Weiwei E ,Jingcai Cheng ,2,*,Chao Yang ,2,*,Zaisha Mao

1 Key Laboratory of Green Process and Engineering,Institute of Process Engineering,Chinese Academy of Sciences,Beijing 100190,China

2 Jiangsu Marine Resources Development Research Institute,Lianyungang 222005,China

A B S T R A C T Article history:Received 20 January 2014 Received in revised form 1 April 2014 Accepted 8 April 2014 Available online 2 January 2015

Keywords:Nickel hydroxide Precipitation Particle size distribution Online measurement Stirred tank Mixing The objective of this work is using the online measurement method to study the process of precipitation of nickel hydroxide in a single-feed semi-batch stirred reactor with an internal diameter of D=240 mm.The effects of impeller speed,impeller type,impeller diameter and feed location on the mean particle size d43 and particle size distribution(PSD)were investigated.d43 and PSD were measured online using a Malvern Insitec Liquid Process Sizer every 20 s.It was found that d43 varied between 13μm and 26μm under different operating conditions,and it decreased with increasing impeller diameter.When feeding at the off-bottom distance of D/2 under lower impeller speeds,d43 was significantly smaller than that at D/3.PSDs were slightly influenced by operating conditions.

1.Introduction

As an important industrial unit operation,precipitation or reactive crystallization of sparingly soluble salts is w idely used to produce fi ne and bulk chemicals,pharmaceuticals,biochemicals,catalysts,pigments,ceramics,etc.[1–5].During the precipitation,a solution of reactant is mixed with another in a reactor[6],and the precipitation product is formed by a chemical reaction follow ed by nucleation and crystal growth[7,8].In most cases,the chemical reaction is fast,or even quasi-instantaneous,and the particle size distribution(PSD)and the mean particle size(d43)are influenced strongly by mixing conditions,which control the rate of instantaneous chemical reactions and influence the distribution of supersaturation in the system.Precipitation has been attracting much attention in research and development.

Barium sulfate precipitation from an aqueous solution is probably one of the most extensively researched test reactions.How ever,the results are not entirely consistent.In semi-batch tanks,Tosun[9]found that feeding two reactant solutions simultaneously produced larger crystals than feeding one reactant to a stirred solution of the other.In double-feed semi-batch precipitation of BaSO4with apropeller agitator,Angeh?fer[10]found that smaller particles were obtained when the feed pipes were near the impeller.By studying benzoic acid precipitation as the model system,?slund and Rasmuson[11]found that under low impeller speeds,feeding near the free surface produced smaller particles compared to feeding near the impeller.However,under higher impeller speeds,this trend was reversed.They also investigated the effect of impeller type on d43.Larger particles were produced with a three-blade marine-type propeller compared to those with a six-blade Rushton turbine.The reason was that at an equal stirring speed,the mean power input of the Rushton impeller was higher than that of the propeller.

How ever,few studies have been performed on the precipitation of nickel hydroxide.As the main active material for the positive electrode in rechargeable batteries[12–18],it is of great significance to investigate the relation ship between electrical performances and the physical characteristics of nickel hydroxide particles,such as size,PSD and morphology.Now,most studies on nickel hydroxide are focused on using different synthetic methods,like using cobalt and other metals as the additive etc.,to test its electrode performances[19–24].For example,Guan and Deng[25]found that the charge–discharge process can take place more easily and more reversibly on the Ni(OH)2electrodes prepared with nanometer spherical particles than that with micrometer ones.Nickel hydroxide with a small crystallite size shows a high proton diffusion coefficient,giving excellent electrochemical performance.A narrower PSD can improve the reversibility of electrode reaction[26].Watanabe and Kikuoka[27]proposed that nickel hydroxide of small size show ed better charge–discharge behavior due to its faster proton diffusion.3D nanostructure with ravine-like surfaces provides an extremely large surface compared to the smooth crystalline surface,and thus promotes the electrical conductivity of the electrodes[28].Sphere-like morphology has been proved to present better electronic behavior,such as higher capacities.Fish-like particles demonstrate improved electrochemical properties with the increase of cycling times[29].From the literature survey,we can draw a conclusion that nickel hydroxide particles with small size,narrow distribution and spherelike morphology behave better electronically.Effective charging of Ni(OH)2electrode to obtain maximum capacity depends on particle size,morphology and PSD.Although the precipitation process plays such an essential role in determining these characteristics,few works on nickel hydroxide investigate the effect of operation conditions on the precipitation process.Therefore,such a study is conducted to better understand the precipitation process of nickel hydroxide.

In view of the complicated complexation–precipitation system of Ni(OH)2(ammonium as a complexation reagent involved)with a high reagent concentration and therefore the extremely high particle number density in slurry,this work aims at investigating the effect of mixing conditions on the PSD and d43in a single-feed semi-batch precipitation of nickel hydroxide using an online measurement technique.Online measurement is more accurate and reliable than by offline procedures,which take a long while so that considerable changes in the sample would occur,and may give misleading d43and PSD information.Moreover,the measured transient PSD data could be used for determination of crystallization kinetics through some mathematical methods.

2.Experimental

2.1.Experimental setup

A schematic of the experimental setup is shown in Fig.1.Precipitation experiments were carried out in a 14 L, fl at-bottomed,airtight,transparent cylindrical tank,which has a diameter of D=0.24 m and height of H=1.3D,equipped with four equally spaced wall baffles.A digital stirrer engine was used to drive the turbine.Two different types of turbine investigated are shown in Fig.2.One is a standard Rushton turbine,and the other is a down-pumping 45°six-pitched blade turbine(PBTD).Both have the same diameter Di=D/3 and impeller off-bottom clearance C=D/3.The feed of NaOH entered through a stainless steel tube with 4.8 mm ID inserted vertically into the tank from the top.The steel feeding nozzle was placed 50 mm off the tank centerline,midway between two baffles,and its off-bottom distance is either CA=D/2 or CB=D/3.

Fig.1.Experimental setup.1—stirrer;2—pH meter;3—sampling valve;4—cascade dilution;5—Malvern Insitec;6—peristaltic pump;A,B—feeding positions;D i=80 or 96 mm,D=240 mm,H=312 mm,initial liquid level h1=D/2, final liquid level h2=1.2D,C=80 mm.

2.2.Precipitation procedure

Ni(OH)2was synthesized via a complexation–precipitation route.The formation of Ni(OH)2in the present system can be expressed by the following reactions:

All chemical reagents used were of analytical grade.Distilled water was used throughout the experiment.A NiSO4solution(6464 ml,0.1 mol·L?1)was mixed with a NH3solution(72 ml,25%,by mass)in the tank at first to form the Ni-complex solution.The precipitation reaction was conducted at constant temperature controlled by a thermostat water bath within ±0.1 °C.Then,6464 ml of 0.2 mol·L?1NaOH solution was added into the tank under continuous stirring within 30 min through a peristaltic pump running at a speed of 215.5 ml·min?1to avoid back mixing[34].When the total feeding time of 30 min was over,the liquid level rose to 1.2D.Since d43and PSD had no further change after the feeding process,we mainly discussed the first 30 min after the NaOH feeding begins.The pH value of the system was closely monitored by an precise online digital pH meter with the accuracy of±0.01.Table 1 gives the experimental operating conditions in this work.

The d43and PSD in the suspension were measured online continuously by a Malvern Insitec Liquid Process Sizer(LPS).The LPS was based on the laser diffraction technique and the PSD was calculated by using the Mie theory.Furthermore,the focusing lens of the LPS-100 mm Fourier lens,which had the measurement range of d43from 0.1 μm to 200 μm,was precise enough for this experiment.One of the most significant advantages of the LPS was the fully automated sampling,dilution and measurement,all in one done automatically as preprogrammed.Moreover,when the small but sufficient amount of sample was taken into LPS,it will be immediately and consistently diluted by running water,which suppressed the further nucleation,growth and aggregation after sampling.This system was capable of measuring and displaying full particle size distribution every second,ensuring the accuracy of d43and PSD.Each sample was measured continuously for 15 s with a very high rate of acquisition.

Fig.2.Impeller parameters:(a)Rushton impeller,and(b)45°PBTD.

Table 1Summary of experimental operating conditions

In this study,about 2.5 ml of the suspension sample was withdraw n from the reaction mixture every 20 s and instantly transported to the cascade diluter.A sampling valve was driven by compressed air.The suspension was diluted at a predefined ratio continuously and then the sample was conveyed to the flow cell for measurement.The effect of dilution can be controlled through the transmission level with higher transmission corresponding to low crystal volume percentage.And the transmission level throughout the whole experiments was above 80%,which ensured no further crystal growth and nucleation.Based on many preliminary tests,a unique standard operating procedure(SOP)has been established.A measurement cycle took 20 s:3 s for sample transport to cell,2 s for value operation and 15 s for sample measurement.The software(RTSizer)was set to update the online results every 1 s.Thus,the averaged PSD from 15 effective measurements in each cycle was recorded as representative data.Fresh water flow ed through the measurement cell and the connecting tubing before each measurement to avoid possible inaccuracy.These procedures made sure that there was no interaction among each measurement.Moreover,at least tw o replicated experiments were performed for each operating condition to ensure reproducibility of results.For the first 3 min,d43was calculated every 20 s,and from 3 to 30 min,d43was calculated every minute.The results were reported as the plots of PSDs and the volume averaged mean particle size,d43(Fig.3).

3.Results and Discussion

In this work,by using the Malvern Insitec LPS,the effect of various operating conditions on precipitation of nickel hydroxide is mainly studied.At the beginning of there action,d43changes significantly and then mildly through the precipitation.It is found that d43varied between 13 μm and 26 μm with different operating conditions.The PSD shifts to a broader one with a lower peak as the stirred speed is decreased.

3.1.Effect of impeller speed on d43 and PSD

Fig.3.Measurement process.1—one cycle:20 s;2—the beginning of measurement(none particles)is 3 s;3—the end of measurement(none particles)is 4 s.

Fig.4.Effect of stirring speed on d43.C NiSO4=0.1 mol·L?1,C NaOH=0.2 mol·L?1,feeding location B,D i=80 mm.

The main parameters studied in the experiments are the impeller speed,feeding location,impeller type and impeller diameter,which are all related with the overall mixing intensity and the micromixing condition at the feeding point.The effect of impeller speed on d43is shown in Fig.4 for three different impeller speeds.At the lower speed of 300 r·min?1,d43increases rapidly with time at the beginning,and after reaching a certain value,it becomes relatively stable.Such phenomenon occurs only at lower speeds when using low er feed concentrations.Under a higher speed of 400 and 500 r·min?1,how ever,the largest d43occurs at the very beginning and then decreases mildly with time.The maximum particle sizes for the three stirrer speeds,i.e.,300,400 and 500 r·min?1,are 25.62,20.13 and 15.16 μm,respectively.Kim and Tarbell[3]found the similar phenomenon when studying the barium sulfate precipitation.Wong et al.[8]measured the barium sulfate precipitation at different stirrer speeds and found that under the operating condition of barium chloride solution being fed at the molar ratio close to stoichiometry,the mean sizeat 150 r·min?1was extremely smaller than that at 50 r·min?1.Higher stirrer speeds result in better micro mixing around the feeding region,making nucleation more favored over growth.In fact,small particles are produced if nucleation is favored.Another reason for this phenomenon might be the influence of the secondary process,since stronger stirring can lead to less agglomeration of crystals or even breakage of formed aggregates.

Fig.5 shows the effect of stirred speed on the final PSD.It is clearly shown that the PSD moves to a smaller size range with the increase of stirred speed.The effect of impeller speed on the PSD can be attributed to the interaction among mixing,crystal nucleation,growth and secondary processes.As explained above,high stirred speeds contribute to improving mixing condition,meanwhile decreasing aggregation.Therefore,smaller sizes and narrower PSDs are produced.

Fig.5.Effect of impeller speed on PSD at 30 min.(C NiSO4=0.1 mol·L?1,C NaOH=0.2 mol·L?1,feeding location B D i=80 mm).

3.2.Effect of impeller type on d43 and PSD

The influence of impeller type is shown in Fig.6.At 300 r·min?1,larger particles are obtained when using a 45°PBTD.?slund and Rasmuson[11]found the similar result in the study of benzoic acid crystallization at 200 r·min?1.Oldshue[34]found that at the same stirring speed,the pow er input of a Rushton impeller was approximately 6 times that of a 45°PBTD.Therefore,a Rushton turbine generates much stronger turbulence.It seems that as the competition betw een nucleation and crystal growth is concerned,the Rushton impeller contributes more to intensifying the micromixing effect,w hich favors for the nucleation.Nucleation promoted by better micromixing spreads rapidly the local supersaturation,and therefore reduces the growth drive force.As a result,relatively smaller particles are produced.Another reason is that the radial impeller(Rushton)generates more intense flow shear,and thus reduces agglomeration of particles.From our experimental results of various operation conditions,aggregation plays an essential role in nickel hydroxide particle growth.

Fig.6.Effect of impeller type on d43.C NiSO4=0.1 mol·L?1,C NaOH=0.2 mol·L?1,feeding location B,D i=80 mm.

Fig.7 show s the effect of stirrer type on the PSD.It is shown that the stirrer type has little effect on the shape of PSD,and the PSD moves to a smaller size range when using a Rushton impeller.The peak value with a Rushton is relatively higher than that with a 45°PBTD.When N=300 r·min?1,the PSD span values[span=(d90? d10)/d50based on the accumulated particle volume]of this tw o impellers,i.e.,45°PBTD and Rushton,are 1.642 and 1.492,respectively.When N =500 r·min?1,the PSD span values are 1.798 and 1.705,respectively.This means that the Rushton impeller has a narrower range distribution.

The effect of impeller type on PSD may be originated from the relationship of power consumption with micromixing efficiency.Since larger power input by the Rushton impeller contributes to better micromixing and a small and concentrated nucleation region,therefore,a great number of smaller particles with a relatively uniform size would be produced.Thus,the PSD is narrower with a higher peak.Under higher stirrer speeds(500 r·min?1),the difference between two peak values is reduced,but the difference of distribution breadth is still significant.This means that increasing the impeller speed can decrease the influence of the impeller type on the PSD.

3.3.Effect of impeller diameter on d43 and PSD

The effect of impeller diameter is investigated by using a 45°PBTD impeller.As shown in Fig.8,at both impeller speeds,the impeller with a diameter of 96 mm generates smaller particles than that with an 80 mm diameter.Increasing diameter contributes to better micromixing as well as more intensive flow shear due to increased power input.The former contributes to the nucleation process and the latter results in lower efficiency of aggregation.These are the main reasons that larger impeller diameters produce smaller particles.At lower stirrer speeds,the impact of impeller diameter is significant.How ever,at higher impeller speeds,e.g.,500 r·min?1,this influence is reduced dramatically.A possible reason for this phenomenon is that under higher impeller speeds,the intensity of mixing and flow shear have already reached a high level,and thus the increase in impeller diameter has little effect on d43.

The PSDs with different impeller diameters are shown in Fig.9.When N=300 r·min?1,the PSD span values of these two impellers,i.e.,80 mm and 96 mm,are 1.597 and 1.598,respectively.When N=500 r·min?1,the PSD span values are 1.543 and 1.717,respectively.The PSDs produced by the larger diameter impeller are broader than those by the smaller one.At 300 r·min?1,the peak values of these two impellers are almost the same.How ever,when at 500 r·min?1,the impeller with a diameter of 96 mm creates a peak value smaller than that with an 80 mm diameter.This is consistent with Subsection 3.2,where narrower PSDs possess higher peak values.Higher peak values and a smaller span value mean that more uniform particles are produced.In a word,the impeller with a diameter of 80 mm generates larger particles with a more uniform size,while the impeller with a larger diameter produces small size particles.

Fig.7.Effect of impeller type on PSD at 30 min.T=30 °C,D i=80 mm,feeding location B,C NiSO4=0.1 mol·L?1,C NaOH=0.2 mol·L?1.

Fig.8.Effect of impeller diameter on d43.T=30°C,feeding location B,C NiSO4=0.1 mol·L?1,C NaOH=0.2 mol·L?1.

3.4.Effect of feed location on d43 and PSD

The influence of feed position on d43is shown in Fig.10,as evidenced by using a 45°PBTD impeller with a 96 mm diameter.Four different operating conditions have been investigated.At lower stirring speeds and higher temperature,significantly larger crystalsare obtained if there actant is fed close to the impeller(feeding location B,C=D/3).The smallest particles are obtained when feeding under the liquid surface(feeding location A).This result was confirmed by Uehara and Armenante[7],w ho used barium sulfate as a model system.At 500 r·min?1,the influence of the feed location is much weaker and almost within the experimental uncertainty.How ever,when comparing the influence of impeller speed[Fig.10(a)]to temperature[Fig.10(b)],the stirrer speed has the dominant impact.Feeding close to the impeller utilizes the high turbulent intensity so that the local supersaturation is rapidly reduced,which displays an essential effect on nucleation.Lower local supersaturation allows more chance for particle growth.Hence,larger crystals are gained when feeding close to the impeller.

Fig.9.Effect of impeller diameter on final PSD.T=30 °C,feeding location B,C NiSO4=0.1 mol·L?1,C NaOH=0.2 mol·L?1.

Fig.10.Effect of feed location on d43.Feeding location B,D i2=96 mm,C NiSO4=0.1 mol·L?1,C NaOH=0.2 mol·L?1.

Fig.11 displays the effect of feed location on the PSD.When N=300 r·min?1,the PSD span values of these two feed locations,i.e.,Point A(D/2)and Point B(D/3),are 1.557 and 1.587,respectively.In contrast,when N=500 r·min?1,the span values are 1.717 and 1.891,respectively.Under a lower speed(N=300 r·min?1),the feed location has as light influence on the PSD.On the contrary,operating at a higher impeller speed(N=500 r·min?1),the feed location has a definite influence on PSD.As shown in Fig.11,when changing the feed location from Point B to Point A,the PSD moves to the smaller size range.Besides,the peak value is also increased,which means that under higher impeller speeds,the PSD is more sensitive to the feed location.Overly,feeding at Point A produces smaller particles with a narrower range.

Since the feed location has little effect on particle size,it is assumed naturally that it also has limited effect on PSD.How ever,the experiment result is not that.A possible reason for this phenomenon is that when feeding at Point B near the impeller,the local supersaturation is easier to be dispersed,which favors for the crystallization growth.Therefore,the produced PSD has a broad distribution size.But for the particle size,since it has little change with the feed location,perhaps there are some other reasons to control particle size.Therefore,further studies are needed to find out the controlling facts.

3.5.Effect of temperature on d43 and PSD

Temperature is an essential factor in studying the kinetics of nickel hydroxide precipitation.Under lower initial concentrations,CNiSO4=0.1 mol·L?1,the particle size decreases with temperature as shown in Fig.12.At higher temperature,T=50°C,smaller particles are produced because the nucleation rate is enhanced very sensitively and more nuclei are generated.Another possible reason for this phenomenon is that the viscosity of slurry decreases with increasing temperature,which facilitates the molecular motion and the formation of crystals,thus increasing nucleation and producing smaller particles[35].

The effect of temperature on PSD is shown in Fig.13.Under initial concentration CNiSO4=0.1 mol·L?1,temperature has little effect on PSD.Generally,the PSD remains to follow alogarithmic normal distribution,except the peak is shifted to a smaller size when temperature increases.

Fig.11.Effect of feed location on PSD.T=30 °C,D i2=96 mm,C NiSO4=0.1 mol·L?1,C NaOH=0.2 mol·L?1.

Fig.12.Effect of temperature on d43.C NiSO4=0.1 mol·L?1,C NaOH=0.2 mol·L?1.

Fig.13.Effect of temperature on PSD.C NiSO4=0.1 mol·L?1,C NaOH=0.2 mol·L?1.

4.Conclusions

Operation conditions have significant effect on the crystal mean size d43and PSD in the single-feed semi-batch precipitation of nickel hydroxide.The impeller speed,impeller type,diameter,feed location and temperature affect the d43of nickel hydroxide,which varies between 13 μm and 26 μm.The following conclusions can be drawn from the experimental conditions used in this work.

(1)For the impeller speed,when N=500 r·min?1,the smallest crystals are produced(12 ＜ d43＜ 15 μm),w hich are around 50% smaller than those when N = 300 r·min?1(24 ＜ d43＜ 26 μm).The PSD slides to a smaller size range with the increase of stirrer speed.The peak values also increase with increasing impeller speed.

(2)For the impeller type,a 45°PBTD produces a comparatively larger crystal size than a Rushton type does.The PSD slides to a smaller range by using a Ruston impeller,meanwhile the peak value increases significantly.A higher peak value and a smaller PSD span value mean that a relatively uniform crystal size is produced.

(3)Increasing the impeller diameter witnesses the decrease in d43and a broader distribution range.Besides,under higher impeller speeds,e.g.,N=500 r·min?1,the peak value decreases significantly.

(4)Feed location has relatively weaker effect on d43when compared

with other operating parameters.Under a lower impeller speed(N=300 r·min?1),significantly larger crystals are obtained if the reactant is fed close to the impeller(feeding location B).How ever,under a higher impeller speed(N=500 r·min?1),two PSDs are nearly overlapped.The PSDs are slightly narrower when feeding at Point A than those at B.In addition,under a higher impeller speed,the effect of feed location on PSDs is relatively obvious.

(5)Generally,the particle size decreases with increasing temperature.The PSDs follow a logarithmic normal distribution,except for a little shift in the peak values.

Nomenclature

C axial coordinate from tank bottom,m

D diameter of the tank,m

Didiameter of impeller,m

d43mean diameter of crystal size,m

H height of the tank,m

h1initial liquid level,m

h2final liquid level,m

N agitation speed,r·min?1

T operation temperature,°C