I N. 0. SCHMIDT and W. S. WISE

Department of Sugar Chemistry and Technology, Imperial College of Tropical Agriculture, Trinidad,,B.W.I. I

Chemical Solution 1 A number of methods for chemical cleaning of evapora- tors are available; EDTA has the widest application, but the choice of method must be based on consideration of the chemical type of scale. Sodium citrate is recommended for the more soluble scales.

Tm formation of scale in sugar factory evaporators is of considerable impor- tance to the industry. By the end of a week's continuous operation, the scale that has accumulated has usually re- duced the rate of heat transfer in the evaporators to such an extent that further working of the scaled evaporator would be uneconomic. Detailed data on the effect of scale on evaporation rates are not available but in one factory (73) using a quadruple-effect evaporator without vapor bleed, in order to main- tain the evaporation rate constant a t 9.0 pounds per hour per square foot of heating surface the steam pressure on the first calandria has to be steadily raised from 10 to 23 p s i . over a period of 7 days. As it is impracticable to work the calan- drias a t much higher steam pressures, the only solution is to remove the scale.

A popular method of descaling is to scrape the scale from the tubes with ro- tary wire brushes or cutters, varigus de- signs of which are commercially avail- able; frequently a chemical treatment softens the scale before the mechanical cleaning. I t is difficult to obtain de- tailed information on the cost of me- chanical cleaning, but a typical value in the British West Indies would be about $0.25 (BWI) per ton of sugar pro- duced. This is not an unduly high cost, and in many factories mechanical clean- ing is preferred. There are, however, objections to mechanical cleaning. Some factory engineers consider that the life of evaporator tubes is shortened by abra- sion, distortion, or fatigue of the tube material as the cleaning tools are passed up and down the tubes. No evi- dence appears to have been published on these points, although Schmidt (9 ) has shown that the abrasion of tube ma- terial caused by a mild wire brushing treatment depends on the composition of the scale and on average amounts to roughly 0: 1 % of tube thickness per crop.

In the bigger factories mechanical cleaning may present difficulties owing

of Evaporator Scale to the time taken to clean the large num- ber of tubes (upward of 1000 tubes per vessel), but this will depend on what other essential work is carried out during the week-end shutdown; the cleaning of the evaporators is not always a limiting factor. Perhaps the most important ob- jection to mechanical cleaning is the fact that, in the British West Indies a t least, it is becoming increasingly difficult to obtain labor to carry out the cleaning, and some factories now using mechanical cleaning are considering 'chemical clean- ing.

An ideal solution of the scale problem would be some means of preventing the deposition of scale in the evaporator tubes; considerable effort, involving either chemical or physical treatment of the juice, has been made to achieve this end, but so far without complete success. Another possibility is to design evapora- tors which either do not scale (5) , or can be descaled more conveniently-e.g., by using external calandrias. However, cost renders this solution unattractive as an immediate remedy.

Under these circumstances the com- plete chemical cleaning of evaporators, already used in some factories, is becom- ing of increasing interest. Some scales contain a high proportion of silica and are readily soluble in hot, strong sodium hydroxide solution. Usually the scale consists of calcium salts with a small pro- portion of magnesium salts, the usual anions being sulfate, phosphate, silicate, and anions of organic acids. As such salts are soluble in acid solution, it has been common practice to use hydro- chloric or sulfamic acid to dissolve evapo- rator scales. The disadvantage of using acid is that the iron or steel body of the evaporator is attacked, and cases have been recorded (7 ) of the collapse of evaporators due to corrosion by acid cleaning solutions

I t has recently been found possible ( 7 7 ) to dissolve the calcium and mag- nesium salts of the scale in alkaline solu- tions of (ethylenedinitrilo) tetraacetic acid (ethylenediaminetetraacetic acid, ED- TA). Factory trials have been carried out on an evaporator and a simple method of regenerating the spent EDTA solutions has been developed (8) . More detailed information was needed on the factors affecting the rate of solution of scale by EDTA.

Ex per i men tal The first experiments were carried out

by circulating test solutions through short

lengths of discarded scaled evaporator tubes (6) , but it was difficult to obtain reproducible quantitative results in this way. The technique was therefore de- veloped of casting plaster of Paris cylin- ders which were rotated at constant speed in the test solutions?

Smooth cylinders, 8 cm. long and 2 cm. in diameter, were cast from dental-grade plaster of Paris in glass specimen tubes, a brass rod being clamped axially in the specimen tube before the plaster was added. The glass tube was subsequently removed by cracking it and peeling from the cast. The flat ends were protected from the action of the test solution by coating with cellulose acetate cement. Cylinders were discarded when diameter was decreased by 2 mm. When not in use, they were stored in distilled water.

For carrying out a run the brass rod was held in a chuck connected by a flexible drive to a reduction gear driven by a */3-hp. induction motor to ensure that the cylinders rotated at constant speed (140 r.p.m.). The brass rod passed through a bearing in a large rubber bung which was fitted into a wide glass tube of volume about 500 ml. The glass tube was contained in a water bath, the temperature of which was controlled to within 0.5' C. In starting a run the test solution was brought up to temperature and then the cylinder, previously warmed in distilled water to the temperature of the bath, was introduced. The motor was then started and thereafter 2-ml. samples of the solution were removed at regular intervals and analyzed for cal- cium by the Schwarzenbach titration method (3) , using Eriochrome Black T as indicator. Immediately after the re- moval of the sample, 2 ml. of a solution identical with the test solution, but not containing EDTA, were added to the test solution to maintain a constant volume.

The test solutions consisted of buffers in which EDTA and sometimes other substances such as sodium fluoride were dissolved. The buffers were hydro- chloric acid for pH values less'than 2, acetic acid-sodium acetate for pH 3 to 6, ammonia-ammonium chloride for pH 8 to 1 1, and sodium hydroxide solutions for pH above 12. All pH values were de- termined using a Beckman laboratory model p H meter. The solutions of mag- nesium-EDTA chelate were prepared by adding the calculated amount of mag- nesium sulfate to a 6OmM EDTA solu- tion. To avoid large variations in sul- fate concentration during the run, all

VOl'. 50, NO. 5 MAY 1958 81 1

solutions, except in the citrate runs, con- tained lLM sodium sulfate.

In the experiments with sodium ci- trate solutions, a fine precipitate of cal- cium citrate appeared during the run. By blowing compressed air through the solution for a few seconds it was possible to obtain a uniform suspension of the precipitate and the samples were then removed by pipet in the usual way. The calcium citrate precipitate dissolved im- mediately in the EDTA solution used for the Schwarzenbach method of calcium estimation, so that the subsequent titra- tion gave the total amount of calcium sulfate removed from the cylinder.

Res u I ts

I t was thought likely that a t constant temperature, speed of rotation, and radius of cylinder, the solution process would be first-order, giving an equation:

(1 1 kA d C / d t = - (C, - C ) V where k is a rate constant, cm. m i n . 7 A is the surface area of the cylinder C is the calcium concentration at time t C, is the calcium concentration a t in-

Vis the volume of solution finite time

The value of C m is the sum of two terms: the nonionic calcium chelated by the EDTA, and C,, the saturation con- centration of calcium sulfate in the buffer solution. If E is the concentration of total EDTA present, the final equi- librium concentration of calcium chelate can be put equal to aE, where cy tends to 1 a t high pH values and to zero at low pH values. Using this notation we have

C, = CYE + C, (2 1 -4 slight correction term must be

added to Equation 1 to allow for the fact that samples were removed from the solu- tion and replaced by buffer solution. In this case the total EDTA4 concentration, E, will itself decrease with time. For convenience we now introduce Y , the concentration of unchelated EDTA- Le., E - C-which was in fact what was experimentally determined. Toward the end of a run, when all the EDTA has been consumed and the solution contains free calcium ions, the value of Y may be- come negative. Assuming that the sampling is carried out continuously at u ml. per minute, we can easily derive the equation : - d Y / d t = -kA/V [(l - a ) E - C,] f

Y ( kA /V 4- v / V ) ( 3 )

In the case where a = 1-Le.: a t pH values where the EDTA chelates virtu- ally a stoichiometrically equivalent amount of calcium ions-this equation becomes - d Y / d t = kA/V X C, +

Y W / V + v / V ) ( 4 )

A test of Equation 4 is shown in Figure 1.

Figure 1. Reaction rate varies lin- early with concen- tration

Initial concentrations of EDTA (mM):

0 100 8 50 @ 40

. 0 0

For these runs a 1 .OM ammonia-ammo- nium chloride buffer was used, containing EDTA at initial concentrations of 0, 40, 50, and 100mM. The temperature of the solution was 40' C. The solution also contained 1.OM sodium sulfate, so that the increase in sulfate concentration due to the dissolving of calcium sulfate did not markedly affect the total sulfate con- centration. Values of -dY/dt were es- timated from successive experimental values of Y and t and Figure 1 shows the relation of -dY/dt to Y , which was taken as the mean concentration over which -dY/dt was measured. It can be seen from Figure 1 that Equation 4 is obeyed. The rate -dY/dt varies linearly with Y, and the results from different initial concentrations of EDTA all lie on the same straight line, which also includes the points obtained when no EDTA was present. The negative intercept on the Y axis gives the values of C,.

In the work reported below, values of k and C, were obtained by drawing graphs similar to Figure 1 and applying Equation 4. In the few cases where a differed sufficiently from unity, Equation 3 was used; under the conditions of the experiments straight-line plots were ob- tained, as a slow sampling rate ensured that E was effectively constant over the time of the experiments. In some cases, values for C m were obtained by rotating the cylinder for some hours in the test solution, no replacement solution being added and no attempt being made to ob- rain rate measurements.

A series of experiments was carried out at different temperatures, to find the activation energy for the dissolving proc- ess (Table I).

From these results, by plotting log k against 1/T, an activation energy of 4.0 kcal. was found for the process of dis- solving calcium sulfate in EDTA solu- tions.

The variation of k with pH is shown in Figure 2 where each point is the mean of at least three determinations. Results

Y mM

were obtained under different conditions: buffer solutions containing no EDTA; solutions containing 60m.M EDTA; solu- tions containing 6Omltl EDTA and 0.5M sodium oxalate, fluoride. carbonate, or phosphate, which form sparingly soluble calcium salts; and solutions containing 60mA.M magnesium-EDTA chelate. As no specific effects were found. the average of the results obtained in the presence of oxalate, fluoride, carbonate, or phos- phate is plotted in Figure 2.

Values of C m under a variety of con- ditions are shown in Figure 3, CCO being the total concentration of calcium which can dissolve in a given solution of EDTA. C m is the sum of the chelated calcium and the saturation concentration of calcium sulfate in the buffer solution.

Discussion of Results

Although there is a certain degree of scatter, Figure 2 shows that the rate con- stant for the solution of calcium sulfate in EDTA solutions is independent of pH and is not influenced by the presence of fluoride, phosphate. carbonate. or oxa- late, all of which form sparingly soluble calcium salts. The same value for the rate constant is obtained in buffer solu- tions containing no added EDTA, except for very acid solutions, when the rate constant increases slightly.

In the experiments using magnesium- EDTA4 chelate a lower rate constant was obtained than for pure EDTA solutions, although equilibrium studies show that calcium ions form almost the stoichio-

Table 1. Variation of Rate Constant with Temperature

Temp., C. k , Cm./Min. 32 40 50 60 70 76

0.110 0.115 0.117 0.166 0.196 0.223

8 1 2 INDUSTRIAL AND ENGINEERING CHEMISTRY

E VA P 0 R A T 0 R DES C A L I N 0

0. 5 L '@\

0.05 t T

I I I 2 4 6 8 IO I2

Figure 2. Rate constant of Figure 3. Relation of cal- calcium sulfate in EDTA solu- cium concentration to pH a t tions is indePendent of PH

PH

infinite time:

@ Buffer alone 0 Buffey and EDTA c) Buffer, EDTA, and sodium oxalate,

fluoride,carbonate,or phosphate 0 Buffer and sodium citrate 0 Buffer and magnesium-EDTA

chelate

A. Buffer alone ' F B . Buffer and 60 mM EDTA C. Buffer, 60 m M EDTA, and 0.5M

D. Buffer, 60 m,M EDTA, and 0.5M

E. Buffer, 60 mM EDTA, and O.5M

Ail buffer solutions contain 1.OM sodium sulfate

sodium oxalate

sodium fluoride

sodium carbonate

metric amount of calcium-EDTA che- late by displacing magnesium from its EDTA chelate.

These results can be interpreted in terms of simple film theory. The fact that a physical rather than a chemical process is controlling the rate of reaction is indicated by the low activation energy found and by the fact that the rate d solution is increased by increasing the flow of liquid past the scale (6). If it is assumed that the reaction between cal- cium ions and EDTA is rapid and takes place in an infinitely thin reaction zone situated within the stagnant film (72), the following equation is readily deduced :

where DE and Dc are diffusion constants for EDTA and calcium sulfate, respec- tively, and d is the film thickness.

Equation 5 is of the same form as Equation 1. The value of k is given by D8/d and is independent of the rate of chelate formation from calcium ions and EDTA, if this is a sufficiently fast reac- tion. The experiments with the mag- nesium-EDTA chelate in which a lower value of k was found suggest that the displacement of magnesium ions from the EDTA chelate by calcium ions is a relatively slow processi

In the case of buffers containing no EDTA, a similar approach gives

dN/dt = DEA/d ( E -k C, Dc/DE - C) ( 5 )

dN/dt = DcA/d (C, - C) ( 6 ) The value of k is given by Dc/d and,

assuming that the diffusion constants for EDTA and calcium sulfate are not mark-

edly different, the conclusion from Equa- tion 6 is that the rate constant, k, for buffer solutions alone should be the same as that obtained with the buffer solu- tion containing EDTA. This conclusion is supported by Figures 1 and 2.

The variation with pH of values found for Cm-i.e., LYE + C,--shown in Figure 3, presents a number of interesting fea- tures which can be explained in terms of simple ionic theory. Curve A was ob- tained in buffer solutions containing no added EDTA. The value of C, is roughly constant from pH I to 11. Be- low pH l the value of C, increases, while above pH 11 it decreases to zero. The decrease is presumably due to the fact that the surface of the cylinder becomes coated with calcium hydroxide in very alkaline solutions; the solubility of calcium hydroxide decreases with in- creasing pH, owing to the common-ion effect of hydroxyl ions. Below pH 1, a significant amount of sulfate exists as bisulfate ions, so that to maintain the solubility product [Ca++] [Sod--] con- stant, the calcium ion concentration be- comes larger-Le., the value of C, in- creases.

Curve B was obtaiqed with solutions containing 60mM EDTA. Over the pH range of 6 to 1 I the value of C is roughly constant. Above pH 11 Cm decreases and tends to the value 6OmM. Below pH 6 CW decreases markedly and by pH 3 curve B joins curve A.

The difference between curves A and B is due to the term LYE. At high pH

values, when a = 1, the EDTA is able to chelate the stoichiometric amount of calcium, here 60mM, while at pH values below 4 the EDTA binds calcium less strongly owing to the competition of hy- drogen ions for the EDTA anions and LY

decreases eventually to zero. The re- sult is that curve B lies parallel to and 6OmM above curve A at pH values above 6, but below pH 6 the difference between curves A and B diminishes, until by pH 3, the curves join together.

Curve C, obtained with EDTA in the presence of 0.5M sodium oxalate, is similar to curve B but joins curve A at a higher pH value than does curve B. This is essentially due to the fact that calcium oxalate is very much less soluble than calcium sulfate. The surface of the cylinder becomes coated with calcium oxalate and solution of calcium from the cylinder is possible only when the con- centration of calcium ions in solution is less than &,ox/ [Ox- -1, where is the solubility product of calcium oxalate. The calcium EDTA chelate is in equi- librium with a small concentration of calcium ions, which, as the pH is lowered, increases because of the competition of hydrogen ions for the EDTA anions. When the equilibrium concentration of the calcium ions reaches the value &ox/ [Ox-- 1, no solution can take place from the calcium oxalate surface of the cylinder-Le, a = 0. Owing to the low value of the solubility product of cal- cium oxalate, the calcium ion concentra- tion becomes equal to [Ox--] a t a comparatively high pH. Curve D was obtained in the presence of 0.5M sodium fluoride; as calcium fluoride is less solu- ble than calcium oxalate, curve D is displaced to a slightly higher pH than curve C.

Curve E was obtained in the presence of 0.5M sodium carbonate. At high pH

VOL. 50, NO. 5 MAY 1958 813

Table II. Solution of Calcium Sulfate in 0.100M Sodium Citrate Solutions at

40” C. and pH 6

Concn., iM C m , *Vl 0 0.110 0.1 0.083 0.5 0.069 1 .o 0.061

Na2S04

values, because of the low solubility ofcal- cium carbonate, curve E is similar to the curves obtained for iolutions containing fluoride and oxalate. As the p H is lowered, however, carbonate ions are converted to bicarbonate ions, and be- cause calcium bicarbonate is much more soluble than calcium sulfate, E joins B a t a p H of about 7.

Results Obtained with Sodium Citrate

While the results obtained from both laboratory and factory experiments have shown that EDTA can be successfully used to dissolve evaporator scales, a pos- sible objection is the comparatively high cost. I t was therefore decided to in- vestigate the possibility of using cheaper complexing agents. Sodium citrate ap- peared to be worth investigating. as it is known to form a complex (7) with cal- cium ions :

Ca+- -+ Cit3- = CaCit-

although in more concentrated solutions calcium citrate is precipitated.

Experiments were carried out in which the plaster of paris cylinders were ro- tated in sodium citrate solutions. As ex- pected, the calcium sulfate was removed from the cylinder and eventually a pre- cipitate of calcium citrate appeared. The formation of the precipitate did not alter the rate a t which calcium sulfate was subsequently removed from the cylinder. Good first-order plots were ob- tained for the solution process and the values of k (Figure 2) are very close to those found for EDTA solutions.

The disadvantage in using sodium citrate is that the complex CaCit- has a much lower formation constant than the Ca-EDTA chelate and sodium citrate does not dissolve the less soluble cal- cium salts. This was found experi- mentally, in that the sodium citrate had no effect on the calcium sulfate cylinder in the presence of added fluoride. The effect was further emphasized by using sodium citrate solutions containing vary- ing amounts of sodium sulfate. The C m values found are shown in Table 11.

The Cm value, which is a measure of the amount of calcium sulfate removed from the cylinder. is reduced by the presence of sulfate ion, because the con- centration of calcium ions in equilibrium with the CaCit- complex is compara- tively high. M’hen the calcium ions

have reached the concentration cor- responding to the solubility product of calcium sulfate, no further solution from the cylinder is possible (cf. the explana- tion of curve C in Figure 3).

I t is obvious that sodium citrate is of much more limited application than EDTA in dissolving scale. However. in experiments (70) on a sugar factory evaporator in which the scale consisted of a high proportion of calcium sulfate. the evaporator tubes were completely cleaned by the treatment.

Application of Results These results are of interest in relation

to the practical problems of evaporator cleaning. Essentially the rate of cleaning will depend on the twofactors, k and Cm. Apart from the case of magnesium- EDTA chelate, the only variables affect- ing k are the rate of stirring and tem- perature. Although an increase in tem- perature corresponds to an increase in k , owing to the low activation energy of the dissolving process. the rate is not greatly sensitive to temperature variations. It is possible to increase the value of k by in- creasing the rate of stirring, but this can- not usually be varied under the conditions of cleaning sugar factory evaporators.

The smaller values of k obtained with the magnesium-EDTA chelate are of in- terest in connection with the use of re- generated EDTA solutions for evapora- tor cleaning (8). Magnesium salts are not in general removed by the regenera- tion process and gradually accumulate in the EDTA solution as the magnesium- EDTA chelate. The formation of this chelate has two adverse effects: the smaller rate constant and the fact that the magnesium-EDTA chelate cannot dissolve magnesium salts from the scale. Therefore. difficulties may be encoun- tered with EDTA solutions which have been used on scales of high magnesium content. This difficulty would be over- come by using a different method of re- generating EDTA based on precipitation of the pure acid (4, 8). However. this process is unsuitable for regular use in sugar factories (8), as calcium sulfate has to be precipitated before the EDTA is precipitated.

The variations in Cm are much larger than the variations in k and their effect on the rate of cleaning is of importance.

The results in Figure 3. B, show that for calcium sulfate scales the rate of cleaning becomes independent of p H above p H 6. It has been the practice to use very alkaline EDTA solutions for cleaning evaporators. but the above re- sults show that this is not necessary. a conclusion confirmed by a factory trial (2). The importance of this result is in connection with the regeneration proc- ess. which involves acidification of the spent solution with sulfuric acid to pre- cipitate calcium sulfate and decantation of the supernatant layer, the pH of which

is then raised by the addition of caustic soda. By not having to raise this pH to a very high value. there is a considerable saving in caustic soda. with a correspond- ing saving in the amount of sulfuric acid required for acidification.

The results obtained emphasize the difficulties of using acids to dissolve cal- cium sulfate scales. The ability of acids to dissolve calcium sulfate depends on the increase of C, at low p H values due to the formation of bisulfate ions in solu- tion. I t can be seen from Figure 3, that the increase in C, is found only at rather high concentrations of hydrochloric acid and is comparatively small, These re- sults correspond to solutions containing 1M sodium sulfate, so that the solubility of the calcium sulfate is somewhat de- pressed by the common ion effect. This does not essentially alter the conclusion that acid is less suitable than EDTA for dissolving sulfate scales. Hydrochloric acid, or even weak acids, can, however, be used to dissolve other scales such as calcium carbonate and this is the basis of the soda ash process of cleaning evaporators, in which the scale is first boiled with soda ash, and thus converted to calcium carbonate, which is then dis- solved in acid.

The results with sodium citrate indi- cate that it is a satisfactory cleaning agent for the more soluble scales, but it may have a restricted application. I t should be of use in descaling rum stills, where the scale is essentially calcium sulfate.

literature Cited

(1) Avalos, M., Keller, A. G., Proc. In- tern. Sac. Sugar Cane Technologists, 9th Congr., in press.

(2) Bennett, M. C., Schmidt, N. O., Wiggins, L. F., Wise, W. S., Intern. Sugar J . 58, 249-52 (1956).

(3) Biedermann, W., Schwarzenbach, G.,

(4) Buckley, G. D., Thurston, E. F., Chem. &Y Ind. (London) 1956, p. 493.

(5) Chandler, J. L., Proc. Brit. W e s t Indies Sugar Technologists, 1957 meeting, in press.

(6) Connolley, F. H., A.I.C.T.A. thesis, Trinidad, 1955.

(7) Hastings, A. B., McLean, F. C., Eichelberger, L., Hall, J. L., da Costa, E.: J . Bid. Chem. 107, 351- 70 (1934).

(8) Holland, I . D., Massiah, B. V.; Meyers, J. C.: Schmidt, N. O., W’iggins, L. F., Wise, W. S., Proc. Brit . W e s t Indies Sugar Technologists

C h k i a 2, 56-9 (1948).

1954, pp. 155-61. (9) Schmidt, N. 0.: Ibid., pp. 141-9.

(10) Schmidt, N. O., unpublished experi-

(11) Schmidt, N. O.? Wiggins, L. F.? IND.

112) Sherwood. T. K.. “Adsorution and

ments.

ENG. CHEM. 46, 867-70 (1954).

Extraction,” p. 194, Mckraw-Hill, New York, 1937.

(13) Springer, H. B., Proc. Intern. Sac. Sugai Cane Technoiogists, 8 t h Congr., 754-65 (1953). RECEIVED for review April 15, 1957

ACCEPTED October 5, 1957 Part of the research program of the British West Indies Sugar Research Scheme.

8 14 INDUSTRlAL AND ENGINEERING CHEMISTRY