42

The performance of a fused magnesium chloride-potassium chloride refining flux, in a paper presentedto the 2012 TMS Conference by John Courtenay*. Michael Bryant**, MQP, reviews.

www.aluminiumtoday.com

Aluminium International Today

Fused magnesium chloride-potassiumchloride refining fluxes are used as anenvironmentally acceptable means forremoving alkali metals and oxides frommolten aluminium than injection ofchlorine gas.

These refining fluxes1 were initiallybased on the classic binary systemmagnesium chloride-potassium chloride,which exhibits two low melting pointeutectics, one at about 55.5% molemagnesium chloride and another at36.5% mole magnesium chloride.

Later a revised binary diagram2 wasaccepted that showed three eutectics withthe two ‘classic’ eutectics and a thirdeutectic occurring at 31% mole.

Today, commercial products are suppliedbased on all three eutectics withmagnesium chloride contents rangingfrom the slightly hypo eutectic 25% byweight up to the hyper eutectic 75% byweight. In addition, a product withenhanced efficiency, containing 25%magnesium chloride with potassiumchloride and a small amount of alkalimetal fluoride is also available.

EconomicsThe major cost factor in the production offused refining fluxes is raw materials andin particular the cost of potassiumchloride. This first became an issue in2008/2009 as demand for potash forworld food production and bio-fuels

increased. Prices are starting once again tomove upwards (Fig 1).

The rising price of potassium chlorideand the likelihood of further increases hasgiven impetus to a programme aimed atdeveloping an alternative flux where thepotassium chloride is partially replacedwith sodium chloride.

This paper summarises the results of athermodynamic study together withlaboratory measurement of viscosity, anddifferential thermal analysis and sodiumremoval casthouse trials of a flux productwhere 25% of sodium chloride isintroduced to replace potassium chloride.

Developing a new fluxIn constituting a new flux it was importantto understand some of the thinking thatoriginally went into devising a refining flux.

Two conventional wisdoms were inplace. The first held that performance interms of sodium removal would be higheras the percentage of magnesium chlorideincreased. Second, that the amount ofsodium chloride permitted in the productmust be below 1%. In reality neither ofthese beliefs is correct as demonstrated bythe theoretical arguments and evidencefrom the series of results andinvestigations presented below.

Influence of % magnesium chlorideon efficiencyTest data was collected under controlledconditions at a number of casthouses.Comparisons were made between a40%MgCl2 and a 60-65% MgCl2

containing product applied manually intwo identical 50t furnaces at a smeltercasting 5xxx alloy.

Results showed the 40% MgCl2 productperformed slightly better than the 60-65%product. In a separate series of tests,results were obtained from production usefor a 40% MgCl2 product and a 25%MgCl2 product at a smelter casthouse. Theaverage % sodium removal for the 40%MgCl2 product was 76% while the resultfor the 25% MgCl2 product was 77%sodium removal.

More recently a study of MgCl2 fusedsalt reagents applied in the salt flux ACD,showed there was no significantdifference in alkali removal efficiency witha 60% MgCl2 or a 75% containingcomposition. Further confirmation camefrom results of an investigation carried outat the Alcoa Technical Centre3 showingthat varying the % MgCl2 between 10%and 90% had no influence on the rate ofsodium removal (Fig 2)

All the evidence from practicalevaluations and published research fromdifferent sources shows alkali removalefficiency is not influenced by the % ofMgCl2 in the fused salt, contrary to somewidely held views.

In an explanation of this Dietze2 hasproposed the concentration of MgCl2 inthe molten salt droplet has little influenceon the kinetics because the rate of saltaddition applied in practise is ten timesthat needed to satisfy the requirement forstoichometric reaction and therefore thereis always an excess present.

A fused magnesium chloridecontaining refining flux

CASTHOUSE

May/June 2012

Fig 1 Price $/t for KCl from 2000-2011 Fig 2 Effect of Magnesium Chloride content in flux on Sodium removal

*Managing Director, MQP**Marketing Manager, MQP

KCI (Spot)

FOB Vancouver1000900800700600500400300200100

00 01 0302 04 05 06 07 08 09 10 11

0

4000

% Mga2 in salt

3000

2000

1000

00 20 40 60 80 100

With salt - normal stirring

With salt - slow stirringWith salt - normal stirring

Na

rate

con

stan

t (cm

3/ s)

mqp_30_AIT_0110 5/29/12 2:50 PM Page 31

May/June 2012 Aluminium International Today

44 CASTHOUSE www.aluminiumtoday.com

Influence of sodium chloride onsodium removalIn terms of cost NaCl would be welcomeas a substitute or partial substitute for KClin fused salts.

The reaction MgCl2 + 2 Na = 2NaCl+ Mg can be considered to move stronglyto the right. If this was not the case thenthe sodium removal process by MgCl2would not be practically effective.Ellingham diagrams show the highstability of NaCl with respect to MgCl2.Therefore it was anticipated there wouldbe no increase in sodium in the aluminiumif quantities of NaCl were introduced intothe flux composition.

The above hypothesis was tested bythermodynamic modelling at IMEAachen4, under Prof Friedrich.

FactSage databases were used toinvestigate the effect of NaCl content inRefinal on the reaction between MgCl2 inRefinal and Na from the aluminium melt.

In terms of selecting the product,composition consideration was given toselecting a ternary composition, with anappropriate addition of NaCl to an existingproven formulation (Fig 3).

The eutectic low melting point area ofthe system can be seen on the diagramand the composition selected was:

– Magnesium Chloride 37-42%– Potassium Chloride 21-26%– Sodium Chloride 27-32%

The thermodynamic study was based onthe following conditions:A standard salt flux composition with 35%MgCl2 + 65% KCl. Additions of 1%; 5%; 10% or 25% NaClinto the flux.

Viscosity was measured by oscyllographicviscometer in conjunction with densitymeasurement, and confirmed with a highdegree of probability the following initialconclusions:

Refinal 352XF has a higher viscosity thanRefinal 350 or 555XF.

Refinal 352XF has a higher change-rateof the viscosity than Refinal 350 or 555XF.The viscosity of all fluxes becomes equal >600°C. The precision of the absolutevalues is difficult to estimate.

Since the viscosities, although quitedifferent at low temperatures, becomeequal at >600ºC it can be determinedthere is no negative effect from theaddition of NaCl as a ternary addition tothe standard binary product in terms ofviscosity.

Differential thermal analysisDTA measurement was carried out on theternary composition, Refinal 555XF andthe result showed a sharp peak at 402.7ºCin the heating up cycle and 405.6ºC whencooling down. The sharpness of the peakindicates eutectic melting (Fig 4).

This data was then compared with theDTA curve for a typical binary composition,Refinal 350 which showed a similar sharppeak at 438.8ºC.

The conclusion was that both productsexhibited sharp eutectic melting pointswith the ternary composition having alower melting point by 30ºC and thatthere was no negative effect from theaddition of NaCl as a ternary addition interms of melting characteristics.

Trials in a casthouseWith the favourable outcome ofthermodynamic calculations and

Starting with an initial Na content of30ppm or 10ppm in the melt andtargeting a final Na content of 10ppm or2ppm respectively.Application of 0.05 – 0.1 % Refinal basedon melt weight.Melt temperatures of 700 and 750°C.

The modelling results showed that aftercompletion there was no sodiumremaining in the melt. Therefore it can beconcluded that up to 25% NaCl can besubstituted for KCl in Refinal 350 andRefinal 352XF without effect on theresidual Na content after treatment.

Given that KCl plays no active part in thealkali removal reaction, substitution orpartial substitution with an alternativestable alkali metal chloride should notinfluence the alkali removal kinetic.

Viscosity measurementsThe objective of the work was to get abetter understanding of the variation inviscosity with temperature andcomposition of the flux with a view toconnecting this with performance inmolten aluminium.

In the context of performance it hasbeen proposed that the beneficialinfluence of fluoride additions to the flux,as in Refinal 352XF, is due to its propensityto reduce interfacial surface tension.

This leads to the formation of smallerdiameter liquid salt flux droplets giving riseto an increase in the surface area forreaction. Prof Friedrich believes thepresence of F ions in Refinal 352 is thereason for the better performance due toF ions reacting with Na better than Cl ions.

The following compositions werestudied in (Table 1):

Fig 3 Ternary system MgCl2, KCl, NaCl showing new salt flux composition

DTA/u/V

exo5

0

-5

-10

-15

-20

-25

-30100 200

Peak: 140.5°C

Red: 1. Aufheizen

Peak: 405.6°C

Peak: 190.1°C

Peak: 372.3°C

Blue: 2. Aufheizen

Peak: 402.7°C

300 400 500 600 700

[1.3]

Fig 4 DTA curve forRefinal 555 XF

Table 1 Compositions

Refinal 350 30-35 60-65 1-3Refinal 352XF 30-35 60-65 1-3 1-3Refinal 555XF 36-41 21-26 26-31 1-3Refinal 750 40-45 55-60

Product MgCI2 % KCI % NaCI % CaF2 %

mqp_30_AIT_0110 5/29/12 2:50 PM Page 32

45www.aluminiumtoday.com

Aluminium International Today

laboratory tests it was decided to proceedwith full-scale casthouse trials at a majorcasthouse in Europe.

Trials were carried out by substituting thetrial product, Refinal 555XF into standardpractice and comparing results to thoseobtained with standard practice using a35% MgCl2 – 65% KCl salt flux.

During 87 furnace preparations anaverage Na-level of 6.5ppm at first time ofbatching was achieved.

The conclusion was the results werecomparable to their standard practice ofapplying Refinal 350.

Subsequently Refinal 555XF has beenadopted into regular production with over100kt of aluminium having beensuccessfully produced.

Summing upNot for the first time widely held beliefshave been challenged by technicalinvestigation.

In the case of the effect of MgCl2content reaction kinetics are of overridingimportance and simply matching thestoichometrically required amount doesnot ensure that the reaction goes tocompletion.

In the case of the effect of NaCladditions, here the thermodynamicsconfirm that, irrespective of reactionkinetic considerations, NaCl cannot be

reduced again to Na. Furthermorelaboratory characterisation of the ternarycomposition has confirmed that there is noadverse effect from the NaCl addition interms of both the viscosity and meltingcharacteristics of the product.

Conclusions– A fundamental study involvingthermodynamic modelling databases hasdemonstrated it should be possible to addup to 25% of NaCl to a MgCl2 – KCl saltflux by substituting KCl without any effecton the residual Na content in the treatedaluminium melt.– A composition, corresponding to theternary eutectic in the MgCl2, KCl,NaClsystem has been produced, characterisedin laboratory testing and trialled on aproduction scale in a large casthouse inEurope.– The results of the laboratorycharacterisation confirmed there were noadverse effects from the ternary addition ofup to 25% NaCl to a MgCl2 – KCl binarycomposition in terms of viscosity andmelting characteristics.– The casthouse results confirmed thethermodynamic prediction and showed asatisfactory performance in terms ofsodium removal, compared with standardpractice, over a monitored series of 87production casts. �

AcknowledgementMQP would like to thank Mr MatthiasRohmann and the management of HDWRheinkalk GmbH for their technical andfinancial support and Professor Friedrichand M.Sc Semiramis Akbari of RWTHAachen for undertaking thethermodynamic study and laboratorycharacterisations.

References(1) G.Beland, C. Dupuis and J.-P. Martin,“Improving fluxing of Aluminium Alloys” LightMetals, 1995, 1189-1195(2) Private communication with Dr. A. Dietze. TUClausthal (3) D.H.DeYoung, “Salt Fluxes for Alkali andAlkaline Earth removal from Molten Aluminium”,7th Australian Asian Pacific Conference AluminiumCasthouse Technology, 2003 (4) Private communication with Prof. Dr.-Ing. BerndFriedrich and M. Sc. Semiramis Akbari, RWTHAachen(5) A.Silney and T.A.Utigard, “Interfacial Tension

between Aluminium, Aluminium Alloys andChloride-Fluoride Melts”

ContactMQP Ltd, 6 Hallcroft Way, Knowle, Solihull, UK, B93 9EW, Tel +44 (0) 1564 200 443Email john.courtenay@mqpltd.comWeb www.mqpltd.com

CASTHOUSE

May/June 2012

mqp_30_AIT_0110 5/29/12 2:50 PM Page 33