Thermometric titration determination of hydroxide and alumina in

cially designed operational amplifier circuit which by gen- erating the first derivative of the thermistor signal, de- tects the end point and stops t...
1 downloads 0 Views 501KB Size
Thermometric Titration Determination of Hydroxide and Alumina in Bayer Process Solutions Eric VanDalen and L. G. Wardl Research Centre, Aluminum Company of Canada, Limited, Arvida, Quebec, Canada

The apparatus determines from 10 to 600 grams/l. hydroxide and from 3 to 120 grams/l. alumina in Bayer Process streams. Essential components include a constant rate buret, a glass enclosed thermistor, and a specially designed operational amplifier circuit which by generating the first derivative' of the thermistor signal, detects the end point and stops the titration. The sample is diluted with sodium tartrate and titrated with 1 . 3 N HCI in a thermos jar. The first titration volume is equivalent to hydroxide; after the addition of potassium fluoride, the second titration volume is equivalent to alumina. The complete titration takes 4 minutes. Suspended solids do not interfere, neither do other constituents in Bayer liquors. The relative standard deviation is 0 . 3 2 % for hydroxide at the 180 grams/l. level, and 0.17% for alumina at the 120 grams/l. level.

Good analytical values for hydroxide and alumina are needed to efficiently operate the Bayer Process, in which hot sodium hydroxide is used to dissolve the alumina from bauxite to give a concentrated solution of alumina. Impurities such as iron oxide, titania, and sand are left behind as insolubles, commonly referred to as red mud. After filtering off the red mud, the dissolved alumina is recovered from the filtrate by allowing it to cool, and then seeding it with aluminum hydroxide. About one half of the alumina precipitates out by this procedure. The Bayer Process control strategy involves constant measurement of the concentrations of hydroxide, which in North America is traditionally ( I ) expressed as grams/l. NaZC03, and of alumina, which is expressed as grams/l. A1203. In some cases the ratio of the alumina to hydroxide concentrations is used for process control. This ratio is related to the degree of saturation of the solution; it has values of less than 0.5 for saturated solutions, and values up to 0.7 for supersaturated solutions, which are fed to the precipitators. Thermometric titrations have been used to determine caustic ( 2 ) and alumina (3) in 2- or 3-component solutions. These methods are not well suited for routine, process control analyses, since the titration must be continued beyond the equivalence point, and a strip chart recording of the titration curve must be inspected to locate the end point. Zenchelsky and Segatto ( 4 ) were the first to point out that thermometric titration curves can be differentiated. They demonstrated this by mechanical amplification of the bridge signal followed by differentiation of the amplified signal in a resistance-capacitance network. Later, a much simpler operational amplifier circuit ( 5 ) was used to generate the first and second derivatives 1 Present address, Aluminum Company of Canada, Limited, Kitimat, British Columbia, Canada.

(1) H . L. Watts and D. W . Utley, Anal. Chem.. 25, 864 (1953). (2) H . W . Linde, L. B. Rogers, and D. N . Hume, Anal. Chem., 25, 404 (1953). (3) W . L. Everson, A n a / . Chem.. 43, 201 (1971). (4) S. T. Zenchelsky and P. R. Segatto, Anal. Chem.. 29, 1856 (1957), (5) M . Gaze, C. H. P y u n . and S-0 Kim, Daehan Hwahak Howiee. 1 4 , 341 (1970).

2248

of the titration curve. Differentiation accentuated the end points, making them easier to locate, but the complete titration curve had to be recorded, which was a disadvantage for routine industrial analyses. The first fully automatic thermometric titrator was developed by Priestley (6). It gave the volume of titrant in digital form, and used a resistance capacitance network to generate the derivative of the titration curve. Several other titration methods have been used for the determination of hydroxide and alumina in the solutions and slurries resulting from the Bayer Process. Watts and Utley ( I ) proposed an acidimetric titration to a phenolphthalein end point, using sodium tartrate and potassium fluoride as complexing agents for the alumina. The method required the prior removal of solids and carbonate, was lengthy, and gave unsatisfactory accuracy and precision. Others have used dihydroxytartaric acid-2,4-dinitrophenyloxazone (Alkalone) and its analogs as an indicator in the photometric titration (7), and as a reagent for the spectrometric determination (8) for hydroxide and alumina using Technicon equipment. These procedures gave varying success. Potentiometric titrations with acid (9) using the glass electrode have been used, but the accuracy was not satisfactory. The thermometric titration procedure proposed in this paper is automatic, the buret is turned off a t the end point, and the volume of titrant is read directly from the buret volume center. Two components, hydroxide and alumina, are determined successively in the same sample aliquot with the same apparatus. Titration times range between 3 and 4 minutes. The procedure can analyze unfiltered slurries, and highly colored solutions. It is not subject to interference by carbonate and any of the other constituents usually found in Bayer solutions. The precision and accuracy are two to three times better than those of the Watts-Utley method.

EXPERIMENTAL Apparatus. The titration apparatus includes a differentiatingcontrol circuit module, a thermistor, a magnetic stirrer, and an inexpensive thermos jar titrating cell. This equipment is now available commercially from Sanda Incorporated, Gypsy and School House Lanes, Philadelphia, Pa., 19144, under the name Thermo Titrator-Alcan Model. The titrant is delivered by a Sargent-Welch, Model C Automatic Constant Rate Burette equipped with digital volume readout. The buret delivers 1.31V HC1 at a rate of 5 ml/min into the thermos jar. Because of the high rate of titrant addition and the large heats of neutralization, no other insulation, equipment for temperature compensation, or thermostating is necessary. The rate of temperature rise in the titration solution is sensed by a glass enclosed thermistor bead, which has 2000-ohm resistance at 20 " C . Since it is used to detect changes in the rate of temperature rise during titration, and not absolute temperatures, it need not be calibrated or have long term reproducibility. In the differentiating-control circuit, the thermistor is connected to the Wheatstone bridge. An operational amplifier am(6) (7) (8) (9)

P. T. Priestley,Analyst i i o n d o n l . 88, 194 (1963). H . Bensch. Aluminium. 43, 360 (1967). R . S. Danchik and R. T. Oliver, Anal. Chem.. 42, 798 (1970) H . L. Watts and D. W . Utley, Anal. Chem.. 28, 1731 (1956).

ANALYTICAL CHEMISTRY, VOL. 45, NO. 13, NOVEMBER 1973

plifies the bridge signal. A second operational amplifier differentiates the amplified bridge signal, generating the first derivative with respect to time. Since a constant rate buret is used, this signal is also the first derivative with respect to volume of titrant added. At the end point, the differentiated signal drops by several volts (minimum 3-4 volts). The magnitude of the differentiated signal is constantly monitored by a comparator circuit. When the differentiated signal falls below 2 volts, the comparator circuit opens a relay which turns off the power to the buret. Volumes of acid titrated are read directly from the volume counter of the buret. Reagents. The 1.3N HC1 is made by appropriate dilution of the concentrated analytical grade reagent. The sodium tartrate solution contains 250 grams of Na2C4H406'2H~O per liter. The potassium fluoride solution is made by dissolving 1000 grams of KF.2HzO in 500 rnl of water, heating if necessary; then cooling and making up to 1000 ml. The K F solution must be just pink to phenophthalein; this is achieved by adding concentrated HC1 or NaOH. For calibration, sodium aluminate solutions of the desired concentration are prepared from carbonate free standard sodium hydroxide solution, into which high purity aluminum metal and sodium carbonate are dissolved. A typical example would be as follows: A commercially sold concentrated standard solution of carbonate-free S a O H , is diluted to give 200 ml of 2N NaOH. Into this is dissolved 13.766 grams of high purity aluminum metal, followed by 12 grams of sodium carbonate. The cooled solution is made up to 200 ml. It contains 130 grams/l. of A1203, and 212 grams/l. of total hydroxide, expressed as Na2C03, and 60 grams/ 1. of Na2C03. Procedure. The sample of Bayer Process solution is quantitatively diluted with sodium tartrate solution so that it contains, in 100 ml, between 1 and 6 grams of sodium hydroxide, and between 0.6 and 2.6 grams of A1203. A 10-ml aliquot of the diluted solution is placed in the thermos titration cell, and further diluted with 7 ml of sodium tartrate solution. The buret tip and thermistor are immersed in the solution, the stirring is started, the appropriate sensitivity is selected, the Wheatstone bridge is balanced, and the titration is initiated. When the titration stops automatically, the buret volume is noted; this is the volume required to titrate total hydroxide. 'Then 10 ml of K F solution is added, the sensitivity is changed, the bridge is balanced, and the second titration is started. When the titration stops automatically, the buret volume is noted; this is the volume required to titrate the alumina. These two titration volumes are multiplied by the appropriate calibration factors t o give the concentration of hydroxide and alumina. The method can also analyze slurries containing sand and red mud. Although the absolute concentrations of alumina and hydroxide cannot be determined, since the exact volume taken for analysis cannot be easily measured, the alumina to hydroxide ratio can be found. A 2-ml portion of supersaturated slurry is placed in the titration vessel, and is diluted with 15 ml of sodium tartrate solution. The titration is carried out as described above, and the titration volumes are converted into hydroxide and alumina concentrations by use of the calibration factors. The ratio is calculated by dividing the alumina by the hydroxide concentration. The titration procedure, exclusive of the preliminary dilution, takes between 3 and 4 minutes.

RESULTS AND DISCUSSION The Chemistry of the Determination. Thermometric titrations were considered for this determination because of the large heats of neutralization that are released when hydrochloric acid is added to Bayer Process solutions. The sequence in which the neutralizations occur, and the heats released, are listed in Table I. Since the heats of neutralization are so different, especially of hydroxide, aluminate, and aluminum hydroxide, a plot of temperature rise against volume of acid added would be expected to give a curve made up of segments of very different slopes, corresponding to the different heats of neutralization. The curve actually obtained unfortunately shows no sharp breaks a t the end points. This is due to merging of one reaction into another, especially aluminate into hydroxide and carbonate into bicarbonate and hydroxide.

0

5

IO

ML

ML OF A C I D

OF A C I D

(b)

(0)

Figure 1. Recorder tracings of thermometric titration curves. ( a ) Normal curve. ( b ) First derivative curve Titration conditions: Titrant: 1.3N HCI added at 5 ml/min. Sample: 10 mi Bayer Process solution, containing 36 g/l. NaOH and 24 g / l . AI 2 0 3 , and 1 2 g/l Na~C03.Dilutant: At point A , 7 ml of 250 g/l. sodium tartrate solution had been added to titrating cell. At point D, 10 rnl of 1000 g/l. potassium fluoride had been added to titrating cell

Table I . Sequence of Neutralizations, and Heat Released, on Adding HCI to Bayer Process Solutions React ion

+ H'+ H 2 0 AI(OH)4- + H+ AI(OH)3 + H20 C032- + H+ HC03-

1. OH2. 3.

4. H C 0 3 -

-

+

-

+

H+ + H20 CO? 5. AI(OH)3 i- 3 H + AI3+ -t 3 H 2 0

A H Kcal/mole

(70) (77) (72) (72) -26.9 ( 7 7)

-13.45 -17.1 -3.55 -2.15

To distinguish between reactions, and to suppress unwanted ones, sodium tartrate is added as the complexing reagent. The qualitative equation for the reaction is Al(OH1,n(C,H4O6)?A1(OH)&C,JI,06),'OH- (1) This reaction liberates an amount of hydroxide equivalent to the aluminate; this liberated hydroxide is chemically indistinguishable from the free hydroxide and both are titrated together to give the total hydroxide content of the solution. The titration curve obtained under these conditions is shown in Figure la. The segment AB represents the neutralization of hydroxide; a t the end point B, the neutralization of carbonate to bicarbonate begins. This reaction liberates much less heat, only 3.55 Kcal/mole compared to 13.4 Kcal/mole for hydroxide; therefore the curve becomes almost horizontal for segment BC. Consequently, because of the difference in slopes of segments AI3 and BC, a sharp end point is obtained a t B. It should be noted that with automatic detection of the end point, the addition of acid is stopped exactly at point B and the conversion of carbonate to bicarbonate, shown in segment BC, never occurs. After the titration has stopped at end point B, an excess of potassium fluoride is added to the solution and a reaction, represented approximately by the following equation, occurs.

+

-

+

(10) 6. C. Tyson, Jr., W . H . McCurdy, Jr., and C. E. Bricker, Ana/. Chem.. 3 3 , 1640 (1961). ( 1 1 ) D. D. Wagman, et a/, Nat. Bur. Stand. ( U . S . J . Tech. Note, 270-3, Washington, D.C.. Jan. 1968. (12) P. G . Zambonin and J. Jordan, Anal. Chem., 41, 437 (1969).

A N A L Y T I C A L CHEMISTRY, VOL. 45, NO. 13, NOVEMBER 1973

2249

~~~

Interfering species Kind

Errors in the determination of

Concn g j l Hydroxide Alumina at 180 at 150 g/l., % g/l., %

Alumina: hydroxide ratio (0.638) %

F - , vo43-, po43-, Sod2-

c03'-

CaC03, Fe203, Ti02 (red mud) Organics

~

Table IV. Precision of Thermometric Titrations

Table II. Effect of Potential Interferences

0.5 to 3.0

-0.06

-0.07

-0.16

60

0

0

0

50

-0.33 -0.11

+0.13 -0.13

+0.31 -0.16

0.3 to 15.0

No. of detns in set No. of separate sets Hydroxide results Nominal concn, g/l. A v re1 std dev, YO Alumina results Nominal concn, g/l. Av re1 std dev Alumina:hydroxide ratio results Nominal value Av re1 std dev

Thermometric titration Watts and Utley

100.0 f 0.2

101.8 f 0 . 4

Al(OH),(@,H,O,),L-

+ 6F-

A1203 at 115g/l,, %

99.8 f 0 . 2

99.4

-+

0.7

A1FG3- n(C,H,O,)?-

3 32

180 0.33

180

not detd

180 0.31

115 0.17

not detd

60 0.30

0.638 0.20

0.638 0.22

0.333 0.30

115

99.8 f 0.2

97.6 f 1.1

+ 30H-

Alumina

Hydroxide

A1umina:hydroxide ratio ( = 0.638),%

Concn, g i l .

Accuracy, %

Concn, g/l.

300 300 300 300 300 180 150 60 60 60 60 60

4-0.23 -0.16

15 30 60 90 120 115 60 3.0 6.0 12.0 18.0 24.0

0 -0.16 -0.23 i-0.27

-0.40 -0.16 i-0.66 i-0.33 -1.7 -1.0

Accuracy, YO +2.0 +1.0

4-0.50 -0.55 f 0.25 -0.34 -I-0.16 f3.33

0 -4-1.66

0 -0.41

(2)

The liberated hydroxide is equivalent to the alumina content, and is titrated under conditions identical to those used for the titration of total hydroxide; thus a similar titration curve describes this titration, as shown by the lines DE and EF, where E is the end point for the alumina titration. Because of the larger volumes used in this titration, as a result of dilution by the potassium fluoride, and by the acid added during the hydroxide titration, the temperature rise and the corresponding signal from the thermistor are lower. Effect of the Differentiating Circuit. The effect of the differentiating circuit is shown in Figure l b . In the first derivative curve, points B' and E' were selected as the cut off voltages to correspond to the end points B and E. In practice, the voltage drop obtained during the alumina titration is increased to be equal to that of the hydroxide titration by increasing the amplification. The control circuit turns off the buret precisely at the hydroxide end point, as soon as the voltage drops down to the value of about 2 volts, before any neutralization of carbonate occurs. Effect of Potential Interferences. In a check of interferences, a series of tests summarized in Table I1 confirmed that none of the commonly found constituents of Bayer Process solutions had an important effect on the results. The effect of organic contaminants was also studied. They were collected by extracting a Bayer Process solution after acidifying it with sulfuric acid. The extract was then added to a solution of pure sodium hydroxide into which a known weight of high purity aluminum metal had been dissolved. The organics did not change the values of hydroxide and alumina found. Accuracy. The accuracy of the thermometric titration procedure was extensively checked. First the results were 2250

3 110

-

Percentage recovery in the determination of Hydroxide at 180 gp., %

3 104

Table V. Accuracy (Found Theoretical Values) on Analyzing Aluminate Solutions

Table I II. Overall Percentage Recoveries Procedure used

Supersatd Saturated Supersatd slurry liquor liquor

Sample type

compared with those obtained with the Watts-Utley procedure ( I ) which was considered reasonably correct. These comparisons showed that thermometric titrations gave higher values for alumina, and lower values for hydroxide, than the Watts-Utley procedure. To determine which method was correct, an extensive series of recovery or doping experiments were carried out. A solution of saturated liquor was filtered, and carefully analyzed by both the thermometric titration and WattsUtley procedures. Portions of this solution were then doped by additions of known amounts of high purity alumina hydrate or of bauxite of accurately known alumina content. These solutions were then digested under pressure a t elevated temperatures to dissolve all of the alumina, thereby yielding series of supersaturated solutions of different alumina and hydroxide content. These doped solutions were again analyzed by both the thermometric and the Watts-Utley procedure and the answers compared with the values calculated from the original analysis and the known weight of alumina added. Table I11 lists the results, expressed in terms of a percentage of the recovered or added alumina and hydroxide. These results show that the thermometric titration procedure gives answers that are very close to the expected values, while the WattsUtley procedure gives high results for hydroxide and low results for alumina with corresponding low values for the a1umina:hydroxide ratio. Precision. The precision of the apparatus and the technique was tested during a ten-month trial period in one Bayer plant. All the determinations were made in triplicate by regular shift analysts working on a three-shift per day basis, under routine working conditions. The relative standard deviation of each set of triplicate results was calculated, and the average relative standard deviation for

ANALYTICAL C H E M I S T R Y , VOL. 45, NO. 13, N O V E M B E R 1973

the ten-month period was calculated by combining the separate sets. The standard deviations on triplicate analyses obtained during this long test period are summarized in Table IV. The testing was done on clear solutions of supersaturated and saturated liquor, as well as on slurries containing suspended red mud, whose liquid phase had a composition corresponding to supersaturated solution. These precisions are much better than those we have been able to obtain by the Watts-Utley procedure. For hydroxide it is about three times better, and for alumina it is about two times better. Application of Method. Apart from the usual saturated and supersaturated Bayer solutions, there are other solutions used in the Bayer plant, in which the concentrations of hydroxide and alumina vary over wide ranges. To show that the thermometric titration procedure was indeed

suitable for the accurate determination of these levels, the experiments partially summarized in Table V were carried out. Solutions of known composition were prepared by dilution of a master solution into which was dissolved high purity alumina. The difference between the calculated concentrations of these solutions and the results found by the thermometric procedure was taken as the accuracy of the method. Thermometric titrations gave satisfactory accuracy over the entire range of concentrations that are encountered in a Bayer Process plant, from 10 to 600 grams per liter of hydroxide and from 3 to 120 grams per liter of alumina. Received for review February 21, 1973. Accepted June 4, 1973. Presented a t the Pittsburgh Conference on Analytical Chemistry and Applied Spectroscopy, Cleveland, Ohio, March 6, 1972.

Adsorption Characteristics of Silver, Lead, Cadmium, Zinc, and Nickel on Borosilicate Glass, Polyethylene, and Polypropylene Container Surfaces Arthur W. Struempler Chadron State College, Chadron, Neb. 69337

Using flameless atomic absorption spectrometry, the container adsorption of silver, lead, cadmium, zinc, and nickel in aqueous solutions has been studied. No single container type proved satisfactory for all ions. Polyethylene containers did not absorb cadmium or zinc. Acidification to pH 2 with HNO3 prevented silver, lead, cadmium, and zinc adsorption on borosilicate glass surfaces. Acidification also prevented adsorption of silver on polyethylene surfaces. Additionally, silver solutions must be maintained in the dark, even under acidified conditions, to maintain stability and minimize adsorption loss. N e w polypropylene containers could not be cleaned satisfactorily for cadmium and zinc studies. Extreme care was necessary to minimize contamination when working with the low ion concentrations detectable by flameless atomic absorption spectrometry.

The loss of trace amounts of metallic ions on container walls during sample collection, handling, and storage of aqueous solutions has been recognized for some time. As technology has advanced to achieve lower detection limits, the need to prevent adsorption loss of ions on container surfaces has become more significant. Adsorption of ions a t low concentrations on container walls appears to be associated with many factors. In a study using radioactive tracers, Robertson ( I ) analyzed the adsorption of 11 elements in sea water on various containers and indicated that optimum conditions for storing sea water were achieved by acidifying to pH 1.5 immediately after collec(1) D. E. Robertson, Anal. Chirn. Acta, 42, 533 (1968).

tion and storing in polyethylene containers. After studying somewhat different ions in hard water, Eicholz et al. (2) reported that it was preferable to use borosilicate glass, rather than polyethylene, to minimize container ion adsorption. Adsorption of silver by containers appears t o vary with concentration, pH, contact time with the container, composition of dissolved salts, type of containers, and complexing agents (3-7). The present work was confined to container adsorption of low concentrations of silver, lead, cadmium, zinc, and nickel ions, in order to develop a workable method for studying trace quantities of these ions in natural precipitation (rain, snow, hail). Three types of containers with different pH values were studied. Temperature and light conditions were also varied for silver. Flameless atomic absorption spectrometry was selected as the analytical method since it appeared to be a sensitive, selective, rapid, and low-cost method per sample. Its value as an analytical method is further enhanced in that only microliters of sample are required.

EXPERIMENTAL Apparatus. A Model 303 Perkin-Elmer atomic absorption spectrophotometer. with a HCA 2000 graphite furnace, was used as the analytical tool. Accessories included a Perkin-Elmer deuteri(2) G. G. Eichholz, A . E. Nagel, and R. B. Hughes, Anal. Chern., 37, 863 ( 1965). (3) W. Dyck, Anal. Chem.. 40, 454 (1968). (4) T. T. Chao. E. A . Jenne. and L. M . Heppting, U S . Geol. Survey, Prof. Paper, 6 0 0 - 0 (1968) (5) R. A . Durst and B. T. Duhart, Anal. Chem., 42, 1002 (1970). (6) F. K . West, P. W. West, and F. A. Iddings. Anal. Chem.. 38, 1566 (1966). (7) F. K. West. P. W. West, and F. A . Iddings. Anal. Chim. Acta, 37, 112 (1967).

A N A L Y T I C A L CHEMISTRY, VOL. 45, NO. 13, NOVEMBER 1973

2251