CROSS-REFERENCE TO RELATED APPLICATIONS

This Patent Appl ication claims priority from Italian Patent Application No . 102022000001697 filed on February 1 ^st , 2022 , the entire disclosure of which is incorporated herein by reference .

TECHNICAL SECTOR

The present invention refers to the field of metallurgy . In detail , the present invention refers to a process for reducing the magnesium content from liquid aluminum alloys . Particularly, the present process concerns liquid aluminum secondary alloys .

DESCRIPTION OF THE PRIOR ART

Aluminum is one of the most widely used metals in the world and, due to its extraordinary combination of product properties , has a wide range of applications in multiple industrial sectors .

An important source of aluminum supply is constituted by the secondary aluminum industry, which is based on the recycling of obsolete aluminum products and, above all , waste and scrap from the industrial processing of aluminum and its alloys .

However, to date secondary aluminum does not meet purity requirements comparable to those of primary aluminum .

For example , aluminum alloys and speci fically silicon- based aluminum al loys used in die casting require a maximum magnesium content of 0 . 1 % . However, often, aluminum scrap to be recycled has a magnesium content of up to 5% , from which secondary aluminum is produced that is not suitable for use in an aluminum al loy for die casting, the magnesium content being higher than the required standards .

In this regard, the problem of removing magnesium from an aluminum bath has been extensively addressed and has produced multiple solutions , but all of them, as will become apparent later, have signi ficant limitations .

The simplest method for controlling the level of magnesium is to mix appropriately aluminum scrap characteri zed by a high magnesium content with those with a low content .

However, the lack of availability of scrap with low magnesium content , associated with their excessive cost , has stimulated the development of chemical operating practices , based on the introduction of speci fic agents that , favouring the formation of more stable and insoluble magnesium compounds in the liquid metal , with respect to the compounds with aluminum, facilitate their separation by sedimentation or decantation according to their density .

The aforesaid practices concerning the introduction of speci fic agents , such as fluxes or other, and aimed at eliminating magnesium from an aluminum bath include :

• gaseous oxidation with oxygen, with air, oxygen/neutral gas mixtures ( e . g . argon) and with carbon dioxide ;

• chlorination;

• the use of fluoride-based flow agents .

Looking at the individual methods listed, their limitations will become apparent .

The removal of magnesium by oxidation, given the high af finity of magnesium for oxygen, results in the formation of magnesium oxide (MgO) and may appear to be the simplest method . However, even at elevated temperatures of the order of 900-1000 °C and prolonged treatment times, magnesium elimination is not satisfactory. In addition, due to the mass effect, the process is accompanied by significant losses of aluminum due to its oxidation to alumina (AI2O3) , so this method has been almost abandoned.

To reduce aluminum losses, air, oxygen or mixtures thereof have been replaced with a mildly oxidising agent, such as carbon dioxide (CO2) . However, even this oxidative treatment did not produce satisfactory purity rates of the secondary aluminum.

Alternatively to oxidative methods, chlorination promotes the selective reaction between chlorine and magnesium thus allowing magnesium to be removed from the alloy in the form of magnesium chloride (MgC12) . Chlorine is introduced into the melt by direct injection of gas, or as a chlorinated organic product.

The objective limitations of this treatment include the formation of significant amounts of gaseous aluminum chloride (AICI3) , in addition to the high cost, both in terms of reagent and plant engineering necessary to reduce the risks associated with the use of this toxic gas, extremely harmful to human health and the environment (chlorine is the basis of the formation of dioxins) and highly aggressive for plant engineering.

Other types of treatments use fluoride-based flow agents. With respect to the series of chlorides, the formation of fluorides and, specifically, magnesium fluoride is more favoured at process temperatures.

In practice, many treatments for removing magnesium in the form of fluoride use commercial products based on mixtures containing aluminum fluoride , such as , for example , flows based on cryolite (NasAlFs) or potassium fluoridealuminate (K3AIF6) .

However, the undoubted ef fectiveness of the treatments with fluorinated salts clashes heavily with environmental aspects : the high vapour pressure of these salts and their ability to react with water vapour, always present in combustion fumes , to give hydrofluoric acid (HF) , results in decidedly high emission values of this acid .

To cope with the limitations of the methods listed above , the secondary aluminum industry has developed alternative processes : one of these uses silica-based reagents ( SiCh ) . The ef fect of si lica on magnesium removal is mainly through the final formation of the spinel (MgA12O4 ) • Although the silica-based process has a number of advantages such as the absence of vapours or pollutants in the magnesium removal process , this has not yet been implemented in an industrial environment but only on a laboratory scale .

It emerges from the technical literature that blowing quartz powder on the bottom of a crucible with an inert gas , eliminates up to 90% of magnesium present in j ust over an hour of process ( about 75 minutes ) , at temperatures ranging from 750 to 800 ° C . The best results were obtained at higher temperatures , with small particle si ze quartz and operating with blowing times that favour greater rise times of the quartz particles in the liquid bath . However, the kinetics of magnesium removal turn out to be very slow . To speed up the reaction rate , some studies have proposed to operate with a bath at temperatures exceeding 900 ° C, by preheating the quartz powder up to 300 ° C before being dosed with an intense stream of argon to favour mixing.

To improve the reaction kinetics, processes based on zeolites (containing more than 50% by weight of SiCt) which are often doped with silica powder and injected into the liquid alloy by means of an inert transport gas are known. In fact, with respect to quartz, zeolites have larger surface areas that speed up the reaction kinetics.

By way of example, the following are also some patent proposals, of which, however, no industrial applications or developments are known.

The Russian patent RU 231802901 presents a method to be applied in general to the metallurgy of non-ferrous metals, and particularly to aluminum alloys, for refining them from gases, oxides and other non-metallic inclusions present. The patent proposes to mix a salt stream consisting of KC1, NaCl and NasAlFs, of very variable composition, with aluminum and silicon refractory oxides. The reactant mixture is prepared in the pasty state by heating it up to a temperature of 730 °C, after which it is fed to the melt to be treated, without however indicating the system thereof. According to this method, the stream interacts with the alloy purifying it from the dissolved gases and non-metallic inclusions. The system, therefore, is not aimed at the de- magnesization treatment, but in general at a generic alloy refining treatment to clean it from gases and inclusions, and not specifically from magnesium or other metals. As a "refractory" reagent, the process uses meta-kaolinite calcined at over 600 °C (preferably above 800 °C) , a heat treatment that is necessary to remove both the moisture and, above all, the water of crystallization always present in its molecule, given its chemical composition: A12Si20s (OH) 4. However, the Russian patent presents a whole series of complications and contraindications , the main ones including : it uses cryolite , a double salt , which by dissociating hot , releases fluorine vapours causing serious environmental damage ; it preheats the reagent mixture composed of salts and meta-kaolin, bringing it to the semimolten state , before being fed to the bath, which in addition to weighing energetically on the treatment costs , entails dangerous plant complications in the steps of subsequent dosing on the bath ; it uses meta-kaolin, a material that in addition to being obtained from quarries , with an obvious depletion of natural resources , must be prepared with an appropriate calcination treatment in order to take away the water of crystalli zation present in its molecule ( to avoid dangers of explosions and hydrogen formation i f it reaches the molten bath) , a process decidedly costly from an energy point of view; among other things , the calcined meta-kaolin, being a double oxide ( 2SiO 2* 120s) is decidedly more stable, with respect to the respective simple oxides and therefore less reactive ; being a process of refining and not of de- magnesi zation of the bath, the dosage of the treatment stream, both as to quantity and quality, referred to the ratio salt mixture and meta-kaolin, depends only on the type of alloy and not on the level of magnesium . The process of the Russian patent , therefore , in addition to not being speci fically dedicated to the removal of magnesium from the liquid aluminum bath, is decidedly costly for the reasons described above , and above all , not very ef fective in the magnesium removal treatment , in addition to presenting obj ective management di f ficulties . US 4097270A discloses a system for reducing magnesium from an aluminum alloy, with an element content of up to 10% , by supplying a solid reagent , for example silica . But the di f ficulty of incorporating the solid particles , which tend to float in the melt characteri zed by a higher density, without being "wetted" by it , forces to operate with strong excesses of reagent ; and the exces s of reagent , in addition to increasing the slag produced, also reacts with aluminum to form alumina, consequently reducing the melting yield . In addition, precisely to increase the limited reactivity of the system, it is proposed to " activate" the reagent , for example by preheating the silica, or to feed the already partially reacted reagent . The strong limitations of the method are obvious .

The Japanese patent application JP 2010275620A proposes a method for reducing the magnesium content of liquid aluminum alloy by adding a powder of volcanic origin, present in Japan in large quantities . It is a very light material , based on silica and alumina, and fed with a si ze of less than 600 pm . However, the material must be prepared, dried to remove moisture , screened to select the desired si ze , and it also contains non-negligent percentages of Fe2Os, which are harmful to the final composition of the bath ( it should be noted that Fe is often a harmful element in Al alloys ) . The system, however, in addition to the limitations dictated by the type and preparation of the reagent , has important kinetic limitations .

However, as already well highlighted, the proces ses currently known for removing magnesium from liquid aluminum alloys with quartz-based reagents , or its compounds , such as meta-kaolin, zeolites or other materials , present obvious problems : most of them need a neutral gas for the transport of the reagent with consequences both on the cost-ef fectiveness of the process , given by the need to use external thermal energy to compensate for the enthalpic losses due to the use of the gas , and on the plant engineering, given by the need to use unconventional furnaces with liquid baths with high masses in order to allow adequate contact times between the particles of the rising reagent and the alloy, the cost of the reagents , mainly in the case of the zeolites or meta-kaolin, and the health hazard, for example with mani festations of silicosis caused by inhalation of powders containing crystalline silica, such as quartz , the need to operate with strong excesses of reagent with respect to stoichiometric needs , without however achieving the zeroing of the final magnesium content , and with consumption of the same aluminum, the ef fectiveness as a function of initial magnesium content : higher ef ficacy for alloys characteri zed by a high magnesium content , the di f ficulty of cal ibrating the feed of the reactant powder, or any mixtures thereof , the need to pretreat the reagents (physical finali zation and/or doping, preheating, calcination) which entails considerable additional costs in addition to speci fic dedicated plants , the di f ficulty of separating the reaction product , for example by adhering the spinel to the bottom of the crucible , the absence of industrial-scale process implementations .

OBJECT OF THE INVENTION

It is therefore an aim of the present invention to provide a process for reducing the magnesium content from a liquid aluminum alloy, particularly from a liquid secondary aluminum alloy, which allows to overcome the limits previously set out and which is ef fective , industrially applicable , fast , economical , and in full respect of the environment and the health of the operators .

In accordance with this aim, the present invention relates to a method according to Claim 1 .

The present method is particularly advantageous in that it uses amorphous silica as an oxidi zing reagent which, with respect to other silica-based reagents ( Si02 ) in crystalline form, such as , for example , quartz powder or other compounds thereof , does not cause environmental and health problems for operators .

In addition, the micrometric si zes of the amorphous silica particles guarantee a greater reactivity of the reagent : the reaction kinetics between silica and magnesium is in fact favoured thanks to the high speci fic surface of the micrometric-si zed particles without any need to increase working temperatures and treatment times .

It follows that , contrary to the use of quartz or other compounds thereof , the process of the present invention does not require preliminary preparations of the reagent ( such as , for example , physical finali zation, drying, calcination, preheating and/or doping) adapted to increase its reactivity, with obvious bene fits in economic, plant simpli fication and operating speed terms .

In addition, given the high reactivity of the reagent , the present invention allows to optimi ze the yields of the treatment , operating with additions of the reagent in stoichiometric or even slightly sub-stoichiometric quantities , thus reducing consumptions and there fore the costs of the treatment , as well as avoiding the oxidation of the aluminum itsel f .

Advantageously the micrometric amorphous silica used in the process of the invention is fed to the liquid aluminum alloy without the need to employ an expensive transport gas . This results in a reduction in the enthalpic losses with consequent energy savings . In addition, the restrictions related to plant engineering are no longer applicable , for example , the height of the liquid bath is now irrelevant to the reaction kinetics . According to the latter aspect , the process of the invention is easily applied to current industrial furnaces , generally characteri zed by liquid baths with low masses , without any need for modi fication or integration of the feeding systems , thus being able to operate on important quantities of alloy .

A further advantage of the process of the invention over the prior art is represented by the fact that its ef fectiveness is independent of the initial magnesium content , thus being able to achieve the zeroing of the magnesium content starting from liquid aluminum alloys with low magnesium content .

BRIEF DESCRIPTION OF THE DRAWING

Further characteristics and advantages of the present invention will become apparent from the detailed description given below, from the examples provided for illustrative and non-limiting purposes and from Figure 1 , wherein :

Figure 1 shows the granulometric distribution of a micrometric amorphous silica used in the process according to an embodiment of the invention .

DETAILED DESCRIPTION OF THE INVENTION

The present invention describes a process for reducing the magnesium content from a liquid aluminum alloy, particularly from a liquid secondary aluminum alloy, comprising the steps of :

( a ) placing said liquid aluminum alloy into a reaction chamber maintained at a predetermined reaction temperature to form a liquid bath in the reaction chamber ;

(b ) adding a salt mixture to the reaction chamber forming a contact layer with said liquid aluminum alloy covering at least a free surface of said liquid aluminum alloy;

( c ) reacting a micrometric amorphous s ilica dispersed in said contact layer with said liquid aluminum alloy; and

( d) slagging said contact layer incorporating magnesium- based reaction products .

The process o f the invention is implemented particularly, but not necessarily, for liquid aluminum-silicon alloys .

Among the steps of the process according to the invention, step ( c ) relates to the most important reaction for reducing the magnesium content from a liquid aluminum alloy . In fact , step ( c ) consists of reacting a silica ( SiCt ) , essentially a micrometric amorphous silica, with the liquid aluminum alloy, obtaining as the main reaction product the spinel (MgA12O4 (s) ) , according to the following reaction ( 1 ) : Mg ₍ i) + 2Al ₍ i) + 2SiO ₂ (s) -> MgAl ₂ O _4(s) + 2Si ₍ i) (1)

The stoichiometric ratio between silica and magnesium is 2:1. In other words, 4.94 Kg of SiO ₂ per each Kg of magnesium are stoichiometrically consumed.

The aforementioned reaction (1) , at process temperatures ranging from 700 °C to 800 °C, typically 750 °C, is characterized by a free energy equal to AG°75o °c = - 85.12 Kcal and by an enthalpy equal to AH°75o °c = -99.95 Kcal, it being therefore a reaction, from the thermodynamic point of view, decidedly favourable to its development and exothermic .

Given the significant exothermicity of the reaction (1) , the process of the invention can advantageously be carried out in the absence of external thermal energy.

Next to reaction (1) , step (c) of reacting silica, essentially micrometric amorphous silica, with the liquid aluminum alloy also leads to the formation of magnesium oxide (MgO(s>) , according to reaction (2) :

2Mg ₍ i) + SiO ₂ ( _S ) -> 2MgO( _S ) + Si ₍ i) (2)

In this case, the stoichiometric ratio between silica and magnesium is 1:2. In other words, 1.40 Kg of SiO ₂ per each Kg of magnesium are stoichiometrically consumed.

The aforementioned reaction (2) , at process temperatures ranging from 700 °C to 800 °C, typically 750 °C, is characterized by a free energy equal to AG°75o °c = - 42,818 Kcal and an enthalpy equal to AH°75o °c = -66.08 Kcal. Based on the thermodynamic values, the reaction (2) is less thermodynamically favoured than reaction (1) .

Therefore, step (c) of the process of the invention leads to the formation of two magnesium-based reaction products: the spinel and magnesium oxide. The latter in decidedly reduced amount with respect to the amount of spinel given the considerable difference in free energy between the reaction (1) and the reaction (2) .

In addition to the above discussed thermodynamic aspects, step (c) also suggests considerations of kinetic type .

The use of micrometric amorphous silica favours the overall kinetics of the reaction (1) , increasing the speed of the kinetically limiting step given by the diffusion of magnesium up to the surface of the micrometric amorphous silica. In fact, the micrometric sizes promote a high specific surface of the amorphous silica, with consequent increase in the number of collisions between magnesium and the exposed surfaces of silica.

Given the low concentration of magnesium in the liquid aluminum alloy, as well as the thermodynamically favourable operating conditions, the kinetically determined step of the reaction is the diffusion of magnesium from the inside of the liquid aluminum alloy up to the surface of the outer silica .

It is known that a chemical reaction controlled by diffusive processes is a function of a diffusion coefficient that depends on many factors, among which, the most important are the relative speed between silica and magnesium that is contained in the liquid alloy, influenced by stirring, the events at the surface of the silica, influenced by the size of the amorphous silica particles, and the properties of the liquid alloy.

The process of the invention acts mainly on the events at the surface: the micrometric sizes entail a strong increase in the specific surface and therefore an increase in the reactive shocks . In this way, the kinetics of the process increase without the need for an increase in the temperature and in the reaction times with energy and plant benefits .

The reaction kinetics are also af fected by the continuous stirring of the liquid alloy, as will be detailed below .

Based on the considerations of thermodynamic and, above all , of kinetic type , the micrometric amorphous silica used in the process of the invention is a highly reactive material and, consequently, it is not necessary to operate with excesses of micrometric amorphous silica with respect to the stoichiometric needs , but the micrometric amorphous silica is measured stoichiometrically or sub- stoichiometrically with respect to the reaction ( 1 ) .

Operating in excess of micrometric amorphous silica is even counterproductive in that , after the initial formation of magnesium oxide due to the oxidation of magnesium, its stabili zation as a spinel , in the presence of excesses of silica, as well as with alumina, could also be carried out by the same silica, with the formation of the compound MgO* SiO2 , or preferably of 2MgO* SiO2 , leading to an increase in the amount of slag, and conversely, a decrease in the available reagent .

The micrometric amorphous silica employed in step ( c ) of the process of the invention has average si zes ranging from 0 . 01 pm to 1 pm, preferably ranging from 0 . 03 pm to 0 . 3 pm .

Preferably, the micrometric amorphous silica taking part in step ( c ) of the process of the invention is a microsilica ( also known as silica fume ) . Microsilica is an ultrafine powder recovered as a byproduct of the production of metallic silicon or iron-silicon alloys in electric submerged arc furnaces. Being therefore a by-product, the use of microsilica in the process of the invention advantageously contributes to the costeffectiveness of the process given the low cost of the reagent, while respecting the circular economy. Microsilica particles are spherical in shape and have the following chemical-physical characteristics : average sizes ranging from 0.03 pm to 0.3 pm, specific area ranging from 13 m ² /g to 30 m ² /g, silica content ranging from 85% to 98% based on the production process from which the microsilica is recovered. For example, in the case of production of metallic silicon, microsilica having a silica content ranging from 93% to 98% with respect to the other oxides present in the microsilica (e.g. calcium oxide) is obtained.

By way of example and considering the reaction (1) , the consumption of microsilica, which is variable according to its silica content, is ranging from 4.5 kg to 5.5 kg per kg of magnesium therefore close to or even lower than stoichiometric .

Given the lower density of the micrometric amorphous silica than the liquid aluminum alloy, silica tends to float on the surface of the liquid alloy to the detriment of its reactivity .

For example, microsilica has a density ranging from 130 to 430 kg/m ³ (depending on the possible densif ication process) with respect to a density of liquid aluminum alloy equal to 2400 km/m ³ . In addition to the low density, micrometric amorphous silica also has a low wettability that favours the aggregation of si lica on the surface of the liquid aluminum alloy, decreasing its reactivity with consequent penali zation on the reaction kinetics and on the reaction yields .

To obviate these drawbacks , the process of the invention comprises step (b ) of adding a salt mixture to the liquid aluminum alloy preceding step ( c ) of reacting the micrometric amorphous silica with the aluminum alloy .

The salt mixture acts as a surfactant additive decreasing the surface tension of the liquid aluminum alloy and thus the energy at the interface .

The salt mixture then forms a molten contact layer on the surface of the liquid aluminum alloy, on which the micrometric amorphous silica, added in the subsequent step ( c ) , is homogeneously dispersed and comes into contact with the liquid aluminum alloy . In this way the reaction between micrometric amorphous silica and magnesium, according to the reactions ( 1 ) and ( 2 ) , is optimi zed .

Preferably, the salt mixture comprises mixtures of potassium chloride (KC1 ) and sodium chloride (NaCl ) , without any addition of fluorinated salts or other low-boiling compounds or potentially source of pollution, for the problems already described . The weight composition of the salt mixture is irrelevant to the result of the process of the invention .

More preferably, the salt mixture is rich in potassium chloride or comprises only potassium chloride . In fact , potassium chloride is a salt that is low viscous at operating temperatures and therefore favours a better wettability of the micrometric amorphous silica .

At process temperatures ranging from 700 ° C to 800 ° C, typically 750 ° C, the salt mixture is slightly volatile , and this allows to keep its quantity substantially unchanged during the process .

In addition to favouring a homogeneous dispersion of the micrometric amorphous silica, the contact layer on the surface of the liquid aluminum alloy contributes to reducing the aluminum losses , minimi zing oxidation of the liquid alloy .

The contact layer in addition to housing the homogeneously dispersed micrometric amorphous silica, incorporates the magnesium-based reaction products resulting from step ( c ) of the process of the invention .

The magnesium-based reaction products comprise the spinel and, to a lesser extent , magnesium oxide , which, being denser than liquid aluminum alloy ( the density of the spinel is equal to 3600 kg/m ³ and the density of the magnesium oxide is equal to 3580 kg/m ³ ) , could not be easily removed on the surface of the liquid alloy .

Therefore , the concluding step of the process of the invention relates to step ( d) of slagging the contact layer incorporating the magnesium-based reaction products .

Optionally, step ( d) is followed by a step of separating the salt mixture and the magnesium-based reaction products from the contact layer .

For example , the step of separating the salt mixture and the reaction products from the contact layer may take place by leaching in water .

The magnesium-based reaction products thus separated from the salt mixture can find easy use in various sectors , while the salt mixture can be reused in the process of the invention .

Steps (b) to (d) of the process of the invention are carried out in a reaction chamber. In fact, prior to step

(b) , the process of the invention comprises step (a) of placing the liquid aluminum alloy into the reaction chamber maintained at a predetermined reaction temperature.

The predetermined reaction temperature is ranging from 700 °C to 800 °C, preferably ranging from 750 °C to 800 °C, more preferably at 750 °C, corresponding to the temperature at which the aluminum alloy reached the melting state in a step prior to the process of the invention.

The reaction chamber preferably comprises temperature sensors (e.g. immersion thermocouples or optical pyrometers) and heating means (e.g. service burners) . The latter are preferably operated when the temperature sensors detect temperatures below 750 °C. In fact, it is generally not necessary to keep the heating means active during the steps of the process of the invention as the exothermicity of the reaction (1) , as well as of the reaction (2) , is sufficient to compensate for the thermal dissipations of the system with benefits in energy terms.

Preferably, the reaction chamber is a conventional or crucible industrial furnace. More preferably, an either tubular or rotary or basin tubular furnace.

As already anticipated during the discussion of the kinetic aspects of the reaction (1) with reference to step

(c) of the process of the invention, in addition to the micrometric sizes of the amorphous silica, the kinetics of the reaction (1) may be further favoured, preferably keeping the liquid bath stirring. In fact, a continuous stirring of the liquid aluminum alloy acts on the di f fusive processes and, speci fically, on the relative speed between micrometric amorphous silica and magnesium, which characteri ze the kinetically limiting step of the reaction ( 1 ) .

Preferably, the liquid bath is kept stirring by rotating the reaction chamber itself or through induced currents or other systems in the case of a static reaction chamber . For example , stirring is achieved by activating stirring means included in the reaction chamber itsel f ( for example electromagnetic stirrers in the case of basin furnaces or series of radially arranged mixers in the case of a rotary tubular furnace ) or gas insuf flation .

Optionally, the aluminum alloy can be chemically characteri zed to determine the amount of magnesium and to establish the amount of magnesium that needs to be reduced . In fact , although the process of the invention can zero the amount of magnesium, it may be advantageous to reduce the magnesium content slightly below the limit value al lowed by the analytical composition of the alloy to be produced as the residual magnesium protects the aluminum from oxidation .

Relatively to the amount of salt mixture to be added to step (b ) , this depends on two parameters : the si zes of the reaction chamber which determines the free surface of the liquid aluminum alloy to be covered entirely with salt mixture , the amount of magnesium to be reduced which determines the amount of micrometric amorphous silica to be reacted in step ( c ) and consequently the amount of magnesium-based reaction products in the form of slag ( step ( d) ) . However, the thickness of the contact layer formed by the salt mixture varies from a minimum ranging from 1 mm to 2 mm, preferably 1 mm, a height necessary to give physical continuity to the contact layer, in the case of furnaces with large free surfaces , like in the case of basin furnaces , to a maximum ranging from 3 cm to 5 cm, preferably 5 cm, in the opposite case instead of crucible furnaces , characteri zed in contrast , by very limited free surfaces . In the case of rotary tubular furnaces , being characteri zed by free surfaces with intermediate si zes with respect to the previous limit cases , the height of the contact layer will be characteri zed by intermediate values , and in any case always proportional to the amount of Mg to be reduced, in order to correctly manage a complete "wettability" of the powdered reagent .

In addition, the amount of salt mixture that determines the thickness of the contact layer is conveniently evaluated, in a close 1 : 1 ratio , with the amount of micrometric amorphous silica to be dosed .

Instead, with respect to the amount of micrometric amorphous silica of step ( c ) , this must be stoichiometric or sub- stoichiometric with respect to the reaction ( 1 ) , i . e . the weight ratio between micrometric amorphous silica and magnesium is not exceeding 5 .

Typically, the micrometric amorphous silica is dosed directly into the reaction chamber without any need for a physical , chemical and/or thermal pretreatment thereof .

In a preferred embodiment of the invention, the stoichiometric or sub-stoichiometric amount of micrometric amorphous silica is reacted in its entirety with the liquid aluminum alloy in step ( c ) . In an alternative embodiment of the process , the total stoichiometric or sub-stoichiometric amount of micrometric amorphous silica is fractionated, so that each fraction is reacted with the liquid alloy in successive steps , each interspersed by a few minutes . In this way, a better dispersion of the micrometric amorphous silica in the contact layer is favoured with consequent better contact between the silica and the liquid aluminum alloy leading to the optimi zation of the reactions ( 1 ) and ( 2 ) . For example , according to this embodiment , the coolings of the liquid aluminum alloy due to the thermal dissipations of the system and to the insertion of the reagent at room temperature are contained thanks to the enthalpic continuity .

In a further embodiment of the process and preferably in the case in which the magnesium content to be reduced is high, steps (b ) , ( c ) and ( d) are repeated in series at least once , fractionating the amount of micrometric amorphous silica with respect to the total stoichiometric or sub- stoichiometric amount to be used in step ( c ) . The fractionated amount of micrometric amorphous sil ica also determines the fractionated amount of salt mixture to be added to the liquid aluminum alloy in step (b ) . For example , after step ( d) of slagging, step (b' ) of adding the remaining amount of salt mixture is repeated, as well as steps ( o' ) of reacting the remaining amount of micrometric amorphous silica and step ( d' ) of slagging the resulting contact layer, incorporating the subsequent magnesium-based reaction products follow .

EXAMPLES

Below are some examples describing the ef fectiveness of the process in accordance with the invention . Several experimental tests were carried out, on different scales, and with different operating modes, linked both to the mode of additions of the reagent and to its amount with respect to the reaction stoichiometry (1) .

The tests have always been performed on liquid aluminum alloys with low initial content of magnesium in order to test the most difficult operating conditions.

1. Chemical-physical characterization of micrometric amorphous silica

The micrometric amorphous silica used in the process according to the invention is microsilica obtained as a byproduct of the production of metallic silicon.

The particle size, the loss on ignition and the chemical composition of the microsilica were determined respectively by laser diffraction, muffle calcination at 950 °C and X-ray fluorescence.

Figure 1 shows the particle size distribution of the microsilica used in the examples.

From the chemical-physical analysis on microsilica, the following characteristics were obtained:

- loss on ignition at 950 °C: 2.5%

- specific surface: 19 m ² /g

- SiO ₂ : 93%

- Fe2O ₃ -Al ₂ O ₃ -CaO: 0.10%-0.15%

- MgO: 0.50%-0.70%

- Na ₂ O: 0.80%-1.00%

- K ₂ O: 2.00%-2.50%

- C: <2.2%

2. Process according to the invention at laboratory scale

The process according to the invention was initially carried out and tested in a low frequency (50Hz) mini-furnace provided with a silicon carbide crucible with a 11-litre capacity .

Exactly 5427 g of alloy of type 48000, of fixed and known chemical composition, and characterized by a Mg content equal to 1.00564%, which fixes its initial content therefore in 54.57 g of Mg, were melted. On the molten alloy, at a temperature of 750 °C, 250 g of a salt mixture formed by NaCl-KCl in weight ratio 85-15 were then fed, creating a contact layer of about 3 mm. Then on the contact layer, 289 g of microsilica (titre SiO2= 93%) , corresponding to the stoichiometric amount to zero the Mg to form the spinel, were dosed.

It should be noted that the reaction kinetics were promoted not only by the micrometric sizes of the microsilica, but also by the induced currents generated by the induction furnace. These, by producing a " fountain" -like movement of the alloy, facilitated the continuous exchange of the alloy on the surface, and therefore the mixing between microsilica and alloy, favouring an intimate contact between the two reagents.

After 1 hour elapsed, the contact layer incorporating the magnesium-based reaction product (spinel and magnesium oxide) was slagged.

The liquid alloy obtained at the end of the test has a final magnesium content equal to 0.03466%, with a percentage reduction thereof well over 95%. Among other things, having sampled the alloy at different times while performing the test, it turned out that already after 15 minutes, a reduction in the Mg content by 37% had been obtained, which increases up to 65% only after 30 minutes. 3. Process according to the invention on a pilot scale The process according to the invention was carried out in a pilot rotary tubular mini-furnace with a capacity of up to 900 kg of liquid charge.

The pilot rotary tubular mini-furnace reaches a maximum rotation speed equal to 3.5 rpm through 4 blades arranged inside the mini-furnace.

The slagging and cleaning operations of the minifurnace were carried out prior to the transfer of 900 kg of liquid aluminum alloy at a temperature of 750 °C in the rotary tubular mini-furnace.

The liquid aluminum alloy was chemically characterized, determining an initial magnesium content equal to 1,208%, which sets at 10,872 kg the weight of the Mg content.

The rotary tubular mini-furnace was operated at the maximum rotation speed, with the service oxy-methane burner switched off for the duration of the entire test, without significant changes in the temperature value.

Through a vibrating channel, 45 kg of salt mixture formed by NaCl-KCl in weight ratio 20-80, were added to the aluminum alloy, reaching a height of the contact layer of about 15 mm.

When the salt mixture uniformly covers the surface of the liquid alloy, 52 kg of microsilica, with the previously described chemical-physical characteristics, with no lumps, were added through a vibrating channel. The weight of the fed reagent constitutes 90% of the stoichiometric amount necessary for the formation of the spinel.

After 1 hour elapsed, the contact layer, which collects the reaction products and the salt mixture, was slagged. The liquid aluminum alloy obtained has a magnesium content equal to 0 . 082 % , demonstrating a reduction exceeding 93% , for the formation also of MgO together with the spinel .

In practice , it has been found that the process of the invention achieves the intended purposes .

In particular, the process of the invention shows excellent performance compared to known proces ses for reducing magnesium from a liquid aluminum alloy both in terms of yield and of reaction times , with the use of low-cost reagents without the need for their prior preparation, in full respect of the environment , plants and operators .