Continuous Double Recrystallization of Light Soda Ashinto Super Dense Soda Ash

Harald Oosterhof,* Jan de Graauw, Geert-Jan Witkamp, andGerda M. van Rosmalen

Laboratory for Process Equipment of Delft University of Technology,Delft, The Netherlands

Received November 19, 2001

ABSTRACT: In this work, a new process is explored for the bulk densification of light soda ash. Advantages of thenew process are that no (crystal) water needs to be evaporated and that a crystalline super dense soda ash is producedat atmospherical conditions. The so-called “mixed-solvent” process is based on the fact that the transition temperatureat which monohydrous and anhydrous soda in contact with an aqueous solution are in equilibrium, can be loweredwhen a second solvent is applied. Continuous double recrystallization experiments were carried out in mixtures ofwater and ethylene glycol: first, light soda was recrystallized into monohydrate below the transition temperature,then the temperature was raised above the transition temperature, and monohydrate was converted into solid,anhydrous soda with a high bulk density. Values of up to 1400 kg/m3 were measured. Further, a significant amountof chloride was removed from the soda during the two recrystallization steps. The potential industrial applicationsof the double recrystallization process are described by Oosterhof et al. (patent application no. EP1998020396319981124, 2000).

Introduction

Soda ash from the Solvay process and the monohy-drate process (also called the “trona process”) has a lowbulk density because the final unit operation in bothprocesses is a calcining step: in the Solvay processsodium bicarbonate (NaHCO3) is calcined:

while in the monohydrate process sodium carbonatemonohydrate is calcined:

This means that in both processes the solid crystallineintermediate is heated to form a porous end-productbecause water (and carbon dioxide) is evaporated fromthe crystal.

Both modes of production have a negative influenceon the quality of the product: the porous crystals arebrittle, which results in breakage during transportationand handling. This causes dusting which is unwanted,especially in the glass industry where the light soda iseasily airborne near hot glass furnaces. Another prob-lem that occurs in the glass industry is that the porosityof the light soda gives rise to air bubbles in the finalglass product. A final disadvantage of the light soda isits low bulk density. Common values are 550-600 kg/m3 for the Solvay process and 750-850 kg/m3 for themonohydrate process. Because of this low bulk density,the transport costs are rather high, since for transportby ship not the mass but the volume of the sodadetermines the costs of transportation. Especially foroverseas transportation of soda, which is about 5 Mton/year from the United States to Europe,1 the volume isthe limiting factor.

Both transport costs and handling problems demanda product with a higher bulk density. Three methodsare reported to increase the density: high temperaturecalcining,2 mechanical compacting,3 and the “monohy-dration process”. During high temperature calcining, thelight soda is heated to temperatures of up to 700 °C fora period of typically 45 min. Under these circumstances,the particles slightly fuse into larger particles with ahigher bulk density. During mechanical compaction,light soda ash is precompacted in a screw feeder andthen metered to a hydraulic roller press that producesflakes with a density of up to 1800 kg/m3 (the crystaldensity of Na2CO3 is 2533 kg/m3). Neither Alexander-werk3 nor Rant4 mentions the bulk density of the finalproduct. The third option which is described by Rant4

is the “monohydration process”: water and light sodaare added to a calciner that is operated around 180 °C.At the entrance of the calciner, the temperature of theslurry is approximately 90 °C, and monohydrate isformed. When the temperature increases along thelength of the calciner, water is evaporated and denseanhydrous soda is formed. This dense soda has a bulkdensity of about 1050 kg/m3. The reported energyconsumption however is high: 400 kg of steam for eachton of dense soda ash.

Notice that each of the processes gives a brittleproduct: none of the particles is appearing solid or wellfaceted. Figures 1 and 2 give two scanning electronmicroscopy (SEM) pictures of both the light soda ashand the dense soda ash from the monohydrationprocess. The difference in particle size and quality isobvious.

In this paper, a new method is presented with whichlight soda ash can be converted into crystalline densesoda ash with a bulk density of up to 1400 kg/m3: theso-called “double recrystallization” process.* Present address: Union Umicore Research, Olen Belgium.

2NaHCO3 f Na2CO3 + CO2 + H2O (1)

Na2CO3‚1H2O f Na2CO3 + H2O (2)

CRYSTALGROWTH& DESIGN

2002VOL.2,NO.2

151-157

10.1021/cg0100336 CCC: $22.00 © 2002 American Chemical SocietyPublished on Web 02/01/2002

Double Recrystallization Process

Sodium carbonate (Na2CO3) is known to have varioushydrates of which the anhydrate is stable above 109 °Cin aqueous solutions. Below this temperature, mono-,hepta-, and decahydrate can develop, depending on thetemperature of the solution. Because a saturated sodasolution has an atmospheric boiling point of 105 °C,anhydrate cannot be crystallized during evaporativecrystallization at (sub)atmospherical conditions.

The influence of ethylene glycol on the transition ofmonohydrate to anhydrate is described by Oosterhof etal.5 A quantitative relation was established in this workbetween the concentration ethylene glycol in the mix-ture and the temperature at which monohydrous andanhydrous soda were in equilibrium.

The use of mixtures of water and glycol for theproduction of dense soda ash is best explained usingFigure 3. In this graph, the solubility curves of Na2CO3and Na2CO3‚1H2O are depicted for a mixture containing70 wt % glycol (on a salt-free basis). The curves arefitted through experimentally determined solubilities.Above approximately 78 °C, anhydrous soda is stable,below this temperature monohydrate.

When (anhydrous) light soda ash is added to themixture at a temperature below the transition point, itwill dissolve until the solution is saturated to monohy-drate. Further addition of light soda will then result ina solution that is supersaturated to monohydrate butstill undersaturated to anhydrate. As a result, themonohydrate will nucleate and grow, while the anhy-

drate will further dissolve. The maximum driving forcefor the crystallization of monohydrate that can bereached is the difference between the solubilities ofanhydrate and monohydrate and is given as ∆c1 inFigure 3. The dissolution continues until all light sodaash is converted into monohydrate. The advantage ofthis recrystallization process is that by precisely adjust-ing the temperature of the suspension the maximumsupersaturation in the solution can be controlled. Asubsequent temperature increase to above the transitionpoint creates a driving force for the reverse process: thenow unstable monohydrate phase transforms into thestable anhydrate phase. This happens again via asolvent-mediated transition, i.e., monohydrate dissolveswhile anhydrate nucleates and grows (at a driving forcemaximally equal to the solubility difference betweenanhydrate and monohydrate: ∆c2). The recrystallizationof soda takes place via the mother liquor.

The proposed process has several advantages. Firstof all, no solvent or crystal water needs to be evaporated,the temperature of the slurry only needs to be altered.The supersaturation (the concentration or chemicalpotential difference between the stable and unstablephase), the driving forces for both recrystallizationsteps, can be adjusted by varying the crystallizertemperatures.

Finally, the two recrystallization steps can be seenas additional purification steps: the light soda oftencontains substantial amounts of sodium chloride, espe-cially when the Solvay process is applied. During bothrecrystallization steps, the unstable phase dissolves andreleases sodium chloride and other impurities that areentrapped in the crystal as either inclusion, occlusion,or in the crystal structure. Subsequently, the impuritieswill be distributed between the mother liquor and thenew growing stable phase. This process is repeatedtwice: first when light soda transforms into monohy-drate and again when the monohydrate recrystallizesinto dense soda. Both recrystallization steps will there-fore contribute to a lower impurity concentration in thecrystal product. In a continuous process, where themixed solvent stream is recycled, an outlet for theimpurities has to be created to avoid the build-up ofimpurities.

Figure 1. Scanning electron microscopy (SEM) photographof porous light soda ash (“LSA”) with a bulk density of about550 kg/m3.

Figure 2. Scanning electron microscopy (SEM) photographof dense soda ash (“DSA”) with a bulk density of about 1050kg/m3.

Figure 3. Schematical representation of the double recrys-tallization process using the solubility curves of Na2CO3

and Na2CO3‚1H2O (approximate experimental values in amixture containing 70 wt % ethylene glycol on a salt-freebasis).

152 Crystal Growth & Design, Vol. 2, No. 2, 2002 Oosterhof et al.

Experimental Section

Setup and Chemicals. Several continuous double recrys-tallization experiments were carried out in a cascade of twocrystallizers that are operated as continuously stirred tankreactors (CISTR). Both crystallizers are coated with “Hallar”to prevent scaling. Light soda ash was recrystallized in thefirst reactor, and monohydrate was recrystallized in thesecond.

Thermostation for the first crystallizer was provided by aLauda C6 heating bath. The second crystallizer that wasalways operated at a higher temperature was heated with amore powerful Lauda K6KS. Accuracy of both thermostats was0.1 °C. Two Lightning (Labmaster SI) stirrers were used forstirring the suspension in both crystallizers. The stirrer speedswere set at 950 rpm which was found to be sufficient to avoidclassification (approximate energy input 3 W/kg).

The influence of both crystallizer temperatures, the weightpercentage solids, and the residence time on the size andquality of the dense soda ash product were investigated; theconversion, bulk density, average particle size, and the purityof the anhydrate crystals were measured. Light soda ash (AkzoNobel, at least 99% pure) was added to the first crystallizerusing a Hethon solids feeder. A Watson Marlow 505DuRLperistaltic pump was used to transport the mother liquor tothe first crystallizer. A mixture of water (double-distilled),ethylene glycol (Merck, at least 99% pure), and dissolved sodaash was used for the mother liquor. The composition of themother liquor mixture was 6 wt % dissolved soda, 0.70 × 94) 65.8 wt % glycol, and 28.2 wt % water, which correspondedwith a salt-free weight fraction, xEG ) 65.8/(65.8 + 28.2) ) 0.70.

Experimental Procedure. At the start of each experi-ment, both the solids feeder and the peristaltic pump werecalibrated to ensure accurate mass flows. Then, both 1.7-Lcrystallizers were charged with the brine and heated. Whenboth setpoint temperatures were reached, the solids feeder andthe pump were started to initiate the experiment. Eachexperiment was carried out for a period of at least 10 residencetimes (based on the second crystallizer). A summary of theexperiments that were carried out is given in Table 1. The lastexperiment (12) was carried out using a crystallizer with aneffective internal volume of 4.0 L for the recrystallization ofmonohydrate into anhydrate; in this case, the slurry residencetimes were 26 and 60 min. In all other experiments, residencetime τ1 was approximately equal to τ2, but differed a little dueto temperature-dependent densities and the slight change inliquid-phase composition because of the water that was setfree during the recrystallization.

Slurry samples were taken from both the first and thesecond crystallizer at approximately 30-min intervals. The sizeof the samples was about 30 mL. After filtration, acetone wasused as a washing medium to remove the adhering mother

liquor. Subsequently, the samples were placed in a centrifugeto remove the washing liquid.

To calculate the degree of conversion (from light soda tomonohydrate and from monohydrate to dense soda), thesamples were weighed and placed in an oven at 150 °C for atleast 24 h. The amount of crystal water in the samples wascalculated from the decrease in mass.

The concentration chloride in the anhydrous product wasdetermined using a high-resolution ICP-MS (Finnigan Ele-ment). Light soda ash was found to contain 0.8 wt % chloride.On the basis of this value, the removal efficiency during thedouble recrystallization was calculated. Furthermore, the bulkdensity of the crystal product was determined by weighing themass of an accurately determined volume of carefully com-pacted product.

Finally, the crystal product was analyzed for its particle sizedistribution. This was done using a Coulter multisizer II withwhich the number and volume of the particles can be mea-sured.

Results and Discussion

The most important experimental conditions andresults are given in Table 2. The figures mentioned forthe bulk density, average particle diameter (both volumeaveraged, m4/m3 and number averaged, m1/m0), conver-sion, and chloride removal are the average of all samplesthat were taken after 10 residence times. Notice thatthe stoichiometric amount of crystal water that isdetermined in the monohydrous samples is alwayslarger than unity, while the anhydrous samples alsocontain at least a little crystal water in all cases. Thehydration numbers mentioned in Table 2 are slightlyhigher than the theoretical values of 1 and 0, respec-tively, perhaps due to adhering wash liquid and incom-plete conversion into anhydrate. The values of V1 andV2 mentioned in Table 2 can therefore be considered asin general too high.

Most valuable information about the final product istherefore gathered by looking at the scanning electronmicroscopy (SEM) photographs that were recorded. Twotypical examples of product from the first and the secondcrystallizer are given in Figures 4 and 5. On the first,needle-shaped monohydrate is visible, together withsome light soda ash that has not been recrystallized yet(middle, right). The other figure shows the product fromthe second crystallizer: hexagonally shaped anhydratein various sizes and dissolving monohydrate that canbe recognized by its needle-shape with rounded edges.

The various types of crystals are well distinguishableas can be seen from those photographs. During the sizemeasurement, however, the Coulter multisizer does notdiscriminate between the various crystals, which meansthat the measured values for m4/m3 (the volume basedaverage diameter) and m1/m0 (the number based aver-age diameter) are based on the total sample, i.e., on allparticles, both monohydrous and anhydrous. However,when the degree of conversion is high, the few dissolvingparticles will have a minor contribution to the averagesize of the final product.

Influence of T2. Experiments 1, 2, and 3 (Table 2)show the influence of a change in the temperature ofthe second crystallizer (T2) on the product.

Most obvious, a temperature of 75 °C (experiment 2)is too low to produce anhydrate. In Figure 3, it can beseen that this temperature is indeed below the recrys-

Table 1. Summary of the Double RecrystallizationExperiments: the Crystallizer Temperatures T1 and T2,the Slurry Residence Time in the Second Crystallizer τ2,and the Weight Percentage Crystals in the Crystallizer,

MTa

experiment T1 (°C) T2 (°C) τ2 (min) MT (wt %)

1 981116 50 80 30 102 981123 50 75 30 103 981201 50 90 30 104 981216 60 80 30 105 981218 70 80 30 106 981223 60 80 30 207 990122 50 80 30 108 990209 50 80 60 209 990212 50 80 15 20

10 990309 60 90 30 2011 990315 60 80 30 2012 990316* 60 80 60 20

a A crystallizer with an internal volume of 4.0 L was used forthe recrystallization from monohydrate to anhydrate during exp12 (*).

Continuous Double Recrystallization of Light Soda Ash Crystal Growth & Design, Vol. 2, No. 2, 2002 153

tallization temperature of about 78 °C. When thetemperature is raised to 80 °C in experiment 1, theconversion from monohydrate to anhydrate is increasedto 34%, according to the measured water content of theproduct. From the picture in Figure 6, it can beconcluded that the conversion is indeed low: a substan-tial amount of dissolving needle-shaped monohydratecrystals can be seen in this sample. The average particlesize of 208 µm is therefore not reliable since it is stronglyinfluenced by the large percentage of monohydrate inthe sample. Duplicate experiment 7 that was carriedout to verify these results did not even result in theproduction of any anhydrate. This lack of conversion isprobably because the crystallization is carried out onlyslightly above the transition temperature, which meansthat the driving force for conversion was too small:

about 0.05 wt % absolute, corresponding to a maximumtheoretical supersaturation of approximately 0.5%. Pri-mary nucleation of the anhydrate is therefore unlikelyto occur. Notice furthermore that a small fluctuation inthe crystallizer temperature can have a relatively largeinfluence on the prevailing supersaturation, because therecrystallization is carried out close to the transitionpoint.

A further increase of the crystallizer temperature to90 °C results in an almost 100% conversion in experi-ment 3. This conversion is also confirmed by visualobservation: Figure 7 shows a lot of small anhydrouscrystals with an average size of about 142 µm. Further-more, substantial agglomeration can be observed.

Experiments 6, 10, and 11 show the same behavior:the highest conversion is obtained at 90 °C (experiment10), but in this case the difference with the experimentsthat were carried out at 80 °C (6 and duplicate 11) ismuch smaller. This might be explained by the higherweight fraction crystals at which these experimentswere carried out: 20 instead of 10 wt %. This higherfraction promotes secondary nucleation which meansthat also at lower driving force a reasonable conversioncan be obtained.

Influence of T1. Experiments 1, 4, 5, and 7 show thatat 50, 60, and 70 °C the conversion of light soda tomonohydrate is complete (within the range of inac-curacy). This corresponds with the visual observationsof the SEM photographs.

Table 2. Summary of the Experimental Conditions and Resultsa

T1 (°C) T2 (°C) τ2 (min) MT (w%) Fbulk (kg/m3) m4/m3 (µm) m1/m0 (µm) hydrate number V1 V2 chloride removal (%)

1 50 80 30 10 1160 208 75 1.11 0.66 283 50 90 30 10 1030 142 73 1.08 0.06 424 60 80 30 10 1290 240 68 1.25 0.10 505 70 80 30 10 1150 243 68 1.15 0.15 586 60 80 30 20 1350 212 86 1.03 0.15 568 50 80 60 20 1040 188 77 1.11 0.23 43

10 60 90 30 20 1050 155 71 1.23 0.59 4411 60 80 30 20 1120 162 67 1.31 0.46 5312 60 80 60 20 1240 202 73 1.44 0.16 61

2 50 75 30 10 640 1.08 1.267 50 80 30 10 980 1.24 1.209 50 80 15 20 820 1.00 0.98

a The nine experiments above the horizontal line yielded anhydrous soda in the second crystallizer, and the three below the line didnot. When no (or little) anhydrate was produced, not all analyses were carried out. The hydrate numbers (V1 for the first crystallizer, V2for the second) is given in mole water per mole anhydrous soda product.

Figure 4. Monohydrate produced with the double-recrystal-lization process (experiment 5, 70 °C).

Figure 5. Anhydrous soda from experiment 4 with a bulkdensity of 1290 kg/m3 (T1 ) 60 °C and T2 ) 80 °C).

Figure 6. Influence of the recrystallization temperature inthe second crystallizer on the quality of the crystal product. Alow conversion is observed at 80 °C and dissolving needle-shaped monohydrate can be distinguished.

154 Crystal Growth & Design, Vol. 2, No. 2, 2002 Oosterhof et al.

The monohydrous samples that were taken at 50(experiment 1) and 70 °C (experiment 5) are depictedin Figures 8 and 9. Although it seems that the sizedistributions look much alike, the Coulter multisizerwas used to measure the particle size distribution of themonohydrous product from the experiments 1, 4, and5. The results are shown in Figure 10: the relativevolume of the product is given as a function of theparticle diameter.

The measured particle size distributions of the vari-ous monohydrous products do not differ significantly butthe quality of the anhydrous product shows largefluctuations from one experiment to the other, as followsfrom Table 2. It is interesting to see that the monohy-

drate crystallized at 60 °C has the smallest averageparticle size. The question now arises: What causes theshape of the solid curve to be different from the othertwo? An answer is hard to give since quantitative dataon the dissolution kinetics of the porous light soda andthe nucleation and growth of the monohydrate in mixedsolvents are not available.

Theoretically, the dissolution rate could either beincreased or decreased when the temperature is loweredfrom 70 to 50 °C: due to the increased concentrationdifference between the stable and the metastable phase,the dissolution rate is supposed to be higher, but alldiffusional processes will take place at a lower rate dueto this lower temperature and the higher viscosity itcauses.

Looking at the final anhydrous product taken fromthe second crystallizer, the average crystal sizes of theproduct from experiments 4 and 5 (carried out at 60 and70 °C) do not differ very much. A slightly highermonohydrate conversion and bulk density are found forexperiment 4. Both results might be due to the higherdriving force for recrystallization and the larger specificsurface area that occur during experiment 4 where T1) 60 °C. SEM photographs of the product are shown inFigures 5 and 11.

Influence of MT. The weight percentage crystals inthe crystallizer directly relates to the yield per volumeinstalled equipment. A higher production rate decreasesthe investment and capital costs per ton product. Forthis reason, experiments were performed using mixtureswith 25 wt % solids (in the second crystallizer, basedon total conversion). However, the experimental setup

Figure 7. Influence of the recrystallization temperature inthe second crystallizer on the quality of the crystal product. Ahigher conversion is observed at 90 °C, but the anhydrateparticles are rather small.

Figure 8. Long, needle-shaped monohydrate crystallized at50 °C.

Figure 9. Long, needle-shaped monohydrate crystallized at70 °C, having almost the same appearance as the product froma similar experiment that was carried out at 50 °C.

Figure 10. Differential relative volume distributions ofmonohydrate crystallized at 50, 60, and 70 °C.

Figure 11. Close-up of the product from experiment 4, withthe characteristic hexagonal shape of anhydrous soda.

Continuous Double Recrystallization of Light Soda Ash Crystal Growth & Design, Vol. 2, No. 2, 2002 155

used was found to be inadequate: it was not possibleto keep all particles suspended in the (simple) benchscale crystallizer and classification and plugging of thereactor outlet took place.

Nevertheless, some indications of the effect of solidconcentration can be derived from experiments 4 and6. With increasing solid concentration, the averageparticle size of the anhydrate is decreased slightly,which is probably due to secondary nucleation that isdependent on the solids density. Furthermore, theconversion decreases slightly, but the bulk density ofthe product is increased significantly: to 1350 kg/m3.On the other hand, the duplicate experiment (11)yielded a product with a lower bulk density. Again, thisis probably due to the inaccuracy in the crystallizertemperature, resulting in an uncertain value of thesupersaturation and thus growth and nucleation kinet-ics.

Higher Conversion. To obtain an even better prod-uct quality, several other experiments were carried out.The influence of the residence time τ2 was investigatedduring experiments 8, 9, and 12. The first two experi-ments were carried out in the unchanged setup, whileduring experiment 12, a 4-L second crystallizer wasused. A larger crystallizer was used to increase theresidence time in the second reactor only (i.e., un-changed residence time in the first crystallizer).

The difference between runs 8 and 9, both carried outwith a 20 wt % solids content in the crystallizer, isobvious: a larger residence time gives a much largerbulk density and conversion. From the experimentsdiscussed above, it was already clear that the recrys-tallization of monohydrate to anhydrate takes placerather slowly (see runs 1 and 7: no or hardly anyconversion). When the residence time is doubled (ex-periment 8), the amount of anhydrate produced isincreased to 77%, and a decreased residence time of 15min during experiment 9 does not yield any anhydrateat all.

From these results, it can concluded that the recrys-tallization of the monohydrate into anhydrate occurs ata low rate at 80 °C. To improve the conversion and theproduct quality, one final experiment (12) was carriedout using crystallizer temperatures of 60 (T1) and 80°C (T2) and residence times of 30 (τ1) and 60 (τ2)minutes. The results of this experiment look verysatisfying: a bulk density of 1240 kg/m3 and an averageparticle size of 202 µm were measured.

Chloride Removal. Table 2 also gives the percent-age of chloride that was removed from the soda duringthe experiments. Removal percentages of 28 to 61%were measured. There is a strong correlation betweenthe size and the purity of the product. Notice that theexperiments that yielded a product with a large averageparticle size and a high bulk density (4, 6, 11, and 12)also contain the lowest amount of chloride. From a pointof view of kinetics, this is understandable since thelowest impurity uptake is realized when the crystal-lization of the anhydrous product is carried out undermild conditions (i.e., at a low supersaturation and thusat low nucleation and growth rate). Particles subjectedto attrition, however, often have an increased impuritycontent. A large particle size stimulates attrition. Ap-

parently, this effect on the purity does not play adominant role here.

Conclusions and Outlook

The results of the above-presented exploratory experi-ments show that it is possible to improve the quality oflight soda ash substantially by subjecting it to a doublerecrystallization step in a mixed solvent. Both the bulkdensity, the purity, and the average particle size areincreased. From the results given in Table 2 and theparagraphs above, the following conclusions can bedrawn:

(i) The best results were found when the temperatureof the first crystallizer was set at 60 °C, while therecrystallization to anhydrate was carried out at 80 °C;a higher value for T2 resulted in the production of smallparticles with a lower bulk density. Because 80 °C isonly slightly above the transition temperature, careshould be taken to avoid temperature fluctuationsduring the crystallization experiments.

(ii) A high solids content in the crystallizer does notinfluence the product properties negatively; the highestbulk density was measured in a mixture containing 20wt % of anhydrate.

(iii) An increased residence time results in a higherconversion, which is quite obvious.

(iv) The chloride removal is found to be dependent onthe “quality” of the crystallization: if the crystallizationis carried out at moderate conditions, the crystal qualityimproves: both bulk density, conversion, and chlorideremoval are increased.

The operating temperatures of both crystallizersdirectly limit the maximum supersaturations (∆c1 and∆c2) at which both recrystallization steps are carriedout. Therefore, these temperatures might be optimizedwith regard to the product quality obtained in theseexploratory experiments at xEG ) 0.70. Also, the con-centration (and even the type!) of cosolvent might bechanged to produce an even better product. To furtherimprove the quality of the soda ash and to increase thelevel of conversion (which was found to be maximally85% at 80 °C), the option of increasing the residencetime in the second crystallizer should be further inves-tigated. Also, the use of a cascade of crystallizers forthe second conversion might be a very promising option.Two or more crystallizers for the production of mono-hydrate may be desirable too because a full conversionof the light soda is important: any light soda that istransferred to the “hot” stage of the process before it isconverted into monohydrate will not be recrystallizedanymore, resulting in a decreased quality of the finalproduct.

Resuming, it can be stated that the proposed double-recrystallization technique is a promising alternativefor the production of super-dense soda ash: the proposedprocess can be carried out at atmospheric conditions andonly relatively simple equipment is required. Theproduct has an average particle diameter of over 200µm and a bulk density of more than 1350 kg/m3.Furthermore, a substantial part of the chloride presentin the light soda feed is removed during the tworecrystallization steps.

156 Crystal Growth & Design, Vol. 2, No. 2, 2002 Oosterhof et al.

References(1) Thomas, R. Int. Bulk J., 1999, 1, 51.(2) Garrett, D. E. In Natural Soda Ash - Occurrences, Process-

ing and Use; Van Nostrand Reinhold: New York, 1992.(3) Alexanderwerk A. G.; Remscheid; http://www.alexander-

werk.com/dens2•en.html

(4) Rant, Z. In Die Erzeugung von Soda nach dem Solvayver-fahren; eine verfahrenstechnische darstellung; Enke, Stutt-gart, 1968.

(5) Oosterhof, H.; Witkamp, G. J.; Van Rosmalen, G. M. AIChEJ. 2001, 47, 602-608.CG0100336

Continuous Double Recrystallization of Light Soda Ash Crystal Growth & Design, Vol. 2, No. 2, 2002 157