feas,b>uty study for the continuous · 4. manganese(li) and zinc(ll) 36 v. sampling system 40 a....
TRANSCRIPT
IVb
SWEM-TR-72.il | „ 1
A FEAS,B>UTY STUDY FOR THE CONTINUOUS
RESEARCH DIRECTORATE
WEAPONS LABORATORY AT ROCK ISLAND
>
Approved for public release, distribution unlimited.
DISPOSITION INSTRUCTIONS i
Destroy this report when it is no longer needed. Do not return it to the originator
DISCLAIMER:
The findings of this report are not to be construed as an official Department of the Army position unless so designated by other authorized documents.
AD
RESEARCH DIRECTORATE
WEAPONS LABORATORY AT ROCK ISLAND
RESEARCH, DEVELOPMENT AND ENGINEERING DIRECTORATE
U. S. ARMY WEAPONS COMMAND
TECHNICAL REPORT
SWERR-TR-72-11
A FEASIBILITY STUDY FOR THE CONTINUOUS AUTOMATED ANALYSIS AND CONTROL OF
PHOSPHATIZING BATHS
February 1972
Pron A1-1-5005A-(Ol)-Ml-Ml AMSCode ^97.06.7037
Approved for public release, distribution unlimited
FOREWORD
This report was prepared by E. P. Perry and R. L. Myers
of the North American Rockwell Science Center, Thousand Oaks,
California, in compliance with Contract DAAFO1 -72-C-O130
under the direction of the Research Directorate, Weapons
Laboratory at Rock Island, U. S. Army Weapons Command, with
S. L. Eisler as Project Engineer.
This work was authorized as part of the Manufacturing
Methods and Technology Program of the U. S. Army Materiel
Command and was administered by the U. S. Army Production
Equipment Agency.
i i
- *.s •- J . I l-l>
TABLE OF CONTENTS
Page No.
Abstract 1
I. Introduction and Scope of Work 2
II. Introduction to the Phosphating Process 3
A. Mechanism of Coating Process 3
B. Composition of Phosphate Baths 3
C. Control of Bath Composition 4
III. Introduction to Automatic Process Control 7
A. Advantages of Automatic Process Control 7
B. Basic Elements of the Control System 7
C. Design of a Data Acquisition System 13
D. System Expansion 16
IV. Selection of Analytical Scheme 17
A. Current Methods of Phosphate Bath Analysis 17
B. Manganese and Zinc Bath Analyses 17
1. Free and Total Acid 17
2. Nitrate Ion 18
3. Ferrous Ion 2k
k. Manganese Ion 25
(a) Ion Selective Electrode 25
(b) Polarography 27
(c) Redox TItrations 30
(d) Complexometric Tit rat ion 30
(e) Colorimetric Methods 30
5. Zinc Ion 31
(a) Ion Selective Electrode 31
(b) Polarographic Methods 32
(c) Titration Techniques 32
(d) Colorimetric Techniques 35
C. Recommended Methods of Analysis (Summary) 35
I i i
Table of Contents (Continued) Page No,
1. Free and Total Acid 35
2. Ferrous Ion 36
3. Nitrate Ion 36
4. Manganese(lI) and Zinc(ll) 36
V. Sampling System 40
A. Obtaining a Representative Sample 40
B. Precision Sampling Valves 41
C. Solution Multiplexing 41
D. Routine Tasks of the Sampling System 42
VI. Control Strategy 44
A. Evaluation of Analytical Results 44
B. Determination of Optimum Reagent(s) 45
C. Quantity of Reagent(s) Added 47
VII. Reagent Addition 50
A. Stirring of the Bath 50
B. Addition of Reagents 50
VIM. Miscellaneous Considerations 52
A. Temperature Control 52
B. Solution Level Control 52
IX. Site Considerations 54
X. Calibration and Maintenance 56
References 57
Distribution 59
DD Form 1473 Document Control Data (R&D) 62
1 v
ABSTRACT
The feasibility of automating production phosphate
coating process baths has been studied under the guidance
of the Research Directorate of the Weapons Laboratory at
Rock Island. The study comprised a detailed examination
of the analysis and control requirements. Continuous
automated analysis and control of phosphate baths are fea'
slble. Analytical methods, sampling system procedures,
and control logic techniques are presented for incorpor-
ation into a prototype system.
I. Introduction and Scope of Work
It is the purpose of this study to investigate the feasibility of con-
tinuous automated analysis and control of phosphatizing baths. The scope of
this report is to define the major problems which would be encountered in con-
structing such a system, recommend approaches to the solution of these problems
and provide a general plan for the implementation of automated control.
The application of corrosion resistant coating is an important phase of
the metal finishing industry. Many techniques have been developed to achieve
this goal. In general, these techniques are based on a chemical or electro-
chemical modification of the metal surface. One specific process which pro-
vides excellent corrosion resistant coatings is the phosphate coating process.
This process was developed in the early 1900's (1,2,3). However, the coatings
produced by the early procedures were difficult to obtain and did not satis-
factorily withstand severe environments. Empirical refinements in the techni-
que were made and by World War II satisfactory coating was being routinely
applied. In the late 19^0's and 1950's, the empirical relationships were de-
fined and studied qualitatively (A,5,6,7)- As a result of these studies, the
optimum concentration for the chemicals in the phosphating baths were known.
The problem of maintaining optimum phosphate bath activity, however,
still remains. As yet, no control system has been built which automatically
controls all of the important bath variables. The lack of continuous control
of bath composition causes a severe problem, since coatings are then produced
which do not meet corrosion resistance tests, e.g., salt spray. The production
of poor coating causes a loss in material and in the manhours required to re-
work the failures.
II. Introduction to the Phosphatlng Process
As a prerequisite to automating any process, the process must first be
understood as it is performed manually. The phosphating process is no excep-
tion. This section of this report will briefly examine the mechanism of the
phosphattng process, the composition of the phosphate baths and the present
method of composition control.
A. Mechanism of Coating Process
A piece of ferrous material which is to be phosphate coated is first
sandblasted, degreased and/or pickled to prepare a fresh clean surface for the
protective coating. The material is then immersed in the phosphating bath.
In the first step of the process, the iron surface is attacked by the acid.
This generates hydrogen gas and ferrous (Fe (II)) ions. The bulk acid concen-
tration must be maintained so that local depletion of the hydrogen ion at the
metal surface will be sufficient to effect the precipitation of the metal
phosphates on the metal surface. The overall reaction which occurs is given
by
Fe + 3Zn(H„P0.) - Zn,(P0,), + FeHPO, + 3H,P0, + H. 2 * 3 * 2(5) *(5) 3 H 2
If manganese is present instead of zinc, both the secondary and ternary man-
ganous phosphates will be formed. The resulting protective coating is a mix-
ture of ferrous hydrogen phosphate and ternary zinc (or manganese) phosphate.
The ratio of the salts present on the surface is dependent upon the ratios of
their concentrations of the ions in the plating solution. Most often nitrate
will also be present in the baths. The nitrate in the bath acts as an accel-
erator to speed up the coating process primarily by oxidizing the hydrogen.
Further details of the mechanism are not important here, but can be found in
a study by GiIbert (7)•
B. Composition of Phosphate Baths
Research conducted on the composition of phosphating solution has shown
that there exist optimum concentrations which achieve the maximum corrosion
resistance of the coating while minimizing the time required and the material
consumed. As stated in Rock Island Arsenal Report 54-2906, the bath composi-
tions should be (in order of decreasing importance):
A. Zinc Base Phosphating Solution
Free Acidity 4.0 points maximum*
Total Acid 24-27 points*
Free to Total Acid Ratio Not lower than 1-6.0
Iron (Ferrous) .40 to .50%
Nitrate 2.0 to 2.5%
Zinc .15% to equal the percent iron
Total Metal Not lower than ,50%
Zinc to Iron Ratio 1-1 to 1-1.5
B. Manganese Phosphating Solution
Free Acid 1.3 to 4.0 points*
Total Acid 24-27 points*
Free to Total Acid Ratio 1-6.0 min. (1-6.5 preferrable)
Iron (Ferrous) .20 to .40% (.30% preferrable)
Manganese .20 to .50%
Total Metal .40 to 1.0%
Manganese to Iron Ratio 1-1 to 1-1.5
Nitrate .3 to 1.0%
*0ne point equals 1 ml of 0.1 NaOH per lOcc sample.
From the above specifications it is clear that the acid and iron concen-
trations are the most important. Although it is not explicitly stated in the
specifications, the difference between the total acid value and the free acid
value is a function of the amount of phosphate present in the solution.
C. Control of Bath Composition
Even though the optimum concentrations of the bath have been known for
sometime, a completely satisfactory procedure for controlling the composition
of the bath to meet the optimum values has not yet been developed. Presently,
a sample is withdrawn from the bath and analyzed using wet chemical techniques.
If the analyses show that one or more constituents do not fall within the
tolerance range, chemicals must then be added to correct the composition.
../«../ . I 1-IX
The type and amount of chemical to be added is calculated from predetermined
tables. A typical table is shown here as Table I. This particular table is
for the zinc phosphate solution and was taken from Rock Island Arsenal Report
5^-2906. Immediately the complexity of the control is apparent. To correct
a given deficiency, the proper amount of one or more of five chemicals must
be added. For example, assume that the results of the analysis are:
Free acid 2.0 points
Total acid 23 points
Ferrous .k2%
Zinc .05*
Nitrate 2.2 %
Comparing these values to the optimum range quoted In the previous section,
one finds that: 1) the free acid is under its maximum values; 2) total acid
is less than the minimum permissible; 3) ferrous is within optimum range; k)
zinc content is low; and 5) nitrate is within optimum range. It is, therefore,
necessary to select a control chemical from Table 1 which causes total acid
and zinc to increase while not changing free acid, iron and nitrate. The
closest response which can be found is that for the modified replenisher. It
is noticed, however, that the addition of modified replensiher will also in-
crease the free acid concentration. The free acid is below its maximum allowed
value so a slight increase will be tolerable. Thus, modified replenisher
would be added to increase the total acid to the middle of the optimum range.
The addition of replenisher would hopefully correct the zinc value and not
change the free acid significantly. If, however, analysis after the addition
showed the same or other deficiencies, a similar decision and addition process
would be necessary.
From the above discussion, it can be concluded that the primary problems
of automatic control of phosphate baths are the analysis of solutions and the
complex problem of reagent addition. This study will concentrate on these two
areas and also include discussion of the lesser problems which are more easily
solved, but nonetheless still important to the successful operation of the
system.
c o
o CO
ai c
CO -C Q. 01 o
CD ol
<TJ CO
o c
01 CO i- 3 oi CD
C o o
-Q CTJ
CD +J 4-i C (0 CD 1- 4J 4-1 C — O z o
c
c c — o IM O
c <D
C 4-1 O C >- O
— o
CO T> 4-1 .— o o \- <
<u -o 0) — u o u- <
<SI
CO CO
E ca
CO
CD CO
01 <D^ 01 4>> «D -C 4> Ol V. — o — C 01
01 CD 01 CD 0) i_ u c
01 OI • CD CD
X 01 01 o CO CO 1- <D CD
0) s O. 1_ 1- E E Q. O U CD CD CO c CD
CO CO ~
Q
01 01 01 01 <D ^—^ <D CD CD 01 4-J 01 oi 01 CD -C CD CD CO CO en CD (D CD !_ •— i_ i_ i_ o •— O o O c 01 c c V
~ " " Q
01 01 <u CU 01 01 CD CD CD <D 1_ <D (D i_ c_> E E O c CO CO c
~ CO co
"
01 01 01 01 co <D CD CD 01 01 01 01 CD CD CO CO CD V CD CD 1- u i_ i_ o O o o c c c 1)
—• — — Q
01 ID 01 CO CD i_ U c
01 <D 01 co CD 1- O c
o < u
01 — c l_
O o •— £ 4-> Q. • — 01 "O o -o -C < Q-
-C U 01 c •—
•— c rsj (D .— i» Q.
(D O- o£ 3 1- TJ >- CO
CO 0) •— -o <+-
c — — O X X> »- o o
— s:
OI OI CD CD 01 01 CO CO CD 4) \~ l_ o O CD ID Q Q
vO
-a c
01 <o CD -o U1 _c .—
— Ol o CO .— -C 1_ -C
o CO 0) 01 •U
E l_ • •— • CO 3 d*? 2 T)
CO 4J LH -a O CD XvO — T3 •— • o -a E CO 01 co •
01 1 — c <D^^ -—^ 1 CD oi 3 01 4-1 4) CO i-> — O CO -C oa o s_ c T> ID Ol o u- CD CO 1_ •— NO — O -o -o o — C"» •»«, c i_ — «^ (0 01 CD CO CD X o Q .— 4J
CO CD -C Q. O co
OI u C U CD 4-1 CD o C <D
OI #— •— 01 1- •— U CO ' C ZD — M M- 01 CO CD • • • i_ -a- LTV vO CJ c
Ol CO Ol CD (0
o c
01 CO 01 CO CO \- CJ c:
u CD c *-> •— CD •— rsl l_ to 4J CO O z:
4-> CO
1
•z.
o Ol Q. c Q- 3 •— •— \-
M -C >- c_> CO 0)
O .— T3 •— c o C — 01 o o O X CO l- 3 1- o
ca — —
• CO X 4-> 0 o • i_ 4-1
> Q- •— CL CD c CO 01 o •—
c co 01 o i_ 4J 1- c — O 3
01 4J
E CD SL • CD 01 Ol —
•— 3 CO «— CD O E — 1- c t-
CD (D O E O C 01 O >« .— — o c \ • C 4-1
— .— <x° o CD ^~ TO
-o Ol O -o — -o • T> O < — o < CO
ro
III. Introduction to Automatic Process Control
Encompassed in the area of most automatic process control systems are the
disciplines of chemistry, chemical engineering, mechanical engineering, elec-
trical engineering and computer sciences. A properly designed system will in-
corporate the knowledge and techniques of each of these fields in order to
achieve optimum results.
A. Advantages of Automatic Process Control
A principal advantage of automatic control is that of improved precision
of control. This results in a more uniform product, reducing the time lost due
to producing goods which fail to meet quality standards. A second feature is
that of constant supervision. The operator is then able to perform other use-
ful functions and the process is less subject to fluctuation caused by operator
mental lapses which so often occur when a human is subjected to a repetitive,
uninteresting task.
The principal disadvantage of automatic process control is the time and
expense required for implementation. As will be considered later in this
study, considerable work must be expended in order to consider every detail
of the operation.
B. Basic Elements of the Control System
The basic elements of any process control instrumentation are the sensor,
the comparator and the corrector (Figure 1). A simple example of this mech-
anism is the temperature control of a heated bath. The sensor can be a ther-
mistor or bimetallic spring. The set point can be set to a resistance or
spring position corresponding to a particular temperature. Should the temper-
ature fall below the set value, a contact could close signifying the results
of the comparison. The closed contacts could then be used to correct the
temperature by turning on an electric heater or actuating a steam valve.
The preceding example was a very simple one. Only one variable was con-
trolled and the control function was implemented with analog signals with the
resultant action being on or off. The extension of process control to more
complex systems requires considerably more elaborate equipment. This is
to
r\ Q£ GO O LU 1— 13
s <c > Q. S 1— O UJ o &1
8
o l_
o
to 10 o O o
3 CT>
especially true of situations where most or all of the variables are inter-
related. The advent of the small or mini-computer has made economically
feasible many process control applications which were previously too complex
for automatic schemes. The power of the mini-computer in these applications
is four-fold. First, the computer can perform computations very quickly.
Second, the ability to make decisions is an integral part of the computer.
Third, the computer can be interfaced to external devices so that they can be
controlled by the computer. Fourth, the computer is easily reprogrammed to
accommodate any modification which may be desired for the process. These
changes are very difficult and expensive to make in a hardwired controller.
Using the mini-computer, three different possible configurations are pre-
sented (Figures 2, 3 and k). In each case it is assumed that the computer has
been preprogrammed to analyze the data and to perform the decision making
process. In the first, the computer acts independently in its decision making
role (Figure 2). This is an open control loop, since operator interaction is
necessary. The operator must input to the computer the readings from the
sensors. The computer will then output the proper control actions to be taken
and it is left to the operator to actually execute these functions. The
second approach (Figure 3) is to have the computer control the analysis, read
the data directly and perform the necessary calculations. Since the operator
must perform the actual corrections necessary, the control loop is still open.
The final approach (Figure k) represents fully automatic control. The control
loop has been closed excluding the human operator. The computer controls the
analysis, reads the data, makes the decisions and controls the reagent addi-
tion. The human operator is needed only for routine maintenance or the remote
possibility of system failure.
Since the phosphate process is quite complex, the remainder of the dis-
cussion will tacitly assume that a mini-computer will be included in the final
configuration. The specification of continuous automated analysis and control
requires that the computer assume a role as shown in Figure 4. The remainder
of this study will develop in detail the specific interactions which will be
requi red.
r~\ <s>
DC UJ UJ Z3 1— _J ZD < a. > s: o 1— o UJ
CO
1 r ae or o > ̂ \ O I— f \ h- c_> UJ 1 s y c^ on V y Lu an v. -^ Q. o O (_>
o 1- t-J c o o 1/1 in <D O o
1- o
\_ n> Q. E O o
tD
M 0)
3
1- O
I o
<_> i
CM
0) l_ =J en
10
<
O CC
O o
o
o <-> in 1/1 a) o o
Q.
oo
I
Of - OS o > ̂ v o 1— f \ H- o LU 1 < J cc ec v J LU cc V» -X Q. o o o
11
z o
oo t—1 cc <c oo o OO z Q > LU z oo o o
I O cc
o o
UJ
O0
t cc o 1— (_> UJ CC CC o
cc o < ZD H- CJ
c o o
l/>
CJ O o \-
Q_
Q- o o
o
E O *->
<c
Q- E O U
i
(1) l_
12
C. Design of a Data Acquisition System
The basic relationship of the computer to analytical instruments has
been described in the previous section. Communication between the devices is
bidirectional. First the computer sends control signals to the instrument
directing it and perhaps assisting in making the analysis. During or at the
conclusion of the analysis the analytical instrument sends data to the computer.
The necessity of communication between the computer and the rest of the system
imposes the stringent restriction that all information must be in the form of
electrical signals (digital or analog). In terms of this analysis, the re-
striction means that titration end points must be transduced to electrical
signals. For example, the computer and analog to digital converter alone
cannot detect the color change of an indicator. If a color change were to be
detected, a spectrophotometer would be necessary to provide an electrical
signal proportional to the intensity of light at a given wavelength. In terms
of sampling and analysis, the restriction means that all valves must be ultim-
ately controlled by electrical command whether directly (solenoid valves) or
indirectly (motor-controlled pneumatic valves).
A more detailed description of the interaction is shown in Figure 5- In
this example, the computer is selecting one of three process streams to be
analyzed. This is implemented by a series of solenoid actuated valves. The
selected stream is directed to the analytical instrumentation. This instru-
mentation may be passive in the sense the computer need only to sample the
data and not control the experiment. Examples would be a thermocouple or a pH
reading. However, for the majority of analyses, the computer must take a more
active role. A typical example of this type is an automatic titrator. This
device must be sequenced by the computer, a chore most normally performed by
the human operator. First, the buret must be filled and the sample must be
moved to the analysis cell. Next the start of the titration must be synchro-
nized with the start of the data acquisition so that the data points of the
titration curve accurately correspond to the volume of titrant addea. Finally,
the computer must stop the titration after the end point and purge the analysis
cell. All of these small steps in the analysis, which the human operator per-
forms as a matter of routine, rrust be designed into the computer interface in
13
STREAM 1 STREAM 2 STREAM 3
CONTRO SIGNALS I I ) I MULTIPLEXER
COMPUTER
T
DATA A/D
CONVERTOR
I TTY
Figure 5 ~ Block Diagram of Automated Arulysis System
14
order to achieve a fully automated analysis.
In a manner similar to the selection of the process stream to be analyzed,
the computer must also decide which of the signals from the analytical instru-
ments are to be converted to digital data at what points in time. The device
which accomplishes this goal is known as a multiplexer. The multiplexer is
generally designed so that any one of the channels can be selected at random
or the channels will be sequentially selected upon receiving an advance com-
mand. The order and frequency in which the computer accesses each channel
will, of course, depend on the experiment being performed.
In order that the signals from the instruments be of any use to the com-
puter, they must be converted to digital format. This is the function of the
analog to digital converter. Analog to digital converters are available in a
multitude of configurations. The options include: conversion time, e.g.,
from less than lysec to 1 sec; accuracy and precision, e.g., from +25% to
0.01%; and full scale ranges, e.g., > 1 KV to < 1OmV. Selection of a partic-
ular converter will also be based on the needs of the system. A conversion
technique which is particularly useful in slower applications (conversion
time > 1 msec) is the use of an autoranging digital voltmeter. In order to
display digital numbers, an analog to digital conversion must be performed.
The coded outputs of the voltmeter are generally available for external devices
such as printers or computers. Autoranging digital voltmeters are particularly
convenient since they can typically convert analog signals over a range of 1
KV to lmV at 0.01% accuracy with a lower limit of lyV.
The data from the analog to digital converter is transferred to the com-
puter upon command of the computer. At this point, the data is available for
calculations. Further information can be obtained from the operator through
the teletype. Status and action messages are also sent via the teletype to
the operator.
The extension to automatic closed loop control is easily accomplished
from this point. It is only required that the computer control the corrective
addition procedure through the use of pumps and valves similar to those used
for the sampling system.
15
D. System Expansion
One of the advantages of developing an automatic process control system
for phosphating baths which involves a mini-computer is the capability of
extension of the control to more than one bath. The rate at which phosphate
baths need to be monitored is anticipated to be very slow compared to the
speed of the computer and analysis system. Once the system is developed, it
would be a straightforward project to extend the sampling system to sample
more than one solution. It would similarly be a straightforward, but more
complex procedure, to extend the computer programs to control more than one
bath. The point of emphasis is that only one computer system would be neces-
sary. Thus, the majority of both development and hardware costs are eliminated
in extending control to several -baths.
16
IV. SELECTION OF ANALYTICAL SCHEME
This section describes the different methods of analysis which were con-
sidered to determine those most suitable for automation of the phosphate baths.
The methods were evaluated by comparing estimates of simplicity, reliability
and precision.
A. Current Methods of Phosphate Bath Analysis
For control of the phosphating baths, analyses for free and total acid,
nitrate ion, ferrous ion, manganese(I I) ion (for manganese bath) and zinc(II)
ion (for zinc bath) are desirable. The present analytical scheme, as described
in Rock Island Arsenal Report 5^-2906, involves a titration with base using
two different indicators to obtain free and total acid. Ferrous ion is deter-
mined by direct titration with potassium permanganate using the permanganate
color as the indicator. Nitrate ion is determined by a somewhat complex redox
method, while manganese is determined by oxidation to permanganate followed
by titration with ferrous ion. Zinc is determined by a precipitation titra-
tion with ferrocyanide. The most important of these analyses for bath control
appears to be the analysis for free and total acid and ferrous ions.
If modified, some of these currently used techniques serve as a basis
for an automated analytical system. For example, a titration for free and
total acid with potentiometric end point detection and a direct potentiometric
titration for ferrous ion with potassium dichromate as the titrant would pro-
vide quite satisfactory automated techniques. The procedures currently used
for nitrate ion, zinc ion and manganese ion, are somewhat complicated, invol-
ving a number of steps, requiring the addition of solid reagents and heating.
Thus, they do not appear to be suited for automation.
B. Manganese and Zinc Bath Analyses
1. Free and Total Acid
As indicated above, a titration with base using two indicators, methyl
orange - xylene cyanole mixed indicator and phenolphtha1ein, is now used to
determine the free and total acids. In the usual definition, free acid con-
stitutes the concentration of uncombined but hydrated hydrogen ions, while
17
total acid constitutes the sum of the uncombined and combined hydrogen ion
concentrations (as in weak acids). While the former could be measured directly
using a hydrogen ion (pH) electrode (activity is actually measured but concen-
tration can be derived if conditions are sufficiently defined), the only way
to obtain the latter is through titration with base. As long as titration is
required to obtain the total acid, it is no more difficult to obtain both free
and total acid by titration and generally much more accurate. Thus, potentio-
metric titration appears to be the superior technique to obtain these values.
For the manganese phosphate solution, the potentiometric breaks are not
very well defined. Figure 6 shows a titration for a 10 ml sample which was
obtained from Rock Island on the 12th of November. This curve is similar to
those shown by Gilbert (4). Since the samples were somewhat aged and not
heated, the quantitative results of titration are not too meaningful. The
shape of the titration curve may also have been affected, but probably not
significantly. Of particular interest is the pH range for the color changes
of the two indicators in Figure 6. These indicator changes, particularly the
second, do not coincide with the analytical (potentiometric) end point. It
would thus seem appropriate to titrate to a given pH value. For the manganese
bath, the free acid range is from 1.3 to 4.0 points (mis.) while the total
acid has an acceptable range from 2k to 27 points (mis.). Because of the
approximate 3 point range which defines the working concentration for both
free and total acid, an uncertainty of approximately 0.5 ml would seem to be
satisfactory, and uncertainties significantly less than this should be poss-
ible if a titration is done to a given pH value (see curve).
For the zinc phosphate solution, the titration curve is somewhat better
defined (Figure 7) but, as with the previous solution, the indicator does not
change color at the break point of the titration. Thus, titration to given
pH values is preferrable here also. The uncertainty of end point detection
would appear to allow adequate control capability, since the free and total
acid concentration ranges are about the same as for the manganese phosphate
solution. Potentiometric titrations have been shown to be quite readily
automated as described in the next section.
2. Nitrate Ion
18
c O
o CO
4) •u <TJ
-C Q- in O -c Q-
C X L_ O
I) > l_ 3
<_>
<0
I
1)
01
19
i—i—'—r 1—r
S o
c o
o GO
-C CL
o -C Q-
c M
I. o
> i- 3
<-> C o
ig
Q-
I
i- 3
u.
Hd
20
Highly selective nitrate ion electrodes are commercially available (8)
for the determination of nitrate ion in the presence of other cations and
anions. One of these electrodes was tested for the determination of nitrate
in the phosphating baths. The potential of this electrode responds to the
activity of the nitrate ion according to the relation
•n = TT° + K In aNQ [1]
where TT° is a reference potential and K is a constant. If the electrode
responds such that K is equal to RT/nF, then the electrode response is
Nernstian. If Nernstian behavior is assumed, differentiation of equation [1]
gives the expression
. _ 0.06 ,AC^ r ,
where the activity coefficients are ignored, and C represents the concentration.
Figure 8 shows a semi-log plot of nitrate ion concentration vs. electrode
potential. The slope of this line if 58mV. Since for nitrate ion n = 1 this
is close to the Nernstian value. From equation [2] we then calculate that
for a 5% change in concentration, the potential will change by 1.3 millivolts.
In actual fact, it is a 5% change in activity, not concentration, so if the
ionic strength of the medium changes, the potential change will be somewhat
different from this. Figure 9 shows a plot of ATT in millivolts vs. (AC/C)
for data obtained by dilution of a zinc phosphate solution. Equation [2] in-
dicates that the slope of this plot should be 26 millivolts. The actual slope
is 27 millivolts. The potential of the electrode is not affected by the pre-
sence of phosphate ion. For example, the electrode potential measured in a -2
10 M sodium nitrate solution was 51mV. Addition of .01 M sodium dihydroger
phosphate did not change that potential, while addition of 0.1 M sodium di-
hydrogen phosphate gave a potential value of 50mV. Because of this repeat-
ability, it is felt that the concentration of nitrate in the zinc bath can be
measured to + 5% relative. The concentration range which is acceptable for
the nitrate concentration in the zinc bath is from 2.0 to 2.5%, corresponding
21
-100
—J
a.
a o a: i— <_>
+50-
+100-
+150 -
+200-
[NaN03] moles/liter
Figure 8 - Response Curve for Orion Nitrate Ion Electrode
22
O >
Figure 9 - Plot of AC/C vs. Air for Nitrate Ion Electrode
23
to a relative range of 25%. It is felt that this precision is probably
sufficient for the measurement. However, some additional testing should be
done to back up this preliminary conclusion. The above concentration range
corresponds to a molarity range from 0.32 to 0.40 M. Because it is recommended
that this electrode not be exposed to high concentrations of nitrate, it may
be necessary to dilute the solution 50-fold in order to improve the measure-
ment. In addition, the slope of the potential vs. log concentration plot in
this range may be somewhat less because of ionic strength effects (see Figure
9).
In addition, since the activity of the ion is measured, the activity co-
efficient must remain constant from solution to solution in order to obtain
concentration from the calibration plot. It is advantageous that the ionic
strength of this bath should not change appreciably for these measurements.
This is realizable since the ionic strength is a function only of the total
number of ions and their valence present, not on any specific ion.
In addition, the electrode shows some response to the hydrogen ion con-
centration at these low pH values (9). If the hydrogen ion concentration
remains sufficiently constant in the bath so this interference can be corrected
for by the calibration curve, or if the response is small, no problem exists.
Data to date suggest this to be true; however, this must be evaluated experi-
mentally in more detail. One of the problems noted with this electrode is
the instability over longer periods of time. A drift in potential occurs
which requires that a calibration technique be added to the electrode operation.
The potential of the nitrate electrode in the manganese plating solution,
although in the right potential "ballpark'1, was found to drift steadily up-
ward with time. The reason for this drift was not determined. Additional
studies will be required to explain this drift and to determine if the elec-
trode can be used to measure nitrate in the manganese bath.
3. Ferrous Ion
The presently used technique for ferrous ion involves the direct titra-
tion with potassium permanganate using the permanganate color as an indicator
of the end point. A modification of this approach which would be suitable
for automation involves the direct titration of ferrous ion using a potentio-
2k
metric end point. No pretreatment of the sample is required. Potassium di-
chromate or eerie sulfate would be preferred for the titratlon since standard
solutions of these reagents are highly stable. The titration curve obtained
with these reagents has a well defined break and a derivative of the titration
curve as obtained by the computer can give a very accurate indication of the
end point. The relative precision of this titration for the ferrous ion con-
centration is expected to be better than 1%.
The concentration range specified for ferrous ion in the zinc bath is
from O.k to 0.5%- For the manganese bath, it is from 0.20 to 0.30%. These
represent relative ranges of approximately 20 and 10%, respectively. Thus,
the precision achieveable by this titration is more than adequate.
A polarographic technique can be used for the determination of ferrous
ion. It has a significant advantage in that either manganese or zinc and
ferrous ion can be determined simultaneously (see below). Unfortunately, the
precision obtainable with this technique is not as good as the titration.
Previous studies have demonstrated that a precision from 3 to 5% could be
readily achieved using automated polarographic procedures. This might be im-
proved slightly. Even the poorer precision could still be used, provided a
proper analysis frequency and control scheme were used. This will be discus-
sed further in the next subsection. It is obvious that if such an approach
were used, the direct titration of iron would not be necessary.
Other techniques could be employed (i.e., colorimetric procedure using
ortho phenanthrophthalein) but would not be as precise and would be equally or
more difficult to automate because of possible additional sample handling.
h. Manganese Ion
A number of techniques have been used for the determination of manganese
(Table 2), but only those which appear to be suitable for automation will be
discussed in detail. The various approaches to manganese determination will
be discussed individually.
(a) Ion Selective Electrode. A technique which has several advant-
ages (if it is feasible for this determination) is the use of an ion selective
electrode. A divalent cation-ion elective electrode is commercially avail-
able. With this electrode, therefore, it should be possible to determine
25
c o l/l
.— >. to 4-J •— 0) 01
in c .— 01 3 • X a) **-* .-—s
o TJ 4-1 X> <u .— O 0) l/l
03 c 4-> 3 C 0J — c E in CT> z: o •— .— 4-J
i_ *. 3 f<ri o o 01 O
o 4-J
03
JQ 03
0) -Q
>~ 0) 3: d
etermi ne sum of
Fe(l1) and Mn(11).
o •
• i c -a c > o OJ o <u . m *-> i- i/> in C 03 .— 4-» o E
X3 C - — O cnxj 03 4-J 4->
C 05 on 3 3 •—
•— 0! «— 03 -* o a> o in l/l 4J L. l/l >. 03
gx. , 4-1
1— OJ
0) 0) 03 l/l E C OJ L. OJ 03 03
— c <u -C OJ oo
O l_ "O 4-> 03 c 4-J
05 X) c
OJ O xi m «4- — o c u • 03 — -a 03 >--a — U u >- L. C
0 03 OJ u 4-J l_ l/l 1/3 03 0) 3 m DI»-
C c U •— 1 X) O — • 03 03 • — c in
•a 03 03 *-i o C
>. 03 E C 0) — O 03 L > 1_ *— in 3 •— _c >--Q 4-J 4-J
3 O L.
o >
en 3 1 , c
03 3
1 0) c O 03 4-> o 4-J E
0) * c
03 in o in
c /l
•o ai x> in 4-J 4-J 0 -C 03 c o E 03 C c o 4-J
03 -— V- 03 in 03 03 cn C Q. 4-» .— •— E 3 l/l c
CL o -Q 0 OJ crx> OJ >. 0 0) o 03 1- 03 03 l_
03 i_ .— 1_ •L. 3 u 03 4-J
2: T3 0) O- CL in 14- C l/l
-C 1 C
4-> 03 03 O 1 i— O-
JO C 03
CL E
X 0)
0) 4-J -— c l_ r— in — 3 3 1/) E 1/1 uj 03 - — — 03 in > CJ E .— 0J l/l
in in -O 3 03 C C 3 o > 03 O O i_ o - —
l_> • — 03 Q_ 4-J in
in 0/
. l_ U . V O XI . u OJ in 4-J a. 2 •— 03 a. OJ c
3 U- in 1_
13 4-1
* — 03 i/i
4-J
0J
c X) T3 i*- 03 c •— E o >- • c 03 \_ ^ 03 3
JD 03 1-
03 O
'V 4-J 3
U Q. o C
3 in 03 03 in .— l/l c X O ^—^ C -C O 3 E u- 3 o 4-J — 03 c — •— o. c 0 3 V. o • o CL •'J .* — 03 ji
\- in 03 1_ in Q. c — 0J u i_ 03 4-J 1- C >. 0 03 4-J *-•* 1- cn 4-J
03 i_ c 0) O c u i- 03 o 03 0J i_ O Ll_ O- u_ < O
0)
lO
X
03
U- l4_ 03
1 Xi XI ^—, q ^ o •— — u 03 c ' . — c cn — c CL 03 4-J 03 -3" — 1 o c X — 03 E '—^ •— 0!
o *— in .— • — l-T O — _C O — 03 •— C c c C O 4-i > o XJ c ' 4J o — 03 O z: z: Z -C <D O c c -—' 3 c 03
1—' u- l_ — • J Q. l_ E o s: 0) O 3: • — c 03 0
M- O u- O —» 03 L. cn E i- \_ OJ U_ 01 O 4-1 O
4-J
L. c o 4>J O
-Q 0 4_* 1) 1_ u5 01 03
_a >/l
3
C C c CL l- u- -V u- -C U cn 4-» • — >~ O o o O L) 03 O i/i ^—v in -—^ • — s~^ >- 4-J 03 u
c . — •*-» 01 i_ -C • — — in 03 — • c E 1_ C 4-» 4-J 4-J .— E 1- u- 03 E o iJ XI — 03 > — o 03 0) o <o <u 03 3 4-J 03 U *•—' .— • — X3 _Q — '—" l- *-t -C 14- — U XI l_ •— 03 c > c o ^ 0J C cn C 4-* o O 4-1 — 4-" 1-3-XJ E < 03 -O s: o \_ ST C Z u o_ — U- 03 — X .— o 0) O 1- \_ c — O OJ _c O in 4J O 4-J c E X Q 03 l/l 03 03 O 03 *-> 0J .-» 03 —- 03 3 OJ
E: 0) LU n- 0J N 1_ •— > o N \- u 03 V- £ c X «- X 03 .— l_ • — •— 3 4-J — 3 - — — 03 3 3 03 3 rg o 4) o -C 4J Q. -C 03 o X) in o cnx> XJ 03 C in 0J l/l XJ in cn
XI *-> XI 4-J 03 E •*-> -t-> 03 — 03 03 o • — — 03 E 03 c C a) u- 03 — -C O .— C Q. X 0J 03 0 u X ^ E 03 03 03 u O 03 a: a] a: 3 Q. o 3 in o
u
E *-> a. o -Q 03
O
in 3:
0J >
4J 4-J E
o o o 1_ l- 1- D. +->
i_ i- \- ^J 4-J 4-J 03 o OJ 4-1 4-» *-> OJ 0J 0J l_ 0J Xi CJ a; 03 03 E £ E cn -— o .— E E E — — • — O 03 1_ _Q — .— -~ 1_ \_ i_ i_ lO 4 J .— > v_ L \- o o 0 03 o ^l 4-» *-> i-j •— <— •— .— c 0J m c
• — • — • — o 0 o o o O O (- h- h- o (- > <-) C3- — UJ d
26 ^3
the sum of ferrous and manganses(I I) cations. No experimental work has been
done with this electrode as a part of this study. Such work Is required to
determine Its ultimate suitability for this analysis. As with the nitrate
electrode, the potential of the electrode Is determined by the activity of the
cations involved. Thus, the same factors as evaluated for the nitrate elec-
trode are important for this experimental program. Such things as ionic
strength effects, uncertainty of potential measurement with subsequent un-
certainty in concentration and nature of electrode response with concentration
will need to be determined. In addition, the selectivity of the electrode to
manganese(I I) ion compared to ferrous ion will have to be determined. With
suitable electrode calibration, the concentration of manganese could probably
be determined directly to +5%. This might be adequate for control purposes
but greater accuracy would be better.
However, the real potential of the divalent cation electrode is not real-
ized in a direct potential determination. A significant advantage is gained
by using the electrode to detect the end points of complexometric titrations.
The divalent cation electrode is not sensitive to EDTA complexes since they
do not have a charge of +2. Thus, although the potential error of direct
measurement may be several millivolts (corresponding to 5-10% change in con-
centration), this error is insignificant compared to the normal potential
change of several hundred millivolts at the end point of the titration
(corresponding to a typical change of concentration equal to 5 or 6 orders of
magnitude). For this reason, the best results will probably be obtained by
using the divalent action electrode for end point detection.
(b) Polarography. While manganese is not the ideal ion for polaro-
graphic determination, it does have some favorable characteristics for deter-
mination. One of the main advantages of the polarographic technique is the
possibility that both manganese ion and ferrous ion can be determined simul-
taneously in the same medium, thus simplifying the analysis. Both normal and
classical polarography, as well as pulse polarography, should be considered.
If significant amounts of ferric ion exist in the bath along with the ferrous
ion, a supporting electrolyte for application of pulse polarography needs to
be selected in which the ferrous-ferric ion couple is somewhat irreversible.
27
For normal or classical polarography, a reversible wave can be used, as long
as the manganese(I I) ion wave is sufficiently separated from the ferrous ion
wave. The concentration of ferric ion which can exist in phosphating baths
has been determined (]Q) and has been found to be highly dependent on the nit-
rate ion concentration and temperature. The greater the nitrate ion concen-
tration, the more soluble is ferric phosphate. In a phosphating bath contain-
ing ]% nitrate, the solubility of ferric phosphate is 0.03 g/1 at 200°F.
This is approximately 1/100 that of the concentration of ferrous ion. At room
temperature, the solubility of ferric phosphate is 1.48 gms. per liter in the
presence of 1% nitrate. However, if a sample is taken at 200°F and it is
separated from all sludge and particulates before cooling, the interference
of ferric ion would still be expected to be negligible. The withdrawal of
adequate samples from the bath without obtaining sludge in the sampling lines
and valves is discussed later. Moreover, it would be expected that the con-
centration of ferric ion would be maintained reasonably constant under all
acceptable bath conditions, because it is determined by the solubility of
ferric phosphate. Therefore, the use of an irreversible wave for iron in the
application of pulse polarography to the determination of ferrous ion does
not seem to be necessary. This needs to be experimentally demonstrated.
Table 3 lists possible candidates for the electrolyte choice for simultaneous
determination of ferrous and manganese ions in the bath. Inspection of the
table indicates that there are several electrolytes which look quite promising
for the simultaneous determination of manganese(I I) and ferrous ions. The
specified range for the concentration of manganese in the phosphating bath is
from approximately .04 M to .09 M. By a fifty-fold dilution of this original
solution, the concentration would be essentially in the optimum range for
polarographic analysis. Because ferrous iron has about the same concentration
range, the determination of both ions would appear possible using the same
dilution.
Previous studies have indicated that it indeed is possible to use a
dropping mercury electrode as an indicator electrode in automated procedure.
However, it is probably the most serious drawback of this technique, and
techniques are under investigation for replacing this electrode with a solid
28
XI c OJ
9) in 4) c 0) en c 05 X
c o
0) c
'i 4)
4-1
4) Q
Q. ft) l_ cn O i_ OJ
"o Q-
w
O 0) c nj
3 E
U O
in
>-
u 0)
in in O
ra
4)
X) OJ
W C o
in
o
-o 4) 0) L. 4-> 3 in T3 <U 0) Cn O Cn O 3 u
-a a) 0) 3 4J C7 in •— 0) c D>-C cn u 3 <u
a) -o c o O
o o 4) ID
— <u Ixi o£
~ + — > — — <u _ ^->s
— L. O _ _ —' 3 v.*' _
<u w J-> 0) w LL ra ro u_ c
<U a: • t r —. LA
f US CM <«*""S • ••• <**"•«* • • — ra — — r^ 0 <D — • "M* — • 1 > — O C v_^ O O
<u 1 x 0) 1 J-" -O LL. Lu <o OJ
4J f 4-> (U 3 a) 3 --^ in U O ^ u O — 0) 3 -O — 3 J3 — in nj — in ro — 4) fO — a> —' £ 0) -IJ c at *-• c OJ X <0 X X (0 X 1/)
c a) x a. t +
H) a>
<u X 4-> a. >^
oa O 0 L. 0
4-1 CM O *; a>
— c •> — r a> r^
LL 4) l_ •—
+ 3 in '
,—v 0) *-> — a) 3
-3 X O
4) oj LL. •
> 4-> <l) ro 1- O 3 •0 in 0 c rtj £ a) 4-< 3- 01 + -i:
4) "
in > 01
CL a) • • < 01 in >> 0) 0) l_ •— i_ in in 4) Ol 3 4) .— .— in O Q- > 3 3 4) 1- CL Q. C 03 l_ in 0)
1— O — 1_ L. cn 0 O O c
O- <—* i» ^~ —— +— IT—
11) 03 — OJ 1 in E — E 4) •— i_ —' L. 1_ -C 3 O <U O O C 4-> a. z LL z z
ra LTl
4) >
La x: 4-J
ro MD vO -—* 00 CM cn — • — •— LA CO -a- O La u\ 0
1— CM 4-1
\£> s 0
1 O
1 O
1 0
1 1
> 4>
0 1
0 1 1
> a) L_
O 1
O 1
O c
c—
1 4) l_ 4)
y- L. 4)
4-< c
4) LL.
+ *
^ 0 4) O O a) c 4) LL. 0) C r~
Lu X LL. LL.
t X 0) 4-1
11 t + + t 4) E
C 0)
csl
4) 4-> nj 1 L- .— 4-J O 1_ c 0J OJ r-
4J CNl m 4) -x.
(D 0 •— 0 Z 01 l_ 0J
z (— z LL. 4) La LL Li- C Lu CM O ra 'i " O CM O 0) 0
4) 4) 4)
SO
e ra
-^ 3:
r Z 4) —' 4-J
ID J_ L.
o
X) 0) l_
in 3 o L. L. 4)
29
stationary electrode.
(c) Redox Ti trations. Perhaps the most common titrimetric method for
manganese is the oxidation to permanganate with an oxidizing agent, then
titration with either ferrous ion or arsenite. There are, however, problems
associated with the employment of the usual oxidizing agents. Thus, common
oxidizing agents such as potassium periodate, ammonium persulfate, sodium
bismuthate, etc. require a filter step, heating of the solution or some other
complication to the sample handling procedure. It has been stated (11) that
manganese(I I) can be oxidized to manganese(VI I) by bubbling with a stream of
oxygen containing 5% ozone. The excess ozone is removed by bubbling with a
stream of C0« or N„. This technique would appear simpler to implement from
a sample handling viewpoint.
Another titration technique which appears reasonably direct is the
titration of manganese(I I) with permanganate in a near neutral pyrophophate
solution (12). In this titration Fe(III) does not interfere. However, Fe(ll)
would be expected to interfere. If Fe(ll) is stoichiometrically titrated with
the permanganate under this condition, the sum of Mn and Fe(ll) could be
determined. If not, a preliminary oxidation would be required to eliminate
this interference.
(d) Complexometric Titration. Although several direct complexometric
titration methods have been suggested (13), only one appears to be sufficiently
simple to be considered for automation. This involves the titration of Mn(II)
by EDTA and masking the interference of iron by reducing to Fe(II) and com-
plexing with cyanide as the ferrocyanide. However, a heating of the solution
is required (lA,15) for complexing the iron, as well as the desirability of
back titration which makes this procedure rather unattractive.
(e) Colorimetric Methods. In general, colorimetric methods have the
disadvantage that if any other colored specie; are present, a spectrophoto-
meter is necessary to discriminate the different wavelengths. If species
which absorb at the same wavelength or particjlates are present, interference
will be found. The most common colorimetric method involves oxidation of the
manganese(I I) ion to permanganate with a suitable oxidizing agent (i.e., per-
sulfate, periodate, bismuthate, lead dioxide, etc.) and measuring the color
30
of the permanganate ion. The same problems apply here as in the oxidiation
step of the titrimetric method. The solution must be heated when most of the
oxidizing agents are used (exception; bismuthate) or else there is a problem
of separation of the excess of the (insoluble) oxidizing agent. This type
of procedure, therefore, would tend to complicate the automated analysis. The
most common oxidizing agent is periodate (16). However, ferric ions are also
said to slow down the oxidation (17), and ferrous ions will be oxidized before
the oxidation with periodate.
Another colorimetric procedure which has been studied by several workers
involves the reaction with formaldoxime. However, iron interferes and the
masking of the iron is not straightforward (18). Other colorimetric procedures
involve the Mn (Ml) ion in sulfuric acid, the Mn(lll) diethyl dichrocarbonate
complex and the green Mn(til) complex in basic triethanolamine (12), but the
methods are considered of less practical importance, and have not been studied
thoroughly. In addition, interference of iron is a general problem.
5. Zinc Ion
(a) Ion Selective Electrode. As described in the previous sub-
section on manganese, the possibility of using a selective ion electrode is
attractive. There is not an electrode which is specific to zinc, but there
is a commercially available electrode which is highly selective to divalent
cations. These two are the only divalent cations of any consequence in the
bath. The selectivity coefficient for ferrous and zinc ions is reported to be
identical (3-5 to 1) (9). The electrode response is, therefore, the same for
both ions, and zinc ion can be used for calibration. The same variables as
discussed previously need to be considered; namely, possible ionic strength
changes in the bath, electrode reproducibi1ity and stability, both short and
long term (which dictates uncertainty in the measured metal ion concentration),
and the effect of hydrogen ion and phosphate ion. If the sum of zinc and
ferrous ions can thus be determined, (ferrous ion concentration can be deter-
mined separately by titration as described earlier) zinc could then be obtained
by difference. In this manner good precision would still be obtained for
ferrous ion, the more important species for bath control, while the precision
of the zinc analysis would be about that of the total divalent cations.
31
As discussed previously, another use of the divalent cation electrode is
for the end point detection of a complexometric titration. In a later section
the specific details of the analysis procedure will be examined.
(b) Polarographic Methods. As described in the manganese subsection,
if the concentration of ferric ion in the bath is small enough to be ignored,
either pulse or normal polarography could be used to monitor simultaneously
both ferrous and zinc ions even if the wave for the iron couple is reversible.
Pulse polarography has the advantage that it is less subject to interference
(e.g., oxygen), to small changes in the drop time of the electrode, and to
bath conditions (e.g., stirring); in addition, neither the solution at the
electrode nor the electrode is changed as a result of the measurement. Thus,
there is more chance of using a solid electrode with this technique.
Table h gives some possibilities for a suitable supporting electrolyte for
the simultaneous determination of zinc and ferrous ions. All those listed
which show possibility involve only the addition of a single reagent mixture,
and requires no heating, or other sample handling to obtain concentration of
both ions. This obviously has advantages for automation provided the concen-
tration measurements can be made with sufficient precision.
(c) Titration Techniques. The classical and most commonly used
technique for the determination of zinc involves the precipitation titration
with ferrocyanide. This technique is currently being used for this analysis.
The ferrous ion present in the original solution must be removed. This is
currently done by careful titration to the stoichiometric end point with
potassium permanganate. The interference of ferric ion can be eliminated by
use of pyrophosphate ion, and manganous ion does not interfere.
If this technique were used in an automated scheme, either ferrous would
have to be oxidized by stoichiometric titration, or through the use of an
oxidizing agent. The use of ozone as suggested earlier, would seem to have
many advantages. Modification of the presently used procedures to eliminate
addition of solid reagents and heating of the solution would be highly desir-
able and would not appear too difficult. Either a potentiometric (19), amper-
ometric (20), or dead stop end point (21) is preferable and would seem to be
adaptable (Table 5)- This titration cannot be recommended for an automated
32
in c o
LO
3
o L.
X) c
O c
o
ID c
'i L-
LO
3
o <u c ID
E
L.
o
10 4)
O L-
•t-J
o 0)
in in O
O-
0)
J3 03
<D L- •u 3 l/> T3 <U (U CT> O O) O 3 1-
CO 0_
-a <o
+-> D- LO •— <D C Ol-C Ol O 3 a)
00 i-
CM
a) -o c o o i- — 4-1 4J o o
— D LU CC
o O . O LA LA LA r^
tfl — LA — <A» — LO —
t LA . o t LA t 0 CM 4-< — 4J O *J LA *J
^-. • <D ^ • IB «-~ • n> -~» • ID — O — O — o — O — 1 o — 1 o — 1 o — IO — c —' c *—' c w- C
QJ 4J M 0) *-> M 0) -M M II UN U_ CO LL. f0 Li. IQ U- ID
t t + t a>*-» a> *-* 0)^-. «>-» C i. — *~s u. — ^-* 1- — *-s L- .*—» O 3 3 3 3 — to — — l/> — — 00 — — L0 4J (0 ' nj • nj —••— (0 *• . <D V C V 01 c (IDC 9) 0) C -a Z b. N X Li- M ST LL. M 3E U- M LU
-0 c • • • CM
a. ID \_ cn O u ID
0 O-
GJ 00
3
<D <u o J- c
-L. IVJ
t + +
a) o) c LL.L1.rs1
a> I •u a. >^
r«-» 0 O i_ O 4J CM O v: 1)
a T5 L.
cn O
T3
E u o
•z.
0 LTV CO
0 CM
rA CO
LA CM rsi
O 0 1
0 1
O 1
Q- o. 03 03 L. L Ol cn O O L- u <u 03
i— .— 0 0
a_ O-
CM
O
<? I
> 0)
>
CO O LL. C
+
—* o
LL. M
t I
0) <u c Li. LL. rvl
II
CA X -x a.
-3" x: n) z *-> -^ ID
C
-3" o
CM O
CM
LA rsi
L.
o
> u
> a)
o c
u
0) C u_ INI
CD •M
(D L- •J L ro h-
<NI 31 m O 2 0J
•z LL
LA U- CXI 0
O CM
o >•
0)
o c tu
o LO L_
a)
LO
I) 3 O"
o 0)
CL
03
cn o
o
a) •M .— a) • z: 00
CM E MD 0 i_ •
i<- Q-
c » U LA ^ V.O 03 CA
33
4J CO ro
4-1
c fa > -a ro in
o c
c o
ro c
'i s_ QJ
•M V Q
1- O
in -o O
XT
c o in
1- ro Q. i o
XI
TJ o a) JZ
u> 4-J <o 4) 4-> 3: c CD 4->
> c TJ V < !_
V. 3 O
a) o c QJ
c: O
— -a 4-1 o Q.-C
— 4J l_ 0) o x I/) 0) 4- Q O
1 <o l_ -C
— 4J O.X: * • E O) c 4-J u <u .- o c N
4-1 •— .— . l/> 4J 0 TJ
— in n Q- •— TJ N • • X •
c u •— TJ .— —— o O a> "O C —
— c 1- QJ I/) in o — 4-1 TO ro ro 4->
ro « -a u i- — c m 1) a) XI c 4-J ro ro .— E 1 u o •- O 4-1 ro <0 ro 0) !_ 1- — l/l M- GO co z • —
ro QJ
-C • c
o o 4-1 .—
4-1
T> D <jj < V o z in
1 1_ .— in • • <u
QJ u TJ o o c c c QJ <u o u 1_ .— 0) 1) 4-J
4- <4- o s_ 1_ ro • t) QJ 1_ OJ
4-J 4-J 4-1 .—
c c X -Q — LU ro
QJ
C o • —
> 1 1 XI 0) • • l_ ^—V I- TJ in (A
4) — fJJ^s — QJ QJ y 14 C i_ u i_ — • i ro QJ (D 4)^- *— 4)*-^ > M- u- 4-1 QJ o 4J 0) U \- 1- 1
C LL. 1- C U_ QJ 0) o -— <u •— O 4-J 4-J 4-J <*- o 4_ — c c QJ 4) o >- <u o -Q .— — > •— > i- ••
Q 4) ai O 4) o c c c E O E O O o o o 0) C >- a> c in 1- 1- 1-
CC QJ -Q o£ 0) ro — — —
-Q
OJ ID
IT)
QJ -a • — c
-C ro 4-1 -^
O i o 1-
c i_
o QJ — H-
ro E !_ 3 4-« •— — in 4-J in
ro 4-J 4-J
CL O CL a.
u — •w 2 • • 0)
• •— E c O O
m in X — ro ro a) 4->
— to • 11 0) a. i_ <c E E E 4J l- ro ro O — o
CO GO O •— UJ
4) 4-J 3 c tr a) MM 4-> C o JZ O- u 0) •
4-1 CL l_ 1
o 4) o o — E 4-1 4-J 4-1 u O CO ro c •— L. U a) L-
fJJ X) • — 4-J 4-1
Q. (0 X) O 0) E a) c CL E < Q — '—'
X
a) GO
E C 3 M in
TJ <U C c ro
E ^-v
L. — 0) —
4-1 —' 0) 0J Q Ll_
4-J in C -— U-
4) O O 1 Di !_ o ro U .— OJ 1) N 03 — i_ <t
in 4-1
QJ
c QJ .— • C QJ O E =i >~ a) > u O CLX: 4-J •— c i_ CL o 4-J — x: QJ >_ ro CL u
NJ <_> C o i_ 0) o CD 0) .—
4- u- N •— o i_ 4) • o O •— (0 1- 3 in a)
x; £ ro in TJ QJ QJ 4-1 i_ .— ro C O in in — O O QJ O i- 3 r> a 2
u
CL 3C
C
•— 4-1
O x:
O
4-1
a. ro QJ QJ
g i_ > -o cn — o — O J-J 1-
V- i_ O 4-J
O ro QJ o QJ
o O Q-
QJ OO LU
LA M3
3^
scheme with much enthusiasm. The method is empirical and the composition of
the precipitate depends on temperature, concentration of reagent and reactant,
addition rate, etc. (22). A false end point is also apparently common when
both indicators and potentiometry are used. While the former problems might
be overcome by suitably standardizing the procedure, the false end point would
be difficult in an automated analysis.
Complexometric titration with ethylenediaminetetraacetic acid (EDTA) is
also rather common for zinc. Ferric and ferrous ions however interfere. Ferric
ion has been masked by glycerol (23). Ferrous ion can be masked with cyanide
(2^) but heat is required. However, cyanide is not recommended for use in an
automated field analysis, and thui, one is left with an oxidation, then masking.
Moreover, end point detection schemes other than indicators have not been
exploited to any extent, and most work has involved a back titration of excess
EDTA.
(d) Colorimetric Techniques. The classical colorimetric technique
for zinc involves the use of dithizone (25). Many interferences exist.
Zincon has also been used for colorimetric determination of zinc (26), but
iron interferes. Other lesser known reagents (27) have been used, but inter-
ferences are also a problem. In addition, with colorimetric techniques, rather
high dilutions of the bath would be required to use colorimetric techniques
and the attainment of the required precision in the analysis may be a problem.
C. Recommended Method of Analysis (Summary)
There are many factors which must be considered in designing an analysis
system. Among these are accuracy, reproducibi1ity, reliability, simplicity,
maintenance, calibration, and cost of development. This section will describe
an analysis scheme felt to be most suited to this application, selected from
the previously discussed techniques. It hopefully will also indicate how and
why certain techniques were selected when others seem just as satisfactory.
Table 6 lists the methods of analysis in order of preference. Obviously, if
any of the principal techniques prove unsatisfactory when tested in the lab-
oratory, the alternative methods would be developed.
1 . Free and Total Acid
Since these variables are probably the most important, this problem was
35
approached first. The free acid could be determined to within several per-
cent by a direct pH measurement. However, there is only one method to deter-
mine total acid and that is by titratlon with base. Therefore, this method
was selected also for free acid due to the added precision of titration with
potentiometric end point detection compared to a direct potentiometric measure-
ment.
2. Ferrous Ion
Here also the most precise results can be obtained by titration with
potentiometric end point detection. Dichromate or ceric(lV) are more suitable
as a titrant than permanganate due to greater stability. The choice between
dichromate and ceric(lV) cannot be made on paper, but will depend on which
gives the better results in the laboratory.
3. Nitrate Ion
Subject to the results of additional laboratory tests (described earlier),
the nitrate ion selective electrode seems to be the best method for nitrate.
It is simple, fast, and requires only the electrode for equipment, since one
of the other titration vessels can be used as the sample container. Should
this method fail, the colorimetric method for nitrate appears to be the next
best approach.
k. ManganeseQ I) and Zinc(II)
Considerable development and equipment costs would be saved if the same
technique were used for both manganese and zinc. Because of increased simpli-
city in sample handling and measurement, the use of direct potential measure-
ment using a divalent cation electrode to determine the sum of Mn and Fe or
Zn and Fe should be investigated. A calibration procedure may have to be
employed. However, if this measurement cannot be made with the required
precision or if the electrode is too unstable, then a complexometric titration
procedure seems very attractive. Costs are further minimized since the pro-
cedure is titrimetric using identically the same equipment as the analysis for
acid and ferrous ion.
An aliquot of phosphate solution is titrated with a basic EDTA solution
in a vessel containing a reference electrode (SCE) and two indicator electrodes,
(a Pt wire electrode and a divalent cation electrode). Initially, as titrant
36
CM
0)
ro c l_ <D
-C -S= -C Q- Q. CL 03 <D 13 l_ >_ l_ Oi Ol en o o O u 1_ i_
o> TO 0)
O o o O- O- O-
i/i _— >- .— V 03 4-> C «j < c
-o o
-C 4-i <D z:
~o 4)
X) c
E O u
(J
o>
o
o 4-J L i_ •— 03 v-i 4-1
1- 1- 0) <U 4-1 4-1 E c E c
1 •— ^—* O O O O •— It X — X — .— r^~ OJ 4-1 ID 4->
i_ X o — (0 — (0 o O CM Q. 1- Q. 1-
-o 1- E 4-> E 4-" o 0) <_> O — O — <_> oc: *~• O 1- o I-
i
o c 1
o i O
o •— O •— — •— 4-1 — 4-J 4-1
L. c 4-1 c c 4-> 0) 0! u 1) <D 4-J U 4-J 4-J
m E c o 4-1 o o Q. O O O- Q- Q_ — *— — H u 4-> 4-< 4-J U ^—* 4-1 O 4-1 CJ
c c <o u .— X > o .— L) — — <U >- <D u o — <D u <U 1_
1_ 4-> 4-1 i_ 4-1 X) *>—* i_ 4-J l_ 4-1
0- o — — OJ 01 <D • — 0) — 0) a. \- Q E a£ o Q E Q E
o <
Ol -J
o 1-
0) T3 in c <1) in t) 03 4-1 3 c
0> o OJ <u 1- L. o O) 0) 4-1 1_ c c 1_ — 0) • — 03
37
is added, the ferric-EDTA complex will be formed (pKF /.,.\ = 25 - t) - When
all the ferric ions are complexed, the potential of the platinum electrode
will change significantly. The Pt electrode is only required if a significant
amount of ferric ion is present in the bath. As noted before, it is not
anticipated that this will be true. As the titration continues the solution
becomes more basic forming the complexation of ferrous and zinc or manganese
(whichever is present). The complex formation constants are sufficiently
close (pKFe(M) = 14.33, PKMn(M) • '3.79, pkZn(n) = 16.50) so that no break
point will be detected as the individual species are complexed in turn.
However, when the end point for the sum of both species is reached, the pot-
ential of the divalent cation electrode will change significantly. Calculated
titration curves for equal amounts of Fe (III), Fe (II) and Mn (II) are shown in
Figure 10. The zinc or manganese concentration can then be calculated from
the amount of titrant used from the first break point (Pt) to the end of the
titration, minus the contribution due to ferrous ion when it was independently
determined.
Serious interference could be caused by other divalent cations. These
species would have to be at concentrations greater than ]% of the zinc or
manganese. However, from available information concerning bath composition,
the problem seems remote. If such a titration procedures does not prove to
be satisfactory, then a polarographic procedure should be investigated.
38
POTENTIAL
DIVALENT CATION ELECTRODE
PLATINUM ELECTRODE
Figure 10 - Theoretical EDTA Titration Curve for Fe(
Fe(lI), and Mn(I I)
39
III),
V . Sampli ng System
The function of the sampling system can be very simply stated. The
sampling system must deliver a representative sample of accurately known
volume to the analytical instruments. The key words in the previous sentence
are "representative" and "known volume". The emainder of this section will
discuss the problems associated with meeting those criteria.
A. Obtaining a Representative Sample
The first prerequisite for obtaining a representative sample is that the
bath composition is homogeneous. Recognizing that the sludge is not properly
part of the specified analysis, homogeneous is meant to imply that the solution
phase is for the most part in equilibrium. In fact, since particulates are
especially detrimental to the reliable and accurate operation of precision
sampling valves, it is quite probable that a filter will have to be installed
at the intake to the sampling line to remove any particulate matter (sludge)
suspended in solution. Studies by Gilbert have shown the processing of normal
work introduces a local concentration gradient at or near the surface of the
work (k). Thus, sampling near the work would produce biased results and
should be avoided. A further aid to achieving homogeneity is stirring, a
topic to be discussed In a later section.
Another difficulty in obtaining a representative sample is that of time
fidelity. If an hour were required to obtain the sample, the results of the
analysis would only be of marginal validity in terms of making decisions con-
cerning corrective additions. To overcome this problem, a recirculating
sampling line should be used. This would require a pump to circulate the
solution through a pipe from the bath to the analytical instruments back to
the bath. If a 1 cm (i.d.) pipe were used, the linear velocity in the pipe
would be approximately 13 meters/min. per liter/min. of volume flow. Thus,
if the instruments were 40 meters from the bath, the sample would arrive in
3 minutes, flowing at a rate of one liter/min. This would seem quite reason-
able in comparison with the time response of the other bath parameters. The
actual sampling valve would be located very close to the analytical instru-
40
merits to minimize the number of precision components necessary.
B. Precision Sampling Valves
An important criterion which should be used to select sampling valves is
that of reproducibi1ity. This satisfies the goal of delivering a known amount.
It is not necessary that the samplinj valve deliver 10.000 ml, but it is
necessary that the valve deliver the same amount every time to better than ]%.
The computer can easily correct for a sample size of 9-37 ml or 11.04 ml, as
long as the amount is known and reproducible.
Sampling valves are available as standard items from several firms.
Specific valves for this project should be chosen on the basis of: (a) deli-
very volume as required by the analysis scheme; (b) resistance to corrosion;
and (c) reliability and wear resistance. For example, one manufacturer (Carle
Instruments) markets a sampling valve which is particularly well suited to
this application. The valve is constructed of tantalum, Teflon, and gold-
plated stainless steel. The delivery volume, which can be from microliter
to milliliter size, is reproducible to at least 0.1%. It could be argued
that such a corrosion resistant valve would not be necessary for this system,
but that stainless steel would be adequate. However, under certain conditions
phosphoric acid will mildly attack stainless steel. This is not considered
to be a severe problem for the switching valves, but for the sampling valves
no corrosion is permissible.
C. Solution Multiplexing
The sampling system should be designed so that any one of several differ-
ent sample lines can be selected. The important advantage of this approach
is that one or more of the sample lines can be a calibration solution. The
computer can then automatically check the calibration of the analytical instru-
ments at regular time intervals. Another advantage of this approach is that
the system is thus easily expandable to more than one bath. The main hazard
of this design is the possibility of cross contamination of the solutions.
This is not severe for the bath solutions, since the amounts involved are
virtually negligible compared to the volume of the bath. However, calibration
solution reservoirs would be much smaller and thus more susceptible to con-
41
tami nation.
A simplified schematic diagram for a sampling system is shown in Figure
11. The stream selector valve (commercially available) connects one stream
to the sampling line and returns the other streams to their respective sources.
The sampling line passes through a series of precision sampling valves which
deliver an aliquot of solution to the analysis vessels. Due to the possibility
of solution contamination caused by the mixing of solutions in the sampling
line, the exit of the sampling line is treated as waste.
D. Routine Tasks of the Sampling System
The sampling system must be capable of more than just obtaining the
sample. An automated analysis requires also that the normal "housekeeping"
functions of analysis must be performed. Included in this category are such
tasks as: draining the analysis vessel, rinsing the analysis vessel, refill-
ing the motor driven buret for the next titration, and other similar functions.
As an example, an automated analysis of the free and total acid would involve
the following sequence, which would be done by the computer through the sample
handling system.
(1) Add appropriate sample aliquot to cell.
(2) Add water to dilute, if necessary.
(3) Begin titration, adding base from motor driven buret and
monitor pH.
(4) Record volume of solution to predetermined pH for first end
point.
(5) Titrate to second end point and record volume to predetermined
pH for second end point.
(6) Dump contents of cell.
(7) Rinse cell with water (or acid, if necesarry) to remove
precipitate and residual sample.
(8) If acid is required to remove precipitate, rinse with water.
(9) Refill buret from reagent storage.
(10) Start again at (1).
kl
ANALYSIS 1
ANALYSIS 2
ANALYSIS 3
WASTE
Figure II - Schematic Diagram of Liquid Sampling System
43
V I. Control Strategy
The analysis and sampling systems have provided the computer with the
concentrations of each of the important species. The next phase is to decide
what action, if any, should be taken. This step is not as straightforward as
it might appear and can be broken down into three separate problems. First,
it must be determined if any species are outside acceptable limits. Second,
if one or more species are not within the proper range, what reagent(s) should
be added? Third, what is the optimum amount of the reagent (s) that should be
added?
A. Evaluation of Analytical Results
As the first step in the control process, the analytical results must be
compared to the optimum values (Section II.B). This comparison must also in-
clude the error of analysis. Thus, the questions which actually must be asked
are: "Is the measured value minus the analysis error less than the minimum
and/or is the measured value plus the analysis error greater than the maximum
limit?"
It is important to note that for all species there will be a concentra-
tion range including the analysis error which will be within the predefined
limits. This establishes a deadband, i.e., there is no corrective addition
necessary for this species. In this application, the deadband is quite valu-
able due to the problems associated with stirring after an addition (Section
VI I.A).
An important optimization will have to be made when defining the maximum
and minimum concentration limits for each species. It is important to narrow
the range since it is quite desirable that the bath never exceed the limits
as delineated earlier (Section II.B). On the other hand, it is also import-
ant that the range be as wide as possible to minimize the number of additions
to the bath. One of the important criteria which will be used as a basis for
this optimization is a comparison between rate at which the bath composition
changes and the rate at which composition errors can be detected and corrected.
The relative time lags are also important since the stability of the control
loop is dependent upon their interaction.
B. Determination of Optimum Reagent(s)
Having determined that a particular species is out of range (preceding
section), the optimum corrective reagent (s) must be selected. The most
obvious approach would be to define a set of simultaneous equations which
quantitatively describe the effect of addition of every reagent on the total
bath composition. These equations could then be solved to arrive at an exact
solution for any and all bath deficiencies. However, it is apparent from this
study that the phosphating bath is sufficiently complex that all these rela-
tionships are not quantitatively understood and moreover are difficult to
obtain. Therefore, another approach must be found.
The simpler and recommended approach is an iterative method. In this
approach, the computer would be programmed to react in virtually the same
manner as a human operator. For example, if the iron content is low, the
operator would go to the manual (or more likely recall from previous exper-
ience) and determine that either Iron syrup should be added or that the solu-
tion should be artificially loaded, depending on the acid values being low or
high. Thus, the computer could be programmed to go through the same type of
decision making process. An example of a flowchart sequence for the decision
making process is shown in Figure 12. This example is quite simplified and
shows the decision process for only two cases. In the first case, all species
(free acid, total acid, and Fe(ll)) are within range and no control action is
necessary. This is seen from descending vertically from START, through the
OK response for each decision stop and finishing at NO ACTION. In the second
step, free acid is low and total acid and Fe(II) are satisfactory. The L
branch from the first decision is taken, followed by the OK responses of the
next two decisions, thus ending at ADD PHOSPHORIC ACID. Initially these
responses would be based on previous results and he experience o' current
operators of the baths. However, during development work the computer program
could be written so that poor decisions could be recognized and modified as
control continues.
^
<_>
1°
ADD
SP
HO
R
AC
ID
) o n: a.
c o
-a
< c a) 01 <t> (U
en
u o
o o 1-
Q.
c O
o 1)
1-
o
rsi
<0 1-
^6
C. Quantity of Reagent(s) Added
Sections V.A and V.B have discussed two aspects of the control strategy.
The final aspect is that of determining the amount of reagent which is to be
added. This element is equally as important as the first two, since the de-
gree of control established over the bath is dependent on the amount added.
If, for example, the proper strategy is selected, but the quantity is wrong,
the concentrations will oscillate above and below the desired value if too
much correction is made or will only slowly approach the proper concentration
if the quantity added is too small.
Much of control theory has been devoted to the topic of loop gain and
stability (28,29,30). Indeed if there were only one variable in the system,
the problem of control would be very similar to that of the servo system of a
recorder. In that case the pen is moved to a position which minimizes the
difference between an internal potentiometer and the applied signal. If the
gain of the servo system is too high, the pen oscillates. If the gain is too
low the pen response is sluggish and does not quite follow the signal. Ideal
response occurs when the gain is at an intermediate values ("critically damped")
This pattern is quite identical to that described for the phosphating bath.
It is the purpose of this section to describe the basic elements of control
with reference to how they will be used in this application. In no way will
an exact procedure be specified. An approach will be given which specifies
the starting point and indicates modifications which are available to over-
come problems which may arise.
The error signal needed to activate the control strategy will, of course,
be based on the results of the analysis. For the purpose of the following
discussion, the error signal will be represented by AC . AC is derived from
the difference between the measured concentration of a particular species (C )
and the optimum values for that species. For example, Fe (I I) has an accept-
able range of from 0.40 to 0.50% in the zinc phosphate solution, from this
range the optimum value could be established as 0.45%. Based on these values,
AC for Fe (II) could be defined as
C - 0.45% for |C - 0.i»5| >0.04%
and 0 for |c - 0.45| < 0.03% .
47
This definition incorporates both the error of analysis (5% relative is
equivalent to 0.09% absolute of 0.45%) and an anticipation factor such that
an error signal is generated before the concentration exceeds the specified
limits. The region where AC equals zero is the deadband for this species.
With no a priori reason to suspect that the corrections to the bath will
have to be rapid or that control stability will be critical, proportional con-
trol seems to be the best method with which to start. In this method the
amount of correction is directly proportional to the error signal. Thus,
ARt = aACt [3]
where AR is amount to be added and AC is the concentration error. a is a
combination of several other constants and will include: conversion factors
for units change as appropriate, volume of working solution, effectiveness
coefficient (e.g., 1 pound changes concentration by 0.1%) based on existing
knowledge, and an empirical weighting factor. The empirical weighting factor
is an important parameter. It should first be set to unity and then modified
to reflect the relative success or failure of previously applied corrections.
Should the proportional method of control be too slow to respond, a
derivative component can also be added to increase the response. Derivative
control could be used alone, but suffers from the fact that long term drifts
are not adequately corrected, since the correction is based on the difference
between the last error signals. Using a combination of proportional and
derivative control, the correction takes the form
AR = aACt + B(AC - AC ,) [4]
t Before proceeding further, two points should be made concerning this anal- ysis and description. First, for simplicity the discussion will reference only one variable. The extension to more than one variable is possible, but is not necessary for illustration here. Second, due to the nature of chemical control, ARt will be assumed to be positive. This is quite permiss- ible, since in reality the sign of a will change if the reagent raises or lowers the concentration of the particular species under consideration.
48
where AC . refers to the error existing at the time of the previous analysis
and 3 is the weighting coefficient for the derivative portion. 3 contains the
same factors as a, but its final value is found by increasing the value of the
empirical factor from 0 to a value which optimizes response. Two points are
worth mentioning. First, if the value of 3 is too large, wild fluctuation
will occur in the bath. Second, the previously determined value of a will
probably have to be decreased to maintain stability.
The most general form of control is proportional, plus derivative, plus
integral. This is expressed as
k AR = aAC + 3 (AC -AC ) + Y £ AC [5]
k=0
where y is defined as 3 and k is the "reset" frequency and is determined
empirically. The integral component corrects for changes in loading which
might otherwise cause long term drifts of concentration which are uncorrect-
able by the other components.
The computer programs developed for the project should be constructed
with form three; proportional, plus derivative, plus integral. At the begin-
ning of the test period a best guess is made for a, 3, and y. As the experi-
mental results become available, the values of the coefficients would be
appropriately modified. This approach would be much more efficient than try-
ing to add, say, an integral component at a later time.
Another problem which has not been considered here, but must be a part
of the development work, is that of initial start of a bath. This may re-
quire a separate program, since the normal program for control will probably
require that the bath concentrations are close to their working range.
^9
VII. Reagent Addition
The problem of reagent addition is a crucial one to the ultimate goal of
continuous analysis and control of phosphating baths. Several factors compli-
cate this section of the system immensely. One problem is stirring to achieve
homogeneity of the baths and another is the physical form of the currently
used correcting reagents.
A. St i rring of the Bath
Regardless of the final form of reagent addition used, the problem of
bath stirring will have to be solved. From available information (4,6), it
appears that the poor phosphate coating produced by a stirred bath is not
caused by the agitation. Indeed this is what is expected after observing the
agitation caused by hydrogen evolution during the course of applying a good
protective finish. The problem seems to be caused by the dispersion of sludge
through the solution which becomes entrapped in the coating mechanism.
Irrespective of whether corrective reagents can be added while material
is being processed or not, a baffle will have to be designed and added to the
bath. Even if the reagent is added between processing, the sludge should not
be dispersed because of the time required for resettling. The baffle must
have the properties that: a) allow the sludge to settle to the bottom of the
tank, and b) permit sufficient interaction between the solution and the sludge
so that equilibrium or quasi-equi1ibriurn is maintained. One possible solution
to the problem is simply to mount a series of slats just above the bottom of
the tank. The sludge could settle to the bottom and if the stirring propellor(s)
were directed at approximately right angles to the slats, the dispersion of
the sludge would be minimized.
The method of stirring (propellors, turbines, etc.) will have to be de-
signed in conjunction with the design of the baffle. This is particularly import-
ant since the effectiveness of the baffle is dependent upon the direction and
velocity of the solution which is moving around it.
B. Addition of Reagents
Just as it was necessary to develop new methods of analysis which are
50
compatible with continuous automated analysis, it may also be necessary to
modify the reagent addition scheme currently in use. For the majority of
cases, the corrective reagents are compatible with automated control (6).
However, there are several notable exceptions as described below.
The reagents which are easy to dispense are homogeneous liquids such as
phosphoric acid, replenisher and manganese and zinc nitrate solutions. If the
mechanical equipment (pump, pipes, and valves) is constructed of the proper
material (e.g., stainless steel, Monel , etc.), it is a straightforward process
to deliver a known amount of compound to the phosphating bath. The delivery
can easily be controlled by solenoid valves actuated by the computer.
The main problems of reagent addition occur in the cases of slurries
(e.g., manganese carbonate) and irregular solids (e.g., steel wool, sheet shot
or scrap iron). In these cases the most cost effective approach is to develop
suitable liquid substitutes rather than design additional equipment to handle
the present reagent form. The development of suitable reagents will be
greatly simplified throujh the use of the continuous automated analysis equip-
ment developed as part of the program. The analysis will be available almost
immediately after equilibrium has been established or before equilibrium, if
that is appropriate.
According to current procedures, it is anticipated that the reagents will
be at room temperature when added to the baths. The main advantages are that
no cost is necessary to preheat the reagents and that many of the reagents
(those with phosphates) have a higher solubility at room temperature. It also
seems reasonable to assume that the relative volume of reagent added versus
the total bath is small enough so that no substantial temperature gradient
wi 11 occur.
A further advantage of an all liquid reagent addition scheme is that it
is easily expanded. After the initial shakedown period when one bath is being
controlled, the delivery system could be expanded to several more baths by
adding suitable feeder pipes and solenoid valves. Additional major invest-
ments would thus be avoided.
51
VIM. Miscellaneous Considerations
In addition to the major problems of automatic analysis and control of
phosphating baths discussed in Sections III to VI, there are several other
variables which also must be controlled.
A. Temperature Control
Temperature control of the bath is very important. The best coatings are
obtained if the bath is operated at temperatures over 200°F. However, if the
temperature exceeds the boiling point of the solution, the composition balance
of the solution is destroyed. Apparently, the changes which occur upon boiling
are so great as to render the solution virtually useless. This necessitates
removing much or all of the solution and adding fresh chemicals to restore the
bath to satisfactory composition. This is expensive in terms of both materials
and downtime.
It is, therefore, necessary to install or implement automatic temperature
controllers on the bath. Since thermostats and Fenwal type temperature con-
trollers are readily obtainable and have histories of proven performance; they
would be the most economical and reliable method of temperature control. It
is important that the data acquisition system include a thermocouple to moni-
tor the temperature of the bath and provide a record of the bath temperature.
An alarm could then be sounded if the temperature of the bath were outside the
acceptable limits. The main point is that temperature control is more suitably
achieved if the computer is not included in the loop, but merely monitors the
performance of the controlling device.
B. Solution Level Control
Current phosphating baths have not been provided with any form of solution
level control. The operators have been relied on to add water to supplement
rinse water in maintaining the solution level against constant evaporation
losses. Furthermore, the phosphate baths do not have drain or overflow pro-
visions. This presents an additional consideration for automatic control.
A level sensor will have to be included in the final configuration of the
bath. Unlike temperature control, level control should involve the computer.
52
Several factors must be considered. A deadband should be included to allow
for the volume change when work is immersed in the bath. A decision should
be made to determine whether corrective reagent additions made will be suffi-
cient to raise the level back to within limits or whether water should also
be added.
It would be most feasible in the automation project to just add a level
sensor to the bath. The sensor should provide a direct measure of level of
the bath. A possible choice for the sensor would be a float connected to a
slide wire potentiometer to give a voltage signal proportional to the liquid
level. If the preliminary tests indicate that an overflow provision Is
necessary, it could be designed and implemented at that time.
53
Site Considerations
The environment of a typical phosphating installation as
set up at Rock Island Arsenal, not unlike other process en-
vironments, is unsuitable from the point of view of process
control instrumentation. The temperature range would be in
excess of 100°F during the summer months and could probably
drop to 50°F - 60°F during the cold winter months. Also, the
area immediately surrounding the bath has a high humidity due
to evaporation loss from the baths operating at 200°F. The
ambient air could also contain corrosive vapors from phos-
phate baths, chromic acid baths, and pickling operations not
to mention sandblasting and degreasing operations occurring
nea rby.
Needless to say, the analytical and process instrumenta-
tion should be located in a more controlled environment if
optimum results are to be obtained. As instrumentation is
currently envisioned, it would occupy very nearly the same
amount of floor space as that currently provided for the
analysis of the bath solutions. Some additional space would
also be required for carboys containing the analytical reagents
and waste materials. Several limitations do arise from this
choice of location being approximately 75 feet from the baths.
However, neither seems to present a great difficulty.
The first problem caused by the distance between the
baths and analysis instruments is that of transporting the
sample. Several solutions to the problem exist. The first
solution is to run the sample lines to the control room in a
stainless steel tube covered with an insulating material such
as asbestos. This introduces a time delay equal to the length
of tubing divided by the flow velocity. However, this would
5k
not seem to be a problem since the composition of the bath
is estimated to change over a longer period of time. If
cooling of the sample during transportation presents a
problem, the transport lines could be steam or electrically
heated to maintain the temperature. The steam could be ob-
tained from tapping into existing facilities. If necessary,
return lines could also be heated in a similar fashion to
avoid any gradients in the bath. However, in view of the
relative quantities involved, this appears to be a very re-
mote pos s i b i1 i ty.
The second problem caused by distance is that of the
transmission of electrical signals. Typically, minicomputer
manuals indicate that peripheral devices should be located
within 25 or 50 feet of the computer. This restriction is
not applicable here since it applies only to pulse type
signals in which the rise time must be less than 100 nsec.
The signals which would be required to go from the computer
to the baths are d.c. levels or level changes which can have
rise times on the order of msec or more. This is permissible
since these transition times are small compared with the
time required to actuate solenoid valves.
55
X . Calibration and Maintenance
One of the main criteria for judging the performance of an automated
\ control system is reliability. It is extremely important that the system be
capable of operating for a reasonable period of time (week, month, etc.) with-
out operator intervention. It is difficult to predict a priori the mean time
between failure of various components. The best approach is to select the
most reliable components when constructing the system.
The routine calibration of the analytical instruments should be done auto-
matically and at specific intervals. Suitable calibration solutions could be
made which would be stable and yet simulate, as close as possible, the actual
phosphate solution conditions. These solutions would then be circulated
through the sample line and analyzed identically to the unknown phosphate
solutions. The results of the analysis would be used to update the normal data
analysis program. In addition, if the standardization data showed drastic
changes from previous results, the operator would be notified and/or an alarm
would be activated.
Routine maintenance for the automated system would not require signifi-
cant amounts of time or material. Typical maintenance tasks would, for
example, include: a) restocking the solutions used for the chemical analysis
and calibration; b) inspection of valves, pipes, and pumps for leakage and
corrosion; and c) lubrication of pumps, fans, etc. as necessary. Also, if
filters were required in the sampling and analysis system, they would also
have to be inspected and changed at regular intervals. The computer and
electronics will most likely be the most reliable components of the system.
Maintenance, other than periodic recalibrat ion, would not be required for
these components except in the case of failure.
56
References
1. Coslett, T. W., British Patent 8,667 (1906) and U. S. Patent 870,937
(1907).
2. Allen, W. H., (to Parker Rust Proof Company), U.S. Patents 1,167,966
(1916) and 1,291,352 (1919).
3. Tanner, R. R. and Darsey, V. M., (to Parker Rust Proof Company), U. S.
Patent 1,888,189 (1933).
4. Gilbert, L. 0., Rock Island Arsenal Report 56-2995 (1956).
5. Gilbert, L. 0., Rock Island Arsenal Report 57"2536 (1956).
6. Gilbert, L. 0., Rock Island Arsenal Report 54-2906 (1954).
7. ASM Handbook Committee, Metals Handbook, Vol . 2_, American Society for
Metals (Menlo Park, Ohio) 1964, pp. 531-547-
8. Orion Research, Inc. Catalog.
9- Personal communication, Orion Research, Inc.
10. Chamberlain, P. G. and Eisler, S., Rock Island Arsenal Report 53-638
(1953).
11. WiI lard, H. H. and Merritt, J., Ind. & Eng. Chem. Anal. Ed. J_4, 482
(1942). . .
12. Copper, M. D. and Winter, P. K., in 'Treatise on Analytical Chemistry",
edited by I. M. Kolthoff and P. J. Elving, Vol. 7, Pt. II, Interscience,
New York, 1961, p. 464.
13- Schwarzenbach, G., Complexometric Titrations, Interscience, New York,
1957, PP. 614-615-
14. Welcher, F. J., "The Analytcal Uses of Ethylenediaminetetraacetic Acid",
A. Van Nostrand, Princeton, New Jersey, No. 5, 1958.
15- Flaschka, H. and Puschel , R., Z. Anal. Chem. 143, 330 (1954).
16. Sandell, E. B. , Colorimetric Analysis, 3rd Ed., Interscience, New York,
1959, P- 605.
17- Ibid., p. 608.
18. Ibid., p. 612.
19. Kanzelmeyer, J. H., in "Treatise on Analytical Chemistry", edited by
57
I. M. Kolthoff and P. J. Elving, Interscience, New York, 1961, Vol. 3,
Pt. II, p. 156.
20. Woodson, A. L., et. al., Anal. Chem. 2k_, 1 198 (1952).
21. Swinehart, D. F., Anal. Chem. 23_, 380 (1951).
22. Kanzelmeyer, J. H., loc. cit., p. 125.
23. Krleza, F., Chem. Abstracts 6_7, 39860p.
2k. Flaschka, H. and Puschel , R., Z. Anal. Chem. ]k3_, 185 (1956).
25- Sandell, E. B., loc. cit., p. 9^5 and p. 9^7-
26. Adachi, T. , et. al., C.A. 6]_, 17582x (1967) •
27. Horiuchi, Y., et. al., C.A. 6^, 7367** (1968).
28. Lee, T. H., Adams, G. E., and Gaines, W. M., Computer Process Control,
J. Wiley 6 Son, New York, I968.
29. Elgerd, 0. I., Control Systems Theory, McGraw-Hill, New York, 1967•
30. Perry, R. H., Chilton, C. H., and Kirkpatrick, S. D., ed., Chemical
Engineer's Handbook, McGraw-Hill, New York, 1963-
58
DISTRIBUTION
Copies
Department of Defense
Defense Documentation Center ATTN: TIPDR Cameron Station Alexandria, Virginia 22314 12
B. Department of the Army
Commanding General U. S. Army Materiel Command ATTN: AMCPP-PI 1
AMCRD-R, Mr. H. Cohen 1 Washi ngton , D. C . 2031 5
Commanding General U. S. Army Tank-Automotive Command ATTN: AMSTA-RCM.l, Mr. Don Phelps 1
AMSTA-RCM.l, Mr. Edward Mortiz 1 Warren, Michigan 48090
Commanding General U. S. Army Weapons Command ATTN: AMSWE-PPR, Mr. F. X. Walter 1
AMSWE-PPR, Mr. G. Hall Rock Is 1 and , I 11 i noi s 61201
Commanding General U. S. Army Missile Command ATTN: AMSMI-IEL Redstone Arsenal, Alabama 35809 2
Commanding General U. S. Army Munitions Command ATTN: AMSMU-MT-MT Dover, New Jersey 07801 2
Commanding General U. S. Army Electronics Command ATTN: AMSEL-PP/P/IED 225 South 18th Street Philadelphia, Pennsylvania 19103 1
59
D ISTRI BUTION
Copies
Commanding General U. S. Army Electronics Command ATTN: AMSEL-RD-GS Fort Monmouth, New Jersey 07703
Commanding General U. S. Army Aviation Materiel Command ATTN: AMSAV-PRI
AMSAV-PPL P. 0. Box 209, Ma i n Off i ce St. Louis, Missouri 63166
Commanding General U. S. Army Mobility Equipment Command ATTN: AMSME-PLA ^300 Goodfellow Blvd. St. Louis, Missouri 63120 :
Commanding Officer U. S. Army Materials and Mechanics Research Center ATTN: AMXMR-TX Watertown, Massachusetts 02172 :
Commanding Officer U. S. Army Aeronautical Depot Maintenance Center ATTN: SAVAE-EFT Corpus Christi, Texas 78^19
Project Manager, General Purpose Vehicles Michigan Army Missile Plant ATTN: AMCPM-GPV-QV Warren, Michigan A8090
Commanding Officer Watervliet Arsenal ATTN: SWEWV-PPP-WP
SWEWV-RDT, Library Watervliet, New York 12189
60
D ISTRIBUTION
Copies
Commanding Officer Rock Island Arsenal
ATTN: SWERI-PPE-531 1 Rock I s1 and , 111i noi s 61201
D i rector U. S. Army Production Equipment Agency ATTN: AMXPE-MT, Mr. H. Holmes Rock Island, Illinois 61201
Commanding Officer Savanna Army Depot ATTN: AMXSV-EN AMC Ammunition Center Savanna, Illinois 6107A
Commanding Officer Fort Detrick ATTN: SMUFD, Mr. H. H. Meier Frederick, Maryland 21701
61
unclass ITiea Secu rit£ Classification
DOCUMENT CONTROL DATA -R&D (Security claaalllcatlon ol till: body ot abatract and indaxlng annotation mutt b* antarad whan tha overall rmport It claaallladj
I. ORIGINATING ACTIVITY (Corporatm author)
North American Rockwell Science Center 1049 Camlno Dos Rlos Thousand Oaks, California 91360
2*. REPORT SECURITY CLASSIFICATION
UnclassIfi ed lb. CROUP
a. REPORT TITLE
A FEASIBILITY STUDY FOR THE CONTINUOUS AUTOMATED ANALYSIS AND CONTROL OF PHOSPHATIZING BATHS (U)
4. DESCRIPTIVE NOTES (Typa ol raport and Inelualwa daft)
o AUTHORISI (Flrat nama, middla Initial, laat nama)
Edward P. Parry R. Lee Myers
B REPORT DATE 7a. TOTAL NO. OF PACES
62 7b. NO. OF REFS
30 Sa. CONTRACT OR GRANT NO.
DAAF01-72-C-0130 b. PROJEC T NO.
Pron Al-1-5005^-(01)-Ml-Ml e.
AMS Code 4497.06.7037 d.
Sa. ORIGINATOR'S REPORT NUMTBER(S)
SC523.1ER
Bb. OTHER REPORT NO(S) (Any othar numbara that may ba atalgnad thla raport)
SWERR-TR-72-11 10. DISTRIBUTION STATEMENT
Approved for public release, distribution unlimited.
II. SUPPLEMENTARY NOTES 12. SPONSORING MILITARY ACTIVITY
U. S. Army Weapons Command Res., Dev. and Eng. Directorate Rock Island, I 11 inois 61201
13. ABSTRACT
The feasibility of automating production phosphate coating process baths has been studied under the guidance of the Research Directorate of the Weapons Laboratory at Rock Island. The study comprised a de- tailed examination of the analysis and control requirements. Con- tinuous automated analysis and control of phosphate baths are feasible. Analytical methods, sampling system procedures, and control logic techniques are presented for incorporation into a prototype system. (U) (Parry, E.P. and Myers, R. L.)
1
DD rout I MOV •• 1473 Unclassified
Security Classification
Unclass1 fled Security Classification
14. KEY WORDS
LINK A LINK C
1 . Phosphat ing
2. Automated Analysis
3. Automated Control
k. Phosphatizing Baths
Unclassified
Security Classification
— — -o •D —
10 tn • O E >~ E t_ O — 0 *-» *-" <0 *-" c 3 C 3 O < < < <J
*> J)
X «J o c — _- CJ
H HO ID «-» O, A XI - o e
O O. O
X) —
E > O —
a c 3 o x •> < < < <J a. a
O o x ac
O- o — O V ^ (Q T) i/i co QC u c o — V 0) V TD o u *-/ -O iA >• O X X) c <_> « u V i- —
41 Z3 a. C V • 1 TJ 3 LA ac O to Z m O.C4 O T3 « *->
m — u_ u Wh.U La o x: « (0 V i_ H o >-co *0 1 o La -C La T) L
Q z W — V» Q. a iA 4J Q V M o -J u — -o X « -C XI to > u i_> o f0 — NU-< L> iy> J-J VI- f0 o 41 — ex: Q- 4) r-< C ID O —I TJ
*- o UJ t- X < - •o •— Xt l> TJ — <s I z -* JJQ- ft) *-» «J t/i tf» -—«. c o — H o a. o * 4) X — nj </t O c — •» -0 t_) o Ll_ U 1 Its E t/l c O La t-l
6 • a£ -o aC c o ^ — O 0> V Q. Q. c e w o Q u i_ - a r irt *-J u T3 « E o O c w» u. z t> c o — i- i m 3 o — V O <_) uui- < S <0 n- — *•» >-N flj TI l_ 3 3 O —
o >- XI u ~ 1 c — — a. cn • TJ c o CO UJ — — NOO o Ha> V >- c c — => — t_ T> r» *J*— c 0 C7» V x: •o
0 «• — t- co >- •J (_) i "—* i ^ c x: -J 3
a — co >- -O E ac -a- >- aaa w +J
m • — —i < - I- -u-\ w *-* **- in
• > >- < • IA 1 (A O *—* — <u ha o 3* • - t- z CO x J. o£ 4) O r^. — o V 4)
a TJ — < X 4-> •S #jf — lA *»•» — o -o « x: >• c -i 1- b O UJ JD 1 o -Q c *-• t- e • • — a < 0 :* • — r^. •— 4) 3 (O
u x — aj LU at z X) «A U 1 ifl 4J v_ < U » — t- C — vl> <D TI "O o
i- — t/» < o - <D *J • < O •J JZ V *J TJ
• <u < X z tA .A 4- lA ••» a. — o C
wi 1) Ji LU o L. 3 O 3 C P+ lA TJ « 1J
VI o u_ H »^i €> o Q.— o <n V o 3 L_ — • t> O :D — >• X v — L. -T -C -C 4-> — IA
Z3 aC aC < < (- x 1- QC — tL. -T h- Q. ^ Q —
SO u o x ^ c 3 a. c v z *g — — u. o
O « fl lit) — v • u •J Xt i* v. —
f^ r- o i-
o z
i- t- O >-i/i -x> i O L X l_ XI La
O z i- — vi a Q. lA *-• o V
o —1 L. — -o x •) J= XI <• > c o
u (_> o t> .— Nb.< k- o» *J «*- TJ o V — ac a_ V r-< c « o -1 T>
»- o UJ t- J < *-o — JQ « "~ •o — IN* X z -X. jio- « w 1) (A lA -~*
u
c a — h- o Q- o <D tt X — (O lA O c — « \D (_> o — U. *J 1 **- E W) C O t- u E a oc T) oc C (J — — O U iq a. a. c E CT o Q La La •» • X «! M U x) <• E o
< o C i« u- z •0 c o •—»-»«» 3 O — w o CJ
o UJ ~ < X * •4- — *J —. T) "• i. 3 2 a ••*
O >- •o u — 1 c — — a at vn •D C Q l^ UJ — — (N O O O >*- « >•
c C — ^ — L rt> r«- o-^-* c o 9 V £ T3
o « — H CO >• t) o 1 —' 1 3 C JC W 3
a. _ Lrt >- X) E an ^r >• — *-* *J
T> . ^ _l < ai L- - IA w w «*- lA
V > >- < a. (A I * o ^* — * k. o 3 « " t- z CO x: -l£ oc v o r- — o « •
Q -a < X *J •J ac — tn CA a -o t» JC ^ C _( H i_ o UJ XI t o XI c *J 1-
E - ra — Q < o j» • — r*- — V 3 *0
i_ x: — CD LU CO z n W» «J 1 lA *J l_
< u * — t- c — >i> 1) <B "D o • L. — CO < o a. T> *•• • -t o 1) X « *-> T3 (fl < X z <A tft 1- lA »*- a — a c *rt V -* UJ o — k 3 O 3 c r*. M TD v <• tfl u Li. 1- ISJ V o Ov— O <T* eu 0 3 i. an Q ftt 0 r> — >•£ » — L--» X X *J >— tA < =) ac ac < < h- X *- OC — Q. -4- H a. >A Q —
u. «J a w
£ a M a o Q. E >- E i- « 0 — 0 <J o k. a « c X 3 C SO «- < < < o
z a o c — — CJ i- a a •a JD -OB
_ X) X) BBB
«J « <A • a->
m *j a» *J — •> x; * IA u O X Q. E >» E u Q.
*A O — 0 u •A
O *J « *-• c o X 3 C 3 O X a. < < < W a. a a.
o
ha .* « o • *-* o 1 • c •> X oc < -J «J • O u X O — f*> C C L- *J a. o — O « « * XJ
UJ CO OC •) c o — « • 4J ZD o U — 1 •J XI tA >-
o X T) C <-> « u • *- — 0 z> a. C • • 1 -o • 3 tA QC O m «-• z •3 — O.CM O w -a •» *J *-» * — LU o MNO L O X • <D o
0 x
La h- O XtA) *0 1 o i> X L. x> ha
o z ha -MIL a. v* w o tl t- o -I L. — -O X • JC XI « >
c O
u t_> o « •— Ca5 U_ < W Ol *J *4- « o • — ac Q. 1) r-< c « o -J XI i. o UJ t- X < --o — JO «
• x> — faj X z • J* J3 O — • «-• • <A tA .-^. iA c Q — h- o O. U • V X — *0 lA u c — •
1 «A c_> o — U. Ja> 1 H- E M c 0 w *J
a: -Doe c o — — O • «J Q- OL c E W o o L. l_ - * X « « U "O • E 0
< 0 C « u. z • CO — 1- 1 M 3 0 — • O o o UJ — •< X *«*- — *J^^ aj • i- 3 9 y *—»
0 >- tj o — 1 C — — a oi »A X) c a CO m • — C4 O O U <*- « >- C c — => — t- «J r*»«j>*-* c o m m x XI
o * — L- V> X « o 1 -— 1 3 c X «-» 3
a CA >- xi E oc -ar >*— *J •J •) • — _J < • h- »m *J> *J H- n 4) > k- < • w 1 tA O —- — • •- O
3k • * t- z <Sl X -J< tt «ON — o • V a -o — < X ** «J or — m f\ - UT) «l X >* c _i h- u o •IJ) l o XI C *J t-
E • « — a < 0 * •» — r- — • 3 « I. X — a. UJ •» Z T» M «J 1 tA *J L-
< u * aaa •- c pa.«jBj AJ * X) O a. — w> < 19 • * *J • < o • X • ** XI AS < X X « (A a- M •*- a.— u c
Ml V J* UJ o — v. 3 O 3 c r^ m -o v <• <
<A U u. 1- M • O Q.— O <n O O 3 a_ — V o ZJ — >*X *) — »- -a- X X ** — |A
3 oc ac < < 1- X 1- •C — o. ^> 1- Q- lA Q —
3 O O X '
— It. o
z. aat o -» u ^
c O
a (_> o * — f) — yc tV « w o LU t- i
•A -O — M X X • JC
c a — >- o a- u «J
o * vX» c_> O — • • OC XI OC c
• O* o o U L.
< 0 C tA u. z «J C 0 u Ul- < » •*•
0 >- •o o — « TJ C a tA UJ — —
c c — 3 — t- -1
o • — h- W» >> • o a. — «/» >- xt E •J • — ••1 < • • > >• < tA
31 « - t- X M£j| O X) — < X w -»
>• c —I H »- O • - -1 - o < o -- x — m UJ •3 Z X)
< a •» — >- c •> ^ u* < (3 - •
• m < x Z W tA W» • JkT UJ o — U 3
•A U u. t- N • O
*r • « o =3 — >-X
C C L. w 0 *> m m -a
c <• O -I xi
— A"* a — c o
— I *J Xi tA >» — (J V U « i- —
• in- 3 tA of O <• O.C4 O *-• XI f) *J *J ror^ut- o x • »J • sj> I O k. X *- XJ
— co a. a. M *J o -O X • X XI -I
(MU-<L. Ot*-*H-«| O
u- *-» i t*~
— — CM O O U »*-
OC — lA #*\
< < t- X I- •>—!--*
•9 « u c — F (A c 0 t-
O • * a. a. *J O TJ -a E 1 0 — • 0 <• L. 3
a. a* 3J O
**- « >^ o ot • X TJ
C X " 3
V — *J *J
*J *J t*- M — *• -- o — o • • — O TJ • X X) c ** h- — « 3 • M * 19 TJ o • « X • *-• TJ
Ifea a.— u c 1A XI •> <• • O 3 k. •*•
X X *-" ^ tA
i~ a. «A o —
ft > m • m a c x r> m r* m 3" — O 9 a -< a —
C O ft rt • ft O. — » — o
to 3 • 9 3 -i » » o • ft — -*
3"-< -Q O •* ft -1 Q- — 3" o w w a w * • ft ft 3 ft 3 3 rt ft a. 01 ft a — O.XI O •<
— O ft -* — 3 — o a rt « -I €J -l
O ft — m — a 3 -< O. o •« o O <-*-** O -i « o -o a x* 3 O 3" ft -l XI O 1 ft 1 ft O r. O TJ — — O 3" O ft ft 1 3 O. rt ft
C ft J3 — -I C 3 ft XT — ft «i ft "I O - rt «
7 a ft ft ft ft 3 3
T> Q. ft ft n I *• O r> (i • r* o Q 3 -* ft ft ft O
•< -t ft o X» O ft 3 ft —— ft
O* — ft) — — 3 -c O ft c M tO - O ~ — C ft ft ft
rt >k It ft 3 C X O ft rt ft 3" — 0 3 3 *< 3 — — ft ft 3 A — rt ft c n ft rt • ft Q. — 4* — O
ft 3 ft 3 3
-i O ft O ft rt — -«»
3"-< •O O ft r» 1 Ov — 3" n v> ft » «• - • ft ft 3 W) 3 13 rt ft a. ft ft 3 — a. xi o -< — O ft -*• — 3 — O 3 rt i/i -i c -1
O ft — ft — 3 3 *< O.
O ft o O ft -* o 1 ft o x* a x* 3 o 3" ft -» XI O -I ft -1 » O ft OT) - — o 3"
O ft ft ~% 3 CL rt ft
c ft X> — T c 3 ft ft- — r» «* ft -1 O - rt ft
3- a ft ft V ft 3 3
X* Q. ft rt i ^ «» o n ft - ft O O 3 -* ft rt ft o "< 1 ft O Xi o ft 3 • — — rt
o- — M — — 3 << o ft C
rt > ft ft ft 3 C X O ft rt ft 3" — O 3 3 *< 3 — — rt ft 3
-O — rt ft C O ft rt ft ft CL — ft — O
ft 3 ft 3 3 -1 ft ft 0 ft ft — -h
3" •< •D O V rt -» Q. — 3" ft </> )/» ft (/> - rt ft ft 3 V» 3 3 rt ft Q. ft ft 3 — O.TJ O *< — O V) -ti — 3 — O 3 rt V) -i (O T
O ft — ft — 3 3 •< O.
O tfl o 0 rt -•t O 1 ft o -0 3-0 3 o 3" rt 1 TJ O -t ft -1 wi O r* O TJ — — n 3"
o » ft "1 3 O. rt ft
C ft -Q — T c 3 ft cr — rt wi ft -i O - rt ft
3- 3 ft ft Wt ft 3 3
XI O. ft rt "1 -1 « o o ft • rt O O 3 -*» rt rt ft o •< I ft O XI O M 3 ft — — rt
o- — w — — 3 *< O ft C Wi ID • o rt —• c
ft 3 C X n ft rt ft 3- —o a a -< a — — r* ft 3 X! — rt ft C O ft rt ft ft ft. — ft — o
ft 3 ft a 3 -i ft ft o ft rt — -h
3"«< X* O ft rt -i a. — 3- ft w» M 0
3 W« 3 3 rt ft CL ft ft a — O.TJ O •<
— O w» -* — 3 — O 3 rt *»
O ft — W — 3 3 -< O. o •» o o t-nn -i • o XI 3 XI 3 0 3" ft 1 T) O I ft T ft O ft O XI — — O 3* O ft ft I 3 O. ft ft
r ft x. — T C 3 ft (T — rt W ft -1 O - r» ft
3- a • * «> <i
3 3 XI O. ft ft T -I ft O O ft • ft O O 3 -* rt rr ft O *< -% ft O XI O ft 3 ft —— ft
O* — «/> — —• 3 -< O ft c ft IB • O ft — c ft O ft a
— — -a
E >- E >- o. w o — O *•» «• -C 4J (0 *-» c O *-• 3 C 3 O X 1) < < < (_> O- CQ
o c — — c-> t- n» a u *•» a ao x> — o e ce. o
o
— — TJ « * TJ —
— rg *-> -
a. IA <D O
E >• E »- O — O ** *-> « *J c O 4-. 3 C 3 O x <tt < < < o a. ao
o c c_» t- «• O => ** Q
— o e ec O
0
i_ J< 0) O
41 *-» o 1 © C JZ QC < —I 1> • O u x i_) — ro C c 1^ 4J
Q- o — O 4) <9 TJ TJ
CO cn at a> c o — 4> 4) 4) z> O u — 1 J_> .a i/i >- — o x TJ c t_> 4) u 4) L. —
• =D a. c 4> • 1 T3 3 tn ce O *D w z <n — Q-csi o 4V "D t> *J 4-»
m — U- o <*\ r^ o 1_ O -C 4) « 4)
o Z
u i- O >-to v£> 1 o l_ Jt L. -a L.
Q a: u — v> Q. a i/» 4J Q 4) 4-J o _J u 1— - O X D -C JQ ID >
C
O
O <_> O • CNI U_ < CT> "*- <n O © i— a: a. 0! r^< C tz a _* T3 u. o Ui 1- J < - -o — XI 4)
UI
UI
V
<J
U <
TJ — <N X z -* -• a — 4) w *J ^ i/i ^-^ C o — t- o Q- u (0 V r — aj Ifl u c — m *X> l_> o U. w i >4- E Ul c o L,
e cc TJ cC c u — — O 4) TJ a. Q- c £ CTt o o U l_ - 0J a: (A 4-J U T3 flj E o o c ui U. z T> c o •— t_ i l/> D o __ 4> Q o
t_> UJ — < I fl> «* — 4-> -—. (0 m L. 3 ^ u —• o >- TJ o — 1 c — — Q- CT>
lA T3 c Q CO UJ — — (MOO o ^4- V >- c c — ZJ — t- <B r^ <_> •—• B o CT» 4) -C "O
0 IB — h- CO >> V o i *—' i r> C x: 4-J 3
Q. _ to >- X) E ce j- >- — 4-1 u iB _J < »- - LT\ *-> •4-
V > V < - lA 1 l/t O ^—% — (0 k_ o Z* © - I- z CO x _* ec t) o r*- — o V 4)
Q TJ — < X w 0 QC — u-\ r<^ — o T3 1) X >• e —i K~ i_ o UJ JD 1 o J3 C 4-J t- E - i) — o «t O 2 <0 — IS. — 4) 3 t) u X — 03 UJ eo z T3 to w I i/1 4-1 U- < u M — t- c i— vO It • TB O
u — CO < o - T) *-* < o 4» -C 4) 4J "O <B ** £ z UI irt L. (/I Q. u c
to © Jf UJ o — 1- 3 O 3 c r^ ui T3 V T3
< lA U u_ 1- M V o a~ O <T* • O D l_ — © o u — >-C D — L- 4P SZ x: w — IA
r> CC CC < < t~ K H- CC — Q. -T t- CL. i/> Q —
\- -^ 4> o
4) •J o 1 4> C X CC < —* 4) o u X O — f*> C c lh 4->
a. CJ — 0 4) <z T) "O LO cn tt 4> c o — 41 4) a =J o a — t u JB V* >• — o X T3 c (_> 4) u 41 i_ —
0) IJ O- C 4) 1 T> 3 <st CC o <D M z: <0 — 0.04 O w tJ « t-l W
T> — u. O c-v r^. O \- O X • <« 4)
0 z
u i- o >.co vj3 1 o c_ X 1- -o L o z l- — to a. a. m *j o V 4-> o —1 1- — - o J: w X Xt m > o *-> o rn rsj U- < en 4-J fl o c
o 13 — CC a. 4) 1^ < c < 0 —I T> 1- o UJ t- 5 < • T3 — -Q •
Ul T) — t^J X z .* -Q Q — 4) *-* 4) in *A •"•
C Q ~ K o o- O •a 4) XI — <n ifl O C — 41
u o <
* ^o <_> O — U- 4-> i u_ E ^ c o la 4->
E a: T> CC c o — — O 4) aj Q. Q. c E cn o o u l_ - n r Ifl 4-J O T3 • E o O c lA u. z « c o — i_ i l/> 3 O ~ 4) O (_>
i_> UJ — < X f0 14- .— *J -^« aj <9 u 3 :* u — o >- X) o — 1 c — — Q. a»
m "O c Q LO UJ — — 04 0 o u >*- 4) >- c c — =3 — t_ nj r^ o *—' c o CTl 4) X •o
O <Tt — l~ CO >• D <-J i •—- i =3 C X 4J 3
a _ <SI >- -a E QC J- >. — u W <T> — _J < - t- » LT\ w w **- i«
4) > >- < * t/i 1 i^ o —. — rt) u o 3 4) » l~ z CO X Jri CC 4) o r*» — O V V
Q -o — < X *J *• cc — LT\ r^ — o -o 4) X >- c ~J t- i_ o UJ O t O J3 c 4-J t-
E - • — Q < o ^ n — r-- — V 3 IB
i- X — CO UJ 0Q z T3 CO - • 1 ^ 4-J b < o 1/1 — 1- C — SJ3 0) tl "O o L. _ Wl < O - 11 t-l < o 4) X 4) 4-1 T3 « < x: z irt M l_ i/^ u- Q- — o c
t/» « ^ UJ o D o 3 C r-. U1 •o 4) «0
< rf> o u. 1- Ki 1) o a. — 0 O^ 4) O 3 L ~ t> o =3 — >-x a> — L. -* X X 4-> — w
=3 CC a: < < i~ X t- QC _ Q- --T H a ^ o —
E >- O —
•BOX c w a. *A
3 0
a x» - o •
a a o
— _ -o T» — U. 4-i ci tfl • «-> — * 4-* ~ *- — « M x * * iB O X (A a g >• G L a.vi < M O — O w «*« X -I O 44 « W C O ~ (J X J C 3 O X «» z a.<<<(_ia.m
- o
•- M l*\
3 5?. * O Ui —
O N TJ e c c *- o • —
a *o —
< u « — I
O X T3 C J Q- C 1)
I- o >«u» Z U O -I U — O O ID —
CC Ok u UJ t- J X 3= • .* h- O Ok U •
o o — ac T3 cc c O Q U I- u. z •« c o
< 3t <•«*- >- -o o —
» >* « o - x> E I < * : - « IV) £J( : x *-< *
»- »- o > < o l » X Tl
. — co < cjj < X UI o U. I-
— I. 3 rsj « o — >*x:
• o — r*\ o — c o
— I <-> « • I "O *
Q.CM O «-• r»\ r«» o i- *o I o
— to Ov -O X •
CM u> < t-
U- 4-1 I 14- o — —
*ffZ N
I c — — «M O O O
I i/> o —. Ct «ON a: - lAfA UJ xt . O
tO *J | — ^O
*J . < o i- M O 3 c r-» ex— o a\ 1) — L. J- •C — O- -*•
c c w. o v m — « • *-• xt n u • 3 ««
•o • Ox©
<D T3 4)
fl * U E «i C o • *v *-• U "^ 3 0 — m >- 3
o a« C X
>.— 4J 4J 4J — * »- — o •
i- -o I o
xi m • m i -J -o •
c — o »- a. a. • E a o
4J 3
O 4)
V X
O • m « T> O X ©
u- a.— iA -O
m o 3
I- Q- 1A Q —
ha -^ • O V 4-1 O
i « c - X ac < -J • • o u X l_> — **» C c L. 4-J
a- o — o © >9 IB TJ CO to ac V c o — © V « z> o u — 1 4-1 XI H >- — O X T3 c <_> 0 u 4) L- —
19 =) a. C « • 1 TJ 3 «* ac o |Q 4J 3S m — O.CM O 4-> T> * 4-1 4-t
m — U- o r^ r»» (_> u 0 X 0 <B 0
o z
i_ t~ O >-to •Xt 1 o \- X L TJ L o Z u — to Q. Q. rfl *j O 0 *-• o _j b ^ -O X 0 X X- * >
c o
o (~> o « — NlL< b a> 4-> Vt^ IB O 4) — tc a. © l*» < c 0 o -J TJ
»- O UJ \~ X < * T) — XI 0
V o u <
TJ — «M X z jjf x» a — © •-* V •A <A ^-> C O — H- o Ok U 0 0 X — <B lA •J C — • • M> <-> O — u w t u- E lA c O U 4-»
E CC •a QC C O — — O 0 0 a. a c E t> o Q b, l_ - 0 X «A «-> O TJ 0 E O O C <A u. z * c o — »- 1 lA 3 O — 0 0 o o Ui — < I <• >*- — 4-1 -—^ « « c 3 ^ u "—
o >~ •o o — I c — — Q. o» M T» C a to UJ — — CM O O u
l*- o >> C c — => — fc- 0 r- o*-' c 0 o» 0 X TJ
o « — »~ irt >. 0 o i •—' i _D c X 4-» 3
a. — to >~ Xt E ac -r >• — 4-> M
m • » -1 < - t- -u> w 4-1 u- ui « > >- < - <A 1 «» o ^—. v— •B u O 3 t> - K z to X .* ac v o r-« — O 0 0
Q TI — < z W 0 CC — LTV r^v _ U TJ 0 X
x C -J K >- o ui xi i o xi C 4-t »- E * « — o < 0 3 0 — p>* — V 3 0 i- JE — m UJ n z -o tO 4-1 1 4J i. < u « — 1- p — vO <0 iB TJ o
l_ — c« < o - m *-» • < o 0 X 0 4-» TJ
m < XL z <r\ *s\ U n *4- CL — o C »/» 0 Ji Ui o — u 3 0 3 C f»» kA TJ V IB
< tf» o u. 1- >SJ « o a— o <T\ c 0 3 c — 0 O =D >*X 0 — U -3" X X 4J m
x> QC a: < < 1- X t- ac ~ o. -T t- a. lA a —
f* > ft • ft 3 C X f> ft r» ft 3" —O 3 a «< a — — r* ft 3 £ — r* m C O a rt • ft Q. — ft — O
• 3 • S 3 -i ft ft o A i-t —• -*.
3"-< •O O «» if -1 O. — 3* ft Ml M A
3 M 3 3 r* ft Q. 01 am — Ok XI o •<
-o w •* — 3 — O 3 r* « -1 «1 -1
O ft — ft — 3 3 *< O. O to O o « -* n -ift o TJ 3 TJ => O 3* r* -l T> O 1 ft -1 V O r* O "D — — O 3* O • O 1 3 ft. r» 9
c a x> — -i c 3 ft D- — i-* *" ft) -» O - r» ft
3- 3 ft ft w a
3 3 X) (X ft) .-t -I 1ft O O • • * o 0 3-* r» 1+ ft o -< -i ft o •O O ft 3 ft — — r*
O" — ft — — 3 «< O • C ft to • o »* — c ft o *• •
rt > ft ft ft 3 C X O ft r» ft 3" — O 3 3 •< 3 — — r* ft 3 A — r* ft C f) a r+ ft ft O. — ft — o
ft 3 • 3 3 -i a ft o ft i* — -*
3"-< -D O M r* n o. — 3- ft ft ft ft ft - a o> ft 3 l/i 3 3 -"• ft O. ft a 3 — Q.-D O *<
— O w» -fc — 3 — O 3 «•» «t -» «a -»
o ft — ft — 3 3 >< Q- O (A O O r» •* O -« a o
T» 3 T3 3 O 3- r» IT) O 1 ft -I ft O ~ o XJ — — O 3- o a » -i 3 Q. r» a
c a x. — T C 3 a o- — r* ft ft *» O - r+ m
TJ Q. ft It I "I «i o o a • I* o O 3 -* r* r* a O •< i • O T> O ft 3 • — — «•
O" — ft — — 3 << O ft C ft 10 • o
rr > ft a a 3 c x O ft r* ft 3" — O 3 3 *< 3 — —• it ft 3 -O — rt at C O a r» a ft Q. — ft — o
ft 3 ft 3 3 -i a ft o ft rf h
3"-< •^ O «» rr -1 Q. — 3" a v> w» a w - a ft ft 3 U» 3 3 r* ft D. ft a a — Q-TJ o <
— O w -«t — 3 — O 3 « «* -l 10 -»
O ft — ft — 3 3 •< O- O «• O O rf -*t O -i a o
"O 3 TJ 3
ft 1 ft O rt O *0 — — O 3* o a a* -i 3 Q. r» ft
c a -o — -I c 3 a o- — r* M ft "i o- rta
3- 3 ft ft ft a
3 3 TJ 0. ft <-r
O O ft • r» O O 3 -h ftftd o << i ft O TJ O ft 3 ft — — r*
tr — ft — — 3 -< o a c ft to • o f» — c an w 3
— r» ft 3 -Q — <-r » C O (» n ft ft o. — ft — o
ft 3 ft 3 3 -I ft ft o ft r* — -»
3-*< T3 O ft r» -» a. — 3- a *• <• • w - ft ft ft 3 W 3 3 r» ft O. ft a 3 — Q.TJ O »<
— o *> -•> — 3 — O 3 r* V» -I « -»
O ft
n ft o O r» -»» O -t ft O
"O 3 TJ 3 O 3" r» -i TJ O 1 ft -t ft O r» O tJ —' — O 3" O • ft I 3 O. r* ft
c a xi — -1 c 3 • cr — rt ft ft T O • rr ft
3- a » * ft a
3 3 TO O. ft r» n -» ft O o • • i* o O 3 -»• rr r* ft O -c -» ft o T> O ft 3
ft — — 3 •< O ft c ft M • O r» — C ft O ft 3