WATER POLLUTION CONTROL RESEARCH SERIES
14010 FNQ 02/72
Electrochemical Treatment
of
Acid Mine Waters
U.S. ENVIRONMENTAL PROTECTION AGENCY
-------
WATER POLLUTION CONTROL RESEARCH SERIES
The Water Pollution Control Research Series describes the
results and progress in the control and abatement of pollution
in our Ration's "waters. They provide a central source of
information on the research, development and demonstration
activities in the Environmental Protection Agency, through
inhouse research and grants and contracts with Federal, State,
and local agencies, research institutions, and industrial
organizations.
Inquiries pertaining to Water Pollution Control Research
Reports should be directed to the Chief, Publications Branch
(Water), Research Information Division, R&M, Environmental
Protection Agency, Washington, D.C. 20J+60.
-------
ELECTROCHEMICAL TREATMENT OF ACID MINE WATERS
by
Tyco Laboratories, Inc.
Bear Hill
Waltham, Massachusetts 02154
for the
ENVIRONMENTAL PROTECTION AGENCY
Project No. 14010 FNQ
Contract No. 14-12-859
February 1972
For sale by the Superintendent of Documents, U.S. Government Printing Office, Washington, D.C. 20402 - Price $1.00
-------
EPA Review Notice
This report has been reviewed by the Environmental Protection
Agency and approved for publication. Approval does not
signify that the contents necessarily reflect the views and
policies of the Environmental Protection Agency nor does
mention of trade names or commercial products constitute
endorsement or recommendation for use.
ii
-------
ABSTRACT
Experimental and analytical evaluations of the direct electrochemical
oxidation of ferrous acid mine drainage (AMD) have shown that this approach is
economically superior to present lime treatment and aeration methods.
Through the use of a packed bed electrode, the size of the oxidation reactor
has been reduced to a stage where the capital investment required for this
equipment can be recovered by cost reductions in latter treatment stages.
These cost savings include:
1. Neutralization with cheaper limestone rather than lime
2. A reduction in sludge settling time due to the better properties
of limestone sludges
3. Reduction of sludge disposal volume.
As a bonus, electrolytic hydrogen, produced during electrochemical oxidation,
should be economically recoverable at high AMD treatment rates.
Preliminary economic estimates of total treatment costs indicate a cost range
from 11^ to 72^/1000 gal, exclusive of hydrogen credits. These costs are
much less than those of present treatment approaches which appear to have
expense rates of from 20^ to $ 2.00/1000 gal treated. In the particular case
of a badly polluted stream containing 2000 mg/1 of CaCC>3 acidity, the total
treatment cost of 31^/1000 gallon for the electrochemical oxidation approach
is less than the reagent cost alone (~ 35^/1000 gal) for conventional lime
treatment.
This report was submitted in fulfillment of Project Number 14010FNQ and
Contract Number 14-12-859 under the sponsorship of the Water
Quality Office, Environmental Protection Agency.
in
-------
Contents
Section Page No,
ABSTRACT iii
I CONCLUSIONS 1
II RECOMMENDATIONS FOR FUTURE WORK 3
HI INTRODUCTION 5
IV BASIC ELECTROCHEMISTRY 7
EXPERIMENTAL 8
ANODIC OXIDATION OF FERROUS IRON 11
CATHODIC PROCESSES 24
TIME DEPENDENT PROCESSES 37
ANOLYTE SEPARATION 42
CONCLUSIONS 44
V ELECTROCHEMICAL REACTOR
CONFIGURATION 47
EXPERIMENTAL 48
ANNULAR FLOW REACTOR 48
FLUIDIZED BEDS §3
EXPERIMENTAL RESULTS 54
ANALYSIS 56
PACKED BED REACTOR 59
RESULTS 61
COMPARISON OF REACTOR TYPES 66
VI PLANT DESIGN AND ECONOMICS 71
ELECTROCHEMICAL REACTOR 71
LIMESTONE TREATMENT 73
AMD TREATMENT PLANT DESIGN
AND ECONOMICS 75
VII REFERENCES 79
VIII GLOSSARY 81
-------
Figures
Page
1. Overall View of Rotating Disk Assembly 10
2. Schematic of Rotating Disk Electrode 12
3. Current-Voltage Scan on a Vitreous Carbon 14
Electrode in 0.02M H2SC-4 at Ambient
Temperature
4. Typical Cur rent-Voltage Scans to 1.5 V 15
(SCE) on a Vitreous Carbon Electrode in
0.02M H2SO4 Containing 0.01M Fe2+
5. Typical Current-Voltage Scan to 0.75 V 16
(SCE) on a Platinum Electrode in 0.02M
H2SO4 Containing 0.01M Fe2+
6. Typical Cur rent-Voltage Scan to 1.2 V 18
(SCE) on a Platinum Electrode in 0.02M
H2SO4 Containing 0.01M Fe2+
7. Typical Current-Voltage Scans on a Vitreous 19
Carbon Electrode in 0.02M H2SC-4 Containing
10-3M Fe2+
8. Relation Between Electrode Rotation Speed 21
and Plateau Currents During the Oxidation of
Fe2+ Concentrations
9. Oxidation of Fe2+ to Fe3+ at a Rotating Carbon 22
Anode in 0.02M H2SO4 Containing Varying
Amounts of Fe^+and Fe2+Species
vn
-------
Figures (Cont)
Page
10. Reversible Potential of the Fe3+/Fe2+ 23
System on Platinum at 35 °C
11. The Effect of Varying the Fe3+/Fe2+ 25
Ratio in Solution on the Cur rent Poten-
tial Curve for Fe2+ Oxidation at a
Carbon Anode in 0.02M H2SO4
12. First Voltage Scan to -0.80 V Freshly 27
Polished on a 316 Stainless Steel Cathode
in Deaerated 0.02M H2SC-4 at Ambient
Temperature
13. Hydrogen Evolution Characteristics on a 29
Passivated (A) and Oxide Free (B) 316
Stainless Steel Surface in 0.02M H2SO4
at Ambient Temperature
14. The Effect of H2SO4 Concentration and 31
Electrode Rotation Speed on the Hydrogen
Evolution Reaction on a 316 Stainless Steel
Surface at Ambient Temperature
15. First Voltage Scan to -0.70 V on a Freshly 32
Polished 316 Stainless Steel Cathode in
Deaerated 0.02M H2SO4 Solution at Am-
bient Temperature and Containing 2 x
10~3M Fe3 +
16. Sixth Voltage Scan to -0.70 V on a Freshly 33
Polished 316 Stainless Steel Cathode in
Deaerated 0.02M H2SO4 Solution at Am-
bient Temperature and Containing 2 x
10-3 M Fe3 +
17. Voltage Scan to -0.90 V on a Stainless Steel 35
Cathode in Deaerated 0.02M H2SO4 Solution
at Ambient Temperature and Containing 2
x 10-3M Fe3 +
Vlll
-------
Figures (Cont)
Page
18. Current Decay During Fe2+ Oxidation 39
in Chemical AMD Containing 10~2M
Fe2+in 0.02M H2SO4 on a Vitreous
Carbon Electrode
19. Current Decay During Fe2+ Oxidation 41
in Uncentrifuged Synthetic AMD Containing
1.04 x 10-ZM Fe2+asWell as 0.73 X 1Q-2M
Fe3 + in H2SO4 of pH 2.72 on a Vitreous Car-
bon Electrode Held at +1.2 V and 23 rps
20. Plot of Fluid Velocity Versus Particle 55
Diameter at 65% Bed Voidage
21. Correlation of the Coefficient of Performance 60
With n the Particle Reynolds Number
22. Packed Bed Data for a Bed Width of 1.08 In 62
and a Bed Length of 1 Ft
23. Packed Bed Data for a Bed Width of 1.08 In 63
and a Bed Length of 2.5 Ft
24. Packed Bed Data for a Bed Width of 0.55 In 64
and a Bed Length of 1 Ft
25. Effect of Flow Velocity on Total Limiting Current 67
26. Settling Times of AMD Sludges 74
IX
-------
Tables
No. Page
1 Conversion Data for a Flow of 1.7 Gal/Hr 49
2 Computer Program for Evaluation of Annular 51
Flow Data Obtained with Pilot Plant
3 Concentration Profile at 1.7 Gal/Hr 52
o i
4 Final Fe Concentrations as a Function 52
of Flow Rate
5 Concentration Profiles for a Flow of 0.1 Gal/Hr 53
6 Fluidized Bed Data 54
7 Computer Program for Evaluation of Fluidized 58
Bed Data
8 Fluidized Bed Calculations 59
9 Effect on Column Performance of Changes 61
in Bed Width
10 Effect of Bed Length on Column Performance 65
11 Comparison of Reactor Configurations 68
12 Comparison of Electrode Systems Via Anode Size 69
13 Capital Cost Analysis for the Packed Bed Reactor 71
Concept (6000 Gal/Hr, 95% Conversion)
14 Capital Charges for the Electrochemical Reactor 72
at Various Conversion Percentages
15 AMD Compositions and Flow Rates 75
16 Plant Investments 77
17 Estimated Operating Expenses for Direct Electro- 78
chemical Oxidation Treatment Plants, ^/lOOO Gal
(Lime Treatment Range 20 i to $2/1000 Gal)
XI
-------
SECTION I
CONCLUSIONS
The following is a list of those conclusions which are relevant to the develop-
ment afa practical electrochemical process for treating AMD.
2+ 3+
i. The anodic oxidation of Fe to Fe takes place on a
carbonr electrode at a mass transport limited rate. The oxidation region: is
about 0.80 V in extent. The potential plateau for this process is shifted ta
more positive values by increasing the Fe^+ concentration although this shift
(<0.2 V) is not sufficient to cause problems in an operational system.
2. Excessive oxidation (high positive potential-s) at the
carbon electrode inhibits the Fe** oxidation. In a practical reactor, how-
ever ^ such high positive potentials will not be reached.
3. Hydrogen evolution occurs on a polished 316 stainless steel
cathode (in deaerated 0.01, 0.02M H2SO4) at potentials more negative than
0.5 V (SCE). These currents are not diffusion limited but are controlled by a
slower electrochemical kinetic step.
4. The cathode in 0.02M H2SO4 is passivated by an oxide film
at potentials more positive than -0.04 V. This film is removed only at poten-
tials more negative than -0.90 V, the normal operating region of the cathode.
5. The diffusion limited back reduction of Fe^+to Fe^+occurs
at potentials more negative than -0.15 V.
6. Soluble ferric iron is reduced to the metal on a 316 stain-
less steel cathode in 0.02M H2SO4 at potentials <-1.0 V. Large currents from
H2 evolution can be supported before iron deposition occurs.
7. A highly selective membrane is available which will pre-
vent the penetration of Fe^+ specie while possessing acceptable ohmic and
structural characteristics.
- 1 -
-------
8. Smooth carbon electrodes show a slow degradation in
performance as the result of poisoning by soluble species. The degradation
is enhanced in AMD water prepared from waste coal. This effect has not
been detected with porous carbon electrodes.
These results were then used as the basis for an experimental and engineer-
ing evaluation of the following prototype electrochemical reactors:
(a) fluidized bed, (b) packed bed, (c) annular flow.
Of these three the packed bed electrode gave rise to the smallest and hence
most economical configuration.
An evaluation of process economics, using a packed bed reactor, indicated
total treatment costs in the range of lie7 to 72c//1000 gal, compared to pre-
sent costs of 2(V to $2.00/1000 gal. Additional cost savings are possible from
the sale of electrolytic hydrogen. Because of transportation and collection
costs, these savings would be realized when handling large volumes of
highly polluted AMD water.
As compared with alternative oxidation methods, the electrochemical pro-
cess is free of the safety hazards associated with radioisotope-induced
oxidation and should as well be free of the temperature sensitivity associated
with biological oxidation methods.
- 2 -
-------
SECTION II
RECOMMENDATIONS
The experimental work discussed in this report has amply demonstrated the
technical feasibility of electrolytically oxidizing ferrous iron in AMD to
ferric. An evaluation of system economics indicates potential cost benefits
are to be realized.
The following are recommended:
I. This process be sealed up to the construction of a full sized
packed bed module
2. The reactor be operated first in the laboratory and next in
field on actual AMD
3. The data so acquired be used to obtain a more precise
evaluation of process economics
4. The applicability of the process be evaluated in terms of
specific treatment sites.
_ o _
-------
SECTION III
INTRODUCTION
Waters which accumulate in mines, and waters draining from mines and
exposed coal formations represent a significant pollution problem which
has been receiving increased attention. Such acid mine drainage (AMD) is
typically highly acidic (pH 2 to 3) and contains significant quantities of
dissolved salts (principally iron sulfates). Although there are significant
variations, this dissolved iron exists mostly in the ferrous state.
Chemical AMD treatment methods typically involve neutralization, using low
cost materials such as limestone or lime, concurrent with an aeration step
to oxidize the ferrous iron to the ferric state. This oxidation is necessary
in order to avoid the high pH required for ferrous iron precipitation.
Air oxidation is a relatively slow process which is highly dependent on the
pH of the water and on the efficiency of the aeration system. Significant ox-
idation rates do not occur until the solution becomes neutral or basic (pH>7).
Therefore, high ferrous AMD is typically neutralized with more expensive
lime rather than limes tone which is too weak a base to produce a pH above
5.5.
The resulting voluminous sludge (typically less than 5% solids) is separated
from the bulk stream and stored. The high residual alkalinity of the sludge
produced with the lime neutralization process adds to the disposal problems
since further leaching of the sludge piles is likely. In addition, -close con-
trol over the neutralization step is needed to avoid overtreatment with the
resultant alkaline stream flow.
Alternative oxidation processes which would proceed rapidly in a highly acid
solution have, of course, been considered. Direct chemical treatment with
nonpolluting oxidants, e.g., hydrogen peroxide, sodium peroxide, or ozone,
involve intolerably high reagent costs for the flow rates that are involved.
- 5 -
-------
Biological techniques, in addition to requiring an aeration step, proceed
relatively slowly, and are adversely affected by temperature. Another
possible approach is electrochemical oxidation. An acid mine water, con-
taining on the average 500 mg/1 of Fe2"1" and 1000 mg/1 H+ and variable
amounts of aluminum, calcium and manganese, is an electrolytically con-
ducting solution. As such, it is capable of electrolysis without resort to
additional salt additives. The electrochemical reaction: :
Fe2+ - Fe3+ + e"
would take place at an inert (nonconsumable) anode. Concurrently, on an
inert cathode, the reaction:
H"1" -f e" - 1/2 H2
would take place, resulting in the generation of 1/2 mol of hydrogen gas for
every mol of iron that is oxidized.
This report describes the technical and economic evaluation of an AMD treat-
ment process based on these reactions.
It will be shown that major cost savings for the entire process can accrue
from the subsequent treatment of nearly 100% ferric mine water produced by
the electrochemical oxidation step. Precipitation of Fe3+ with cheaper lime-
stone produces a denser sludge (10 to 20% solids) with a faster settling time.
Thus, in addition to savings in reagent costs, capital and operating outlays
for sludge separation and sludge disposal can be lower than when lime is
used as the neutralizing agent.
In essence, the costs of the increased capital investment required for the
direct electrochemical oxidation of ferrous mine waters will be defrayed by
lesser capital investment and operating expenses in later treatment stages.
Such a treatment process would have the additional attraction of generating
by-product electrolytic hydrogen. It will be shown that, in certain situations,
this hydrogen can be sold to further defray treatment costs.
- 6 -
-------
SECTION IV
BASIC ELECTROCHEMISTRY
Upon the electrolysis of acid mine drainage water, the following electro-
chemical processes are possible for the major soluble ionic species: *
1. At the negative electrode (the cathode):
2H+ + 2e" ^H2 E = -0.35 V (1)
Fe2+ + 2e~ - Fe E = -0.79 V (2)
Fe3+ + e" ^ Fe2+ E = +0.52 V (3)
2. At the positive electrode (the anode):
Fe2+ ^ Fe3"4" + e~ E = +0.52 V (3a)
2H2O - O2 = 4H+ + 4e" E = +0.88 V (4)
2H0O ^ H0O0 + 2H1" + 2e" E = +1.41 V (5)
Ci ft fj
2HSO" - S00" + 2H"1" + 2e" E = +1.8 V (6)
TI £t O
9^O~ ^ *s DZ -I- 9^" F - -1-1 7 V ^7^
f-nD\J jt * Or*^Q i ttC J-. — ~JL» IV V I /
4 2 ^
The electrical potentials listed are thermodynamic values versus a satura-
ted calomel reference electrode (SCE) and apply for concentrations of 0.02 N.
The sign (+ or -) is that which results from the use of the standard reduction
potential convention for half-cell reactions.
The desired anodic process is reaction 3a and the desired cathodic process
is reaction 1. All other processes are to be considered parasitic for the
development of a viable AMD treatment.
- 7 -
-------
For all practical purposes, it is possible to disregard reactions 5 through
7 since these occur at potentials more positive than those readily accessable
in dilute aqueous electrolytes. The current (reaction rate) -potential de-
pendencies of the remaining reactions were determined experimentally.
EXPERIMENTAL
2-f- (2) 2+
AMD water varies in Fe concentration and pH ,0.02M in Fe andinH2SC>4
was taken as representative. In the initial studies, the experimental con-
ditions were simplified by using pure FeSC>4 (FeSC>4 • 7 H2O, Fisher Scien-
tific Co.) and pure H2SO4 (suprapure grade, Brinkmann Instruments, West-
bury, N.Y.). Subsequently, the results were confined in "synthetic" AMD
water produced in the laboratory by draining water through waste coal. The
characteristics of this water were described in the Final Report for Project
14010 DLI, contract 14-12-560.
Normally this effluent solution from the waste coal was cloudy yellowish brown
in color and varied in total soluble iron concentration from ~ 4 x 10~^ M to
~2x 10~2M (both Fe2+andFe3+). The solution was generally pH2.7. When
these water samples were left for extended periods of time, the Fe2+concentra-
tion usually decreased and Fe3+ increased, presumably from oxidation of the
former by dissolved O2. (The participation of bacteria in this process was not
studied.) Thus, over a 5-day period, initial concentrations of 1.7 x 10~2MFe2+
and 1.2X 10-2MFe3+changed to 1.4x 10~2M Fe2+and 1.3x 10'2M Fe3+. Some
iron was apparently lost through precipitation during the period. Ferrous and
ferric concentrations were determined for this and subsequent experiments by
the standard dichromate techniques using adiphenylamine sulfonate indicator. ^
In most experiments, this "synthetic AMD" was first centrifuged for 10 min
before use, in order to remove solid residues. This treatment caused only
a small drop in concentration of both Fe species; the solution was then clear.
However, a yellowish brown residue settled out in the electrochemical cell
over a period of days. Similarly when the solutions were passed through a
0.45-H Millipore filter, the cloudiness was removed but again a light yellow
residue settled out in the deaerated cell over several days. This latter
treatment also slightly reduced the concentration of both Fe species in
solution.
2+ 3+
The values for both Fe and also Fe in these solutions are somewhat
greater than might normally be found in AMD in streams, 3 i.e., after
natural dilution. In some experiments, therefore, this synthetic AMD was
diluted with 0.02M H2SC>4 before use.
The measurements of solution conductivities were carried out using a con-
ductivity bridge (model no. R.C. 16B2, Industrial Instruments, Inc., Cedar
Grove, N.J.). Conductivities of a series of solutions containing 10~2 to
- 8 -
-------
10 M Fe in 0.02 and 0.01M H2SC>4 were measured. The presence of Fe
had minimal effect on the conductivities of the solutions. Typical values
ranged from 104.6 ohm-cm (10~6 M Fe2+ in 0.02M H2SC>4) to 183.5 ohm-cm
(10-6 M Fe2+ in 0.01M H2SO4).
Carbon was selected as the anode since it is inexpensive, can be obtained in
tubular form, is very resistant to chemical attack, and has a relatively wide
potential range over which it is electrochemically stable3 (i.e., does not
evolve O2, H2, or dissolve). As cathode, 316-stainless steel rather than
pure iron was chosen, since stainless steel is more resistant to corrosion
in dilute acids.
Initial experiments on a carbon anode were aimed at confirming its stabili
under applied potential, and determining the potential range over which
can also be oxidized. In the latter context, it was necessary to determine the
potential region in which Fe2+ is oxidized at a mass transport limited rate,
i.e., in the absence of any electron transfer or other limitations. A rotating
disk electrode^ was used in these studies. This configuration is particularly
useful since the hydrodynamic boundary layer setup is well characterized.^
Thus, in a purely mass transport limited situation, currents for Fe2+ oxi-
dation can be calculated at given rotation speeds of the assembly, using the
equation derived by Levich:^
= 0.62 nFD2/3 TT r2 u~1/6 co1/2 cfe (8)
where n = number of electrons involved in the process
F = the Faraday constant
o
D = diffusion coefficient (cm /sec)
r = radius of the disk (cm)
v = kinematic viscosity of the solution (cp/cnrVgm)
o> = rotation speed of the disk (rad/sec)
c, = bulk concentration of reactive species (mol/1)
L. = limiting current (mA)
An overall view of the rotating disk electrode setup is shown in Fig. 1. The
equipment consists of a sturdy stand on which a 1/15-hp Bodine motor and a
precision ball bearing for the 0.025 in. rotating shaft were mounted. The
motor speed was controlled by a Minarik speed control (SL-52). The coupling
of the motor and the electrode shaft was accomplished by a nonslip belt. The
rotation rate of the electrode was continuously monitored by frequency modu-
lation resulting from magnetic coupling of an electromagnet with an iron gear
mounted on the rotating shaft. This signal was amplified and displayed on a
frequency counter (Hewlett Packard model no. 5221A). Electrical contact to
the disk electrode was accomplished through a mercury pool in the top of the
rotating shaft.
- 9 -
-------
Fig. 1. Overall view of rotating disk assembly
-10-
-------
The electrochemical cell was made of a 50-mm O-ring joint with a 30 mm
thick Teflon cover. The Teflon cover contained tapered holes to accommo-
date a liquid seal in the center with a 29/40 standard tapered joint surrounded
by four 10/30 joints for gas inlet (N2) and outlet as well as the counter- (pla-
tinum) and reference electrodes (calomel). A Teflon bell was fixed to the
rotating shaft. Mercury was used as sealant liquid to prevent 63 from con-
tacting the solutions.
Fig. 2 is a diagram of the actual electrode. The main feature of this elec-
trode is the vitreous carbon anode (Atomergic Chemetals Co., 584 Mineola
Avenue, Carle Place, L.I., N.Y.), which is press-fitted into hot Teflon and
contacts the upper mercury pool via a simple spring. A rotating disk elec-
trode was constructed inhouse from 316-stainless steel rod.
The electronic equipment employed was a potentiostat (Wenking model no.
61TRS, Brinkmann Instruments, Westbury, N.Y.) supplying ±1 A at ±15 V
with a rise time of ~ 10~5 sec. This was used to control the potential between
the calomel reference and carbon anode using a platinum counterelectrode.
A Tyco-built function generator (TLI no. 661207) was used to apply a linear
voltage sweep (1 V/min and 0.1 V/min) to the system, and current-voltage
curves were recorded on an x-y recorder (model no. R-100, Houston Omni-
graphic Corp., Bellaire, Texas). Voltage drop (i-R) in the system was
measured on an oscilloscope (Tektronix type 561A with 2A63 and 2B67 plug-
ins) by interrupting an applied constant current. IR was compensated for
during the measurements by a suitable amplifier setup. Potentials were
measured and are reported versus an SCE.
ANODIC OXIDATION OF FERROUS IRON
The basis of this AMD treatment concept is being able to oxidize electro-
chemically ferrous iron to ferric at a rapid rate, i.e., at a rate limited only
by the mass transport of ions to the electroactive surface. Specifically it
was necessary then to determine:
1. In what potential region was the anodic oxidation of ferric iron
diffusion limited
2. At what potential did the oxidation of water to ©2 occur
3. In which potential region the carbon electrode itself was
stable.
Solutions employed were: (1) 0.02M H2SO4 to examine the limits of stabilitv
of the carbon anode in these solutions, and (2) 0.02M Fe2+ to 5 x 10~4 M Fe2+
- 11 -
-------
-Teflon bell for liquid seal
-Teflon
Stainless steel shaft
V
-Teflon
•Stainless steel shaft
Contact spring
- Electrode
Fig. 2. Schematic of rotating disk electrode
- 12 -
-------
2-L 3_i_
in 0.02M H2SO4 to examine the oxidation of Fe to Fe at the anode. Solu-
tions were deaerated with N2 prior to scanning the electrode. Rotation speed
of the electrode was varied from 3.0 to 50.6 rps.
Fig. 3 shows a typical voltage scan (0.10 V.min) on a carbon anode in de-
aerated 0.02M H2SO4, At 0 V (SCE), a small (2-/iA) negative current is ob-
served which becomes positive and increases steadily (on scanning positive)
until it reaches a value of +16 juA at 1.2 V (SCE). Above this potential, the
current increases more rapidly with increase in E, corresponding to the on-
set of 02 evolution (+1.2 V -» +1.6 V SCE). The reverse scan from these
potentials does not follow the forward one exactly, but remains slightly below
it. The electrode is stable in the cathodic direction to -0.90 V (SCE), at which
point H2 is observed. The relatively high value of background current ob-
served during the scan is attributed to the double layer charging process as
E is increased, rather than from impurities in solution.
2+
Solutions containing 0.01M Fe were scanned between the general limits
-1.2 V and +1.5 V (SCE). A typical sequence is shown in Fig. 4 where the
detailed scan history was from -0.80 V held for 20 min, then immediately
to +1.5 V. The reverse scan from +1.5 to 0 V was then carried out. Negative
scans (< 0.0 V) were completed at 1 V.min, while all the data in Fig. 3 were
at 0.102 V/min. Electrode rotation speed was 23.4 rps.
On the reverse scan from +1.5 V, a prewave is in evidence. However, the
figure shows clearly that no prewave is found on the forward scan to +1.5 V.
It was found that, if the electrode was cycled between 0.0 V and +1.5 V, the
prewave became more marked. If the electrode was cycled from 0 to -1.2 V
and back again, or held at -0.8 V for 20 min before carrying out the scan to
positive potentials (> 1.2 V), then the prewave was not observed. Provided
the electrode was not scanned beyond +1.2 V, the prewave did not reappear.
It was easily shown that this prewave was not to be due to solution impurities.
It was necessary then to show that the prewave was, in fact, characteristic
of a carbon anode solely, rather than some extraneous effect, and, more
important did not necessarily result in a permanent degradation of the elec-
trode. Therefore the experiment was repeated, but using a rotating platinum
disk as anode; the electrochemical properties of this metal and its oxides
are well established.
2+
Fig. 5 shows the oxidation of 0.01M Fe on a rotating Pt electrode in 0.02M
HoSO4 using a scan speed of 0.102 V/min and a rotation speed of 23.4 rps.
Also shown is the reverse scan on the electrode. The overall currents shown
are greater than on the C-electrode, purely because of the greater area of
the platinum electrode. What is most evident is the fact that the prewave is
not found on platinum. Attempts to reproduce the phenomenon observed on
the C-electrode by varying scan speed and rotation speed were not successful.
- 13 -
-------
S
o
U
+ 25
o
-0.4
-0.8
-1.2
Potential, V versus see
Fig. 3, Current-voltage scan on a vitreous carbon electrode
in 0.02M H2SQ4 at ambient temperature (scan rate = 0.10 V/min)
-------
Ul
I
250
o
c
CD
o
1
+ 1.6
+ 1.2
+0.8
Potential, V versus SCE
+0.4
0
Fig. 4. Typical cur rent-voltage scans to 1.5 V (SCE) on a vitreous carbon
electrode in 0.02M H2SO4 containing 0.01M Fe2+ (scan rate = 0.10 V/mih,
rotation speed = 23.4 rps)
-------
1.2
e
o
c
-------
Basically, it was assumed that the wave was caused by a change in the nature
of the electrode surface during scanning. The most likely change is the for-
mation of a surface oxide at high positive potentials and it is known, at least
for materials like platinum, that surface oxides inhibit the oxidation of Fe2+
to Fe3-+. Therefore, an attempt was made to induce the prewave by forming
platinum oxide on the electrode. With our electrochemical system, in effect
calibrated by this better established process, it would then be possible to
interpret the data for carbon.
In Fig. 6 data equivalent to that in Fig. 5 are shown on a Pt electrode, except
that in Fig. 6 the electrode was scanned out to +1.2 V (SCE) where an oxide
is known to form on Pt.6 (In Fig. 5, the scan was only made to +0.75 V (SCE)
where little Pt - O is formed.6) Whereas the two plateaus in Fig. 5 almost
coincide, those in Fig. 6 differ markedly. On the forward scan of Fig. 6 (0
to 1.2 V), the current rises to the plateau region at > 0.40 V and reaches the
plateau at +0.60 V. On the reverse scan (1.2 to 0 V), the fall from the pla-
teau occurs almost immediately and is very sharp below +0.& V. The current
reaches a minimum at +0.5 V and then increases until it meets the original
forward scan (at +0.4 V) which it then follows back to the starting point (0 V
SCE). Since at potentials of the order of 0.4 V (SCE) no oxide is present on
a Pt surface,6 we can thus conclude that on the reverse scan the oxidation of
Fe2+ is inhibited by Pt-O formed at high potentials and that the original be-
havior of an oxide free surface is again restored at lower potentials, as is,
in fact, indicated by the current increase at +0.5 V.
Thus it would appear that a slowly reduced oxide is formed on a C-anode at
> +1.2 V (SCE) which inhibits the oxidation of Fe2+ to Fe3+. This inhibition
of the reaction is reflected by a shift in the diffusion limited plateau for the
reaction to more positive potentials. Provided the electrode is not cycled
beyond +1.2 V, no such inhibition occurs. Holding the electrode at -0.8 V
will reduce this oxide.
Having established that the limits of the electrode were satisfactory, a series
of E versus i curves was measured in solutions containing varying Fe2+ con-
centrations. Typical data in 10"3M Fe are shown in Fig. 7 at varying rotation
speeds. Clearly, the current is insignificant to +0.30 V (SCE), when a sharp
rise to a plateau is observed [the latter is attained close to +0.50 V (SCE) and
extends to +1.30 V (SCE),. when ©2 evolution then sets in]. The slightly
sloping character of the plateau is probably a reflection of the increase in
background current as the voltage is made more positive. Clearly, the
plateau current increases with increase in rotation speed, as would be ex-
pected from a purely mass transport limited process.
2+ -2
The above results, and others at varying Fe concentration (10 to
5 x 10"^M) are plotted as plateau current (corrected for background) versus
- 17 -
-------
1.2
g
o
c
0)
a
0.6
0
I
+ 1.2
+0.8 +0.4
Potential, V versus SCE
0
Fig. 6. Typical current-voltage scan to 1.2 V (SCE) on a platinum electrode in
0.02M H2SO4 containing 0.01M Fe2+ (scan rate = 0.10 V/min, rotation
speed = 23.4 rps)
- 18 -
-------
CO
i
s
o
_
0)
U
50
25
0
1.6
1.2
0.8
Potential, V versus see
0.4
0
Fig. 7. Typical current-voltage scans on a vitreous carbon electrode in 0.02M
H0SOA containing 10"% Fe2"1" (scan rate = 0.10 V/min)
ti 4
-------
1/2
(co) . As may be seen from Fig. 8 in every case a linear relationship is
obtained, with the extrapolated current passing through the origin. This is
exactly what would be expected for a purely mass transport limited process.
Moreover, assuming a diffusion coefficient for Fe2+ of 5 x 10~6 cm* /sec
and a value for the kinematic viscosity of 2.15 (u ~l/6) cp/cm3/g (this is
the value for 1M H2SO4, 9 and is probably slightly high) , then, in 10~2M Fe2+
solution, a diffusion limited current of 252 jit A would be expected at a rotation
speed of 13.5 rps, assuming a one-electron transfer process [Eq. (2) ] . The
measured value was 230 MA. We conclude from the above results that the
plateaus shown correspond to the mass transport limited oxidation of Fe2+
to Fe3"1".
2+ 2+
The oxidation of Fe was examined in solutions which varied in both Fe
content and also Fe3"1" content. Rotation speed of the electrode was varied and
the oxidation currents were recorded at +1.2 V (SCE). These currents,
corrected for background are shown in Fig. 9 for various ratios of Fe3+/Fe2+.
In all cases, the oxidation currents were directly proportional to the square
root of electrode rotation speed. Calculated values are shown on the graph
also for each Fe2 + concentration and they agree very well with experimental
values. As the figure shows, the presence of Fe3+ in solution does not affect
the kinetics of oxidation process.
3+ 2+
The reversible potential for the redox couple Fe /Fe can be calculated
directly from the Nernst equation, ? thus:
where:
E = En + . In (9)
redox 0 nF a_ 2+ v '
Fe
3+ 2+
EO = standard equilibrium potential of the Fe /Fe couple
n = number of electrons involved in the reaction (= 1 here)
a^ 3+/a_ 2+ = ratio of the activities of the oxidized and reduced
Jr1 Q Jr* 6
species in the reaction (we have taken concentrations
only here)
It is clear from this equation that the redox potential will change in propor-
tion to any change in activity ratios. This is shown graphically in Fig. 10.
As the ratio Fe^/Fe"1" increases, the redox potential shifts more positive.
For example, when the Fe3+/Fe ratio in solution is increased from 10~2
to 10 the reversible potential of the changes from approximately +0.41 to
+0.57 V (SCE) . This is, for present purposes, a negligible change.
- 20 -
-------
400
c
-------
400
360-
320
_ 280
LU
or
tr
CO
o
240
200
160
120
80
40
X *
' X
-3 a*
5 X 10 MFe
-S 2 +
10 MFe
I
10 12 14 16
ROTATION SPEED, ou? (rad/sec)*
2+ 3+
Fig. 9. Oxidation of Fe to Fe at a rotating carbon anode in Q.02M
containing varying amounts of Fe3+ and Fe^+ species fy, l
Fe3+; o , 10"3M Fe34";*, Calculated Currents)
- 22 -
-------
Activity ratio Fe /Fe
Ferric ion in excess
I0'2 10"3 ICT"4
Ferrous ion in excess
3-(- 2-f
Fig. 10. Reversible potential of the Fe /Fe system on platinum at 25 °C
- 23 -
-------
7
It can be shown that the equation expressing the relationship between current,
electrode potential and ion concentration also involves the log term of Eq. (9).
In the present experiment, it was difficult to evaluate changes in potential
over a wide range of Fe3+/Fe2+ ratios since the oxidation current for Fe2+
becomes quite small at < 10~3M Fe2+ concentrations, even at the highest
electrode rotation speed. However, a 130 mV negative shift was observed in
the plateau potential for the data in Fig. 9 as the Fe^ VFe24" ratio was varied
from 1/1 to 1/10.
The above is illustrated more clearly in Fig. 11 where are shown two current-
voltage curves on a carbon electrode in 0.02M H2SC>4 containing approximately
5 X 10-4M Fe2+ and 5 X 10"4M Fe2+ + 5 x 10~2M Fe3+, respectively. The
activity (concentration here) ratio varies from an overwhelming excess of
Fe2+ (i.e., no Fe3+ added) to a 100/1 excess of Fe3+. Data were obtained
on a freshly polished electrode which was held at -0.90 V for 5 min initially
to reduce any oxide present. The first scan after this procedure was not
well defined and current flow commenced at +0.50 V reaching a plateau at
+1.0 V. Thereafter, the electrode was scanned from +0.20 to +1.40 V and only
a slight shift in the curves towards more negative potentials was observed
on each succeeding scan. Scan A in Fig. 11 is a typical one for a concentra-
tion of 6.7 x 10~2M Fe2+. The current is substantially zero to near +0.35 V
when a sharp increase is observed to a "sloping plateau" at +0.50 V, which
continues to +1.40 V. Curve B in Fig. 11 illustrates the effect on the curve
of added Fe3+ (5.43 x 1Q-2M). The Fe2+ content here was 5.50 x 10~4M.
The current is negative until +0.55 V when a sharp increase is observed to
a plateau at +0.65 V, which continues to +1.40 V. That is, a 0.150 V shift to
more positive potentials is observed by adjusting the Fe^^/Fe2"1" ratio from
no Fe3+ content to a 100/1 excess of Fe3+. The diffusion currents for Fe2+
oxidation can be calculated in both cases. Hence, in 6.7 x lO^M Fe2+ solu-
tion at a rotation speed of 23 rps, a limiting current of 20.5 juA would be
expected, compared to 15.2 fiA in 5.5 x 10~^M Fe2"1" solution at the same
electrode rotation speed. At +1.2 V, the observed values, corrected for
background, are 26 juA and 15 juA, respectively. It is thus clear that
in solution does not affect Fe2+ oxidation even at high excess of Fe3+ and
there is a workable potential region to use for Fe2+ oxidation even when the
2+ ratio in solution changes considerably
CATHODIC PROCESSES
H2 Evolution: The described cathodic process is the reduction of hydrogen
ion to nydrogen gas, i.e.,
+ 2e~ -» H
- 24 -
-------
DO
CJ1
40
30
LU
£ 20
o
UJ
!= 10
CO
o
Q.
CURVE B
START
SCAN
+ 0.2
+ 0.6
+ 1.0
+ 1.4
POTENTIALS vs. S.C.E.)
3+ 2+
Fig. 11. The effect of varying the Fe /Fe ratio in solution on the current
potential curve for Fe2+ oxidation at a carbon anode in 0.02M H2SO4
(Curve A, no Fe3+ present; Curve B, 100/1 Fe3+/Fe2+ Ratio)
-------
The current-potential dependence for this reaction traditionally has involved
a pronounced dependance on the chemical composition of the electrode sur-
face. For nonelectrochemical reasons, 316-stainless steel was chosen as
the anode material. It remained to be determined whether this decision was
indeed a workable one. As a first step, the formation and reduction of sur-
face oxides were studied on the 316-stainless steel cathode in 0.02M H2SO4.
On freshly polished stainless steel, an open circuit potential (Eoc) was ob-
served in deaerated solution in the region of +0.110 V (SCE). A typical
current-voltage scan at 0.03 V/min is shown in Fig. 12. The current is very
small (< 5 MA) from Eoc to +0.80 V (SCE); above this, a sharp increase in
current with increase in potential is observed which reaches a maximum
typically in the region of +350 M, at 1.1 V (SCE). At more positive potentials,
the current decreases slightly to a minimum before rising sharply again at
0.04 V (C-2 evolution).
The reverse scan on the electrode to more negative potentials follows the
forward one closely and the current is insignificant to -0.40 V. Below this
potential, the current then increases rapidly to about 1 mA at -0.80 V. When
the scan is again reversed, some hysteresis in these high currents is ob-
served as evidenced by a 0.004 V shift in the return scan to more noble
potentials.
The above experiment on this electrode was repeated several times, each
time scanning to more negative potentials. The most striking differences in
the curves so obtained are that the H2 currents on the electrode (< -0.40 V)
during the forward and reverse scans show greater hysteresis as the reverse
scans are carried out to more negative potentials. Secondly, an anodic peak
appears at -0.040 V when the scan is first made to < -0.80 V, and it increases
in size as the scans are carried out to more negative potentials.
The obvious explanation for these results is that, when the stainless steel is
first immersed in the 0.02M H2SO4, it spontaneously passivates, i.e., the
observed Eoc of +0.11 V is actually within the passive region for the alloy in
these environments. Riggs^ shows polarization curves for 316-stainless steel
in 10M H2SO4 in which the active to passive transition occurs over the range
-0.40 to 0.0 V (SCE). This spontaneous passivation does not occur in 1M
H2SO4 where the observed Eoc is -0.350 V. When the electrode is passivated
in the latter environment, the active to passive transition is observed in the
active to noble (positive) potentials scan but not in the reverse direction. This
is obviously what happens also in 0.02M H2SC>4 and indicates that the Fe2C>3,
C^Og passivating film is difficult to reduce on the surface once it is formed.
The results do indeed show that quite negative potentials (< -0.80 V) have to
be reached before significant reduction of the passivating film occurs.
- 26 -
-------
500
+ve
0
LL)
CC
cr
Z>
o
-500
-ve
-1000
START SCAN
1
1.2
+0.8 +OA 0
POTENTIAL (V. vs S.C.E.)
-OA
-0.8
Fig. 12. First voltage scan to -0.80 V freshly polished on a 316 stainless steel
cathode in deaerated 0.02M ^SO^. at ambient temperature (scan
rate = 0.102 V/min)
- 27 -
-------
The slight peak observed at > +0.80 V (in the transpassive region) is most
likely due to some gross and relatively porous oxide layer (or hydroxide or
sulfate) on the surface. In any event, corrosion of the surface protective
oxides occurs readily here and no protective ability is obtained. Secondly,
and more important, stainless steel in 0.02M H2SO4 environments is effec-
tively anodically protected and will thus corrode only at an infinitesimal rate
« 0.5
The divergence of the Ho curves on the two scans can be explained by the
presence of the oxide fikn. Obviously the H2 currents at equivalent potentials
are greater on the "bare" surface than on the oxide coated one, as would per-
haps be expected.
In order to evaluate this point, the electrode was potentios tatted at -1.0 V
for 10 min to remove any surface oxides on the steel. Once we had carried
out this reduction step, the electrode was held sufficiently negative so that
the oxides were not again formed on the surface. The current-potential
behavior was then measured by potentiostatting the electrode in multiples
of 0.05 V steps negative of the start of the Tafel region [generally -0.50 V
(SCE) ] . After 1 min when the currents were generally steady, the electrode
was returned to the start of the Tafel region and left for 5 min. After this,
the procedure was repeated and the current measured at the next potential
value.
In Fig. 13, we show a comparison in the E versus i curves obtained in 0.02M
H2SO4 on a stainless steel surface passivated (with oxide) and oxide free.
The Tafel slope on the oxide free surface (138 mV/decade) is greater than
on the passivated surface (100 mV/decade) . At a given value of potential,
currents on the oxide free surface are greater than on the passivated sur-
face. For example, at -0.50 V, the relative values are 37.0 x 10~6 A/cm2
and 4.5 x 10~6 A/cm2, respectively. At < -0.75 V, the curves converge.
This is as expected since the oxide on the passivated surface will be reduced
in this potential region.
There is a tendency for the E versus i curves to flatten out to a plateau region
at < -0.70 V. To determine if this represented the onset of mass transport
limited H2 evolution on the electrode, the following, experiments were carried
out. At a rotation speed of 4.1 rps and assuming EX""1" = 7.45 x 10~^ cm2/seclO
in 0.01M HoSO^ solution, a diffusion current for H+ to the disk of 11.7 mA/
cm2 was calculated. The measured value at < -1.0 V was 13 mA/cm2. What
was puzzling however was that, at the higher rotation speeds examined (up
to 51.6 rps) , the currents did increase slightly but not in the calculated
manner. Indeed at 23.8 rps and 51.6 rps, the current in the plateau region
for H2 evolution was constant. Also, no well defined plateau was observed
- 28 -
-------
10
10'
,-3
o
LL)
O
I0~5
D '^°
o"
I
I
-0.5 -0.6 -0.7 -0.8 -0.9
POTENTIAL (V. vs S.C.E.)
-1.0
Fig. 13. Hydrogen evolution characteristics on a passivated (A) and oxide
free (B) 316 stainless steel surface in 0.02M H2SO4 at ambient
temperature
-------
2 2
but instead the current increased from 14.0 mA/cm at -1.2 V to 62 mA/cm
at -2.5 V. In line with mass transport limitations, currents were greater in
0.02M H2SC>4 solution.
_3
These data were repeated in the same solutions and in 10 M IH^SC),. The
results, summarized in Fig. 14, show exactly the same trend. Deviations
from the Tafel regions occur at less negative potentials and the "sloping"
plateaus are reached at lower current levels as the acid concentration is
reduced. In these plateaus, currents tend to be independent of rotation speed
of the electrode. The most likely explanation for the data is that mass trans-
port is not limiting but rather that an adsorption step or electrochemical dis-
charge step limits the overall rate of the reaction.
3+
Back Reduction of Fe : An obvious complication in the total process for
Fe24- oxidation is that Fe3+ produced at the carbon anode may be swept over
to the steel cathode and reduced back to Fe2+ or indeed to Fe. Already shown
werethe potentials where significant H2 currents can be attained on a steel cath-
ode. Additional information is also required on where (i.e., at what potentials)
significant reduction of Fe species can take place also on this cathode. If
these two regions overlap, then it may be necessary to shield the cathode from
Fe34- oxidized at the anode by using some type of membrane.
34-
A preliminary scan on a rotating steel electrode to determine where Fe is
reduced to Fe2+ andFeis shown in Fig. 15. Data were obtained in 2 x ICT^M
Fe^"1" solutions in 0.02M H2SO4 at a scan rate of 0.102 V/min and a rotation
speed of 51.1 rps. On the forward scan (40.20 V to -0.70 V), the current
does not increase significantly until -0.20 V when a rise to a small plateau is
observed near -0.60 V. Above this, currents increase sharply. Reversal of
the scan at this point leads to a well defined plateau from -0.60 V to -0.20 V
which then falls sharply between -0.20 V and 0 V. The current in this plateau
region is 716.7 juA. Assuming a diffusion coefficient for Fe^+ of 2 x 10~6
cm2/secll and taking a rotation speed of 51.1 rps, then from the limiting
diffusion equation to a rotating disk, a current of 758 juA would be expected
from a one-electron transfer process namely, Fe^+ 4- e~ -» Fe2+. We attri-
bute the plateau in Fig. 15 to this reaction then.
The data on Fig. 16 illustrate the sixth scan on an electrode which had also
been left at 40.140 V for 1 hr before the scan and indicate widely different
current voltage characteristics. Here reverse and forward scans are much
closer together and show no plateaus. In addition, the current levels are
smaller than in Fig. 15. Thus, the currents at -0.60 V are 340 juA on for-
ward and reverse scans respectively, compared to 420 j^A for the first scan
on the electrode (Fig. 15).
The explanation for these observations lies simply in the fact that the pro-
tective oxide coating is also formed on the electrode here [the Eoc is 40.200 V
- 30 -
-------
OXIDE FREE/' 0'02 M H2S04
SURFACE ^3 |Cf3M H2S04
PASSIVE/2 °-OIM
SURFACE U 0.02 M H2S04
-0.5 -0.7 -0.9 -I.I -1.3
POTENTIAL (V vs S.C.E.)
-1.5
-1.7
Fig. 14. The effect of H2S04 concentration and electrode rotation speed on
the hydrogen evolution reaction on a 316 stainless steel surface at
ambient temperature
-------
0
-200
UJ
cr
cr
ID
O
uo
-400
-600
START
SCAN
I
0.2
0
-0.2 -0.4
POTENTIAL (V. vs S.C.E.)
-0.6
Fig. 15. First voltage scan to -0.70 V on a freshly polished 316 stainless
steel cathode in deaerated 0.02M H2SO4 solution at ambient tem-
perature and containing 2 x 10~3M Fe3+ (scan rate - 0,102 V/min)
-------
OJ
UJ
+ 0.2
0
-0.2 -0.4
POTENTIAL (V. vs S.C.E.)
-0.6
Fig. 16. Sixth voltage scan to -0.70 V on a freshly polished 316 stainless
steel cathode in deaerated 0.02M HSCU solution at ambient tem-
perature and containing 2 x 10'^M
(scan rate *= 0,102 V/min)
-------
3+
(SCE) in Fe solutions] and is not removed by scanning to -0.70 V. As the
number of scans between 40.20 and -0.70 V is increased and the time of
immersion of the electrode at positive potentials is increased (^ +0.20 V),
the protective oxide film is thickened and thus its inhibiting effect on the
pe3H- -* Fe2+ reduction process becomes more pronounced.
To confirm these conclusions, the electrode was scanned more negative and
the effect of this on the FeS* -» Fe2+ reduction plateau was monitored. It
was found necessary to scan to between -0.80 and -0.90 V before a well de-
fined plateau was observed on the reverse scan to more noble potentials. In
all cases, when the electrode was brought to +0.20 V before each scan, a
hysteresis was observed between the forward and reverse scans at > -0.30 V
and amounting to ~ 0.150 V. The scans on the two plateaus coincided at
< -0.30 V. This is shown clearly in Fig. 17 where the negative scan was to
-0.90 V. It is interesting to note that a slight dip occurs at -0.225 V in plateau
obtained during the reverse scan on the electrode. This could be due to a
superimposed oxidation of Fe (to Fe2+) plated onto the electrode at -0.90 V
but, more likely, it is caused by anodic oxide formation on the electrode.
In any event, it seems that oxide at this stage is not quite so inhibiting on
the Fe3+ reduction process since the plateau is again observed at -0.20 V
before it drops off to zero current between -0.150 and +0.20 V. Obviously
also, once all the Fe2+ is oxidized off the surface (at +0.20 V), the inhibiting
effects of the oxide film again came into play. It is worth noting finally that,
if the negative scan is only carried out to -0.80 V, then the fall in current in
the reverse plateau occurs between -0.20 and 0.9 V. Also, the dip in this
plateau at -0.225 V is not observed here. Both facts provide further evidence
of the stability of the protective oxide at negative potentials and of its effect
on the reduction process when it is present on the surface in the absence of
Fe2+. In line with earlier data in 0.02M H2SC>4 only (i.e., no Fe3+ present),
these results show that the oxide is not substantially reduced at until ^ -0.90V.
This is also why the reduction plateau for Fe3+ -» Fe2+ (-0.150 to -0.60 V) is
not observed until potentials are more negative than would be expected from
the literature value of +0.340 V (SCE)3 (10"3M Fe3+ solution).
Having clarified the processes operative on the stainless steel electrode in
these environments, a final check was required that the plateau between ~- 0
and -0.60 V on the oxide-free electrode was indeed due to the reduction of
Fe3+ to Fe2"1" at a diffusion limited rate. The Fe3+ concentration in 0.02M
H2SO4 was thus varied from 2 x 10~4M to 3.7 x 10~3M and cyclic scan ex-
periments were run on the electrode at various rotation speeds for each con-
centration. In all cases, the electrode was scanned to -1.0 V (the current
here was generally 5.9 mA) before measuring the current on the reverse
plateau (oxide-free electrode). The data are a linear function of rotation
speed as would be expected from a diffusion limited process on a rotating
disk electrode. Also, the current at a given rotation speed is directly pro-
portional to Fe^+ concentration. For example, at oA/2 (rotation speed) V2
- 34 -
-------
0-
START
SCAN
OJ
VJ1
-200
z
LU
DC
<£
ID
O
UJ
g-400
-600
SCAN
TO -0.90V
1
+0.2
-0.2 -0.4
POTENTIAL (V vs S.C.E.)
-Q6
Fig. 17. Voltage scan to -0.90 V on a stainless steel cathode in deaerated
0.02M H2SC>4 solution at ambient temperature and containing 2 x 10 M
Fe3"1" (scan rate = 0.102 V/min)
-------
1/2 -43+
equal to 14 rad/sec ' , the current in 2 x 10 M Fe is 42.5 juA; it is 470
in 2 x 10"3M Fe3+; and it is 990 juA in 3.7 x 10"3M Fe3+. Based on the latter
current and concentration, currents of 535 HA in 2 x 10~3M Fe3"1" and 53.5
in 2 x 10~^M Fe3"1" solutions would be expected.
The diffusion limited current to a rotating disk electrode can readily be cal-
culated. At oA/2 equal to 14 rad/sec*/2, we would expect current of 874.9 j
472.9 fiA in 3.7 x 10"3M Fe3+, 2 x 10"3M Fe3+, and 2 x 10~4M FeJ+ solutions
for a one-electron reduction process to Fe2"1". The observed values are 990,
470, and 42.5 juA, repsectively, in these concentrations. The agreement is
thus good and indicates that the reduction plateau on the "oxide free" stainless
steel electrode from -0.150 to -0.60 V is due to diffusion limited reduction of
Fe3+to Fe2"1" species.
Deposition of Metallic Iron; It has been demonstrated that reaction
(Fe3"1" -> Fe2"1") occurs over the range -0.15 to -0.60 V at a diffusion limited
rate prior to onset of H2 evolution. In the next experiments, the cathode
was scanned generally between 0.0 and -1.0 V in an attempt to isolate a
diffusion limited plateau for Fe3 reduction to the metal. Electrode rotation
speed was varied. In 2 X 10~3M Fe3+ at 23.4 rps, assuming a diffusion co-
efficient for Fe3"1" of 4.1 x 10~6 cm2/sec, H a diffusion limited current of 1.23
mA for reduction to the metal would be expected. In some scans, a slight
plateau was observed at < -0.80 V. This was not, however, reproducible
though it occurred generally at current levels of 0.90 to 1.4 mA. It was not
strikingly dependent on rotation speed. It was therefore not possible to con-
clude with any degree of certainty whether or not Fe3+ is reduced to the
metal at < -0.60 V on a stainless steel cathode. Essentially H2 evolution on
the cathode interferes with a detailed examination of this potential region.
The next experiments used a cylindrical 316 stainless steel electrode of
area 3.6 cm2 which was potentiostatted for 90 hr at -1.0 V (SCE) in 0.02M
H2SO4 containing 4 x 10"2M Fe3+ The electrode was brightly polished be-
forehand and was also weighed. At -1.0 V, a constant current of 23 mA was
recorded. The diffusion limited current of a dissolved electroactive species
to a stationary electrode is given by the equation: 5
nFDCA , %
i = 5— (10)
where:
C = concentration of moles/cc
A = area of the electrode (cm2)
5 = thickness of the diffusion layer (normally 0.05 cm in
unstirred solution)
-36 -
-------
Based on the equation, a current of 3.5 mA would be expected for reduction of
Fe6* to Fe in 4 x 10"2M Fe3+ solution. Over a period of 90 hr, this would
result in the deposition of 200 mg of Fe on the electrode. During the actual
experiment, the entire electrode surface became dark and a weight'increase
of 11 mg was observed. Though this is significantly less than the calculated
value, it can be inferred that reduction of Fe3+ (and Fe2+) can occur on a
316 stainless steel cathode at potentials of the order of -1.0 V (SCE). Dif-
ferences in calculated and observed weight change during deposition may be
attributed to H% evolution in the same potential region interfering with Fe
deposition and consequently with weight of Fe deposited. In a practical
setup, it is evident that quite large currents (large overpotentials) can be
supported on the cathode before Fe deposition takes place. Normally then,
this will not be a problem.
TIME DEPENDENT PROCESSES
The experiments described to this point have been entirely short term. It
is important to realize, when evaluating such data and when trying to repro-
duce these results, that superimposed on the phenomenon discussed, there
will always be a slow deactivation of the electrode surface from the absorp-
tion of trace impurities. Examples of this deactivation are given below.
Before proceeding to this information, it is important to realize that the
effects observed are general ones, common to most heterogeneous processes
involving solid surfaces, i.e., a small concentration of surface active material
is sufficient to rapidly deactivate a smooth solid surface. The classical
solution to this problem is to increase surface area. Porous fuel cell elec-
trodes are longer lived than smooth electrodes; porous petrochemical cata-
lysts are longer lived than smooth catalysts. Most impurities reach the
solid surface by diffusion and consequently the geometric area of the elec-
trode is the important parameter in this respect. With highly porous elec-
trodes, these impurities build up at pore mouth, while the desired reaction
takes place slightly within the pore on clean surface.
•
Since this is so, one can ask the question: Why use smooth electrodes at all?
Data of the type required to unravel the various reaction mechanisms are
easiest to interpret with smooth electrodes. In other words, an idealized
smooth vitreous carbon electrode was deliberately chosen for these studies
to simplify investigating the electrochemical mechanistic aspects of the
processes. When used in the form of a rotating disk, mass transfer con-
ditions to this electrode were then exactly characterized. Unfortunately,
- 37 -
-------
the very feature of the electrode that makes this possible, i.e., its smooth-
ness, renders it highly susceptible to poisoning. Fortunately the time avail-
able before the poisoning process sets in had been sufficient to permit the
acquisition of meaningful data.
To this point, the discussion may appear to be an excuse for a possible in-
soluble problem. Considerable credibility is given to the surface area pro-
tection argument from the data in Section 5 of this document. It will be shown
that no deactivation is observed for high surface area carbon electrodes ex-
posed for days to identical solutions which deactivated smooth electrodes in
hours.
Fig. 18 shows a prolonged current transient in chemical AMD water containing
10-2M Fe2+ in 0.02M H2SO4. The calculated limiting current for the Fe2+
oxidation to Fe3+ is 300 j^A. The initial experimental current at -1-1.2 V is
335 juA, slightly higher than expected. More important, however, is the
observation that this current is not sustained for any length of time but begins
to decay rapidly and continues to do so, reaching a plateau at 212 j^A after
14 hr immersion. This corresponds to a decrease in current of 36.7% over
this time period.
At this point, the electrode was "cleaned" by pulsing it to -0.90 V and leaving
it there for 5 min. An initial current at +1.2 V was 384 juA which fell to 330
after 0.5 hr. Thereafter, the current decreased to a plateau at 260 A*A after
14 hr. These data are also shown in Fig. 18. The decrease in current here
from the initial value of 384 yA corresponds to a 32.3% fall. The final pla-
teau current is 48 AtA greater than in the first case.
Following a similar cleaning procedure, the electrode was then potentiostatted
at +0.90 V for 67 hr. An initial current of 300 pA. fell to 223 juA after 14 hr and
then to 165 ft A after 66 hr. After 14 hr the decrease was 25.7% and after 66 hr
the current was 55.0% of the original value. Cleaning the electrode at negative
potentials is apparently more effective than doing so at positive potentials.
_o
Similar data were obtained in a solution of chemical AMD containing 5 x 10 M
Fe2+ as well as 5 x 10"2M Fe3+ in 0.02M H2S04. At +1.2 V, the initial
current was 1360 )LtA and this decreased to a plateau value of 940 ptA after
14 hr, corresponding to a 30.9% fall. The calculated current for Fe2+ oxida-
tion in this solution is 1500 ^ A which is slightly larger than the experimental
value. The current rose to 1250 j^A at this point after the electrode was
"cleaned" for 3 min at -0.9 V.
Clearly then, on the smooth electrode, even in pur,e chemical AMD solutions,
there is a fall in Fe2+ oxidation current with time, which, in the worst ease,
-38 -
-------
380-
340
<300
=1.
260
I
(JO
UJ
OC.
OC
cn
°
I80
140
100
0
8
10 12
TIME (hrs)
14
16
18
20
22
24-
Fig. 18. Current decay during Fe oxidation in chemical AMD containmg
10~2M Fe2+ in 0.02M H2SO4 on a vitreous carbon electrode (A,
initial decay; B, after cathodic cleaning step)
-------
reaches 55% of its original value after 36 hr in the plateau region. Since the
electrode can be cathodically regenerated (-0.90 V), the indications are that
either some surface oxide accumulates on the surface and inhibits the oxida-
tion process or that small quantities of impurities (cathodically reducible) in
our solutions diffuse to the electrode and have a similar poisoning effect.
Data were then taken for various synthetic AMD waters; current transients
were measured in each case with the electrode held at +1.2 V. Fig. 19
shows the transient over a 66 -hr period in synthetic AMD of pH 2.72 and
containing 1.04 x 10~2M Fe2+ (the calculated oxidation current is 312
and 0.73 x 10~2M Fe3+. The solution was not centrifuged. The initial current
is 320 J*A and falls to 24 jtxA after 66 hr. The decrease after 14 hr was 50.6%
and after 66 hr it was 92.5%. So in synthetic AMD the same phenomenon is
observed as in chemical AMD but to a much larger extent. Now, is the effect
due to soluble species or to the suspended solids?
_o
In a centrifuged solution of synthetic AMD containing 2.16 x 10 M Fe and
1.89 x -2M Fe3+, the initial current was 616 M (648 J^A calculated) falling
66 MA after 14 hr, i.e., an 89.3% decrease. Thus, the phenomenon still
occurs, even in these solutionse Cathodic cleaning at -0.90 V only resulted in
an increased current to 380
Since the solutions prepared above were more concentrated than actual AMD
and contained more solids, a centrifuged synthetic AMD solution was diluted
with 0.02M H2SO4. The resulting solution was 2.29xlO~3MinFe2+and 2xlO"3M
in Fe3+. The calculated limiting current for Fe2+ oxidation in this case is
69.4 juA. The transient current at -1-1.2 V in this solution fell from 70 fJ-A to
53 juA after 14 hr (a 24.3% decrease) and thence to 22 juA (a 68.6% decrease)
after 66 hr. Again the same phenomenon was observed but again to a lesser
degree.
In summary, it is evident that on the vitreous carbon electrode employed for
these electrochemical investigations, the current for Fe2+ oxidation to Fe3"1"
decays with time of immersion of the electrode in both chemical and synthetic
AMD, but in the latter solutions the rate of decrease is generally less if the
originally prepared solution is centrifuged and/or diluted. Thus, after 14 hr
at -K).9 V or -fl.2 V in chemical AMD, an ~ 30% fall in current is observed
compared to a 50.6% fall in uncentrifuged synthetic AMD and 24% in centri-
fuged and diluted synthetic AMD. After 66 hr, the equivalent decreases were
45%, 92.5% and 69% in chemical AMD (10" 2M Fe2+), uncentrifuged synthetic
AMD (1.04 x 10~2 and 0.73 x 10~2M Fe3+) and centrifuged and diluted syn-
thetic AMD (2.29 x 10'3M Fe2+and 2 x 10"3M Fe3+), respectively.
It was also found that an increase in rotation speed of the electrode in chemi-
cal AMD causes a sharper drop in current with time. In centrifuged synthetic
- 40 -
-------
-2 UJ
— 00
— O
xsm
Fig. 19. Current decay during^Fe" ' oxidation in uncentrifuged synthetic AMD
containing 1.04 X lO^M Fe2+as well as 0.73 x 10"^M Fe3+ in H2SO4
of pH 2.72 on a vitreous carbon electrode held at +1.2 V and 23 rps
-------
AMD containing 1.88 x 10~2M Fe2+ and 1.62 x 10~2M Fe3+, the rate of faU of
current was 1.9 j^A/min at a rotation speed of 12.1 rpsj at 35.1 rps, the rate
of fall was 6 juA/min. This, together with the observations that the electrode
can be cathodically cleaned at -0.9 V (the optimum time here appears to be
5 min) suggests that the electrode is being poisoned by a soluble reducible
impurity. Formation of some surface oxides may also contribute to the ob-
served decreases. The indications are that the poisons can be removed by
cathodic pulsing of the electrode. Application of large (2.0 V) anodic over-
voltages certainly does not enhance performance of the electrode.
ANOLYTE SEPARATION
3-H 2+
It has been shown that reduction of anodically generated Fe back to Fe
occurs on a 316 stainless steel cathode at < -0.15 V (SCE). This finding has
also been confirmed in a laboratory scale reactor. Because of this, a mem-
brane is required to separate anode and cathode; the following properties
are necessary:
1. Acid stability
2. Low ohmic resistance „
3. Substantial impermeability to dissolved Fe
4. Permeable to Fe2+
One candidate material (1-mil thickness) was obtained from RAI Research
Corporation (L.I. New York).
Acid Stability: To test for this property, a sample of the membrane was
immersed in 0.02M H2SO4 and examined at various intervals up to 1 month
total immersion. The examination was a simple visual one and in addition a test of
mechanical strength was carried out using an Instron Tensile Tester. The latter
gives a value of the utlimate tensile strength of materials by destructive
testing (i.e., pulling the membrane apart and measuring the load required to
do this). A simple visual examination showed no change in the membrane
after immersion. An ultimate tensile strength of 3666.7 psi was indicated
for the immersed specimen which is to be compared with 3895.2 psi for a
sample of the membrane which had not been immersed. It is concluded that
the membrane is stable in these acid solutions.
Ohmic Resistance: This property was evaluated with a conductivity bridge
(model no. R.C. 16B2, Industrial Instruments, Inc., Cedar Grove, N.J.).
A plexiglass conductivity cell was constructed which had provision for in-
sertion of the membrane to divide the cell in two. When positioned, no con-
ducting path between two Pt electrodes was available in the 0.02M H2SO4
solutions on either side of the membrane, other than through the membrane.
- 42 -
-------
The resistivity of the 0.02M IH^SO^ was measured in the absence and in the
presence of the membrane. In the former case, a resistivity of 96.2 ohm-cm
was indicated and in the latter case the value increased slightly to 106.4 ohm-
cm. A previously determined value for the resistivity of 0.02M H2SC>4 was
104.6 ohm-cm. It is clear then that the membrane has a very low ohmic re-
sistance and thus will not significantly increase iR loss in a practical cell in
which it is present.
Q_|.
Impermeability to Dissolved Fe : The same conductivity cell was also used
for these measurements. The membrane was used to separate solutions of
"Fe3+-free" 0.02M H2SO4 on the one side and a solution of 4 x IQ-^M Fe3+ in
0.02M H2SO4 on the opposite side. This setup was left to stand for 11 days
and the 0.02M H2SO4 initially containing no Fe3+ was then tested for Fe3+ by
titration with K2Cr2C>7 using diphenylamine sulfonate as indicator. Though
the test is sensitive to 5 ppm, no Fe3+ species were found. It is concluded
that the membrane is impermeable to Fe3+.
2-t.
Permeability to Fe ; This property was tested in the same cell. One side
of the cell contained 0.02M H2SO4 and the other side contained 10"2]VI Fe2+
dissolved in 0.02M H2SO4. After 8 days, the 0.02M H2SO4, originally con-
taining no Fe^"1", was titrated for Fe2"+ with K^G^Orj, using diphenylamine
sulfonate as an indicator. The titration showed that Fe2+ had diffused through
the membrane to give a concentration of 2 x 10~^M
To further test the selectivity of the membrane, 0.02M H2S04 was placed on
one side of the cell, with a mixture of 5 x 10~2M Fe2+ and 5 x 10" ^M Fe3+ in
0.02M H2SO4 on the other side. After 11 days, titrations indicated that
whereas no Fe3+ had diffused through the membrane, the concentration of
in the H2SO4 which had originally been iron free was now 8.2 x 10"4M Fe2+.
It is apparent then that not only is the membrane impermeable to Fe3+, but it
also selectively discriminates between Fe3+ and Fe2+ species, allowing the
latter to diffuse through. A computer analysis had been carried out previously
on the distribution of the various soluble iron species present in chemical AMD
as a function of pH. This had shown that the dominant ferrous species is Fe2+
while, depending on pH, the principle ferric iron species are the Fe(SC>4) +
complex and the bare Fe3+. The amount of ferric iron present as hydroxyl
complexes is a pronounced function of pH; a pH change of 2.58 to pH 1.8
lowers the concentration of Fe2(OH) 24+ (mol % total Fe) from 10.6 to 0.72%.
Most likely the membrane distinguishes between the two Fe species on a size
basis alone, i.e., the ferric iron is much larger because of its complexation
in solution and does not permeate through the membrane.
Performance in Synthetic AMD; The membrane was next tested in synthetic
AMD prepared in the laboratory from waste coal (containing 1.15 x 10~2M
Fe2+ and 9.65 x 10"3M Fe3+ at pH 2.67) . On one side of the cell was placed
- 43 -
-------
fresh 0.02M H2SC>4 and on the other was placed the above, uncentrifuged,
synthetic AMD. At the end of 9 days, the fresh 0.02M H2SO4 was tested for
iron content. This solution originally containing no Fe2+ and no Fe3+ was now
found to contain 4.25 x 10" 4M Fe24" but still no Fe3+. The results of the ex-
periments in chemical AMD are confirmed. Also, over a 32-day period in the
synthetic AMD, no disintegration in mechanical properties of the membrane
was found. It is concluded that the membrane is an emminently practical
material for use as a separator for anode and cathode in an operational elec-
trochemical AMD unit.
The following is a list of those conclusions, based on the work described in
this section of the report, which are relevant to the development of a practi-
cal AMD treatment process.
2+ 3+
1. The anodic oxidation of Fe to Fe does take place on a carbon
electrode at a mass transport limited rate; the oxidation region is about
0.80 V in extent; the potential plateau for this process is shifted to more
positive values by increasing the Fe3+ concentration; this shift (< 0.2 V) is
not sufficient to cause problems in an operational system.
2. Excessive oxidation (high positive potentials) at the carbon elec-
trode will inhibit somewhat the Fe2+ oxidation; this will not cause a problem
in a practical reactor, since such high positive potentials will not be reached.
3. Hydrogen evolution occurs on a polished 316 stainless steel cathode
(in deaerated 0.01, 0.02M H2SO4) at potentials more negative than -0.5 V
(SCE) ; these currents are not diffusion limited but are controlled by a slower
electrochemical kinetic step.
4. The cathode in 0.02M IH^SC^ is passivated by an oxide film at
potentials more positive than -0.04 V; this film is removed only at potentials
more negative than -0.90 V the normal operating region of the cathode.
3+ 2-1-
5. The diffusion limited back reduction of Fe to Fe occurs at
potentials more negative than -0.15 V.
6. Soluble ferric iron is reduced to the metal on a 316 stainless steel
cathode in 0.02M H2SO4 at potentials =s -1.0 V; large currents from H2
evolution can be supported before iron deposition occurs.
- 44 -
-------
7. A highly selective membrane is available which will prevent the
penetration of Fe3+ specie while stillpassing Fe2+; the membrane has
acceptable ohmic and structural characteristics.
8. Smooth carbon electrodes show a slow degradation in performance
as the result of poisoning by soluble species; the degradation is enhanced in
AMD water prepared from waste coal; the degradation has not been detected
with porous carbon electrodes.
- 45 -
-------
SECTION 5
ELECTROCHEMICAL REACTOR CONFIGURATION
The economics of any large scale electrolytic process are dependent upon the
capital costs-of the electrolysis cells. Obviously, for a given set of electrode
materials, the smaller the required electrode area, the less expensive will
be the system. For the total process being considered here, i.e.,
24" 4" 34-
Fe 4- H 4-electrical power -»Fe + 1/2 H0
z
the overall area of the electrodes is determined by the rate of diffusion of
Fe2+ to the anode-electrolyte interface. To a first approximation, this dif-
fusion rate is proportional to the concentration of Fe2+ in the bulk electrolyte
and to the diffusion coefficient of the hydrated Fe2+ ions. The rate is inversly
related to the thickness of liquid boundary adjacent to the electrode surface.
The diffusion coefficient is an inherent property of the AMD solution and is
not amenable to external control. The concentration of Fe^+ is fixed by the
specific AMD water being treated and by the desired conversion factor. Thus
the only variable available for experimental manipulation is the distance Fe2+
must diffuse in order to react on the electrode surface. (In this treatment,
mass transport mechanisms such as convection are considered as techniques
for reducing the diffusion boundary layer and are essentially specified by the
reactor configuration.)
Three reactor configurations (simple parallel electrodes, fluidized bed elec-
trodes, and packed bed electrodes) were evaluated in the context of minimizing
this diffusion boundary layer and hence the eventual size of an electrolytic cell.
Specifically, the investigation of each configuration involved the reduction of
experimentally measured reactor performance to an analytical equation. Then,
choosing a "typical" AMD concentration and an acceptable conversion factor,
these equations were used to compare the reactor types via a minimum geo-
metric anode area.
- 47 -
-------
EXPERIMENTAL
All the experimental engineering investigations were performed in a small
pilot plant constructed entirely of glass and 316 stainless steel. The elec-
trolytic reactor was formed inside a 4 in. glass pipe lined with a 1/2 in.
thick graphite anode; a smaller column was available for preliminary studies
of fluidization phenomena. The system was designed so that variations in
reactor length and electrode geometry could be made without excessive
downtime.
At 100 A (the current capability of the power supply) the maximum possible
AMD flow (0.02M Fe2+) for 100% conversion is 48.5 gal/hr. A high speed
Teflon gear pump, together with a variable bypass stream, provided a
smooth, nonpulsating flow from a 200 gal polyethylene tank. Flow through
the reactors was controlled and measured by valved rotameters below the
columns.
All experiments were performed using a chemical AMD solution containing
either 0.01 or 0.02M Fe2+and 0.02M H+. The stock solution was prepared
and stored in the 200 gal tank incorporated in the system. The principal
system variables were flow rate and current. Iron (total and Fe^+) was
determined via standard techniques. A periodic check was made on the stock
solution to determine if any Fe2+ oxidation (from dissolved air) was occurring
spontaneously; no Fe%+ was found.
For all runs, the Fe2+ conversion was determined after the effluent had
reached a constant concentration. Although it was desired to measure the
rate of hydrogen production by means of a wet test meter, it was found that
the gas production rate (at these low currents) was too far below the range
of the instrument to permit reliable readings.
ANNULAR FLOW REACTOR
The basic annular flow reactor contained a 3 in. diameter carbon anode and
a 2.875 in. stainless steel cathode. The effective column length was 4 ft
(120 cm). The cathode was wrapped with a single layer (0.010 in.) of porous
polyethylene felt in order to prevent shorting to the close spaced anode.
Table 1 presents the data obtained for a typical test run. The flow rate was
1.7 gal/hr. The current was varied step wise and the indicated final Fe2+
concentration was measured. Cell voltages are not presented since they
- 48 -
-------
represent mostly iR losses in the current leads and are thus specific to this
single experimental setup.
Table 1. Conversion Data for a Flow of 1.7 Gal/Hr
2-L
Current, Final Fe Current
Amperes Concentrations, M Conversion, % Efficiency, %
0.5 0.0174 12.5 100
0.75 0.0162 19 100
1.00 0.0153 24 93
2.00 0.0136 31 60
3.00 0.0130 35 49
The increased ferrous oxidation at the higher currents (1 ampere and above)
results from the oxidation of Fe2-l- by oxygen evolved at the carbon anode.
Note that, although higher Fe^+ conversions are now possible, the process
is relatively inefficient.
These data were then compared to the results predicted by applying a suitable
mass transfer equation to the experimental reactor configuration. A corres-
pondence in results would establish the validity of the mass transfer equation
as used for the system under study. The general equation^ for the diffusion
limited current in annular flow (for a one-electron process) is:
L. = 0.85 FD.C
lim i
where
(1 - k) rD.X
1/3
(ID
F = 96, 500 A-sec/equivalent
D. = diffusion coefficient of species i
C = local bulk concentration
oo
< V> = flow velocity
r = radius of the outer tube
X = distance from the leading edge
k = radius of inner electrode/radius of tube
- 49 -
-------
— fi o --
The diffusion coefficients used were 5 x 10 mol/cm -sec for Fe and
2 X 10-6 mol/cm2 -sec for Fe^+. It was shown in the rotating disk experi-
ments (Section 4) that these coefficients reasonably well described the dif-
fusion of Fe2+ and Fe3+, in 0.02M
The quantity of 0 is a tabulated geometric parameter, which describes the
effect of different annular configurations on the limiting current. As can be
seen, the limiting current is directly proportional to the bulk concentration
and inversely proportional to the cube root of the distance from the up-
stream edge of the annulus.
For the type of annular flow being considered, the limiting current is
essentially infinite at the upstream edge of the anode where a fresh solution
is brought in contact with the electrode. The limiting current drops as the
hydrodynamic boundary layer increases in thickness and as the bulk con-
centration begins to fall. In the lower part of the column, the growth of the
boundary layer limits the maximum current. In the upper portion, the
boundary layer reaches a relatively steady thickness and the change in the
bulk concentration controls the limiting current.
For a given annulus geometry and flow rate, the above equation can be solved
using an incrementally determined bulk concentration. The bulk concentra-
tion is computed by a simple mass balance and the value used to compute the
new limiting current. The entire concentration profile for a column of
arbitrary length can be rapidly calculated using a computer.
The computer program (written in DCAL) shown in Table 2 was used to
provide analytical confirmation of the experimental data obtained from the
pilot plant.
The program first requires the major input variables (step 1.2) : the flow
rate in gal/hr (FLOW) , the radius of the anode (ROUT) . After computing the
flow area and velocity, the program requests the value of the geometric
parameter 0 (PHI) in step 2.2. In step 2.45, the initial concentration of Fe2+
(ICONC) is converted into a charge per unit volume basis. The initial
limiting current is calculated in step 3.1 according to Eq. ( 10)described above.
Step 3.15 takes into account the existence of the ferric to ferrous back re-
duction which, as shown, occurs at the cathode. Step 3.2 computes the quantity
of charge passed in the first centimeter of column height using the previously
calculated limiting current. Steps 3.4 and 3.5 compute a new concentration
to be used in the next iteration. The calculations continue until the concen-
tration profile for the entire column is developed (steps 2.6 and 3.6) . The
computed concentration profile for a flow rate of 1.7 gal/hr is shown in
Table 3 below.
- 50 -
-------
Table 2. Computer Program for Evaluation of Annular
Flow Data Obtained with Pilot Plant
1.1 FLOW.= Plf RTN = 0, RnUT = Pi, P.HI-0
1.?. DEMAND FLOW, RIN, POUT
1 .25 AREA = PI*((Rni!TA?.)-(RJ.NA?.))
1.3 VELOCr (FLOW*. 1.K3) /AREA
1 .4 KAP=RIN/RO!JT
1.5 DO PART 2
2. 1 TVPE KAP, AREA
2.2 DEMAND PHI
2.3 TCHARGErPi, 131,11=?!, ITLI^f , PCHAPGE = P!, MCHARGE^P, I
2.4 DEMAND DIPT.CnNC, ICON
2.^5 ICHARG.E: jcoMC*?^5PPi?!p'^*f:;.^5*ARFA
?.5 TYPE "CONCENTRATION PROFILE AT GALLONF/HOUR
HEIGHT (CD CONCENTPATIOMC^I) CURRENT C1A
2..f DO PART 3 -FOR DIPT rDIFT PY ! TO 12R
3.1 I2LI^r((2^. I*COMr.*PHI*CDIFTAC-.333)))*C(V?:i.nC/C C 1 -KAP )*RO'!T*
K.5^)) -.333))
3. 15 I3LI^I=((13. l*CICONC-CONC)*PH-I*'(Dr?TAC-.333.)) )*((VELOC/(( 1-
3.2 PCHARGEr 5H82*P J*POUT* (I ?.LI v|- I 3LJ.^!)* ( VELOC "- 1 . P1)
3.3 TCHARGErTCHARGE+PCHAPGE
3.4 fOCHARGErlCHARGE-TCHAPGE
3'. 5 CO NC = ( I C ON* NCH AP GE ) / 1 C/H A R GE
3.6 TYPE IN FOR1 J? DIPT, I 0C*P!*CONC , 1 HPiPi*! P.LI^l , IF FP
3. .7 TYRE IN FORM ?. ? I TLI *1/ 1 2'7!*287 1 IF FP DI?T/].20=0
FOR.1 1 s
7,7,7, 7.7,7,7.7, 7,. 7,7,
FOR "I 2r
REQUIRED CURRENT: %%.%%
- 51 -
-------
Table 3. Concentration Profile at 1.7 Gal/Hr
Distance,
cm
10
20
30
40
50
60
70
80
90
100
110
120
Concentration,
M
0.0192
0.0188
0.0184
0.0181
0.0178
0.0175
0.0172
0.0170
0.0168
0.0166
0.0164
0.0162
The predicted final concentration of 0.0162M agrees with the maximum ob-
served conversion at 100% current efficiency. Further confirmation of the
applicability of the analytic equations used was obtained by extending the
data to include a wide range of flow rates. These results are shown in Table
4. Here again, the agreement between the experimental and analytical
values is excellent.
2+
Table 4. Final Fe Concentrations as a Function of Flow Rate
Flow,
gal/hr
5
2.5
1.7
1.0
0.5
0.26
0.0
Observed
0.0184
0.0173
0.016
0.016
0.0132
0.010
0.007
24-
Fe Concentration, M
~ «—^^
Predicted
No Back Reduction
0.018
0.0168
0.0160
0.015
0.012
0.0092
0.00
Back Reduction
0.018
0.0170
0.0162
0.0154
0.013
0.0096
0.0071
-52 -
-------
O i
The effect of the competing back reaction (the reduction of Fe on the cathode)
is apparent. The predicted maximum conversions indicate that this back re-
duction of Fe3+ does not become of major significance until the total conversion
reaches 50%. Since conversions in excess of 90% are desired, it is apparent
that a means of suppressing the competing back reduction is required. As is
shown in Table 5, where the computer predicted concentration profiles for a
flow of 0.1 gal/hr are compared, conversions of 75% or more can be obtained
in relatively short reactors operated at low flow rates once the ferric to
ferrous back reduction is suppressed.
Consideration of several alternative approaches to eliminate this reactor
inefficiency led to an evaluation of ion selective membrane materials. A
polyolefin based material produced by RAI Industries was found to fulfill the
basic requirements of stability in dilute sulfuric acid solutions, low ohmic
resistance (i.e., permeability to solvated protons and anions), reasonably
low cost, and a low permeability for hydrated ferric ions.
Table 5. Concentration Profiles for a Flow of 0.1 Gal/Hr
Distance, cm
10
20
30
40
50
60
70
80
90
100
110
120
No Back Reduction
Concentration, M
0.0154
0.0130
0.0113
0.0099
0.0088
0.0079
0.0072
0.0065
0.0059
0.0054
0.0050
0.0045
Back Reduction
Concentration, M
0.0157
0.0137
0.0124
0.0114
0.0107
0.0102
0.0097
0.0093
0.0090
0.0088
0.0086
0.0084
FLUIDIZED BEDS
In a fluidized bed, the continuous agitation of glass beads suspended in the
interelectrode space by the AMD flow reduces the thickness of the diffusion
boundary layer (5) and thus increases the limiting current density. A further
advantage of a fluidized system lies in the elimination of the hydrodynamic
boundary layer which is present in a simple flow system.
- 53 -
-------
EXPERIMENTAL RESULTS
The experimental configuration was similar to that described above. The
carbon anode was 2 ft long and had an internal diameter of 3 in. The stain-
less steel cathodes were 0.84and 1.90 in. in diameter. Since wall effects on
fluidization velocity are not important, these variations in annular size permit
the fluidization of a wider range of bead sizes than could be possible if the
reactor geometry were fixed. All experiments shown were performed with
commercial grade glass beads, (Gataphote Corp.). The particle size given
thus represents an average value.
Note that the bead size and the liquid velocity cannot be independently varied,
since each size bead requires a specific liquid velocity for fluidization. The
fluid velocity at 65% bed voidage ranges from 0.05 cm/sec for 35 ju beads to
2 cm/see for 375 M beads. A graph of fluid velocity (at a measured 65% bed
voidage) versus particle diameter is shown in Fig. 20; the observed velocity
varies as the average particle size to the 1.54 power. These results are in
excellent agreement with the data of Saxonl2 which indicate that a value of
1.59 should apply at constant bed voidage.
Table 6 presents the experimental data for the treatment of chemical AMD
(1000 mg/jf Fe2+, 2000 mg/£ H"*")-. The actual current passed through the
reactor was consistent with the theoretical current calculated on the basis of
the observed ferrous to ferric conversion. For each experimental point, the
reactor was operated at constant current until the effluent Fe^+ concentration
reached a steady value.
The quantity of glass beads in the bed was adjusted so that at the required bed
height (2 ft) bed voidage would be 65%. This value has been shown to result
in optimum mass transfer rates. Note that in the case of the larger (and
hence more mobile) beads, conversions are now approximately a factor of
20 times greater than a simple annular reactor of equal anode area.
Table 6. Fluidized Bed Data
Final Fe2+
Flow Rate, Bead Current, Concentration, Conversion,
gal/hr Size, ju amps M %
34.3 375 9 0.017* 12
20 200 5 0.0176f 12
1.2 35 1 0.0112f 44
*(initial 0.0193).
t(initial 0.02).
- 54 -
-------
Slope = 1.54
1000
Particle
100
100
10
O.I
10
u
4>
in
\
E
o
o
T3
CD
CD
5*
ID
CO
1.0
a:
Fig. 20. Plot of fluid velocity versus particle diameter at 65% bed voidage
-------
Since in these experiments no membrane was used around the cathode, the
back reduction of ferric ion must be considered in evaluating the effectiveness
of the apparatus.
In these experiments, conversions approaching the desired 95-99% were not
obtained. According to the subsequent analysis, this was primarily because
the reactor length was too short for the flow rates being employed. It was
shown subsequently that in the case of the 35 ju beads, a column 320 cm long
would have been needed to produce a Fe conversion above 95%. The alter-
native of using bead sizes below 35 n was not feasible due to entrainment of
the very fine particles.
ANALYSIS
Mass transfer correlations for liquid fluidized beds are not available. We
have therefore used the following theoretical model of the configuration to
assist in analysis of the experimental data. In a fluidized bed, the local
limiting current density is presumed to be proportional to the local concen-
tration of the reacting specie. In other words, although the height of the bed
affects the overall conversion, the local current density is not a function of
position. It is, therefore, possible to represent the local limiting current
by an equation of the form:
ILIM = FKON • CONG (12)
where:
2
ILIM = the local limiting current (amps/cm )
FKON = a proportionality factor which is constant for a given
reactor geometry, a bead size and flow rate
CONG = the local concentration (mol/l)
This equation should be compared to Eq. (13) below, which expresses the
limiting current in stagnant solutions Eq. (5).
ILIM = n F5D CONG (13)
where:
n = number of electrons transferred
F = 96,500 AMP-SEC/equivalent
- 56 -
-------
D = diffusion coefficient
6 = boundary layer thickness (cm)
g
CONG = concentration (mol/cm ).
Note that if a value for the coefficient of performance (FKON) is known, a
value for the equivalent boundary layer thickness in the fluidized system can
be computed as shown below.
6 - nFD (14)
1000 • FKON
24- 3-f
For the reaction Fe -» Fe + e , Eq. (13) reduces to:
8 - °-5 x 1Q"3 ,«x
FKON t10'
Using the experimental data shown above, values for the constant FKON were
calculated via an iterative computer program (Table 7) developed specifi-
cally for this purpose. Since for each point the flow rate and the final and ini-
tial Fe^+ concentration are known, the unique value of FKON which satisfies
these boundary conditions can be established. In addition, the reverse pro-
cedure, i.e., predicting concentration profiles for other size reactors, can
also be readily accomplished.
FKON is first given an initial value which is used to perform an incremental
material balance along the length of the column. Successive estimates of
FKON are made until the predicted value of the final ferrous concentration
coincides with the experimental value. The program is written in DCAL. In
Part 1, the cathode size, the anode diameter, the average particle diameter,
the flow rate, and the bed voidage are requested. The flow area and the fluid
velocity are then computed. Part 2 requires the reactor length and the initial
and final ferrous concentrations. Step 2.45 converts the initial ferrous con-
centration into the quantity of charge present in the first centimeter of the
column. Part 3 performs successive material balances up the column. The
local current density is calculated in step 3.1 which also corrects for the
back reduction of Fe3+. In Part 4, the computed value for the final ferrous
concentration is compared with the experimental data. Part 5 computes the
particle Reynold's number after the proper value of FKON has been deter-
mined. The values so obtained are shown in Table 8.
- 57 -
-------
Table 7. Computer Program for Evaluation of Fluidized
Bed Data
1.2 DEMAND RIN, ROUT, DPART, FLOW, PVOID
1.25 FAREA=CPl*((RnUTA2)-(RIN'k2)))*RVOin
1 .3 VELOCr ((FLOW*. 1 f,3 )/FAREA )
1.4 DO PART ,2
2.1 TYPE IN FORM 1: FAREA, VFLOC
2.3 TCHARGE=0 PCHAPnE=0, NCHARGE=0f ILIM=0
2.4 DEMAND LENGTH, ICONC, FCONC
2.^5 ICHARGE=ICONC*C>6500*.FAREA*6;.45
2.5 TYPE IN FORM 2r FLOW
2.6 FKON=.02
2.7 DO PART 3
3.0] DI?T =1, TCHARGE=0
3.02 CONG:ICONC
3.1 ILLMr(FKON*CONC)-(FKON*CONC*(1/2.5)*((IGONC-CONC)/GONC)*RIN/POUT)
3.2 PCHARGE:5080*PI*POUT*ILIM*(VELOC"-1)
3.3 TCHARGEtTGHARGE+PCHARGE
3.4 NCHARGE:JCHARGE-TCHARGE
3.5 CONGr(IGONC*MGHAPGE)/ICHARGE
3.51 DIPT =T)IPT+1
3.52 TO PART 4 IF PIST=LENfiHT+l
3.53 TO 3.1
4.1 FKON=FKON+.005 IF CONC>FCONC
4.101 TO 5.001 IF CONC
-------
Table 8. Fluidized Bed Calculations
Particle Reynolds
Flow, Bead Size, Numbers,
gal/hr ju FKON VD P/JU
34.3 375 0.315 8.58
1.2 35 0.0055 0.02
20 200 0.20 1.70
Fig. 21 shows the relationship between the calculated values for the perfor-
mance coefficient (FKON) and the experimentally determined particle Reynolds
number as shown on the graph. The data are well correlated by Eq. (16) .
FKON = 0.17 N15 (16)
The upper limits of the above equation are not clear, since the flow rates
obtainable in the apparatus were not sufficient to fluidize larger beads. How-
ever, based on other work concerning heat transfer in fluidized beds, such
a simple relationship is unlikely to hold above a Reynolds number of 100.
The values for the equivalent boundary layer thickness are seen to be less
than 0.001 cm at the higher Reynolds numbers. Since a value of 0.05 cm is
representative of stagnant systems, 5 the abrasive effect of the glass beads is
clearly evident from both an experimental and theoretical standpoint.
PACKED BED REACTOR
In this configuration, the entire anode space was filled with porous, con-
ductive particles in direct contact with the electrode wall (the "feeder" elec-
trode) . The intent was to extend the electroactive surface into the bulk of
solution thus minimizing the average diffusion required by the ferrous iron.
The exact distance over which the electrode potential was effective remained
to be determined. It is to be expected, a priori, that the total exposed elec-
troactive surface would be greater than in either of the previous configurations
considered.
The anode bed was formed from 4 x 10 mesh carbon granules within a 3 in.
diameter carbon cylinder. Bed heights of 1 ft and 2.5 ft were used. The
stainless steel cathodes (perforated in order to allow the generated hydrogen
- 59 -
-------
o
i
o
fe
^
cu
o
ctf
s
0)
"o
c
• I—I
o
\)
o
u
1.0
0.1
FKON = 0.17 NT
0.28
0
0
0.1
10
Particle Reynolds Number,
100
Fig. 21. Correlation of the coefficient of performance with ^ the particle
Reynolds number
-------
bubbles to leave the reactor without obstructing electrode area) were cen-
trally located along the length of the carbon bed. Cathode diameters of 0.83
and 1.90 in. were used in order to investigate the effect of bed width. For the
data shown, the cathodes were wrapped in a single layer of a ferric imper -
meable membrane.
For each experimental point, the column was operated at constant flow rate
and current until a steady effluent concentration was obtained. The Fe2"1"
stock solution (0.01M Fe24, 0.02M ^804) was stored in a 200 gal polyethy-
lene tank. The time for each point varied from 5 hr at 1 gal/hr to less than
30 min at 20 gal/hr. Iron concentrations were determined via standard
titration techniques.
RESULTS
Figs. 22, 23, and 24 illustrate the basic operating data obtained on packed
bed reactors. The results are plotted in terms of the observed ferrous to
ferric conversions versus the applied current (expressed as a percentage of
the theoretical current required for complete conversion). Data are given
for flow rates of 1, 2, 5, 10, 15, and 20 gal/hr. The data in Fig. 22 were
obtained at a bed length of 1 ft and a bed width of 1.58 in. Fig. 23 represents
runs carried out at the same electrode spacing but with a bed length of 2.5 ft.
The effect of changes in bed width from 1.58 in. to 0.55 in. is shown in Fig. 24
for a bed length of 1 ft.
Below a critical mass transfer limitation for each flow rate, the steady state
concentrations fall on the 100% efficiency line. When this point is exceeded,
most of the excess applied current is consumed in the anodic generation of
oxygen. Some of this oxygen produced will oxidize additional Fe2"1", but this
process is relatively inefficient as can be deduced from the low slope of the
operating lines deviating from the 100% efficiency area.
Table 9 compares the mass transfer limitations observed with the 1.08 in.
and 0.55 in. bed width. These data were obtained from Figs. 22 and 24.
Table 9. Effect on Column Performance of Changes in Bed Width
Observed Conversion, %
Flow Rate,
gal/hr 0.55 in. width 1.08 in. width
2 50 47
5 28 27
10 22 20
15 20
20 18
- 61 -
-------
% CONVERSION
I
ro
o —
Fig. 22. Packed bed data for a bed width of 1.58 in. and a bed length of 1 ft
-------
CO
DC
Id
o
o
S5
100
90
80
70
60
50
40
30
20 —
10 —
0 -
/
100% EFFICIENCY
v 20 gol/hr
o 15
o 10
A 5
• 2
D |
20 40 60 80 100
% THEORETICAL CURRENT—*-
Fig. 23. Packed bed data for a bed width of 1.58 in. and a bed length of 2.5 ft
- 63 -
-------
z
o
CO
cc.
LJ
>
Z
o
o
90
80
70
60
50
40
30
20
10
•100 %
EFFICIENCY
1
20 40 60 80 100
% THEORETICAL CURRENT *•
Fig. 24. Packed bed data for a bed width of 0.55 in. and a bed length of 1 ft
-------
Since slightly higher conversions are exhibited by the bed which contains less
carbon, it must be presumed that much of the carbon contained in the wider
bed is electrochemically inactive. The width of this active carbon zone
determines the maximum electrode separation and hence the total area of the
stainless steel required in the final reactor.
At a given AMD flow rate, the actual velocity in the bed containing less carbon
is higher because of the reduced flow area. Therefore the higher conversions
shown in Table 9 are to be expected as a result of increased local turbulence
in the fluid surrounding each carbon particle. The next size parameter is bed
length.
Table 10 compares the data obtained from the 1 ft (Fig. 22), and 2.5 ft (Fig. 23)
columns. The data represent the maximum conversion obtained at 100% current
efficiency, i.e., the mass transport limitation of the columns at each flow rate.
Table 10. Effect of Bed Length on Column Performance
Flow Rate, Observed Conversions, %
gal/hr 1 ft bed 2.5 ft bed Exponent
2 47 66 2
5 27 50 2.5
10 20 39 2.5
15—33 —
20 — 31 —
If, as expected, the diffusion limiting current for ferrous oxidation is pro-
portional to the ferrous concentration, the effect of bed length on the overall
conversion should be given simply by:
(1-c)1 = F (17)
where
c = conversion in a column of unit length
i = number of unit lengths in the column
F = final conversion observed
Computed values, (obtained by plotting the observed conversions versus bed
length on log-log coordinates) for the exponent £ were given in Table 10.
- 65 -
-------
Since the values determined experimentally are close to 2.5 (the ratio of the
column lengths) Eq. (17) should be applicable in predicting the performance
of columns of arbitrary length.
Fig. 25 presents the mass transport limited current plotted against flow rate
for the region where the overall conversion was relatively constant. These
data were obtained from Tables 9 and 10. For liquid flow through packed beds,
standard correlations ^ predict a slope of 0.69. The experimental data ob-
tained indicate a value of 0.67. The excellent agreement obtained allows
justifiable scaleup to beds operating at higher superficial velocities.
COMPARISON OF REACTOR TYPES
A summary of the pilot-plant data for each of the three reactor configurations
considered is shown in Table 11. All the reactors had the same basic inside
diameter of 3 in. Since the data points chosen are at currents where signifi-
cant conversion via oxygen production was not occurring, the currents passed
represent the absolute amount of Fe2+ that was removed.
It is difficult to compare the three systems directly from these data since the
concentrations, flow rates, and reactor sizes are different. However, it is
evident that the simple flowthrough reactor cannot compare with the other
configurations. In comparing the fluidized bed with the packed bed, it must
be noted that the fluidized bed current includes a factor of 2 for the increased
Fe^+ concentration, a factor of 1.5 (3.40-4) for the increased flow rate, as
well as factor of 2 for the reactor length.
A more effective comparison of the three systems can be given in terms of
the anode electrode area needed to handle a practical fluid flow. For this
purpose, treatment to 95% conversion of the following AMD water was
considered: 1000 mg/4 of ferrous iron, 0.01M H2SO4 at 6, 000 gal/hr.
The electrode area was computed from the experimentally based extrapolation
equations derived above. These comparison data are given in Table 12.
- 66 -
-------
<0
10
9
8
7
6 ~~
o 5
UJ
o
t 4
o
o
UJ
cc
a:
^
o
Bed
Width
A 1.08 in.
© .55 in.
X 1.08 in.
I
SLOPE = 0.67
I
5 10
FLOW RATE, Gal/Hr
15
20
30
Fig. 25. Settling times of AMD sludges
-------
Oi
CD
Reactor Type
Flowthrough
Fluidized bed
Packed bed
Table 11. Comparison of Reactor Configurations
Initial Fe2+
;entration,
M
0.02
0.02
0.01
Flow Rate,
gal/hr
5
34
10
Conversion, %
8
12
20
Reactor Length, ft
4
2
1
Current P
amf
1
9
2
-------
Table 12. Comparison of Electrode Systems Via Anode Size
System Area, sq ft
Parallel plate 50 X 103
Fluidized bed 6.2 x 103
o*
3.25 x 10
Packed bed 1.74 x
*Based on value for FKON corresponding to fluidizing
1 mm beds
•fArea of stainless steel cathode.
A few more details regarding Table 12 are desirable. The value for the paral-
lel plate system is based on a minimum partial electrode spacing of 0.005 in.
since the closer the electrodes, the more efficient the reactor. It is assumed
that a ferric iron impermeable membrane can be used; it is also assumed that
there will be no problem in feeding solution into the electrolyte gap. Any
departure from these assumptions will increase the size of the reactor.
3
The value of 3.25 x 10 sq ft for the fluidized bed is based on being able to
fluidize 1 mm glass beads. The bead size is felt to be the upper practical
limit. At 65% bed voidage, the Reynolds number is 100 and the coefficient
of performance is 0.6.
The design point for the packed bed reactor was the observed conversion of
17.5% at 20 gal/hr. The electrode area figure in Table 12 is for a reactor
consisting of two series connected vertical tanks, each 8 ft high by 3 ft. Each
tank is partitioned into 35 flow channels, 1 in. wide and treating 170 gal/hr
of AMD.
Clearly then, of the three systems, the packed bed reactor requires the
smallest amount of electrode area and should therefore be the least ex-
pressive. It is the economics of this system which will be discussed in the
following section.
- 69 -
-------
SECTION 6
PLANT DESIGN AND ECONOMICS
The technical basis of the process has been presented in previous sections
of this report. Sufficient data have been acquired to justify a first order
evaluation of system economics. This is presented below.
Two key items are treated: (1) the cost and configuration of the electro-
chemical reactor, and (2) the basis for the limestone treatment credits.
Finally the economic picture for the entire process is presented.
ELECTROCHEMICAL REACTOR
As the base level design of the packed bed reactor, treatment of 6000 gal/hr
of AMD containing 500 mg/4 Fe*+ is considered. The ferrous concentration
was to be reduced to 5% of its original value. The design was based on the
pilot-plant data given in Tables 9 and 10. At a flow rate of 20 gal/hr (a
superficial velocity of 4.7 gal/hr-in?), a conversion of 17.5% was observed
in a 1 ft column. Possible performance improvements to be derived from
operation at higher flow rates were not considered. From Eq. (17), a total
column length of 16 ft will be required to accomplish a 95% conversion at
this flow velocity. The reactor would be constructed in two series connected
vertical tanks, each 8 ft high by 3 ft wide. Each tank is partitioned by per-
forated stainless sheet into 35 flow channels, 1 in. wide, each treating 170
gal/hr of AMD. Each unit is thus about 5 ft long for a total reactor volume
of 240 ft3. Table 13 summarizes the capital cost estimation procedure.
- 71 -
-------
Table 13. Capital Cost Analysis for the Packed Bed Reactor
Concept (6000 Gal/Hr, 9&% Conversion)
Reactor:
Stainless steel tank $10,000 Vendor quote
Carbon bed 2,100 53
Membrane 1, 400
$ 13, 50a
Net capital cost: 2.6^/1000 gal treated (10% charge rate)
Due to the modular nature of electrochemical devices, the capital charge
of 2.6^/1000 gal is generally applicable to all situations where a conversion
of 95% is desired. Table 14 summarizes the capital costs for conversion of
90% and 99%, i.e., for other initial Fe2+ concentrations.
Table 14. Capital Charges for the Electrochemical Reactor
at Various Conversion Percentages
Conversion, Capital Charge,
% £71000 gal
90 1.9
95 2.6
99 3.8
Not included in Table 14 are the initial costs associated with the AC- DC
rectifying and control circuits required for the operation of the oxidation
reactor. These costs, which are considered below in the detailed economic
evaluation of the proposed AMD treatment scheme, are sensitive to both AMD
flow rate and initial Fe2+ concentration, as well as to the desired conversion.
Since depreciation charges for electrical equipment cannot be directly com-
pared with the costs in Table 14, they are not included at this point.
In addition to the capital charges- described above, the only other cost peculiar
to the direct oxidation concept is the power cost associated with the oxidation
reaction. Labor and maintenance charges are not significant and would not
be expected to alter the economics of the overall treatment system..
- 72 -
-------
LIMESTONE TREATMENT
The additional costs associated with the electrochemical oxidation step can
only be compensated by cost reductions in other portions of the treatment
scheme.. As wilt be described below, these cost reductions accrue from the
elimination of aeration equipment, the use of cheaper limestone rather than
lime to precipitate the iron, and a reduction in equipment size and disposal
problems due to the denser more rapidly settling sludge produced by the
limestone treatment of high, ferric content water.
2-f
The present technique for treatment of AMD containing substantial Fe con-
sists of lime neutralization with concurrent aeration. The resulting sludge is
passed into either a clarifier- settler or a settling pond to permit the sludge
to thicken. The capital costs of this separation process are the most signi-
ficant in the AMD treatment system (the choice between a clarifier-settler
or a settling pond is usually made on the basis of local topography).
The parameters which permit sizing of the separation equipment (or settling
pond) are the settling rate of the sludge and its final solids content. Both of
these values can-be readily estimated in the laboratory. Fig. 26 presents
data on settling rates for limestone treatment of synthetic high ferric mine
water. Results are shown for systems containing 0% and 5% ferrous (a total
concentration of 1000 mg/I Fe and 2000 mg/JE H- was maintained). In all
cases, the synthetic mine water was neutralized with finely divided CaCOg.
For purposes of comparison, and in order to duplicate current practice,
similar tests were carried out on a 50% ferrous solution which was precipitated
with lime and concurrent aeration. Sludge settling times were determined
by allowing the sludge to thicken in graduated cylinders.
As may be expected from data available in the literature, the sludge settling
rates observed with limestone treatment of pure ferric mine waters are
relatively rapid. A solids content of 6% and a sludge volume of less than 10%
of the original AMD volume were observed after 1 hr. Lime neutralization
of a mixed ferric/ferrous system produced a much slower settling sludge
with a lower solids content. In our experiments, a solids content of 3% and
a final sludge volume of 17% were observed after 1 hr. Although literature
data vary extensively, limestone sludges are reported to14 have ultimate
settling volumes up to 10 times less than lime sludges.
The results observed with limestone precipitation of a mixed ferrous (50 mg/1)
/ferric (950 mg/1) mine water were significantly less attractive. Although
the ferrous concentration in the supernate was lowered to 7 mg/I, the sludge
produced was of a much lesser density (14% of total solution volume). In order
- 73 -
-------
100
90
80
CoC03 Treatment of Pure Ferric AMD
Solids Content 6% Final Volume 9.5%
CaC03 Treatment of 5% Ferrous AMD
Solids Content 5% Final Volume 14%
Ca(OH)2 Treatment of 50% Ferrous AMD
Solids Content
3%
Final Volume
17%
70
60
a>
o>
•a
0)
E
a>
u
L.
0)
a.
40
30
20
10
_L
10 20 30 40
Time, min.
50
60
Fig. 26. Settling times of AMD sludges
-------
to ensure meeting stream discharge standards and to minimize sludge disposal
volume, the effluent from the oxidation reactor will be designed to contain no
more than 5 mg/f Fe .
Based on the above data, the clarifier and/or settling pond requirements for
limestone treatment of ferric mine water, containing less than 5 mg/f Fe2+
may be estimated. Primary settling, performed in a conventional clarifier-
settler with a 1 hr residence time, will discharge 10% of the total stream flow
to a settling pond/storage basin for final compaction and disposal. The use of
a primary clarifier, by reducing the sludge volume and increasing its solids
content to 6 to 10% by weight, facilitates sludge disposal in shallow lagoons
or abandoned mine shafts. The costs associated with the final storage volume
required are, of course, dependent on the initial acidity and iron content of
the AMD.
AMD TREATMENT PLANT DESIGN AND ECONOMICS
Since AMD varies widely in composition as well as flow rate, three flows and
three ferrous iron concentrations were selected (Table 15) in order to
represent a variety of possible situations.
Table 15. AMD Compositions and Flow Rates
2+
FIow Rate, Fe Concentration, Total Acid,
gal/day mg/1 mgCaCOg/4
0.25 x 106 1000 2000
1 x 106 500 1000
6 x 106 50 500
If the iron concentration and total acid are not independently varied, a total
of 9 design cases must be considered.,
The basic treatment scheme is the same in all cases. Under the conservative
assumption that no ferrous iron will precipitate during limestone treatment,
the oxidation reactor is designed for a final maximum Fe24" concentration of
5 mg/4. Since the flow of mine drainage will vary to some extent, even on
a daily basis, a holding pond with a controlled output has been provided. For
plants with a low flow rate, the holding pond is also used for AMD storage,
thereby reducing labor charges since continuous plant operation would not be
necessary.
- 75 -
-------
A limestone slurry is produced by loading a tumbling mill with bulk limestone
and providing a flow of water to give the slurry concentration desired. Lime-
stone containing 75% CaCO^ costing $ 5/ton F.O.B. was used as a basis for
estimating equipment size and operating costs.
The AMD from the holding pond is fed to the electrochemical oxidation reactor.
Following the.oxidation, the limestone slurry is added to the ferric mine water
in a simple neutralization reactor. In the absence of ferrous iron, the pre-
cipitation should be rapid and in a form favorable to rapid settling. A primary
clarifier is used to separate the rapidly settling dilute sludge (about 10% of
the total stream flow) from the iron free supernate. The underflow from the
primary clarifier is sent to settling lagoons for final compaction and storage.
The economic recovery of the hydrogen produced by the electrochemical
oxidation is dependent both on the quantity and iron concentration of the AMD
being treated. Location of the AMD site and transportation and packaging costs
must also be considered. Although hydrogen generation as a byproduct of
AMD treatment can never be economically self sustaining, a significant savings
in overall treatments should be obtainable at the high AMD flow rates.
The investment costs for the AMD treatment plants are listed in Table 16.
Sizing of the process equipment was based on operating times of 8 hr/day
for the 250, 000 gal/day rate, 16 hr/day at the 1,000, 000 gal/day rate, and
continuous operation at 6,000,000 gal/day. Holding pond capacities were
adjusted to provide 30 hr retention volume. Process stream flows are thus
41,600 gal/hr, 104, 000 gal/hr and 250, 000 gal/hr, respectively. Since the
capital costs associated with final sludge disposal are too variable to be
included without reference to a specific location, this investment require-
ment has not been included in the preliminary analysis.
The estimated operating expenses (exclusive of final sludge costs) are
shown in Table 17. For 500 mg/i Fe2+, the operating costs range from
20^/1000 gal at 6, 000, 000 gal/day to 55071000 gal at 250, 000 gal/day.
Comparable figures for current approaches to AMD treatment are not readily
available. In the one case where operating data were available, ^ a lime cost
alone of 13^/1000 gal was reported for a plant treating ~ 3, 000, 000 gal/day
of ~ 200 mg/1 Fe2+, 700 mg/1 H+ AMD. Reported values for total treatment
costs (including capital charges) appear to range from 20^ to $ 2/1000 gal
treated.16
The proposed approach is strikingly superior in the treatment of badly pollu-
ted streams at relatively high flow rates. Conventional lime treatment of a
stream containing 2000 mg acidity/I will result in a lime cost alone
~35(zf/1000 gal. Electrochemical oxidation, followed by limestone neutralization,
results in a reagent cost (treatment power -I- limestone) of only 19^/1000 gal
- 76 -
-------
Table 1&. Plant Investments
AMD Flow,
gal/day
250, 000
1, 000, 000
6, 000, 000
Fe2+ Conv,
mg/i
1000
500
50
1000
500
50
1000
500
50
Holding
Pond,
M$
1.5
1.5
1.5
6
6
6
30
30
30
Oxidation
Reactor,
M$
160
140
70
400
351
175
951
844
422
Electrical
Equipment,
M$
42
25
8.4
84
50
17
168
100
18
Limestone
Mill,
M$
16
6
4
26
10
6.5
39
15
9
Instru. +
Controls,
M$
3
3
3
4
4
4
6
6
6
Piping +
Tanks
M$
4
4
4
5
5
5
6
6
6
Clarifier
Settler
M$
22
22
22
33
33
33
47
47
47
Equipment
Total,
M$
251.5
200
113
558
459
240.5
1247
1048
538
Site
Preparation,
M$
58.5
50
27
142
121
59.5
303
262
132
Total Cost,
M$
310
250
140
700
580
300
1550
1310
670
-------
at 6,000,000 gal/day. Total process costs for the electrochemical oxidation
concept are only 31^/1000 gallons. Thus, even without consideration of
possible savings resulting from the sale of byproduct hydrogen, the direct
electrochemical oxidation approach is cheaper than present treatment methods.
Without reference to a specific site location, credits for hydrogen production
can only be estimated. Treatment of 1000 gal of AMD containing 500 mg/£
Fe^+ will result in the generation of 15 ft3 of hydrogen. Optimized electrolytic
hydrogen plants can produce H2 at about 30^/100 ftr. Shipping charges will
range from 20^ to $ 1/100 ft^ depending on distance and method.•*•' If the by-
product hydrogen can be sold at a credit (after collection and packaging costs)
of 40^/100 ft3, the credit to the treatment process would be 6^/1000 gal of
AMD treated. At AMD flow rates of 6, 000, 000 gal/day, this byproduct return
would represent a savings of 30% on total treatment costs. For streams con-
taining 1000 mg/£ Fe2+, the savings approach 50% of total costs.
Table 17. Estimated Operating Expenses for Direct Electrochemical
Oxidation Treatment Plants, ^/lOOO Gal (Lime Treatment
Range 20^ to $ 2/1000 Gal)
Flow Rate, gal/day
Fe2+ mg/4
X C- 9 11 Ig,/ 3L
Acidity, mg/1
250, 000 1, 000, 000 6, 000, 000
50 500 1000 50 500 1000 50 500 1000
500 1000 2000 500 1000 2000 500 1000 2000
Treatment Power, 5 V 0.5
Plant Power
Limestone
Labor H- Overhead
Depreciation
Total Costs
0.5
3
2
16
15
5.3
3
4
16
27
11
3
8
16
34
0.5
3
2
8
8.2
5.3
3
4
8
16
11
3
8
8
19
0.5
3
2
2
3
5.3
3
4
2
6
11
3
8
2
7
37 55 72 22 36 49 11
20 31
Basis: Power at
Depreciation at 10% of investment
Limestone at $6.67/Ton (100% basis)
Labor -1- overhead at $5/hr
Plant On-Stream 8, 12 and 24 hours day respectively
- 78 -
-------
SECTION VII
REFERENCES
1. Latimer, W., "Oxidation Potentials, " 2nd Ed., Prentice-Hall Publ.,
N.J. (1952).
2. Hill, R., "Mine Drainage Treatment-State of the Art and Research
Needs, " U.S. Dept. of the Interior, FWQA Cincinnati, Ohio (December
1968) p. 8.
3. Zittal, H., Miller, F., Anal. Chem., 37, 200 (1965).
4. "Corrosion Resistance of Metals and Alloys, " Ed. F.L. LaQue and
H. Copson, Reinhold Publ. Corp., N.Y. Chapt. 15 (1963).
5. Levich, V.G., "Physiochemical Hydrodynamics," Ed., N.R. Amundson,
Prentice-Hall Inc., Englewood Cliffs, N.J., Chapt. 2 (1962).
6. Schuldiner, S., Roe, R., J. Electroch. Soc., 110, 332 (1963).
7. Potter, E., "Electrochemistry," Cleaver-Hume Press Ltd. Chapt. IV
(1961).
8. Riggs, O., Corrosion 19, 180 (1963).
9. Newman, J., Ind and Eng Chem, 60, 4 (1968).
10. Vetter, K., "Electrochemical Kinetics, " Academic Press, New York,
London, 537 (1967).
11. Klatt, L., Blaedel, W., Anal. Chem., ^0, 512 (1968).
12. Saxon, J., Fitton, J., and Vermeulen, T., AICHE Journal, 1£, 120
(1970).
- 79 -
-------
13, LeGaff, P., I & E. C., 16, 10 (1969).
14. Lovell, H.L., "The Control and Properties of Sludge Produced From
the Treatment of Coal Mine Drainage Water by Neutralization Process, "
Third Symposium on Coal Mine Drainage (1970), Pittsburg, Pa.
15. Draper, J.C., Third Symposium on Coal Mine Drainage (1970),
Pittsburg, Pa.
16. Steinberg, M., Treatment of Acid Mine Drainage by Ozone Oxidation,
EPA Contract No. 14-12-838.
17. Hydrogen, "Chemical Week" May 19, 1962.
18. Hamilton and Simpson, Quantitative Chemical Analysis,
Macmillan Co., 1964.
19. Treybal, R. E., Mass Transfer Operations, McGraw Hill, 1955.
- 80 -
-------
SECTION VIII
GLOSSARY
N Normality, i.e., equivalents per liter
n Number of electrons involved with charge transfer process
F The Faraday constant (96,500 Asec/equivalent)
2
D Diffusion coefficient (cm /sec)
r Radius
3
v Kinematic viscosity (cp/cm /g)
co Rotation speed (rad/sec)
C Bulk concentration of reactive species
i,. Limiting current
SCE Saturated calomel electrode (a reference electrode)
E Standard equilibrium potential
E Reversible electrode potential
R Gas constant
T Temperature (°C)
A Electrode area
5 Thickness of diffusion layer
< V> Flow velocity
- 81 -
-------
SELECTED WATER
RESOURCES ABSTRACTS
INPUT TRANSACTION FORM
1. Report No.
2.
4. Title ELECTROCHEMICAL TREATMENT OF ACID
MINE WATERS
3. Accession No.
W
5. Report Date
6,
%8:
7. Author(s)
R. Jasinski
L. Gaines
5. Organization
' 12. Sponsoring Organization
15. Supplementary Notes
Tyco Laboratories, Inc.
Bear Hill
Waltham, Massachusetts 02154
'. • Report No*
*^ -,, .
/ "'-. •. ji* .... .4. • ...v. .,-," **,-.'< •.-.
70. Project No.
14010 FNQ
/. Contract/ Grant No.
14-12-859
13. Type of Report and
Period Covered
16. Abstract Experimental and analytical evaluations of the direct electrochemical oxida-
tion of ferrous acid mine drainage have shown that this approach is economically
superior to present lime treatment and aeration methods. Through the use of a packed
bed electrode, the size of the oxidation reactor has been reduced to a stage where the
capital investment required for this equipment can be recovered by cost reductions in
latter treatment stages. These cost savings include:
1. Neutralization with cheaper limestone rather than lime
2. A reduction in sludge settling time due to the better properties of limestone sludges
3. Reduction of sludge disposal volume.
As a bonus, electrolytic hydrogen, produced during electrochemical oxidation should
be economically recoverable at the higher AMD treatment rates.
Preliminary economic estimates of total treatment costs indicate a cost range from
lie7 to 72071000 gal, exclusive of hydrogen credits. These costs are much less than
those of present treatment approaches which appear to have expense rates of from 20^ to
$2.00/1000 gal treated. In theparticular case of a badly polluted stream containing
2000 mg/4 of Ca COS acidity, the total treatment cost of Sic7/1000 gallon for the elec-
trochemical oxidation approach is less than the reagent cost alone (~35c//1000 gal) for
conventional lime treatment.
17a. Descriptors
Mine Drainage
Oxidation
Electrolysis
17b. Identifiers
Treatment
Electrochemistry
Electrodes
Permselective membranes
17c. COWRR Field & Group O5D
18. Availability
19. Security Class.
(Report)
20. Security Class.
(Page)
Abstractor Lewis Gaines
21. No. of
Pages
Send To:
22. Price
WATER RESOURCES SCIENTIFIC INFORMATION CENTER
U.S. DEPARTMENT OF THE INTERIOR
WASHINGTON, D. C. 20240
Institution
Tyco
WRSIC 102 (REV. JUNE
913.261
------- |