materials

Article Characterization and Interpretation of the Aluminum Zone Refining through Infrared Thermographic Analysis

Xiaoxin Zhang † , Semiramis Friedrich *,† and Bernd Friedrich

IME Institute of Process Metallurgy and Metal Recycling, RWTH Aachen University, 52056 Aachen, Germany; [email protected] (X.Z.); [email protected] (B.F.) * Correspondence: [email protected]; Tel.: +49-241-80-95977 † These authors contributed equally to this work.

 Received: 27 September 2018; Accepted: 16 October 2018; Published: 19 October 2018 

Abstract: High purity metals are nowadays increasingly in demand to serve in electronic, photovoltaic, and target materials industries. The zone refining process is the most common way to achieve high purity in the final step of metal purification. Zone length and growth rate are the main parameters that control the zone refining process. To determine these values, information about temperature profiles in the molten zone is necessary due to its direct correlation with these values. As the determination of this profile is not practically achievable in the present, the novel approach of applying an infrared (IR) camera during the zone refining of 2N8 aluminum is the focus of the investigation in this work. The whole temperature profile of the region near the molten zone was recorded by IR camera during the entire running process. The zone length and the rate at each thermographic image shooting moment were successfully extracted by thermographic analysis. Results showed that both factors varied significantly, which is in contrast to the assumption in literature about their stability while running under constant input power and heater movement velocity, though noticeable purification took place in all of these experiments. However, the impurity concentration during refinement fluctuated remarkably. This was well-demonstrated by the tendency of variation in crystal growth rate attained in this work. These results provide a better understanding of the mechanisms of zone refining with an inductive heater and contributes to the optimization of the process.

Keywords: zone ; infrared camera; process control; zone length; crystal growth rate; aluminum

1. Introduction Zone refining is the most common method used in the production of most high purity metals (purity greater than 5 N). Its mechanism is based on the difference in solubility of impurities in the solid and liquid state. The purification takes place by redistributing the impurities at the freezing interface when one or a series of molten zone(s) move through a solid bar. Many experimental and theoretical investigations have been conducted in recent decades with focus on improving refining and/or production efficiencies [1–15]. Two general categories, as characterized by different emphases, can be summarized from these investigations. The first is in realizing the target within the frame of the impurity distribution profile by optimizing experimental parameters such as zone length, zone movement velocity, number of passes, etc. [1–10]. The second is a focus on knowledge of the molten zone, such as zone shape, zone length variation, and the position of the molten zone, by taking heat transfer and heat balance into consideration [11–15]. The shape of the freezing interface affects the dislocation density and the impurity segregation [12]. The zone position and zone length variation are

Materials 2018, 11, 2039; doi:10.3390/ma11102039 www.mdpi.com/journal/materials Materials 2018, 11, 2039 2 of 12 closely correlated with the freezing interface movement (crystal growth) and the dissolving capacity of impurities in the molten zone. It is well known that zone length affects not only the ultimate concentration distribution of impurities but also the rate of its approach [6]. Applying longer zone lengths in early passes followed by shorter lengths in later passes [6–9], or constantly adjusting the zone length in an optimized relation with the position of the freezing interface [10], will be helpful in improving refining efficiency. However, the research on zone length shows that this is simultaneously affected by many experimental conditions such as heating power, zone movement velocity, crucible size and material, charging material, and the method of cooling. All of these result in the challenge of regulating the zone length. In the case of having high thermal conductive materials for both charge and crucible, a stable zone length is difficult to maintain, as in, for example, the, refining of cadmium in graphite crucibles [16]. However, graphite crucibles are commonly used as a heat susceptor during the refining of magnetic-non-coupling materials (such as aluminum) with induction heating zone melting equipment. Therefore, the molten zone of aluminum under such operational conditions is supposed to be unstable as well. However, it has been shown in our preliminary work [17] that the application of a thermographic analysis system (e.g., an infrared camera) can deliver a variety of new information to ease the complexities in the measurement of zone length. Based on these facts, the objective of this work is, firstly, to illustrate the molten zone movement process in horizontal zone refining of aluminum through thermographic analysis, and then to investigate the zone length variation and crystal growth rate under different combinations of power and zone movement velocity. Additionally, the relationship of impurity concentration distribution after refinement (refining efficiency) with crystal growth rate and zone length will also be studied in focus.

2. Experimental Procedure The trials were conducted in industrial scale horizontal zone refining equipment provided with a single inductive heater capable of generating up to 45 kW with a maximum frequency of 10 kHz, while applying a graphite crucible of 100 cm in length. An infrared (IR) camera (InfraTec GmbH, Dresden, Germany) was deployed to record the refining process in the form of thermographic illustrations.

2.1. Experimental Design Two different series of experiments, regarding the heating power and the heater movement velocity, were designed in this work, as shown in Table1. As the zone refining process generally demands very low velocities in order to allow an effective segregation of impurities, the movement velocities of the heater were determined here as 1.2 mm/min as well as 0.8 mm/min (for the crucible used in this work, leading to periods of around 14 h/run and 21 h/run). Lower velocities were not examined due to the huge time consumption. A total of 9.8 kW of power is the minimum required to provide a molten zone along the bar, i.e., when the power is lower, the molten zone will disappear in the middle of the process or the metal cannot be melted. Due to an additional heating generated from the graphite (susceptor) mass at the end edges of the crucible while zone melting, the starting position of the heater was set at 20 cm after the edge. This assured a similar and stable heating effect along the investigated area. The experiments using a velocity of 0.8 mm/min ended at the moment when the melting interface of the molten zone arrived at the end of the crucible. In order to investigate the variation of zone length and crystal growth rate in the end section (unidirectional region), experiments using a velocity of 1.2 mm/min were run over that moment and finished until the heater was located at the end of the crucible. A sketch of the refining process is illustrated in Figure1, where one pass was conducted for each experiment under 400 mbar Argon atmosphere. The initial material used was 2N8 aluminum initially doped with 0.1 wt % Fe as well as 0.1 wt % Si. Materials 2018, 11, 2039 3 of 12 Metals 2018, 8, x FOR PEER REVIEW 3 of 12 Metals 2018, 8, x FOR PEER REVIEW 3 of 12

Table 1. Experimental parameters. Table 1.1. ExperimentalExperimental parameters.parameters. Power Heater Movement Velocity PowerPower Heater Heater Movement Movement Velocity 11 kW 1.2 mm/min 0.8 mm/min 11 kW11 kW 1.2 mm/min 1.2 mm/min 0.8 mm/min 0.8 mm/min 9.8 kW 1.2 mm/min 0.8 mm/min 9.8 kW9.8 kW 1.2 mm/min 1.2 mm/min 0.8 mm/min 0.8 mm/min

Figure 1. Sketch of the zone refining process. Figure 1. Sketch of the zone refiningrefining process.process. 2.2. Data Analysis 2.2. Data Analysis 2.2.1. Thermographic Data Analysis 2.2.1. Thermographic Data Analysis A typical calibrated thermographic image image at at a certain time can be seen in Figure2 2a.a. ItIt clearlyclearly A typical calibrated thermographic image at a certain time can be seen in Figure 2a. It clearly shows thethe temperaturetemperature distribution distribution along along the graphitethe graphite crucible crucible and charge. and charge. The position The position of melting/ of shows the temperature distribution along the graphite crucible and charge. The position of melting/freezingfreezing interfaces interfaces and the and region the ofregion molten of molten zone can zone also can be also distinguished. be distinguished. To calculate To calculate the exact the melting/freezing interfaces and the region of molten zone can also be distinguished. To calculate the exactinterface interface positions, positions, the temperature the temperature values values along alon theg moltenthe molten zone zone were were exported exported to formto form a plot a plot of exact interface positions, the temperature values along the molten zone were exported to form a plot oftemperature temperature against against the the distance distance in pixel in pixel as shown as shown in Figure in Figure2b. The 2b. region The region with a with temperature a temperature higher of temperature◦ against the distance in pixel as shown in Figure 2b. The region with a temperature higherthan 660 thanC can660 be °C defined can be as defined the molten as zonethe molten when talkingzone when about talking pure aluminum. about pure The aluminum. length (in pixel) The higher than 660 °C can be defined as the molten zone when talking about pure aluminum. The lengthof this region(in pixel) and of the this positions region and (in pixel)the positions of melting/freezing (in pixel) of melting/freezing interfaces in relation interfaces to the in beginning relation to of length (in pixel) of this region and the positions (in pixel) of melting/freezing interfaces in relation to the barbeginning have been of the converted bar have to been length converted and position to le inngth cm and using position MATLAB in cm (R2015a, using MathWorks,MATLAB (R2015a, Natick, the beginning of the bar have been converted to length and position in cm using MATLAB (R2015a, MathWorks,MA, USA). Natick, MA, USA). MathWorks, Natick, MA, USA).

Figure 2.2.An An exemplary exemplary thermographic thermographic image image in zone in refining zone ofrefining aluminum of (aluminuma) and the corresponding(a) and the Figure 2. An exemplary thermographic image in zone refining of aluminum (a) and the correspondingtemperature distribution temperature profile distribution in the molten profile zone in the along molten the barzone (b along). the bar (b). corresponding temperature distribution profile in the molten zone along the bar (b). 2.2.2. Impurity Concentration Analysis 2.2.2. Impurity Concentration Analysis 2.2.2. Impurity Concentration Analysis A sample can be divided into three regions after refinement as shown in Figure3. They are A sample can be divided into three regions after refinement as shown in Figure 3. They are the the unprocessedA sample can region, be divided the refined into three region, regions and after the refinement quick crystallization as shown in region Figure (while 3. They power are the is unprocessed region, the refined region, and the quick crystallization region (while power is turned turnedunprocessed off). Numerous region, the samples refined alongregion, the and refined the quic regionk crystallization as well as some region along (while the areapower of theis turned quick off). Numerous samples along the refined region as well as some along the area of the quick crystallizationoff). Numerous were samples taken along and analyzedthe refined on theregion cross-section as well as via some the along spark spectrometrythe area of the method. quick crystallization were taken and analyzed on the cross-section via the spark spectrometry method. The Thecrystallization concentration were distribution taken and analyzed of Fe and on Si the in thecross-section refined region via the was spark assessed spectrometry to reveal method. the refining The concentration distribution of Fe and Si in the refined region was assessed to reveal the refining concentration distribution of Fe and Si in the refined region was assessed to reveal the refining efficiency. The impurity concentration in the quick crystallization region indicates the intensity of efficiency. The impurity concentration in the quick crystallization region indicates the intensity of

Materials 2018, 11, 2039 4 of 12

Metals 2018, 8, x FOR PEER REVIEW 4 of 12 efficiency. The impurity concentration in the quick crystallization region indicates the intensity of thethe accumulationaccumulation of impurities.impurities. The attained concentrationconcentration value at every position is defineddefined as CCSS, and the ratio CCSS/C00 (where(where C C0 0isis the the initial initial concentration) concentration) represents represents the the refining refining efficiency. efficiency. The lower thethe valuevalue ofof CCSS/C0,0 ,the the higher higher the the refining refining efficiency. efficiency.

Figure 3. 3. TheThe form form of the of sample the sample after refinement after refinement and the positions and the used positions for impurity used concentration for impurity concentrationanalysis. analysis.

3. Results and Discussion 3.1. Characterization of the Zone Refining Process 3.1. Characterization of the Zone Refining Process Figure4 illustrates the positions of melting and freezing interfaces during the continuous Figure 4 illustrates the positions of melting and freezing interfaces during the continuous movement of the heater along the aluminum bar. The exact values of these positions have been movement of the heater along the aluminum bar. The exact values of these positions have been attained from thermographic analyses. As a general rule, after the heater is switched on, a specific attained from thermographic analyses. As a general rule, after the heater is switched on, a specific length of molten phase is formed around the heater. The two portions of this length, on the left or right length of molten phase is formed around the heater. The two portions of this length, on the left or side of the heater, were changing continuously while the heater moved along the bar. Consequently, right side of the heater, were changing continuously while the heater moved along the bar. the zone length, i.e., the difference between melting and freezing interfaces, was changing in the Consequently, the zone length, i.e., the difference between melting and freezing interfaces, was meantime (even in the case of consistent power and velocity), as shown in Figure5. The zone length to changing in the meantime (even in the case of consistent power and velocity), as shown in Figure 5. be considered here corresponds to the time period of 40–550 min. This is because, in the first 40 min, The zone length to be considered here corresponds to the time period of 40–550 min. This is because, the molten zone started forming and expanded along two opposite directions (see Figure4) without in the first 40 min, the molten zone started forming and expanded along two opposite directions (see crystallization, and the molten zone in the last 100 min was in the unidirectional crystallization region. Figure 4) without crystallization, and the molten zone in the last 100 min was in the unidirectional Without changing the heating power, the zone length could be seen as quite large at both ends of the crystallization region. Without changing the heating power, the zone length could be seen as quite bar and relatively stable with smaller values in the middle of the bar. This is most probably due to large at both ends of the bar and relatively stable with smaller values in the middle of the bar. This is the lower thermal dissipation through solid aluminum and inside the crucible when the molten zone most probably due to the lower thermal dissipation through solid aluminum and inside the crucible is located close to one of both ends—with one available dissipation direction—in comparison to the when the molten zone is located close to one of both ends—with one available dissipation locations in the middle of the bar with two available dissipation directions. After starting at the 20 cm direction—in comparison to the locations in the middle of the bar with two available dissipation position (see Section 2.1), the charge has a much shorter heat dispersion to one side—because the directions. After starting at the 20 cm position (see Section 2.1), the charge has a much shorter heat portion of the bar is much shorter—and a much longer one on the other side. However, these portions dispersion to one side—because the portion of the bar is much shorter—and a much longer one on change when the heater moves to another end of the charge. Furthermore, it can be seen from Figure4 the other side. However, these portions change when the heater moves to another end of the charge. that in the case of zone refining of aluminum, the heater was always located in the middle of the Furthermore, it can be seen from Figure 4 that in the case of zone refining of aluminum, the heater molten zone. This is attributed to the low Prandtl number of aluminum (low ratio of viscosity to was always located in the middle of the molten zone. This is attributed to the low Prandtl number of thermal conductivity), preferentially resulting in symmetrical molten zones [11]. aluminum (low ratio of viscosity to thermal conductivity), preferentially resulting in symmetrical molten zones [11].

Materials 2018, 11, 2039 5 of 12 Metals 2018,, 8,, xx FORFOR PEERPEER REVIEWREVIEW 5 of 12

Figure 4. PositionsPositions ofof the the melting melting interface, interface, freezing freezing interface, interface, and and heater heater against against time time for the for theexperiment experiment with with11 kW-1.2 11 kW-1.2 mm/min, mm/min, showing showing that the that freezing- the freezing- and melting and melting interfaces interfaces moved movednonlinearly. nonlinearly.

Figure 5.5. Zone length variation during the refiningrefining process derived from thermographic imaging for anan exemplaryexemplary experimentexperiment withwith 1111 kW-1.2kW-1.2 mm/min.mm/min.

Another measurement—attained indirectly via thermographicthermographic images at everyevery imageimage shootingshooting moment—is thethe crystal crystal growth growth rate. rate. It isIt the is summationthe summation of heater of heater movement movement velocity velocity and the freezingand the interfacefreezing movementinterface movement velocity in thevelocity thermographic in the thermographic image as the IR cameraimage movesas the simultaneouslyIR camera moves with thesimultaneously heater. The position with the of heater. the freezing The position interface of inthe the freezing thermographic interface imagein the thermographic is actually data image attained is directlyactually fromdata anattained IR image, directly and itsfrom differential an IR image, coefficient and its against differential time is coefficient its movement against velocity time in is theits image.movement It can velocity be seen in in the Figure image.6 that It acan significant be seen variationin Figure of 6 crystal that a growthsignificant rate variation takes place of duringcrystal thegrowth process. rate Ittakes should place be during pointed the out process. that the It tremendous should be pointed growth out rate that at the the end tremendous (after around growth 600 minrate inat the this end experiment) (after around was due600 min to the in significant this experiment) reduction was of due induction to the significant heating. At reduction this time, of the induction molten zoneheating. had At already this time, arrived the at molten the end zone of the had crucible, already i.e., arrived the unidirectional at the end crystallizationof the crucible, appeared, i.e., the andunidirectional the heater crystallization was close to the appeared, end of crucible. and the Therefore,heater was the close graphite to the mass end of functioned crucible. asTherefore, heating resourcesthe graphite decreased mass functioned as the heater as moved heating forward. resources Inaddition, decreased the as mass the ofheater material moved was significantlyforward. In reducedaddition, (a the typical mass zone of meltingmaterial phenomena), was significantly leading reduced to an accelerated (a typical solidification zone melting in phenomena), this area. leading to an accelerated solidification in this area.

Materials 2018, 11, 2039 6 of 12 MetalsMetals 2018 2018, ,8 8, ,x x FOR FOR PEER PEER REVIEW REVIEW 66 of of 12 12

FigureFigure 6. 6.6.Variation VariationVariation in position inin positionposition of the of freezingof thethe freezing interfacefreezing ininterface thermographicterface inin thermographicthermographic images and theimagesimages corresponding andand thethe crystalcorrespondingcorresponding growth rate crystalcrystal against growthgrowth time raterate in the againstagainst case oftimetime moving inin thethe the casecase heating ofof movimovi coilngng to thethe the heatingheating end of coilgraphitecoil toto thethe crucible endend ofof (thegraphitegraphite red dotted crucible crucible line (the (the represents red red dotted dotted the line line movement repres representsents velocity the the movement movement of the heater). velocity velocity of of the the heater). heater).

TheThe freezingfreezingfreezing interfaceinterfaceinterface isis thethethe place placeplace where wherewhere the thethe targeted targtargetedeted purification purificationpurification occurs. occurs.occurs. InIn orderorder tototo interpret interpretinterpret impurityimpurityimpurity concentration concentrationconcentration profiles profilesprofiles after afterafter refinement refinementrefinement so as tososo improve asas toto theimprimpr refiningoveove thethe efficiency, refiningrefining a mathematicalefficiency,efficiency, aa correlationmathematicalmathematical of zone correlationcorrelation length and ofof zonezone crystal lengthlength growth andand rate crystalcrystal with growth growth the freezing raterate withwith interface thethe freezingfreezing position interfaceinterface is very desirable. positionposition Asisis very very seen desirable. desirable. in Figure 4As ,As the seen seen distance in in Figure Figure of the 4, 4, the freezingthe distance distance interface of of the the fromfreezing freezing the beginninginterface interface from from of the the the bar beginning beginning is not linearly of of the the proportionalbarbar is is not not linearly linearly to time. proportional proportional Therefore, to theto time. time. relationship Therefore, Therefore, of the zonethe relationship relationship length and of of crystal zone zone length growthlength and and rate crystal crystal with freezing growth growth interfaceraterate withwith position freezingfreezing is interfaceinterface different positionposition from their isis differentdifferent relationship fromfrom (see theirtheir Figures relationshiprelationship5 and6 ) (see(see with FiguresFigures time. This 55 andand result 6)6) withwith is showntime.time. This This in Figure result result7 is .is shown shown in in Figure Figure 7. 7.

(a)(a) (b)(b)

FigureFigure 7. 7. ZoneZoneZone lengthlength length ((aa)) ( andaand) and crystalcrystal crystal growthgrowth growth raterate ( rate(bb)) asas ( b a)a functionfunction as a function ofof thethe position ofposition the position ofof thethe freezingfreezing of the freezinginterface.interface. interface.

3.2.3.2. DependencyDependency ofof thethe ZoneZone Length Length on on Power Power andand and HeaterHeater Heater MovementMovement Movement VelocityVelocity Velocity TheThe relationships relationships ofof the the zonezone zone lengthlength length with with the the positionposi positiontion of of thethe the heaterheater heater andand and thethe the positionposition position ofof of freezingfreezing freezing interfaceinterfaceinterface are areare represented representedrepresented in Figuresinin FiguresFigures8 and 898 . and Bothand 9.9. figures BoBothth show figuresfigures similar showshow zone similarsimilar length zonezone variation lengthlength tendencies. variationvariation Thetendencies.tendencies. diagram TheThe of the diagramdiagram zone length ofof thethe zoneversuszone lengthlength the position versusversus thofthee the positionposition heater ofof (or thethe versus heaterheater time) (or(or versus hasversus so time)fartime) been hashas the soso farfar been been the the only only way way to to show show the the zone zone length length vari variationsations when when no no IR IR camera camera is is available. available. However, However, zonezone lengthlength versusversus thethe positionposition ofof freezingfreezing interface—onlyinterface—only possiblepossible whenwhen usingusing anan onlineonline IRIR

Materials 2018, 11, 2039 7 of 12

MetalsMetals 2018 2018, ,8 8, ,x x FOR FOR PEER PEER REVIEW REVIEW 77 of of 12 12 only way to show the zone length variations when no IR camera is available. However, zone length versus the position of freezing interface—only possible when using an online IR camera—is more camera—iscamera—is more more meaningful meaningful because because it it provides provides a a novel novel chance chance to to enable enable investigation investigation of of the the effect effect meaningful because it provides a novel chance to enable investigation of the effect of zone length on ofof zonezone lengthlength onon refiningrefining efficiency.efficiency. InIn generageneral,l, thethe zonezone lengthlength inin allall experimentsexperiments showedshowed aa refining efficiency. In general, the zone length in all experiments showed a decreasing and then an decreasingdecreasing and and then then an an increasing increasing tendency. tendency. The The se selectedlected characteristic characteristic values values (listed (listed in in Table Table 2), 2), increasing tendency. The selected characteristic values (listed in Table2), however, were different. however,however, were were different. different. As As a a short, short, intermediate intermediate conclusion, conclusion, higher higher power power induces induces a a greater greater zone zone As a short, intermediate conclusion, higher power induces a greater zone length, as normally expected length,length, asas normallynormally expectedexpected duringduring zonezone meltinmeltingg process.process. InIn addition,addition, lowerlower heaterheater movementmovement during zone melting process. In addition, lower heater movement velocities result in further delays velocitiesvelocities resultresult inin furtherfurther delaysdelays inin thethe positionposition ofof thethe freezingfreezing interfaceinterface whilewhile thethe zonezone lengthlength in the position of the freezing interface while the zone length reaches the minimum level. In the reachesreaches the the minimum minimum level. level. In In the the ca casese of of a a movement movement velocity velocity of of 1. 1.22 mm/min, mm/min, the the zone zone length length at at the the case of a movement velocity of 1.2 mm/min, the zone length at the end was greater than that at the endend waswas greatergreater thanthan thatthat atat thethe beginning,beginning, whwhileile thethe oppositeopposite waswas truetrue forfor 0.80.8 mm/min.mm/min. TheThe beginning, while the opposite was true for 0.8 mm/min. The percentage difference between maximum percentagepercentage difference difference between between maximum maximum and and minimum minimum zone zone lengths lengths in in this this work work was was quite quite large, large, and minimum zone lengths in this work was quite large, ranging from 25% to around 36%. However, rangingranging from from 25% 25% to to around around 36%. 36%. However, However, higher higher power power seems seems to to be be beneficial beneficial in in reducing reducing this this higher power seems to be beneficial in reducing this difference. difference.difference.

(a)(a) (b)(b)

FigureFigureFigure 8.8. 8. ZoneZone Zone lengthlength length variationvariation variation versusversus versus theth thee positionposition position ofof of thethe the heaterheater heater withwith with aa a heaterheater heater movementmovement movement velocityvelocity velocity ofof of 1.21.21.2 mm/minmm/min mm/min ( a(a)) )and and and a a a heater heater movement movement velocity velocity of of 0.8 0.8 mm/min mm/min ( (b(bb).).).

(a)(a) (b)(b)

FigureFigureFigure 9.9. 9. ZoneZone Zone lengthlength length variationvariation variation versusversus versus the the positionposition position of of thethe the freezingfreezing freezing interfaceinterface interface forfor for aa a heaterheater heater movementmovement movement velocityvelocityvelocity ofof of 1.21.2 1.2 mm/minmm/min mm/min ( a(a)) )and and and a a a heater heater heater movement movement movement velocity velocity of of 0.8 0.8 mm/min mm/min ( (b(bb).).).

Metals 2018, 8, x FOR PEER REVIEW 8 of 12

Table 2. Comparison of characteristic values of zone length variation by changing process Materialsparameters.2018, 11, 2039 8 of 12

11 kW-1.2 9.8 kW-1.2 11 kW-0.8 9.8 kW-0.8 Characteristic Values Table 2. Comparison of characteristic values ofmm/min zone length variationmm/min by changingmm/min process parameters.mm/min

Peak of zone length at the beginning (Lp1) 11 kW-1.2 28 cm 9.8 kW-1.2 22 cm 11 kW-0.8 32 cm 9.8 kW-0.8 28 cm Characteristic Values Peak of zone length at the end (Lp2) mm/min 32 cm mm/min 28 cm mm/min 26 cm mm/min 23 cm

MinimumPeak of zone zone length length at the (L beginningmin) (Lp1) 28 cm 24 cm 22 cm 18 cm 32 cm 22 cm 28 cm 19 cm Peak of zone length at the end (L ) 32 cm 28 cm 26 cm 23 cm 1-(Lmin/Lp1) p2 14.3% 18.2% 31.2% 32.1% Minimum zone length (Lmin) 24 cm 18 cm 22 cm 19 cm min p2 1-(L1-(Lmin/L/L) p1) 14.3% 25.0% 18.2%35.7% 31.2%15.4% 32.1%17.4% Position1-(Lmin /Lof p2freezing) interface for Lmin 25.0% 24 cm 35.7% 21 cm 15.4% 49 cm 17.4% 50 cm Position of freezing interface for Lmin 24 cm 21 cm 49 cm 50 cm 3.3. Crystal Growth Rate Variation 3.3. Crystal Growth Rate Variation Zone length variation is the result of process window changes. The resulting changes of the Zone length variation is the result of process window changes. The resulting changes of the freezing interface leads directly to a variation of the crystal growth rate, which strongly affects the freezing interface leads directly to a variation of the crystal growth rate, which strongly affects effective distribution coefficient (keff). This can be seen in Figure 10 as the dependency of keff of the the effective distribution coefficient (k ). This can be seen in Figure 10 as the dependency of impurity, Fe, in aluminum on crystaleff growth rate [18]. This correlation is based on the keff of the impurity, Fe, in aluminum on crystal growth rate [18]. This correlation is based on the Burton-Prim-Slichter (BPS) model and shows a dramatic rise of keff while achieving the crystal Burton-Prim-Slichter (BPS) model and shows a dramatic rise of k while achieving the crystal growth growth rate of 1 mm/min. The following equation represents theeff effective distribution coefficient as rate of 1 mm/min. The following equation represents the effective distribution coefficient as well as well as the BPS model: the BPS model: k k kkeff==   (1) + ( − ) −Vδ−Vδ (1) k+k 1−k1 expk exp DD wherewhere kk isis the equilibrium distributiondistribution coefficient,coefficient, D is the impurity diffusion coefficient coefficient in the melt, δ thethe thicknessthickness ofof thethe diffusiondiffusion boundaryboundary layerlayer atat thethe solid/liquidsolid/liquid interface, and V is the molten zone movement velocity.

Figure 10. The relationship of keffeff withwith crystal crystal growth growth rate for th thee impurity, Fe, in aluminum [[18].18].

As mentionedmentioned inin SectionSection 3.1 3.1,, the the crystal crystal growth growth rate rate at at every every position position of of freezing freezing interface interface can can be attainedbe attained by by analyzing analyzing the the data data from from the the IR IR camera. camera. These These results, results, together together withwith thethe correspondingcorresponding impurityimpurity distributiondistribution profiles,profiles, areare shownshown inin FigureFigure 1111.. The crystalcrystal growth rate fluctuatedfluctuated constantly around the heaterheater movementmovement velocity (red dasheddashed lines).lines). The fluctuationfluctuation rangerange was especially large at thethe beginningbeginning ofof crystallizationcrystallization (around(around thethe firstfirst 1010 cm).cm). This was supposedlysupposedly because of the thermalthermal loss in thisthis regionregion duedue toto thethe “cold“cold neighborhood”neighborhood” at the beginningbeginning of the bar andand shortlyshortly before the heating starting area (see(see FigureFigure1 1).). ThisThis probablyprobably ledled toto aa fastfast solidificationsolidification oror aa veryvery highhigh crystal growth rate inin thisthis areaarea whilewhile departingdeparting thethe heatingheating coil.coil. In contrast, Figure 1111cc showsshows thethe lowest fluctuation scope in the region after 10 cm, which indicates that higher power and lower heater movement velocity could result in more stable crystal growth rate. Metals 2018, 8, x FOR PEER REVIEW 9 of 12

Materialslowest2018 fluctuation, 11, 2039 scope in the region after 10 cm, which indicates that higher power and9 oflower 12 heater movement velocity could result in more stable crystal growth rate.

(a) (b)

(c) (d)

FigureFigure 11. 11. Comparison between between the the concentration concentration distribu distributiontion profile profile of impurities of impurities and andthe crystal the crystalgrowth growth rate for rate different for different experiments: experiments: (a) 11 kW-1.2 (a) 11 mm/min; kW-1.2 mm/min; (b) 9.8 kW-1.2 (b) 9.8 mm/min; kW-1.2 (c mm/min;) 11 kW-0.8 (c)mm/min; 11 kW-0.8 (d mm/min;) 9.8 kW-0.8 (d) mm/min. 9.8 kW-0.8 mm/min.

Materials 2018, 11, 2039 10 of 12

These phenomena will surely have an influence on refining efficiencies. For all experiments, the high crystal growth rates at the beginning were related to higher impurity concentrations, i.e., lower refining efficiency. Almost each of the peaks or valleys in the crystal growth rate curves corresponded to a drastic increase or decrease in the concentration distribution curves. Surprisingly, a valley takes place at the region of around 60 cm in all cases; the concentration of solids should have continued increasing due to the high accumulation of impurities in the molten zone. This can be explained by the low crystal growth rate that existed in that region. Taking Figure 11a into consideration, when the freezing interface arrived at 50 cm, the crystal growth rate started to decrease. This was probably due to the fact that the heating dissipation was decreasing in that area as the melting interface reached the end of the crucible. However, the mass of graphite involved in the induction was kept the same, meaning the same power input to the molten zone was to be expected. Hence, a heat balance could have taken place at the freezing interface, resulting in its slow movement velocity. In contrast, when the freezing interface passed 57 cm, the melting interface had already arrived at the end of the crucible. Therefore, an increase of crystal growth rate happened, which was possibly due to the sudden disruption of the heat balance. However, the crystal growth rate soon decreased again, which was caused by the participation of the edge mass of graphite in the crucible in the electromagnetic induction and providing additional power. Furthermore, the overall graphite involved in induction sharply decreased, resulting in a drastic increase of crystal growth rate. In addition, as stated in Section 3.2, a greater zone length can dilute the impurity concentration more extensively, resulting in lower CL and hence lower CS values. However, a clear correlation between impurity concentration profiles and zone length curves cannot be established when observing Figures9 and 11. This could be due to the offset of the influence of crystal growth rate on refining efficiency. Nevertheless, Figure 11 generally shows that the higher power and lower heater movement velocity (Figure 11c) results in higher refining efficiency, while lower power and lower heater movement velocity (Figure 11b) results in the opposite. This is the joint result of the zone length (affected by power and movement velocity; see in Section 3.2) and the crystal growth rate.

4. Conclusions and Outlook An IR camera has been applied for the first time in zone refining of 2N8-pure aluminum with dopants Fe and Si in order to characterize and interpret the zone refining process. The relationships of the three characteristic positions of freezing/melting interfaces and the heater as well as zone length and crystal growth rate have been derived live through online thermographic analysis based on the temperature information near the molten zone. The influences of heating power and heater movement velocity on zone length variation and crystal growth rate, with respect to only one zone pass, could be derived. Furthermore, interpretations of impurity concentration distribution profiles have been conducted in correlation with the crystal growth rate. The main conclusions drawn are the following:

1. The application of an IR camera is a good way to control and characterize the zone refining process, which can derive the current position of the molten zone, the size of zone length and the actual crystal growth rate online. 2. Zone length varies with the tendency to firstly decrease and then increase for all power and heater movement velocity combinations (see Figure9). The percentage difference of zone length between maximum and minimum can be depressed by increasing the power. Lowering the heater movement velocity delays the freezing interface position where the zone length becomes minimal. 3. Effective refinement (50% to 75% Fe/Si-reduction) of aluminum happens for all experiments with only one zone pass, but higher power and lower velocity result in the highest refining efficiency. However, the impurity concentration distribution profile after zone refining fluctuates along the bar in each case. Materials 2018, 11, 2039 11 of 12

4. Crystal growth rate is not equal to the heater movement velocity and instead, fluctuates around the heater movement velocity. The fluctuations of the impurity concentration profile after refinement mainly results from the variation of crystal growth rate.

The large and unstable variation of zone length and crystal growth rate reveals the challenge to improve the refining efficiency during zone refining of good thermal conductors such as aluminum, especially when using inductive heaters. The realization of an online connection between the infrared camera and the control system of the equipment, which allows for automatization to adjust the power as well as the movement velocity in order to always keep an instant zone length, is strongly recommended for its huge potential to significantly improve refining efficiency.

Author Contributions: B.F. was the principal investigator. X.Z. and S.F. conceived of and designed the experiments. X.Z. performed the experiments. S.F. and X.Z. analyzed the data. X.Z. and S.F. wrote and edited the manuscript. Acknowledgments: The authors would like to thank CSC—China Scholarship Council for the financial support of Xiaoxin Zhang. Conflicts of Interest: The authors declare no conflicts of interest.

References

1. Cheung, N.; Bertazzoli, R.; Garcia, A. Experimental impurity segregation and numerical analysis based on variable solute distribution coefficients during multi-pass zone refining of aluminum. J. Cryst. Growth 2008, 310, 1274–1280. [CrossRef] 2. Hashimoto, E.; Ueda, Y.; Kino, T.; Hashimoto, E.; Ueda, Y.; Purification, T.K.; Purity, U.; Journal, A. Purification of Ultra-High Purity Aluminum. J. Phys. IV Colloq. 1995, 5, 153–157. [CrossRef] 3. Nakamura, M.; Watanabe, M.; Tanaka, K.; Kirihata, A.; Sumomogi, T.; Hiroaki, H.; Tanaka, I. Zone Refining of Aluminum and its Simulation. Mater. Trans. 2014, 55, 664–670. [CrossRef] 4. Yang, G.; Govani, J.; Mei, H.; Guan, Y.; Wang, G.; Huang, M.; Mei, D. Investigation of influential factors on the purification of zone-refined ingot. Cryst. Res. Technol. 2014, 49, 269–275. [CrossRef] 5. Ghosh, K.; Mani, V.N.; Dhar, S. A modeling approach for the purification of group III metals (Ga and In) by zone refining. J. Appl. Phys. 2008, 104, 024904. [CrossRef] 6. Spim, J.A.; Bernadou, M.J.S.; Garcia, A. Numerical modeling and optimization of zone refining. J. Alloys Compd. 2000, 298, 299–305. [CrossRef] 7. Burris, L., Jr.; Stockman, C.H.; Dillon, I.G. Contribution to mathematics of zone refining. J. Miner. Met. Mater. Soc. 1955, 7, 1017–1023. [CrossRef] 8. Pfann, W.G. Zone Melting, 2nd ed.; Wiley: New York, NY, USA, 1966. 9. Ho, C.; Yeh, H.; Yeh, T. Numerical analysis on optimal zone lengths for each pass in multipass zone-refining processes. Can. J. Chem. Eng. 1998, 76, 3–9. [CrossRef] 10. Rodway, G.H.; Hunt, J.D. Optimizing zone refining. J. Cryst. Growth 1989, 97, 680–688. [CrossRef] 11. Chang, C.E.; Wilcox, W.R. Heat transfer in vertical zone melting of poor thermal conductors. J. Cryst. Growth 1975, 28, 288–294. [CrossRef] 12. Kobayashi, N. Power required to form a floating zone and the zone shape. J. Cryst. Growth 1978, 43, 417–424. [CrossRef] 13. Roussopoulos, G.S.; Rubini, P.A. A thermal analysis of the horizontal zone refining of indium antimonide. J. Cryst. Growth 2004, 271, 333–340. [CrossRef] 14. Wang, J.H.; Kim, D.H. Numerical analysis of melt/solid interface shape in zone melting recrystallization process. J. Cryst. Growth 1997, 173, 201–209. [CrossRef] 15. Louchev, O.A. The influence of natural convection on the formation of a molten zone under optical heating. J. Cryst. Growth 1993, 133, 261–266. [CrossRef] 16. Munirathnam, N.R.; Prasad, D.S.; Sudheer, C.H.; Rao, J.V.; Prakash, T.L. Zone refining of cadmium and related characterization. Bull. Mater. Sci. 2005, 28, 209–212. [CrossRef] Materials 2018, 11, 2039 12 of 12

17. Curtolo, D.; Zhang, X.; Rojas, M.J.R.; Friedrich, S.; Friedrich, B. Realization of the measurement and control of zone length via implementation of an infrared camera during a zone melting refining process. Appl. Sci 2018, 8, 875. [CrossRef] 18. Zhang, X.; Friedrich, S.; Friedrich, B. Investigation of Influencing Parameters on Zone Melting Refining of , Part One Impu-rities: Iron and . Proc. EMC 2017, 1, 327–334.

© 2018 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).