the use of properly selective membranes, -values of 180
q g
13 C 100 '
a 1 amp.hr,i.e. 82r? ~ of the (theoretically Possible maximum can
The use of neutral rinsing liquids is necessary to attain the
above mentioned q-value.
:with greater current densities q declines. This decline is
dependent upon the membranes used and
upon the degree of movement
of she dialysate.
The diffusion of sodium chloride through a selective membrane
is much less
than through anon-selective one.
D. Tentative experiments on chloride removal at the anode in a two-,
cell with non-reversible electrodes.
Several 'des
along experiments were carried out in
an electra-
dial sis cell converted into a two-compartment cell de'
scription of
section VI.A.1 .
The dia hra used was ac
.. .: ellophane membrane in order to regent
gm r ..... , . .
any flow of liquid between the tt'io compartments. In princaple.of course
glass fibre cloth, saran cloth and the li
key can be used as diaPhr ams,
while moreover a negatively charged dia
gm phra could regent themovement
prevent of `anions
towards the anode Cf. . section `II.A .
Unlike the desalting of the dial sate i
. g y in a three-compartment cell
the desalting now is only brought abo
ut by electrode processes Cf.
section III.F , viz. formation of chlorine at the anode. Therefore the
removal of chloride ions ma two-comi artme
p nt cell is very dependent
on the nature of the anode'. SThen
the chlorine is rapidlYcarried off
the active chlorine content in the anol to will no
y not grow high.
In a number of 'tentative experiments Only the i
y nfluence of the
pH and the chloride c
ontent of the anolyto tivere examined. To this.
purpose diluted solutions of sodium chloride
circulating through the
anodic compartment circ. rate 6 1/h) were desalted`for o ~
one hour. The
experiments were carried out at constant chloride concentrations (30,
20 and 10 me C1_1 res
q pectively)and at constant PH s about 1. 7y 5.0
and 7.0 of the anolyte. A solution of sodium hydroxyde 20 me 1 was
dosed in the cathodic compartment dos. rate 0.15 1/h). From these
experiments one could calculate the number of grammes of sodium chloride
removed er 100 am
P p hr in?d?esalting at a definite const
ant, chloride
content therefore i
ndicated by q c The main details and red
ults
P
 of these, only tentative, experiments are comprised in Table XVI.
?-The results show that under the circumstances chosen a qc -value
P
of 90 g NaC1 100_amp.hr.-i.e. about 40% of the theoretical maximum-
1
can be attained ,_and that q (c) increases:
a., with an increasing. chloride content of the anolYto
s
b. with a higher acidity of the anol te: this increase is
Y Y larger
at.a lower chloride.content.
C. by aerating the circulating anolYte.
As the. nature and the roughness of the anode surface and the
current density; have a considerable effect on chlorine formation at
the anode, it seems probable that the value of 90 g NaCl 100 amP.hr
,
can be increased. still further by operation under optimum conditions
E. Several desalting experiments on potassium chloride solutions in a
two-com artment.cell:with reversible silver-silver chloride .
electrodes 2
1. Method. ?.
The elec R
trolysi,~.cell,is built up of two halves of plexiglass
with milled out e
xcavatjons measuring 10x10x0.2 cm. These de-
pressions contain two silver-silver chloride electrodes 10x10x0.1
-cm :which were thoroughly. ungreased beforehand while in addition
.to thisthe cathode had been anodically chlorinated for an hour
~-p
at a current density of approx. 2.5 ma/cm2 in a 4% -potassium
chloride solution. Theoretically therefore this electrode can now
function for one hour as a reversible electrode if a current
r
density of :approx. 2.5 ma cm2 ~.s used. In practice however, it
proved to be :usable for half an hour..
..The,, currents supply to the electrodes is. effected through
.
erforations i
P in the plexiglass, the liquids are fed and..run off
through three ho],es in the to and three holes in the bottom o
p. of
each half cell. .
1
.Cf. experiment no. 54.
2 .
These experiments were suggested by Prof.Dr J.Th.G.Overbeek.
The two cell halves are separated by a cellophane membrane 0.02 mm
thick,
while for t
he various experiments; they- were wi
Y th
solutions of potassium chlorine pro analysis, with c
_ oncentratiais
varying between 28 and 14 meq Cl 1. These liquids were
circulated
during electrodialysis by pumping at speeds of 0
.9 72.1 h. .
When current passes.silver chloride is formed on the od
an a and
the anodic liquid is desalted. On the cathode on t
. he othere hand -
silver chloride is converted into silver, an,d in the cathodic com-
partment salt
accumulation occurs. By reversing the olarit
g P y from
time to time and simultaneously interchanging the two liquids, con-
tinuous desalting of one of these liquids was obtained:
Determination of the chloride concentrations
.. was e
ffected con
ducti ..
metrically as there were no other salts in the solution and
the pH remained in the region of the neutral point viz. between
7 and 6.7 for the cathodic li uid
q and between 7.O and 5.7 for the
'anodic
liquid.
Results of the experiments.
a. Energy consumption and desaltin `effect.
The results of a number of experiments are given in Table
XVII. As the desaltin did not cove
. g r the range from 28 to 8 me q.
C1/]. on any occasion we shall try to deduce the order of
magnitude oft 3
he kWh consumption per mfrom the experiments.
p It a ears t
pp hat for the ranges 27-17 me C1-land 19-
q _ 6
me q, 1 the aver
q age amounts of energy needed for. a decrease
in concentration of 1 meq C 3
1 1 and expressed in kGh m are.
0.06 and 0.04 respectively. Nov
the energy consumption for de-
salting from 27-6 meq C171 is taken to
be composed of the
amounts corresponding to.the steps 27-18 and 18- -
,, ? an 6 meq C1 1.
Hence it becomes: .
x 0.0 3
9 6+ 12x00 ~. 1.Ok"J
4 hm
Extrapolation to the range 28-8 -
. meq C1 71 yields approximate-
ly 1 kWh m3 and a desalting time of
3M hours for 500 ml,
We have applied this method of calculation and have not
extrapolated
the result of every experiment directly to the
 desalting range 28-8 meq Cl l as extrapolation ,is not.. allowed, if
it has not as n t first been proved that the degree of desalting
is in-
dependent of concentration. The extrapolation now ? n r applied by us is
done over a very small range.
For potassium chloride solutions -values can be
q -values of
132 g KC1 100 amp.hr, which would correspond to q21aC1103 g NaCI
100 `amp .hr. This conversion is not entirely correct as the trans-
ference numbers of sodium
and potassium ions are note the same.
Nevertheless they do give an idea of
. the magnitude of q if sodium
chloride solutions had been desalted.
b..Trend of voltage and adhering of silver chloride.
--
the starting voltage of 0.3 v rapidly increased in these
experiments Cf. Graph 18 it was necessary to limit this as far
as possible-so as to keep ` down energy consumption-
and not allow
t to exceed a-certain value. Therefore, as soon as the voltage
exceeded this value reversion of polarity was applied.
The `voltage increases in consequence of worsened adhering to
the electrode of the silver chloride formed at the anode..
Adhering
appeared to be better 'when:
?
the silver plates are thoroughly degreased before anodic
chlorinationr ;.
a network of grooves with 1 x 1 mm squares and 0.
5 mm deep
is scratched in the electrodes
r
the
silver chloride 1 er is thinner. I
~Y, If greater current
densities aroused the intervals between the reversal of the
polarity must be much smaller in order to keep the p voltage below
the
above-mentioned value. rlith lower current densities however
the counteraction of diffusion
call always be greater.the voltage
can be kept constant for longer periods by higher rates, but an
increase cannot be suppressed.
reversal of the polarity is less frequent; the oftener it is
done, the show
ter the interval between two reversions will have
to be Cf. e.. ex
g .p 65, Table XVII and Gra h 1
p 8.
C. L embranes.
Particularly with higher rinsing rates it is very ufficult
to prevent contact between the cellophane membranes and the
electrodes, With the use of thicker membranes e.
. g. 0.09 mm with
the aim at retarding the diffusion, it is
g practically impossible
to avoid this contact without using membrane supports.
d. The silver content of the water. .
The quantity of silver ions in the desalted . solution was not
determined, so it is.not known whether
the desalted product is
suitable as drinking water. ,
e. Resulvi
The results can be summarised briefly as follows.
In a two-compartment cell with reversible silver-silver
chloride electrodes r desalting of half
a litre of water from 28
to 8 meq C1 1 caw be achieved
with an energy consumption of ap-
prox. 1 ktilh m3, provided various measures
are taken to make ad-
hering of the silver chloride as good as possible. This desalting
takes about 3 hours.
Silver-silver chloride electrodes will probably not stand up
to regular use, and with desalting as
above the cost of electrodes
will be considerable.
F. Discussion of the results.
1. Ex eriments
performed with the two-compartment cells.
With regard to the results of these experiments it should
be ointed out t
A hat. industrial use of the two-compartment cell
with non-reversible electrodes for d
esaltin water
g . with high
chloride concentrn
gtion (e.g. ea wato
r is justified. Thesug-
estion
g as attractive, since no heavy demands are
vY made upon `
the membranes or diaphragms. Further research in this
. direction
is very desirable.
Desalting in a two-compartment cell with reversible elec-
trodes opens up possibilities only if redox-electrodesscan be
found, of which the operating costs are not excessive. This is.
not so for silver-silver chloride electrodes.
 2. Experiments performed with the three-compartment
cell.
De"salting with the use of non- or only slightly selective mem-
branes and more or less acid and/or alkaline rinsing liquids will
not be considered owing to the low -values and the consequently
q high
energy consumption. Desalting with the use of selective membranes
offers better prospects at least for desalting brackish water.
On the assumption of a kWh rice of Fl. 0.0
P 7 an energy con-
sumption of 3 kWh m3 for desaltin water from 1000 g mg C1 1 to 300
mg 1/l l still leads o 3
8 t a cost of Fl. 0.~ ~1 m for. energy only. The
cost of apparatus, operation etc.. are roughly estimated to
amount to
approximately six times the above mentioned value. The greater
part
of these costs is formed by the cost of the magnetite anodes
and the
membranes. Thus we arrive at the conclusion that the total costs will
amount to approximately Fl. 1.50 m3 desalted water.
In view of the average rice of potable water
pride in the Netherlands
viz?. Fl. 0.20 _ Fl. 0.15 m3 it will be clear that industrial desalting
in a single three-compartment cell is not economically justif iad.
The factors determining energy consumption have been dealt with
at length in section IV. A low terminal voltage as well
as a high
q-value are necessary for ensuring a low energy consumption. Under
conditions which are also industrially
practicable high q-values
have already been attained viz. up to 82% o
P of the theoretical maximum.
A substantial further reduction of the energy consumption by reducing
the current density is not possible, so that the only solution is to
reduce the effect of the terminal. voltage fac
for on energy consumpt-
ion.
The terminal voltage can be reduced to
a limited extent by
working with thin compartments. From the results shown on page 91,
it appears that more
than 60% of p . terminal voltage is due to the
overvoltage at the electrodes. If reduction of this percentage were
energy consumption possible, of less than 3 kl;h m could be achieved.
According to data in the
g literature roughening the electrode surface
would result in a decrease of the overvoltage
. of approx. 0.5 v i.e.
in a 10% reduction in energy consumption. However, even this amount
is too small.
The only remaining possibility is increasing the number of
dialysate compartments per set of electrodes. If between two electrodes
one more unit consisting of two membranes, a dialysing and a rinsing . ,
compartment is interposed the total terminal voltage o '
according ~ the
data or pag. 91 ,. becomes 4.8 + 1.45 = 6.25 v. The portion accounted
for by overvoltaga is thus reduced to 48.?
Supposing a total of hundred of these units per then t
g P he
portion of the voltage drops at the electrodes is-only - whereas
for two hundred units it becomes 1% per cell. With
p p the compartment
thickness and current density given in experiment 49 see Appendix ;8
e ?~ .3
th re would be no point in reducing energy consumption per m3 by using
more than 200 units per cell. 1'lith such an arrangement the number of
magnetite anodes required will be reduced considerably viz. by a
factor 100 or 200. In addition the greater part of the membranes will
not be exposed to
. the corroding influence of extreme pH-values and
or chlorine which involves a longer life of the membranes. Hence
~ the
~'cost?'of apparatus, etc. will be considerably lower then the preliminary
estimate made on ^
Pg 98 for the case of single three-compartment cells.
The possibility of economically realising electrodial Ytic desalting
of water in a multi-compartment apparatus therefore seems to exist.
For this reason we give below an approximate calculation
of the cost
o 3
prie of water thus obtained for a capacity of 48,000 m per day,
which is about one tenth of the Netherlands' estimated requirement
in drinking water in the year ar A.D. 2000.
3. A tentative calculation of the cost rice of electrodial tic
desalting of brackish water from 1000 to 300 mg C1- l for a
3
capacity of 48000 m per d in a multi-compartment apparatus
with the use of selective membranes and magnetite anodes.
. The following calculation is based'uPon the conception of a
multi-compartment apparatus, subdivided into cells, units and corn-.
partments Cf. Fig.6 . In accordance with Fig.
. 6 the smallest items
are the compartments. Each unit consists of one rinsing compartment
and one dialysate compartment, inclusively two membranes one posi-
tively charged and one negatively charged). A set of '
unite together
with two electrodes and one extra rinsing compartment constitutes a
 -
cell. The total apparatus consists of a number of such cells.
In the calculation the number of units per cell is 'chosen at
200, whilst data of experiment 49 Cf. Appendix 8 are used.
.a. The required cell area.
The required cell area may be estimated if it is assumed
that as in experiment 49'a q-value of 180 g NaCl 100 amp.hr can
be attained. In this test the effective cell area is 580 cm2.
60 I.
From the data a capacity of - x 2 = 2,18 1/h is found.
55
The following area of each membrane species positive and
negative is therefore required for a capacity of 48s000 m3Per
day
1
48,000,000
24
x 2.18 x 0.0580 m`
53
500 m2.
The maximum width of the cell area is for the time being
determined by the maximum obtainable width of cellophane, so
that it cannot exceed 0.40
in. The height of the cells will pro-
bably be chosen in such a manner as to ensure that desalting is
effected in onloney pass through the compartment, provided the
further consequences thereof -such as concentration differences
within the compartments,
etc. -are acceptable, which would have
to be determined later. For a capacity of 2.18 1/h the circulation
rate was 13 1/h for a cell height of 30 cm. In order that a single
13
passage would suffice a cell height of 2.18 x 30 cm = 1.80 m or,say,
2 m would be needed.
The effective cell area thus becomes 2 x 0.4 = 0.8 2
m and
. 00
the'total;number of units 67,000> which in the case of
0.
200 units Per cell amounts to 335 cells.. This number of units
,.
per cell applies `only to the membrane spa~. spacing used in the ex-
periment. In the case of smaller membrane spacing the number of
units per cell will have to be larger so as to attain the same
percentage of voltage losses at the electrode per unit.
b. The cost of the membranes.
The.low current density applied is one of the reasons why
a very large membrane area is required.
For an effective cell
area area of 200 x 40 cm one should allow for a gross membrane area. of
say230x5O cm.For 6,000 units
1 a length of 67,000 x 2.30 a00
54, m is re_
u:red for
q each of the two kinds of membrane i.e.
a total of 308,000 m. At a cost of Fl. .-- per 100
75 P m this involves an ex-
penditure of F1. 231,000,_ excluding the costs o
g of preparing. the
membranes. If we estimate these costs at say F1. x.25 per straight
metre, we arrive at Fl. 77,000.- for the aforementioned 3
o8,aoo m.
The total costs for one set of membranes then amount to ~a rounded-
off figure of Fl. 310,000.-.
Hence it follows that the life of the membranes is a vital
factor in the cost-price of the water, since
the, membranes1 portion
of the total costs
at the current density selected become
,~ , s ac-
ceptable only if a prolonged life is ensured.
This fact is clearly
demonstrated in the following table= .
Life of
Total production
Total costs of
Costs of the mem
membranes
of desalted water
the membranes
-
branes per m3
in m3
desalted water'
1 week
336,000
P1.310,000 ,-
Fl. 0.93
4 weeks
1,344,000
1 3i0,a00,_
" 0.23
. 24 weeks
8 064 000
>
I,
31a,aao,-
0.04
c. Energy Consumption,
The current strength required per cell for a current density
of 2.6 ma cm2 and a 'cell area of $000 cm2 works out at 21 amps.
The terminal voltage per cell aver . ages '200
x 1.45 + 0.35 +
+ 3 = 295 v. Per cell a direct current power averaging 6.2 kW is
taken up. The total direct current power for 335 cells becomes
2080 k7, so that energy consumption' of the desalting amounts to
2080 3 . .
-, kWh /m or ,n at mnro +ie
48 000
24
 Approved For Release 2002108128 : CIA-RDP82-00041R0001001
- 102-
a. The costs of the magnetite electrodes.
M s.
For the manufacture of flat magnetite electrodes a price of
2
Fl. 200.-- per m magnetite area is reckoned (this refers to the
combined area at both sides).
By combining two cells the area on both sides of the magnetite
2
anodes can be used. The total anode area is 335 x 0.8 2 m = 268 m .
The price of a complete set of anodes is therefore Fl. 54,000.-. With
a life of 24 weeks the magnetite costs are 0.68 Dutch cents per m3.
.The price of the iron cathodes is neglected
e. Energy consumption b the pumps.
Assuming that the total amount of water has to be pumped to a
height of 10 m, with an overall pump plus motor efficiency of 50%a
the consumptionper m3 is 20,000 kgm or 0.06 kwh m3.
f. Wages operators.
This cannot yet be estimated. On the assumption that 20 r will
be required for one shift, the total is 60 men at hourly v/ages of
Fl. 1.25 and an eight hours' working day. Add to this an overhead of
200p % for mana$ement> administration> worshops, social insurance etc.
300
.., and the total amount of daily wages becomes x 60 x 8 x Fl. 1.25 =
100
= F1. 1800.-. These costs thus amount to 3.75 Dutch centsper m3.
g. Depreciation costs.
At this stage the capital investment cannot be specified, since
the correct design of the apparatus, for instance, has not been
finally determined. However, to get some idea of the extent of the
various items, it seems reasonable to assess the total capital in-
vestment on installation, buildings, laboratory, etc. at Fl.3:000,000.
Calculating 011 the basis of 10 years' depreciation the costs of de-
preciation for an annual production of 17.5 mill. m3 wa1
ger is 1.7
Dutch cent per m3.
h. Maintenance of the installation.
Allowing for an annual amount of Fl. 50,000.-
for
maintenance
of the installation, incl. lubricants, maintenance and cleaning
material, this works out at roughly 0.3 Dutch cent per m3.
Approved For Release 2002108128 : CIA-RDP82-00041R0001001'60001-0
- 103-
?0 Summary of the cost price factors discussed.
In the ,
preceding sections the principal factors deters
cost price of desalting brackish water have been dealt wit mary of these amounts will be found in the
following table;
is based on the assumption that electric power can be obt
a
Fl. 0.07 per kVh.
Survey of o eratin costs and calculating of
cost
Price per m3 w
, ater
Annual production 17.5 mill. m3 water
I
I Costs for
I
I
Fl. per
Year
Dutch cents
per m3
Perc?
of t
coE
.
Electric energy:
Direct current
f
18.2 mill.kl'dh
1,275,000.-
7.28
1
Power 1.05 mill.
kv ih
73,500.- 1
0.42
Membranes 4 weeks
4>030>000.-
1
23.--
6
Magnetite 6 months
116,000.-
0.68
j Wages
656,000.-
3.75
1
Maintenance
50,000.-
0.29
1 Depre Q
I ca.ation
i
3009G00.-
1.72
Total
6s500,500.-
3.14
?
10
The cost prase accordingly amounts to Fl. 0.37 per m
fore the advantage over the use of a single three-co
mpartme;
is obvious. In the latter the cost rice amounts to F1. 1.
p 5
(C f. 98).
The costs of the magnetite aodcare no longer of i
mp
With regard to the cost of maintenance and to some extent
depreciation, it is obvious that they do not yet affect the
costs.
0bvw1 membrane and energy costs
Y constitute the lar
items ar' any further reduction o
~ Y of the already attractive c
price has to by sought by reduction of these costs.
 Approved For Release 2002108128 : Cl
- 104 -
P82-00041 R0001 00160001-0
- 105-
The membranes account
, for more than 60% of the cost price, so
that we wonder
whether it might be possible to operate with. greater
current densities whICh wou
ld enable the membrane area to be reduced.
On the. other hand in
that case the energy consumption
will be higher.
However further reduction of the width. of cells might bring down
energy consumption. Further reduction of membrane costs is possible
if they are given a to
nger life and are made of ales
s expensive base
material.
As to the reduction of energy costs it should be
pointed out that
the figure given for direct current energy - 1 kWh .. per m is based
on previous experimental results and hence. can be considered reliable.
Needless to say the most important feature is the industrial product-
Ion of membranes which keep their original selective
. .., properties even
when they are used for
longer periods. This subject
has hardly been
examined yet, which is also the case with the construction of the
multi-compartment electrode
alysis apparatus on an industrial al scale.
The price of a kilowattho
ur is also of great importance. ?There
they are available at low
prices, it would be more economical to
use
more kWh's and a smaller membrane area higher current density). It
is thought that minimum total costs will be realised if membrane
and
power costs are roughly equal.
BRIEF SULiMARY.
The problem of water supply, already existing in some countries and
imminent in other countries is responsible for the fact that many coun-
tries are actively seeking an economical method
of preparing water for
domestic, industrial and agricultural use from sea or brackish water.
The principal existing methods have been examined and their costa
have been compared. In addition a number of surveys have been made of
patents relating to these methods.
The various investigations and calculations
appertaining to electro.
dialytic desalting of water and the ensuing processes at the electrodes
have been thoroughly examined.
As the energy consumption appeared to be
a vital economic factor all
elements governing energy consumption in e1e
g gY ctrodialysis have been
thoroughly studied.
Of major importance in this respect are the charge efficiencies in
the various membranes of the ions to be removed. With the aid of theore-
tical considerations it has therefore been
ascertained what conditions
affect these efficiencies.
In view of the prevailing confusion definitions have been au8Bested
of current-, current siren ht- current density-, charge- and coulomb
efficiencies.
Experimental research has been performed into the electrodialYtic
desalting of water '
g in a single three-compartment cell by means of non-
selective and more or less selective membranes with the use of acid,
alkaline or neutral rinsing liquids.
In
addition a number of ex e.Yiments has been carried out cancorning
desalting in
a two -compartment cell apparatus with non-reversible
electrodes and with reversible silver-silver chloride
electrodes.
i~~fi
Wth the help of the results obtained the construction of a multi-
compartment apparatus with selective membr
apes has been recommended and
a tentative cost calculation has been made for a capacity of 48000 m3
desalted water per d
ay
. 3
According to this calculation it appeared possible to des
ale I m
water from 1650
to 500 mg sodium chloride. Per litre at a cost of
F1. 0,37 per m3vihile further reduction of this
~ figure seems passible.
Approved For Release 2002108128 : (IIA-RDP82-00041R000100160001-0
 Approved For Release 2002108128 : CIA-RD
Approved For Release 2002108128 : CIA-R
Aultman, i;'.W.
2. Ano
nymus.
3. Chapman.
Calder, R.
14. Steinbach, A.
15. Akeroyd, E.I.
16. Showell, E.B.
17. Inoue, S., c.s.
Geologie en Mijnbouw 11, '141 1
_., 949..
Recent gxoundrrater investigations in the
Netherlands Monogr "
aphy Research in
Holland" series, Elsevier Publishing
Com ,Amsterdam New York, 1946
De I,ngenieur 6 4, A18.1, A195 (1952).
Mech. Eng. ?, 1039 (1930)
Paper read before. Inst. of Chem:
Engrs. by. 17, 1949 (unpublished
Rapport inzake de Gouvernements t7ater-
leiaing op curagao en Aruba (Rijksinsti-
tiuuti voor Drinkwatervoorziening,`Den
Haag, 1937.
Report on the Government Water Supply
on Cura ao and Aruba. Government In-
stitute for Vdater Supply, The Hague,
(1937)).
Eng? and Science Monthly 10, 8 (1947).
y , .
Process for removing dissolved salts
.. ........
from the liquid solvent.
U.S.?; 2,510,186.
29Aug. 1950,
Chemie I
ng Techn, 2 , 296 (1951).
Chem.
and Ind. 1951, 1187.
J. Am. -Vater Works Ass. 4~, 522 (1951).
Bu1.Inst. Tzekoku Jinzo Kensho
Kaisha Ltd. 1, no. 1, 34 ;
1.949~_
C.A. 4?, 49a 1951.
..
 20, Ziemba, J
22. Brigge, R.E.
Bull. Inst. Tiekoku Jinzo Kenaho Kaisha
Ltd. no 1, 39 (1949)4
C.A. 4477i (1951)?
Japanese patent 1 89340 (March 31, 1949)?
Food. Eng. 239 42 (1951)?
Trans. Chalmers Univ. Technol.,
Gothenburg 3 (1950);
C.A.'S, 3970 (1951)?
Civil. Eng. 16, 312 (1946).
24. Ionics Incorporated "Permionic membrane demineralisation"
(Cambridge Massachusetts) Bulletin no. 2, February 1952.
25. Anonymus.
26. Anonymus.
30. Manegold, E., Kalauch,C.
C'
31. Aten, A.H.
2. St berge
3 am r, P.
Chem. Eng. News ~, 693.(1951)?.
Daily News Bulletin U.S.2.5., March 3,
X552.
Science News Letter 61, 100 (1952)?
Dialysis and electrodialysls.
In: R.E. Kirk and D.F. Other.
Encyclopedia of chemical technology,
vol. 5, 1-26.
The Interscience Encyclopedia,
Inc. New York, 1950).
Dialysis and electrodialysis.
In: A. Weiszberger. Technique of
organic chemistry-Vol. III, pg.313'-
361. Interscience Publishers
,
New York, London 1950).
Elektrophorese, Elektro-Osmosf,,
Elektro-dialyse in Flussigkeiten
Th.Steinkopf, Dresden, 1931).
ease '2
35. Prausnitz, P.,
Reitstotter, J.
36. Billiter, J.
37. i11ig, K.
38. Illig, K.
39? Gerth, 0.
40. Hellweg, K.
41. Jackel, \'l.
42. Jalowitz, E.
43. Becker, J.
44? Jackel, Vt.
45. Sarrot du Bellay,
46. Patin, P.
47. Marie, L.R.
48. Bartow, E.,
Jebens, R.H.
49. Behrman, A.S.
50. Bartow, E.
51. Beuken, D.L.
52. 2hukov, I.I.
German P. 383,666; 394,360; 395e7529
39794; 498,48, 531,155; 579,023.
British P. 211,562;
French P. 619.0808 227x970?
British P. 352,103 S19313
43,695 11935~
U.S.P. 2,093,770 (1937).
nitch P. 29:194 (1933).
Ztschr.f.Angew. Chem.9 7085 (1926.
.Siemens Ztschr. 8 (6), 349 (1928)?
Siemens Ztschx. 12, 211 (1932)?
Die Y;arme.. 283 (1933)?
Ztschr. V.D.I. fl, no. 5, 132 (1939)
Ztschr.#.d.Brau and Malzindustrie:(1935)
Gembrinus. nr. 7.
Chem. Apparatus fl,
Gosundh. Ing. ~, 114 (933)?
La IZature nr 2761, 462 (127)?
Chimie et Ind. 1, 205 (1928)?
Science et Ind. 12, 96 (1928)?
Ind. Eng. Chem. 22, 1020 (1930
Ind. Eng. Chem. 7S, 1229 (1927)?
3. Am? hater Works Assoc. 22, 1115(1930)?
Ons iijdschrift (1933) 333?
3. Appl. Chem. U.S.S.R. 12, 613 (1946)?
C.A. 4873 (1947)?
Sliss. VerSffentl. sus dem Siemens
Konzern Q, 339 (1930? II).
Ind. Eng. Chem. ~}, 743 (1941).
 Jakovkin.
Nernst, W.
a
Sand, J.
Sand, J.
Luther, R.,
,. Mac Dougall, F.H.
60. Pedler, A.
61. Pebal.
62 Toerster > F.,
Muller7 B.
63. Billiter J.
64. Allmand, J.,
Cunliff4 P.Vi.
f
Maddison, R.E.W.
65. Sp~.nks, Porter, J.M.
66. Taf el .
67. Knobel, Caplan.
68.- Azzam, Bockriss.
69 Goodwin, Wilson.
70. Bockrsss, Parsons.
72. Moertazajen, A.
- 119.
Chem. Eng. Progress 43, 691 1947)?
3. Russ: Ph;/s. Chem. Soc. 32, 673 (1900).
Z. Phys. Chem. 48, 601 (1904).
2. Ptys. Chem. Q, 465 (1905).
Z.-Phys. Chem. ~, 477 (1906)i 62+
199 (1906) J. Am. Chem. Soc. 896 (1920).
L ebigs Ann. 231, 144 (1885)?
Ztschr. Li Elektrochem. 8, 429 (1902)?
J. 9m. Chem. Soc. ~, 264 (1934)?
z. Plays. chemie Q, 641 (1905).
Trans. Am., Electrochem. Soc. 4}, 55,
(1923)?
Nature j6r, 403 (1950)?
Trans. Am. Electrocliem. Soc. 4Q,
173 (1921).
Trans._Faraday Soc. ~, 916 (1949)?
J. AppL. Chem. U.S.S.R. 22~ 572 (1949)i
C.A. 44, 52d (1950)?
J. Plays. Chem.. U.S.S.R. 1247 (1929)9
Morn. Inst. Chem. Acad. Sci. Ukrain.
S.S.R. 6, 45 1939)s 3, 527 (1941);
c.A., 91544 (1939), C.A. 3j, 33509
(1.943)?
75. Bockriss, Parsons.
76. Bockriss9 Conway.
770 Hickling, Salt.
78. Hickling9 Salt.
79. Goodwin, Knobel.
80. Hickling, Salt
81. Hickling, Salt.
82. Hickling9 Salt.
83. Hio
kla.ngs Salt.
84. V7etterholm.
85. Birett9 Vt.
86. Cupr.
87. Piontelli..
88. tiVells.
89. Tammann.
90. Teorell, T.
91. Meyer-Sievers.
92. Planck, M.
94. Nernst, Vt.
Trans. FaradaSoc. 4, 860 (1948)?
Nature j59, 711
Trans. Faraday Soc. , 3, 350 (1941).
Trans. Faraday Soc.'
8, 474 1942.
Trans. Am. Electrochem. Soc. , 617 1920
Trans. Faraday Soc. , , 319 1941
Trans. Faraday Soc. 6, 1226 1940
Trans. Faraday Soc
Trans. Faraday Soc. ~, 861
German Pat. 701,803 (1940).
Chem. Listy. 32, 215 (1938).
C.A. 32, 6955( (1938). -
Atti acad. I,incei Classe sci. Pie. rat.
nat. .?:Lr 581 (1938).
Trans. Am. Electrochem. Soo. 22
(1972), -
Ztsahr. f. Elektrochem, ~, 460 (1950?
Hely. chin. Acta. 1Q, 649 (1936).
 95. Guggenheimf E.A.
6.
9 Stenler V.V.,
Sirak, I,I
Davidson,
9? G.F.
9
98. ears
h, J,F,,
Wood9 D.C.
99Moore, E.V1.
3. PAj.s. Chem. 33, 842 (1929)?
Trans. ilectrochem. Soc. 68, 1-27 (1935)
(Preprint).
J'. Textile Inst. 4, T59 (1948).
3. Soc. Chem. Ind. , 79 (1933),
1
Costs for
Method
Diesel fuel oil
Power
Chemicals
'Total 3
2 cent kvni
South
cost.
9.5 cent/1
11.7 cent/1
Califor-
,
200:12)
.
220:12)
nia
I Vanourcom-
4
pression
54
63
125
distillation
.
Freezing out
125 .
Ion-exchange
500
Electrolytic
process of
~
s 5
Bri
g
90
1 3
The cost is shown in Dutch cent perm produced water. Prices of
'craw materials" are given in the table in various columns.
2
tiiater to fuel ratio see text).
3
Cost at sea level, excluding cost of distribution. Allowing for pump-
ing g from sea level, labour, interest, amortization, road factor,
distribution and so forth, the actual average cost might 'total more
than three times the bare fuel, power or chemical costs.
4
,iris figure originates from Leicester 8
5
See section II.A.4.a.
A. T . A.-T . IT. 0.
T.A. No 270.
Table I.
08128 ; CIA-RDP82-00041 R000100160001-0
IA-RDP82-00041 R0001 00160001-0
 Approved For Release 2002108128 : CIA-RDP82-00041R000100160001-0
1
See note 1, table I.
2
See note 2, table I.
3
See note 3, table I.
4
These figures originate from Aultman
A.T.A.-T.N.O.
T.A. No 270.
Table II.
Comparison of operating costs of va pour compression distilling
units and conventional steam heated evaporators 1
1,8
~
evaporators
I steam
vapour compression stills
Costs of
~
single
double
triple
quadru-
electrical
diesel drive
~
effect
effect
effect
pie eff.
drive and..
with exhaust
!
heating
heat exchangers
steam 9
based on cost of
1079
555
388.
304
21
21
steam at 0.9/kg
condensor water,
based on cost of
water at 26 m3
84
42
26
21
-
-
electric power 9
based on cost of
power at
4,4/kWh
12
8
8
8
_ 73
fuel 9
based on cost of
fuel at 9.5/1
(1:200)2).
544
id. at 11.7 l(1;220)
63
water and energy
1175
605
422
333
94
84
.
total costs 3
125k)
Approved For Release 2002108128 :JCIA-RDP82-00041 R000100160001-0
 Approved For Release 2002/08/28:
CIA-RDP82-00041 R000100160001-0
The mobility of different ions at 25"C and at
a voltage drop of 1 volt/cm, ex ressed i 04
p n l cm.sec
36.2
OH
20.5
K'
7.6
;,1'
7.9
Na+
5.2
so4'
8.3
Increase + 2% per ?C rise in temperature
The hypochloric acid concentrations in
the anolvte for v 1.
arious conditions.
Px
~ [cc]
inn
p
in atm.
[HC10
in 10n
2
0.1
1
2.150
2
0.1
0.1
0.215
2
0.000007
0.0001
2.150
i
0.000001
0.0001
0.215
2
0.03
0.0001
0.00002
~ 7
0.03
i
0.0001
7.2
See calculations section III.B.
2
A.T.A.-T.N.O.
T.A.No 270
Table III,IV.
Approved For Release 2002108128 CIA-RDP82-00041 R000100160001-0
 Approved For Release 2002108128 : CIA-RDP82-00041R000100160001-0
'1
According to Billiter a potential difference smaller
then 20 volts is generally most favourable, preferably
between 8 and 16 volts.
A.T.A.-T.N.O.
T.A.Na 270.
Table V.
The electrodialytic desalting of v+ater with application
of a hydrostatic ressure diffeme c
p e 36
Conditions
Water rich in carbonate
and sulphate
Carbonate-, sulphate-
and
chloride- contain-:n8 water
Average voltage per cell v 1)
~
1.3.
75
13,75
18
Current density ma cm2
0.5
1.5
5
--
1.5
5
Hydrostatic overpressure
! mm head of water
26
58
200
Diaphragms
linen
linen
ceramics
Quantity of water moved per
hour through the diaphragms 1
0.58
1.6
25
Quantity of water purified
per hour
p 1
10
10
10
Dry residue y ue of the water mg 1
untreated
159
59
620
62
1320
Dry residue of the water mg 1
free of chloride
treated
9
8
124_...
Energy consumption
,
in -Vh m3 mg salt
44
32
37
Approved For Release 2002108128, CIA-RDP82-00041 R000100160001-0
 Approved For Release 2002108128 : CIA-RDP82-00041R000100160001-0
The potential of magnetite electrodes in some
solutions according to "ells 88
t
solution of
pH
p
potential v
sulphuric acid 1 n
+ 0,3
+ 091
-
potassium chloride 5%
+ 5.5
+ 0.40
sodium hdro de 1
3r XY n
+ 1
_ 4
- 0.
22
E' of some cells with a magnetite electrode
according to Tammann (89)
~
cell
EEC`
-Fe30q
Fe304/ZnSO4 soln. 4n/Zn
+ 1.20
+ 0.40
Fe304/PbC12 soln. satd./Pb
+ 0.60
+ 0.41
Fe304/Cuso4 soln. 2n/Cu
+ 0.14
+ 0.44
Fe3o4/Ag2504 aoln. eatd./Ag
- 0.30
+ 0.39
A.T.A.-T.N.O.
T.A.fio 2'j0.
Table VIA VII,
Approved For Release 2002108128 : CIA-RDP82-00041R000100160001-0
 Approved For Release 2002108128 :flA-RDP82-00041 R000100160001-0
Is
I.~
H
SJ~
H
H
.J .q
Cq 0 II
cr, ?ri
II .
~
+
, t4 p1
LI'
0
i
N
0
U
N.
U
II
+
UI to
0) q) a)
C) m
H H
C
H
H H
C) H
U ,0 U
A C)
E n
o a) rh y )
E E 0
,0 N a) 'd N a)
a) U 4D a) 0
Cd
H `ii. UI II U)
N UI U)
0 0 a U U p,
Approved For Release 200210812 : CIA-RDP82-00041R000100160001-0
S 5'.
r- 'd
0)
N :
U .r?
... 4a
a)
~ b
r0
H Lr\
0\
a) r_s y.l
0
UI W UI
a) ,/` cd
H
s~ 0 0)
+'
.'Ik rn a)
. C;
a+ .c
0 10 0)
?
b o
a) 43
p4 .r # m ,a
H ~- a
/.l jam) 'r+..+ .0 .: - . ;?
-P. S
y f~
0 0o
0) Cd y
a) ,q ~.
cd v .,1
0 >~ 0. is 4i
+' p U N . 0
H ?r14
,'..I' a"i a N
.~ a)
H 0 + 'p. ~..
0 0 p I
4_ H. U)
43+
43 0
p4 a) 0
0 H I
m a
!S a)
a) a) II H t~
`-d ,a 0
43 0.
~! + UI
H
0
C
p
+3
(4)
0 Cl)
m
v-I
43
0)
m
U)
v
HI H 0
2 Cd a a
C4
01,0
+, +cd
U) U)
I'. pi
0
.- N
A.T.A.-T.A.O.
'P11.. 1io 270.
Table VIII.
 Approved For Release 2002108/28 : CIA-
ii
DP82-00041 R000100160001-0
0
r1
ri
?
0
A
<
k
?
?
^
0
o
0
b
I
I
0
o
`o
1
~o
6
n '
r1
N1
N
M
a
r
0
R
M
.v
n
?-
M
0'
..
0
N
0
O
v.
0.
0'
V
a
1
o
'
eo
?
w
,v.
~-
F
0
I-
N
?
D
I-
~
?-
e1
? .
?-
?-
?
?-
I-
I-
?
N
i
0
m
o
0
0
0
0
0
0
0
0
0
0
p
N
I-
N
ti
?
M
? r??
0??A
5
U
$4 4
b
.0
o
?0
0)
0)
o
.0
.0
~0
.
VN/v
.1
.."
_
b
14
,-
0
oD
0
'
0.
?
u\
M
0
0
~0
?
C
d
.a
0
,v
.
,
..
w
1
r
0
b
N
0
ti
t
?
,.
m
I
'
?
'
A
a
P
m
0.
0
CD
t0
0
0
0
0.
0.
..
h
r
2
~
a
M
. .
a
,n
( ?
,.
I
n
? A
?'\
N
?- ?
It\ ?
i- ?
Ifs ?
,- ?
N ?
-
A. ?
IA ?
h ?
U1 ?
N ?
h
p
O.0
0 0
ti
O ? 0
o
?
?
?
O q
?- ?
o a
?- ?
? 0
,- ?
0
.- ?
? 0'
.- ?
.;. ?
3
,- ?
?
p
.-
?. 0
o v
O
O
d
0
b
0
O b
?
i
..
:
4
o
.
0)
0)
0)
~
O
0
0
0
0
o
E
1
y
j
$
r
M
M
M
M
M
.
M
r
M
9
r
.
?
M
M
M
?
M 0
If ?
N?
? 0
N?
? 0
IA ?
N?
? 0
?
+
0
IA ?
~i p
In ?
~? ?
IN ?
N?
IA ?
I U
IA ?
04
?
?
N .4
O d
o b
o b
I
1
?
0 q
O v
b
?
0
?
?
0
4
r .1
O
b
?
,.
N
0
0
0o
0
0
0
9
e
v
?
0
U
4g
U
. o t
0]
~1 p
t) 01
'~ oo
V 01
.y p
T
,y
? N
M RI
a
M pt
*
M
4 I
!
i
V
M
U
k
4
i7
4
?
?
to
X
!
Oa
M
M
M M
M0. ?
r
n
M M
I-
N
P1
?f
IA
Vf
j.
01
O
Approved For Release 2002/08/28 : CIA-RDP82-00041R000100160001-0
T.1Jo. 270.
T?bl? U.
 Approved For Release 2002108128 : CIA~RDP82-00041R000100160001-0
0
?d
r,
.
1-4
r4
?I4
14
l
-4
ri
V
e
d
0
0
0
0
0
0
1
0
..4
o
rl
rl ?
0
a
oo
A
0
.,
N
?
?
~y
ti
?
-
~,
! a

5_
g
.1
.
U
V
^
I
aC
0
o
0
0
g
V
r?1
^
I
v
,_
5-
\ ?
S
.
? 0
.
.0
I-
?
054
?
a
~
.a
^
~
1
E
1
11+
U
i
o
&
I
5- 0
R0
m
O
P1
8
0
^
v-
p4
ma
n .
s
.
5-
I
a^
I +
+
I +
I +
+
o?
I1 M
00 I:
Iq M
A
a.
ate'
0 O
r1R w
g29
0 0
~r 5
0
M.
0
?
?
~4
a
2
r
N
..
w
^
y
- , 4' '
~~???
a0 .4.
. M -?
S ,4
9}' M 0
N +r
M M
.
U p.
G N. ? ..
0 ?i4. .4
gujso
..
o v' ~-
T.l.Io? 270.
Ttbi. ZI.
Approved For Release 2002108128 : C1A-RDP82-00041R000100160001-0
 Approved For Release 2002108128 : CI -RDP82-00041R000100160001-0
H
0
p,
0
H
U)
?rI
i'
fti
r~
a)
'a
.r?
H
U
N
a)
b
N
a)
H
a)
E
a)
?1-1
4)
U
a)
rl
a)
N
41
0
a)
r1
a
0
U
cd
0
l~
Qi
b
~n.wwxxwl ~rx
~ .M.NMMYIM
YMMw
MiNw~FMM
~MNM1Y~
NwxM
wwwW~
r
O
?
In
0
0
1
I
I
'
?
n
\
r
o
0
E
0
O
?rI
_x~.,.xx.
..
x .
...,.,
.
.
-
r
.?
?
?
O
0
t
I
t -
ap
N
o
d
d
d
r
O
?
O
w-._-...
-
~
?rl
...
If'
1~
?
-
r
N
lf1
U1
d
00
\
d'
d'
N
N
r
r
?
0
U)
U)
a) n
0.
U1
t11
0
1
rh
-
O
rn
Uv
?
?
1
1
?
?
N
?
t-
?
.~
0
0
0
0
0
0
...,>r....,........
1.
.,,..
.f-
.- - _
.
A
a)
'rl
U
r1
N
0)0
n
U1
I
H
~
0
ti
ap
op
o
'
>
+
r
r
r
r
Nl
d
.,1r,a
N
Cl)
~- v
a) cH a) ?
c40
0
r
a)
t
4D
N
?
cP
4
r
.0
U1
\0
111
~ ai
40
a)
N-
?
I
M
?
r-
?
Nl
r
r
O U
0
0
0
?
0
1-
0
k
.~
a)
4)
I
....
-
a)
U
H
rl
p
,a
?
?
E
?rl
-
:
0
0
0
a)
E
a)
W
?
;o
?
H
H
a)
'4
0`
0
N
?,
a) r
H rt
P 4::.
9H
. a1.
E a
4.).
?r l.
U
q.4 cd
0
rd
. , C)
0
,,-I ?,.I
cd
U) ar
a)
b
H
4.) -0
C)
0
a)
,-1"
k ~
00
W U
m ~.
H
?r4
bc ?
~ U
.i
C 4+
H C) U)
11 cd
0 HU
r~
~b
0)0)
H +)
0
4i r1
r? 0.rl
U b
a ,a
? Id +'
0~ U ?rl
a)
~ m
A.T.A-T.N.O.
T.A. No 270
Table RII.
Approved For Release 2002108128 : CIA-RDP82-00041R000100160001-0
 Approved For Release 2002108128 : CIIRDP82-00041R000100160001-0 '
Influence of current densitYon 4cpsckwith the use
of membrane combination ARX 102 - CR 51 1
exp. current
no density q30-20
ma cm2
23
1.9
17
24
3.1
16
25
5.8
26
10.0
a(20-1o) I n(30
1
cell A2; middle compartment 6 mm wide;
circa rate d' -10
zalysate 8-10 1/h.
h
A.T.A.-T.t1.0.
T.A. No 270.:
Table XIII.
Approved For Release 2002108128 : -RDP82-00041R000100160001-0
 Approved For Release 2002108128 : Cl
Influence on q of current density and stirring of the dialYsate
with the use of different combinations of membranes _1
eX
.
' '
1 current
~ density
, Q
f g NaCl/
dialysate
anodic
no
1.
~ (ma/cmz
f 100 arnp.hr
circ.
stirring
membrane
27
~ 1.9
174
-
_
A19
{ 28
1 2.4
168
-
-
A19
29
I 2.8
150
_
A19
30
i 3.3
90
I =
-
A19
31
~ 6.5
132
_
_
A19
32
~ 10.8
132
-.
A19
33
40
126
- .2.)
A 19
~ 34
2.7
192
+
A19
35
6.5
l
180
-
+
A19
36 ~
10.0
150 -
-
+
A 19
37
1,9 I
156
-
-
nax 44
38 ~
10.8 (
102
-
ARX 44
39
2.7
210
=
+
dR7C 102
~ 40
6.5
186
-
+
ARX']02
41
10.8
132
-
+
ARX 102
42
10.8
54
-
-
ARX 102
cell A 1; middle compartment 31 mm wide;.
cathodic membrane CP 51
in this experiment the middle compartment was 11 mm wide.
A.T.A.-T.N.O.
T.A. No 270.
Table XIV.
4-RDP82-00041 R000100160001-0
Approved For Release 2002108128 :CIA-RDP82-00041R000100160001-0
 Approved For Release 2002108128 : C RDP82-00041R000100160001-0
x_
43
0
0
0
r
N
0
P4
0 0 0 0
0, c0 U\ r-
r' N N c-
0
N
0
0 {~
4i
N td
Es
4i
ld Cl)
U1 d I(\ I- N
rI i
N
r r n
br
N
N rl
0(55
U)
43
4i 0
>7 'r?
4.4
0 ~
?ri b
(Si
td
O N
+'O
?rl ~.
40 0
4),-'
cd p
E~
+)
PH
m`
c Do
4D i~ ri
e- r- 'd' d' r- r
4)r-4
P HU
. ?rl fd
f-fl m N tv r~ r'
0
N
Cl)
Cl)
43
0 Cl) 4l
s~ P 0
0 .rr H
OHO..
0
i3 v
4.3
?rl
U) .
P1 M
'rl U) U)
0 .--' 0)4-4 ,f-I ?ri
0 p% ..#;::f' r r
(k U) CSiF.4 v Lr U1
H O N H 0 ? ?
A +) cd ,4) ri H a
0 .rl 0 D ?rl
0 0
E4-s.. E 4-s 0 0
0\ a~
r '-
H
0
0
. H H I 1
1(1 lh N '0 rd
N o0
1" r
.. L '0 ti m
,- ~t .-
r
I
U)
H
H
m
0
A.T.A.-T.N.O.
T.A.o 2']0
Table XV.
Approved For Release 2002108128 : CIA-RDP82-00041R000100160001-0
 Approved For Release 2002108128 : Cl k-RDP82-00041R000100160001-0
Influence of chloride concentration and pH of the anolyte on the
removal of chlorine at the anode in a two-compartment
cell with non-reversible electrodes
Q(Op) ,
(g NaC1/100 amp.hr)
Exp.
no.
chloride
content
me 1
active chlorine
content end
me41
pH
aver.
52`)
30
15
53
30
5
7,0
543)
30
0.4
1.1
552)
20
5
1.7
56
20
5?1
57
20
7.0
582)
10
4
1.9
59
10
4.2
60
70
7.4
1
a diluted solution of sodium chloride circ. rate 6 1/h
dos. rate catholYto NaOH. 20 me41 0.15 1 h.-
2
Anolyte; a diluted hydrochloric acid solution.
rate 100 1/h to decrease its active chlorine content.
3
outside the cell air was passed through the anolyte
A.T.A.-Z.N.O.
T.A.No 270,
Table XVI.
Approved For Release 2002108128 : CIA-RDP82-00041R000100160001-0
 Approved For Release 2002108128 : Cl RDP82-00041R000100160001-0
a
N
p
S
.. ,-
a tp
pp
N %O ;
w
oaf
m m
I
a
$
\'
a~i
o~
d v
~+1
*
0
,?
0
1
'o
O
toN
0
-
'
ql
~o
N r{ (;
?4'
q a
N
IA
N
.t ?0
..I
~
0
0
0
00
w I
0
0
0
00
-
?
yy
y, M
MM .
4 k
r
NS
d
V
It%
b
V
om- r
p
t
NO
0
0
0
00
U
.,4
..
N
o Ir
O
aD
~o N
N M
O B
g
ri R
N1
N
N N
f~ f+)
N
d~a
o.
$
H
NH
H N M
M M
b 0
r?
N
r -
N '- i-
?- ~
0
V
rl
l~,Q
.4\
v
?
0
N
'-
N
N
N N
t.- !-
0
M
n
1
U
1
0 e
q m0
N
N
~p
N
~p
N N
8~
0
M
v
d
o _
M
tT
It1 111
-.-
.
0
?
0
.
0
? .
00
n
I
yy~
0
N
N
O
e ti
,
~
~
I
0
9
o,
00
N
N
a
i
r
0?
,-
oti
,
"N
0
a
v
o _
K1
C
G G
-..
?
0
1
0
.
0
. .
00
?
?
i. .. ..
5- N M
T.L.Ao.270
Tdblo XPII.
Approved For Release 2002108128 : CIA-RDP82-00041R000100160001-0
 Approved For Release 2002108128 : CIA-RDP82-00041R000100160001-0
400
U 300
3250
200
150
ioo
50
(Total cost
1
\ Power cost
I I __ --
Amortization cost ? 15 yr
Operating cost*
34
*Labour, maintenance, supplies, etc,
10 11 12 13 14 15 16 11 18 19
Capacity m' er hour -
~ P ) --r
EFFECT OF CAPACITY ON OPERATING COSTS
A. T,A,?T,N,O,
TA-no 270
GRAPH 1
Approved For Release 2002108128 : (~4-RDP82-00041R000100160001-0
 Approved For Release 2002108128 CIA-RDP82-00041 R000100160001-0
0.88
0.84
0.42
0.28
zs,
20?
0.9
o.s
0.7
0,6
o.s
-1
i
p0
?
20
24 ma
60a
20o0
4 ma/
ma /
600
amp. hrs
GRAPH 2.THE COURSE OF THE SODIUM SULPHATE
REMOVAL AGAINST THE NUMBER OF AMP.
HRSAT DIFFERENT TEMPERATURES
0.4
0.3
60
/cm ?
cm 2
cm
16 32 48 64
%z SO
4
20G
iz
16 20 24
current density (ma/cm
GRAPH 3. RELATION BETWEEN REMOVAL
OF SODIUM AND SULPHATE
IONS, CURRENT DENSITY AND
TEMPERATURE
THE EXPERIMENTS OF HOFFMAN
60
12 - 16 2(1 24
-
current density (ma/cm2)
GRAPH 4. RELATION BETWEEN ACIDITY
OF THE DIALYSATE, CURRENT
DENSITY AND TEMPERATURE
A,T.A.?T.N.O.
T.A.?no 210
GRAPHS 2,3 & 4
Approved For Release 2002108128 : CIA-RDP82-00041R000100160001-0
 Approved For Release 2002108128 : CIA-RDP82-00041R000100160001-0
v
w
a
b
c d
? 5 -4
-3 .2 . 1
0
----~ log i (amp,),
a, in nitrogen atmosphere
b, in contact with air
C. air passed through (low rate)
d. air passed through (high rate)
INFLUENCE OF OXYGEN ON THE HYDROGEN OVERVOLTAGE AT A
Hg-ELECTRODE IN HYDROCHLORIC ACID (in)
A.T.A.-T,N.O.
TA-no 270
GRAPH 5
Approved For Release 2002108128 : CIA-RDP82-00041R000100160001-0
 Approved For Release 2002108128 : cA-RDP82-00041R000100160001-0
Effective potential
DIAGRAMMATIC EVALUATION OF THE EQUATION OF TEORELL FOR
THE INDIVIDUAL CURRENT DENSITIES CONNECTED WITH THE VARI-
OUS ION SPECIES'AT PASSAGE OF CURRENT AND EQUAL TOTAL
CONCENTRATIONS IN A SPECIFIC CASE
in the membrane (my)
+116
A.T.A.?T. N. 0.
TA-no 270
GRAPH 6
Approved For Release 200210812 CIA-RDP82-00041 R000100160001-0
 Approved For Release 2002108128 : C'4-RDP82-00041R000100160001-0
i8 =pos.
GRAPH 7, CURRENT DENSITY EFFICIENCY OF THE CATION IN A TWO-ION SYSTEM
AS A FUNCTION OF CONCENTRATION DIFFERENCE(c2- cl) ON EITHER
SIDE OF THE MEMBRANE
2 ma/cm 2 1 ma/cm
3 ma/cm 2
9 ma/cm 2
21 ma/cm 2
a: ma/cm 2
-iz
-8
?4
i 0.30
nCI
+4
.f8
+12
(cl - c2) k 1O
9 q )
GRAPH 8. CURRENT DENSITY EFFICIENCY OF THE CHLORIDE ION IN A TWO-ION
SYSTEM (HCI) AS A FUNCTION OF THE CONCENTRATION DIFFERENCE.
(cl- c2)
b
GRAPH 9, SCHEMATIC DRAWING OF THE C CONCENTRATIONS OF THE IONS IN A
THREE-ION-SYSTEM
CURRENT DENSITY EFFICIENCIES AND ION CONCENTRATIONS IN A
SIMPLE"MEMBRANE
Approved For Release 200210812 : CIA-RDP82-00041R000100160001-0
A,T,A.?T.N.O.
TA-no 270
GRAPHS 7,8 & 9
 Approved For Release 2002108128 : IA-RDP82-00041R000100160001-0
Concentration dialysate
30
20
io
(meq CI /I)
Exp, no 1; anolyte containing 0.01 n H zSO
a
Exp, no 2: anolyte containing 0.02
n H2S0a
Exp, no 3: anolyte containing 0,03 nH2S0a
Exp. no 4: idem, rinsing rate increased twelve fold
1 2
ACCELERATION OF DESALTING BY DECREASING THE ACIDITY OF
THE ANOLYTE (CONDITIONS CF TABLE IX).
~ I. Time
(hours)
5
-- amp.hr
A.T.A.-T.N.O.
TA-no 270
GRAPH 10
Approved For Release 2002108128' : CIA-RDP82-00041R000100160001-0
 Approved For Release 2002108128 IA-RDP82-00041 R000100160001-0
io
Concentration dialysate
(meq CI /I)
30
0
z
0
.
_
_
_
_
_
_
-
_
_
_
_
_'
N
.
4
1
8
Exp. no 7,8: bakelitised anodic
cellophane membrane
Exp, no 9: untreated anodic
cellophane membrane
L
2
3
Time
(hours)
1 2
5
---~r- amp,hr
DESALTING CURVES OBTAINEDINEXPERIMENTSWITH BAKELITISED A.T.A.-T.N,O.
AND NON?TREATEDANODIC CELLOPHANE MEMBRANES(CONDITIONS TA-no 270
CF TABLE IX) GRAPE; 11
Approved For Release 2002108128 : CIA-RDP82-00041R000100160001-0
 Approved For Release 2002108128
-
E~
'0
Lt)
CIA-RDP82-00041 R0001 00160001-0
0
0
Vl
;
0
C
a
I-
.a
u
0
-c
o
u
vuvvm
C C C C -
00000
aaaaa
00000
~,'mvvm
U U U U U
E
m
-
o I-
U
00
d 0 u d
o v
0
C-_ P 0
a
a
?C
0 U
-
?:- a
a
0 ]-
v0
--
` au-
u
_o
u aQ
4-
C
'z: 0
Nth ~f ~D
v C
?- ,_ .- ,_ '-
a
x
W
DESALTING CURVES OBTAINED IN EXPERIMENTS WITH ANODIC A,TA-
T,N,O,
CELLOPHANE MEMBRANES TREATED IN VARIOUS WAYS(CONDITIONS TA- no 270
CF TABLE X) GRAPH 12
Approved For Release 2002108128 : CIA-RDP82-00041R000100160001-0
 Approved For Release 2002/08/28~IA-RDP82-00041 R000100160001-0
n
N ~
_ _/ _ 7-- _-
0
_I/ri
G ~
E~
l
/ :
-/
II y
I __:i-i__
C
(N
0
U
x
0
`
y
0
0 0 C
D
0 0 0
2
C C _C
-
aaao
000-
In
G G v ~?
V
U U U Q
C
0
V
n
E
G
:
d G
~a
0. 0?
v 0 0 G
?0
C C
.-
L
C
.
-
0
a ?_ ,_
o
o ~, m
0 Q1
u _a
.o u
f
c
.` 0
NN.o3 0.
a
x
W
DESALTING CURVES OBTAINED IN EXPERIMENTS WITH CATHODIC
A,T,A,?T,N,O,
CELLOPHANE MEMBRANES TREATED IN VARIOUS WAYS (CONDITION
TA-no 210
CF TABLE X) GRAPH 13
Approved For Release 2002108128 : CIA-RDP82-00041R000100160001-0
 Approved For Release 2002108128 IA-RDP82-00041 R000100160001-0
..
E~
__
__
__
it
(-
I
~- o
.c
__
__
__
N
i
n
;
p
__
__
__
__
__
a
7
i
I
/
M
Qt
C
.
'
/
C
~
I
/
~ C
C ?Y
,N 0
!
J
/
/
I
I
:
av
N
N
0 d
< a
?a
< }
a
3 _.
.-
O N
,..
-
0 0
ji
/
-
-
/
_
-
_
_
M
/
_
_
_
_
_
-
I
I
RETARDATION OF DESALTING WITH NON- AND SLIGHTLY PERM E? A, T, A, T,N,O.
ABLEMEMBRANES BY USINGNEUTRAL RINSINGLIQUIDS
(C0NDITIONS TA-no 270
CF TABLES IX AND X) GRAPH 14
r
a
E
0
Approved For Release 2002108128 : CIA-RDP82-00041R000100160001-0
 Approved For Release 2002108128 :GIA-RDP82-00041 R000100160001-0
Concentration dialysate
(meq.CI /I)
3.1 ma/cm
10.0 ma/cm 2
amp. hr.
RELATION BETWEEN CHLORIDE CONCENrRA
TION OF THE DIALY. A,T.A,?T,N U
SATE AND THE NUMBER OF AMP,HR. AT '
DIFFERENT CURRENT TA-no 270
DENSITIES (CELL: A2i MEMBRANES: ARX 102
CR 51) GRAPH 15
Approved For Release 2002108 : CIA-RDP82-00041R000100160001-0
 Approved For Release 2002108128 :FIA-RDP82-00041R000100160001-0
Desalting effect
(g.NaCI/100 amp.hr)
218
(max.
180
120
60
Wit
hout st
With
9 10
Current density
(ma/cm I
RELATION BETWEEN THE DESALTING EFFECT AND CURRENT A.T.A.-T.N,O.
DENSITY WITH AND WITHOUT STIRRING OF THE DIALYSATE (CELL TA?no 270
A2; MEMBRANES: ARX 102 - CR SI) GRAPH 16
Approved For Release 2002108128 : CIA-RDP82-00041R000100160001-0
 Approved For Release 2002108128 : CIA-RDP82-00041R000100160001-0
Desalting effect
(g NaCI/100 amp, hr)
218
(max.)
I80
izo
60
With stirring
Without stirr
ing
to
Current density
(ma/cm 2
RELATION BETWEEN THE DESALTING EFFECT AND CURRENT A.T.A.T.N.O.
DENSITY WITH AND WITHOUT STIRRING OF THE DiALYSATE (CELL TA-no 210
A2; MEMBRANES: A 19 ? CR 51) GRAPH 17
Approved For Release 200210 8 : CIA-RDP82-00041R000100160001-0
 Approved For Release 2002108128 : CIA-RDP82-00041R000100160001-0
>.o

L
-
-
~v
---
_
)
__
.. u
_
_
~~
ou
o>
N
av
-
_
-
-
_
--
-
k
-
r
4'_
( 'l'. P.v i . F + f(v4.F 2 RT, p , v +
R
11
This is a differential equation of the type.
From this it follows:
In the latter equation we multiplied by A + 1 on the understanding that:.
A.T.A.-T.N.G.
T.A.No. 270
Appendix 7.
 Integration of (13) gives
A+1
1)w + B = 0 C= integration constant
. C t(L+1)w+B} '
14) ;, C {(A + i)+ B.(} '-
BY writing down equation 14 for compartment 1 and for compartment
2 the integration constant can be eliminated. Unless
(+ i)>+B. r
i.e. C = G7
This is9 we should note, the very case if we convert the three-ion-
system into a two-ion-system see appendix 79 section III). Only in this
event therefore the following elimination is no ion er permissible:
g
In this:
~ 1
,R l.u
A= +1 and
2 }2
B = - v
Substituting this in 15 and introducing the symbol Z having the
meaning:
Z=
i.
~(y -.. t~ 1)D
we find the following implicit expression for current density efficiency
of the :R'-ion. .? .
i C ~R Z+ 1)
A.T.A.-T.Id.O.
T.A.NO 270
Appendix 7.
21
III. Conversion of three-ion-to two-ion-system.
If O, equation 15 would have to pass over into the
+
equation for the two-ion system. In such case K = u + v.
v-u -.
Substituting in 15 of l(
2
then
1 U
(X2-Y?D
aS ~12~ Y~
Equation (17) however means: A + 1 = o k this is in conflict with
, h
condition (13a).
The solution 17 can therefore be introduced into the conversion
The correct solution for the tyro-ion-system is found by
(13)-(13a).
substituting the expression N v - u - in (12). Vie then obtain:
~
v - u+ Av - u-+H =0
from which as a variable is eliminated. Equation 18
gives at once the solution which, after substitution of A and B
acquires the following forma
2.D .D v
In this D = and = n R - transference number).
? + vu
D +D
Solution (19) has already been obtained in the direct treatment
tme
Y t a nt
of the two-ion-system.
This solution is now just identical however to the equation
14a (Compare equation 14a with equation 18), at least for a two-
ion-system. Proper elimination of the integration constant is im-
possible see equation 15 because this constant becomes 03 .
This is therefore the explanation why equation 16 is no longer
applicable to the two-ion-system.
A.T.A.-T.N.O.
T.A.No 270.
Appendix '/.
^
 Approved For Release 2002108128 : C1-RDP82-00041R000100160001-0
DatesApril 3 rd,19
Electrodialysis experiment nr 49 Z)
lpparatua 82a z)
supplied
Carried off
ate(l/h)
Anolyte
29 meq BaCl/l
not measured
5
doe.
Catholyte
30 meq HaC1/1
not measured
5dos.
Dialysate
2000 g 30 neq AaCl/l
1975. 8 ~0.2 megq 0H71
ciro.
,
i[nbranee
Anodic: T.A.O.- l 21'
Cathodic: T.N.O.- A 22'
Current
Current
Energy
Time
Temperature
strength
Voltage
consumption
oonsum ti n
(min.)
(?c
(v)
(emp.hr)
(1Phr)
0
74
1.5
4.5
10
14
1.5
4.6
0.25
1.13
20
14
1.5
4.7
0.25`
1.16
30
14
1.5
4.8
0.25
1.19
40
14
1.5
4.9
0.25
1.21.....:.
50
14
1.5
5.2
0.25
1.26
55
14
1.5
5.5
0.13
obi
Total
1.38 6.67
dverag
4.8
Energy
consumption 6
62 !hr, eo W(
30,9) .6.62
~ . 3.3 k1Th 3
by ezt
apolations b$
. p(28,8)
.1
dmp.hr
consumption 1
38, eo 4(30
) ^ 180 g HaC
100 amp.hr
z) Thi
see of the
ompartmentes
1 mm
2
M~
tans area
s
500 cm
2
G1ur
ent density
:
2.6 ma/an
d.T.A.-T.B.O.
T.A.No. 270
Appendix 8.
Approved For Release 2002108128 : CIA-RDP82-00041R000100160001-0
 Approved For Release 2002108128 : CIA-RDP82-00041R000100160001-0
Date:Ygv 26 X1952
Electrodialysie experiment nr 50
Apparatus: B $
Supplied
Carried off.
Eate(l/h)
Anoiyte
970 g 30 meq Nac1/1
535 B (86 meq Cl}/1
0.37
(38 meq H /1
dos.
catholgte
1890 g 29 meq HaC1/).
1890 8 ~27 meq Ci /1
1.08
30 meq 0H/1
dose
Dialysste
2000 g 30 meq NaC1/1
1960 g 8 meq Ci71
T
e
H'F/1
10
Ci
m
q
ro.
Membranes
anodic: T.N.O.- A 40
cathodic: T.N.O.-A lib
Current
Current
Eaergy
Time
Tem erature
strength
Voltage
consumption
consumption
(min.)
(?C)
.(amp.)
(v)
(amp.hr.)
(wiir)
0
15
1.5
4.1
_
_
5
15
1.5
4.1
o.125
0.513 .:,
to
15
1.5
4.2
0.25
0.518
15
15
1.5
4.3
0.25
0.531
20
15
1.5
4.2
0.25
0.53
25
15
1.5
4.2
o;125
0.526
30
15
1.5
4.7
0.125
0.526
35
15
1?5
4.3
o.125
0.526
40
15
1.5
4.3
o.125
- 0.538
45
15
1.5
4.3
0.25
0.538
50
15
1.5
4.3
0.25
0.538
55
15
1.5
4.4
0.125
0.543
60
15
1.5
4.4
0.125
0.550
65
15
1?5
4.4
o.125
0.550
70
15
1.5
4.4
o.125
0.550
75
15
1.5
4.5
0.25
0.557
80
15
1.5
4.5
0.125
0.563
85
15
1.5
4.5
0.25
0.563
90
15
1.5
4.6
o.125
0.568 -
95
15
1.5
4.6
o.125
0.575
1o0
15
1.5
4.8
0.125
0.581
toy
15
1.5
4.8
0.125
0.593
Total
2.625 11.49 '-
Average
4.5
Energy
onsumption 11
49 W+hr, so
(308) 11.
9 x } . 5,7 na/m3
by eztr
olation: 9g
W (28,8) _
.2 kWh/m3
dmp.hr
onsumption 2.
25, so 4 (30,
) - 99 8 NaC
/100 amp.hr;
Pam?r
) (indir.) .
50%, (P~(3
r$) 5%?
A.T.A.-T.N.O.
T.A.-No 270
Appendix 9.
Approved For Release 2002108128 : CIA-RDP82-00041R000100160001-0
 Approved For Release 2002108128 : CIA-RDP82-00041R000100160001-0
Dates May 23r ,195
Electrodialyeie experiment nr 51
Apparatus: H2$
Supplied
Carried off
Rate(l/h)
anoiyte
30 meq xaci/i
3 meq a+/i
aos.
Catholyts
30 meq AaCi/1
4 meq 0K/1
8
dos, ,
Dialyeate
2000 g 30 me? NaCl/1
7940 g ~0.2 meq H %1
circ.
Membranes
anodic: T.N.O,- A 40 J
cathodic: T.N:O.- A 17b
Time
Temperature
Current
strength
Voltage
Current
consumption
Sher
consumption
(mi.)
?C)
(amp.)
(o)
(amP?hr)
(1fhT)
0
18
1.5
4.1
5
18
1.5
4.3
0.125
0.525
10
18
1.5
4.4
0.125
0.543
15
18
1.5
4.4
0.25
0.550
20
18
1.5
4.4
0.125
0.550
25
18
1.5
4.5
0.125
0.557
30
18
1.5
?4.5
0.125
0.563
35
19
1.5
4.6
0.125
0.568
qo
19
1.5
4.6
0.125
0.575
45
19
1.5
4.7
0.125
0.581
50
19
1?5
4.8
0.25
0.593 -
55
19
1.5
5.0
0.25
0.613
Total
1.375 6.22.
Average
4.5
Snergq a
neumption 6.
1P hr, so W
30,10) 6.2
z ~ . 3.1 k /m3
by extra
olstions.Fg
W(28,8) . 3.
kWh/m3
9mp.hr o
neumption 1.
Ss 80 4(30,1
) '174 (8 N
1 100 m p.hr
A.T.A.-T.N~G.
TA.-No 2'/0
Appendix 10.
Approved For Release 2002108128 : CIA-RDP82-00041R000100160001-0