Adsorption of Zn, Cd, V, Ba, Cu, Mo, Ni, Cr, Li and Pb to silicon and
aluminium reduced AOD-slag
Edvin Elmroth
Bachelor thesis 15hp, 2017-11-26
Supervisors: Karlsson, Stefan; Sjöberg, Viktor
Examinator: Mattias Bäckström
Abstract
During production of steel, slag is formed as a by-product. The process of
steelmaking involves usage of additives such as chromium or vanadium as reactants to optimize and produce high quality steel. Vast amounts of slag are formed and there is a continuous search for applications that can make good use of the slag. Currently the use of slag in cleaning of metal polluted waters is researched and promising results has been found for many different types of slags. In this work two different AOD-slags has been used as sorbents for some selected elements (Zn, V, Cr, Mo, Pb, Li, Cd, Ba, Cu, Ni). The main difference between the two slags is the reducing agent that has been used, aluminium and silicon. This results in slags with different adsorption properties. The aluminium reduced slag show tendency for better adsorption capacities in general for the tested elements (Zn, V, Cr, Mo, Pb, Li, Cd, Ba, Cu, Ni), with a few exceptions. The buffering capacity of the materials were high, shown by the fact that final pH reached nearly 11.5 independent of the start pH (varied between 2 and 8). The adsorption process was rather quick and 24 minutes contact time was in most cases sufficient to reach equilibrium. For several of the elements e.g. Lithium, the maximum capacity of the slags was not reached even though a load of 3,07 mg Lithium was added per gram of slag.
Keywords
Content Abstract 2 Keywords 2 Content 3 Introduction 4 Metals 4
Steel production and slag formation 5
Adsorption theory 5
pH 7
Aim 7
Materials and Methods 7
Material preparation 7
Titration for capacity determination 8
Adsorption test solutions 8
Macro metals 8 Time dependence 8 Sample preparation 8 Instrument 8 Data modelling 9 Freundlich 9 Langmuir 10
Results and discussion 11
Observations of material behaviour 11
Macro metals 11
Saturation 12
Titration 12
Time dependent adsorption 13
Adsorption with different initial pH 14
Adsorption modelling 15
Adsorption for the slag, with an accumulated scale for each of the portions added. 17
Surface capacity 20
Theoretical values versus experimental 21
Conclusions 22
Acknowledgements 22
Introduction
The need for clean water is always a constant topic around the world. An ability to be able to clean the water from heavy metals with simple means would be of high
interest since different technical innovations and human lifestyles utilizes heavy metals. Thus, the environment has an increased content of many metals. Steel is an important material in today’s society and many tonnes are produced each year, along with slag. The ability to utilize the otherwise growing amount of steel slag would be of interest if it can be used to clean water systems.
Metals
Many rivers and surface waters have shown increased concentrations of heavy metals, above expected background levels (cf. Nriagu et al 1979, Stumm, Baccini 1978, Miaosi et al 2015). The metals included in this study are known for their harmfulness to life. Some of them are important in steel production (chromium, vanadium, molybdenum) and others are well known for their toxicity like lead and lithium that are present in many ores (cf. Yasemin, Zeki 2007). During the production of steel, the slag is stored in the open, often on designated waste sites, where wind and weather can affect it. Thus, there is a possible leakage of chromium,
molybdenum and vanadium to the environment. It is therefore of importance to be able to remediate waters contaminated by the slag, as well as waters containing natural high concentrations of metals. A material with good adsorption properties is therefore of interest. Optimized steel slag could be a possible candidate for this purpose since it has been designed to have a high adsorption capacity and chemical stability already from the steel making process.
There are different ways of preforming metal analysis from classical gravimetric methods to modern high-end mass spectrometry. All methods have their own pros and cons but today the choice is often made according to time efficiency why
instruments capable of multi-element analysis is therefore often chosen. As a result, inductive coupled plasma (ICP) -instruments have typically been the number one choice for several decades. Recently, the microwave plasma (MP) running on nitrogen was introduced to the market, as a cost-effective alternative. The ICP uses a radiofrequency coil to ignite an argon plasma, resulting in a higher energy and ionization of the sample than the MP that relies on a microwave induced ignition of the nitrogen gas. Because of the lower ionization energy for nitrogen, the MP has a lower energy content and thus a lower potential for ionization of analyte elements. However, the ICP suffer from high running costs due to a consumption of argon gas of at least 18 L min-1, compared to the MP that runs on nitrogen that can be
extracted from the air by a low cost gas generator. A few years ago, Agilent Technologies introduced their new Microwave Plasma Atomic Emission Spectrometer (MP AES 4100). It has been found to have detection limits and performance not far from ICP-OES instruments after optimization (cf. Berg 2015, Karlsson et al 2015). Even though matrix effects may limit the performance
sometimes, its low running cost, accuracy and robustness makes it suitable for analysis of complex matrices that are common in geochemical studies. (cf. Wang et al 2017). It must be kept in mind however, that for most elements, emission lines from excited element atoms are superior to ion lines since the latter suffer from a higher matrix dependency.
Steel production and slag formation
During 2010 Sweden produced 1373 Mkg of metallurgic slag, of which 105 Mkg was Argon Oxygen Decarburization (AOD) steel slag (Jernkontoret, 2012). Outokumpu Stainless, the sample provider in this study, uses an electric arc-furnace for scrap metal melting at their production site in Avesta, Sweden. From the arc furnace, the melt is poured into the AOD-converter, where the scrap metal is refined into stainless steel (Jernkontoret, 2012). This process has three major steps, as described in Jernkontorets Handbok (Jernkontoret, 2012):
1. Decarburizationon 2. Reduction
3. Desulfurization
During the decarburization process, free chromium in the steel melt reacts with oxygen, which produces chromium oxide (Eq.1). Further, the chromium oxide reacts with carbon in the steel melt resulting in carbon monoxide and free chromium (Eq.2). 4 Cr + 3 O2 2 Cr2O3 (Eq.1)
2 Cr2O + 6 C 6 CO + 4 Cr (Eq.2)
Even though the reaction equations express equilibrium, there will be a decrease of chromium in the melt, since some of it will be contained in the metal lattice in the steel. During the reduction, a species with a higher affinity for oxygen than chromium is added, resulting in a reduction of the Cr2O3. This is done with either silicon or
aluminium. Addition of facilitation agents such as lime (CaO) and fluorspar (CaF2),
results in easier breakage of Cr2O3 in the melt, and helps to keep the slag in a liquid
state, as well as lowering the amount of slag that is formed. The desulfurization process removes sulphur from the melt. It is best described as a net reaction where sulphur reacts with lime, as CaO, and produces CaS and O. The CaS becomes a part of the slag and the oxygen is released as gas from the melt. To get to a specific optimal predetermined property of the slag, aluminium or silica can be added to remove some of the oxygen from the steel melt.
Adsorption theory
In general, adsorption represents several, in principle, different mechanisms where a dissolved species becomes associated with a solid surface. Hence, there are several different ways of adsorption for each adsorbent as well as adsorbant. A physical bond is usually an electrostatic interaction between the solute and surface of the material. This bond is electrostatic in character, and ranges from rather weak van der Waals-forces to strong ionic bonds. The van der Waals-force is an interaction
differences in electronegativity induces a weak attraction between the positive and negative regions in two molecules. A chemical adsorption is when a covalent bond is formed between the solute and the surface species of the material (Drever, 1997). An important feature of the surface concept is the “zero point of charge” (ZPC or pHZPC), i.e. the pH where the pH dependent charge is neutralized. At low pH,
protonation will favour a net positive charge while at high pH, deprotonation will result in a net negative charge (Drever, 1997). These properties are highly specific for each kind of material and ranges from pH 2.5 for quartz up to pH 8 for several metal oxides, including aluminium.
Figure 1. A schematic overview for the DDL (Diffuse Double-Layer) theory (Bayesteh, 2015).
The Diffuse Double-Layer (DDL) model is commonly used to describe the surface of a material. The concept of the DDL is that the surface of the material has an outer layer that is in contact with a solution phase. The outer layer has either a net positive or negative charge depending on the charge of the surface. At pH above pHZPC the
charge has a net negative charge that is balanced by electrostatically bound cations on the surface and an accumulation of negative charge in the outer sphere. Hence, the solute will interfere with the outer layer of the surface by increasing the order of ions in solution. Therefore, a solid surface does not only result in a direct interaction with counter ions but also on the orientation of ions as a function of distance, as seen in figure 1. If the material is physically porous, the adsorbed ions will also migrate into the matrix of the material, in principal driven by the concentration gradient on the surface. If the material is compact, there will be a migration in fissures, but it will take a considerably longer time to reach equilibrium (Drever, 1997). If the time is extended even further, usually on a geological time scale, displacement in the lattice will occur.
Material adsorption properties can be described by different mathematical equations, considering features of the surface and the solute. The choice of model is typically made according to the available information. The isotherm is a convenient and
common way to mathematically describe the adsorption process, even down to a mechanistic level. Primarily two isotherms are used to provide this information, the Langmuir and Freundlich, see below. This means that a relatively simplified
mathematical model can be used to describe a partition of several components in quite complex systems.
pH
The solution pH has a high impact on the adsorption behaviour of AOD steel slag (cf. Repo et al, 2015). The results in their study suggested that at pH 1, the adsorption was around 50% for lead and around 20% for cobalt of the solution concentrations. In the pH range 2-3 the adsorption had increased to 80-90% for lead and 70-80% for cobalt (Repo et al, 2015). The pH is also of importance for formation of solid
hydroxides (cf. Liberto, 2011), either in solution or on the surface. Precipitation reactions are sometimes mistaken for adsorption processes. Although such
reactions can have the same impact on a metals solid/solution distribution they are principally different. To form solid hydroxides the critical solution pH depends on the properties of the element, Cr3+, Pb2+, Cu2+, Hg2+, Ni2+, Zn2+, Mn2+, Ag+, Cd2+, Ba2+ and Ca2+, all form solid hydroxides at element specific pH of; 4.2; 4.8; 6.0; 3.0; 7.5; 6.2; 7.0; 10.6 and 8.0, respectively (Liberto, 2011). In order to make realistic
conclusions concerning the reactions it is also important to include the rather
different chemical conditions at the solid interface in relation to the bulk solution. For example, the orientation of ions according to DDL model will result in a pH at the interface that typically is 1 to 3 units lower than in the bulk, for many materials. Aim
The purpose of this project is to experimentally elucidate the adsorption efficiency and capacities of two different steel slags, with metal concentration, pH and time as variables. Mathematical modelling of the adsorption mechanisms and saturation conditions are used to better understand the binding to the surface of the material. Materials and Methods
Material preparation
During the experiments, polypropylene (PP) lab-ware was used to minimize interactions between metal ions and surfaces. Test tubes holding 15 and 50 ml, respectively (Sarstedt ®), PP -syringes (VWR-line) and 0.2 μm PP-filters (VWR-line) were used. The slags were crushed and grinded using a mortar and pestle and then sifted. The size fraction of 0.9-2.0 mm was collected and used throughout the whole experiment.
Titration for capacity determination
Determination of acid capacities of the slag was conducted as batch titrations with proportions of 0%, 20%, 40%, 60%, 80%, 100% relative to concentrated (12 mol L-1) HCl, diluted with deionized water (DI) of 18.2 MΩ quality. In the titration, a liquid to solid ratio (L/S) of 10 was used, with 4 g material and 40 ml leachant. Time intervals for equilibrium measurements were 4 hrs and 23 hrs, respectively. The titration was performed on an end-over-end shaker, with n=1 for the batch titration.
Adsorption test solutions and experiments
The following salts were used to prepare single element stock solutions containing 1000 mg L-1 of each element in 1% HNO3: CdCl2 (Merck, 98%), NH4VO3 (Merck,
99,5%), Na2MoO4 (Fluka, 99%), Na2CrO4 (Merck, 99,5%), ZnCl2 (Merck, 98%),
Pb(NO3)2 (Merck, 99,8%), Ni(NO3)2 (Merck, 99%), Ba(NO3)2 (Merck, 98%), LiCl
(Merck, 99%). The multi metal mixture used in the experiments contained 80 mg L-1 of the individual elements in 1% HNO3.
For adsorption experiments, 4h was used for equilibrium and at a liquid to solid ration (L/S) of 10. The maximal adsorption was determined by consecutive additions of the metal mix, using a concentration of 10 mg L-1. The aqueous phase of the sample was then decanted before addition of the following portion of the metal mix. For the most optimal initial pH, the metal mix was pH adjusted (pH 2, 4, 6 and 8) and then added to the slag, using a 4h equilibrium time with n=2. The kinetic study was done on the initial pH which had the best adsorption efficiency.
Macro metals
Determination of water soluble macro metals released from the slag (Ca, Al, K, and Na) was done by batch leaching with DI at L/S 10, for 24 hrs at room temperature. Time dependence
In this experiment, only the initial pH 7 was tested. A L/S 10 was used, 4 g slag with 40 ml test solution, and equilibrated on an end-over-end shaker.
Sample preparation
Before analysis with the MP-AES, each solution was filtered through 0,2 μm and acidified with 100 μL concentrated HNO3 per 10 mL sample. The same ratio for
acidifying was used for addition of internal standard. Certified Standards (AVS TITRINORM from VWR), with concentration of 1.000 g L-1; Y, La, Lu was used as internal standard, each at a concentration of 1.0 mg L-1 in the sample solution. Instrumentation
An Agilent 4210 MP AES was used for metal analysis. Monthly monochromator calibration was performed, with a certified calibration solution. Before each sample sequence the wave guide and nebulizer flow were optimized. For tuning, certified standard material, from Merck, containing Al, As, Ba, Be, Ca, Cd, Co, Cr, Cu, Fe, K, Li, Mg, Mn, Mo, Na, Ni, Pb, Sr, Zn and V in 1% HNO3 was used, at concentrations of
0,1; 0,5; 1, 5 and 10 mg L-1. As ionization inhibitor, 1.25 g L-1 CsNO3 was added in a
second flow line to the nebulizer. Tubings of 0.19 mm and 0.38 mm diameter were used for sample uptake and ionization inhibitor, respectively. Both liquids were mixed in a y-piece mounted after the peristaltic pump. A pump speed of 15 rpm was used during the analytical cycle. In between samples, rinsing was performed with 1% nitric acid for 90 sec at a pump speed of 50 rpm. For sample uptake was 90 sec used at 50 rpm followed by 30 sec of stabilization before the analytical cycle. For quality assurance, after every 5 samples, a standard was analysed.
Principal anions (column, solution phase, flow, detection)
The release of principal anions (fluoride, chloride, nitrate and sulphate) with impact on the solution conditions was evaluated by leaching with DI for 4 hours in three replicates for each slag. Quantification was made with a 830 IC Metrom Ω equipped having a Dionex IonPac ™ AG12A RFIC™ 4x50 mm guard column and an AS12A RFIC™ 4x200 mm for analytical separations. The pump speed was 1 ml min-1 and the mobile phase contained 0.3 mM NaHCO3; 2,7mM Na2CO3, in DI. Ion
suppression was used with conditioning solutions consisting of DI and sulphuric acid (4 ml L-1 H2SO4), respectively. The tubing was PharMed® BPT NSF-51.
Data modelling
Saturation conditions
Estimates of the saturation state were modelled from analytical data of the bulk compositions using Visual Minteq. The carbonate modelling assumes equilibrium with atmospheric CO2.
Freundlich
The Freundlich isotherm has the following mathematical expression: 𝑚𝑖 𝑎𝑑𝑠 = 𝐾𝑓𝑚𝑖(𝑠𝑜𝑙𝑛 )𝑛
Where, 𝑚𝑖 𝑎𝑑𝑠 is the concentration in mg g-1 of adsorbed species on the surface, 𝑚𝑖(𝑠𝑜𝑙𝑛 )𝑛 is the concentration of the same species in to solution at equilibrium, expressed as mg L-1, 𝐾𝑓 is the distribution coefficient, n is a specific constant, that
of n is in the original form usually less than 1 implicating a chemical bond. In this kind of assumption of the material, it is modelled as a heterogeneous surface with
different available bonding energies. A heterogeneous surface will have a finite amount of possible binding sites, making a graph of adsorbed species (i) versus added concentration of species (i), that flattens out, the higher the total concentration of the added species (i) is. This is because the strongest binding sites will be
saturated first, giving the highest coefficient value, followed by the weaker sites that will be reflected by lower values of the coefficient. For competitive bonding,
increasing the concentration of one solute will be reflected by lowered values for elements that are competing for the same sites.
The Freundlich Isotherm can be derived into a linear form: log 𝑞𝑒 = log 𝐾𝐹 +
1
𝑛 log(𝐶𝑒)
Where 𝑞𝑒is the concentration of metal adsorbed on the surface (mg g-1), 𝐾𝐹 is the
distribution coefficient, 1
𝑛 is used to describe how favorible the adsorption was for
either a chemical or a physical adsorption, 𝐶𝑒 is the concentration of the soulute in soulution at equlibrium (mg L-1). If the ratio of 1
𝑛 = 1 then the adsorption is linear, less
than 1 is indicative of chemical bonds and greater than 1 implies physical bonds that are subjected to competition with other solution species. (Kamala et al, 2005)
Langmuir
The Langmuir isotherm uses another approach to describe the adsorption process. It was first developed to describe binding of a monolayer of gas to a surface, but it can also be extended and used for dissolved species in aqueous systems (cf. Stumm, 1992).
𝑉𝑎𝑐𝑎𝑛𝑡 𝑠𝑖𝑡𝑒 + 𝑖 = 𝑜𝑐𝑐𝑢𝑝𝑖𝑒𝑑 𝑠𝑖𝑡𝑒 𝐾𝐿=
𝑚𝑖(𝑎𝑑𝑠 ) 𝑚𝑖(𝑠𝑜𝑙𝑛 )𝑚𝑣𝑎𝑐𝑎𝑛𝑡 𝑠𝑖𝑡𝑒𝑠
Where, 𝐾𝐿is the distribution coefficient, 𝑚𝑖(𝑎𝑑𝑠 ) (mole L-1) is the concentration of the adsorbed species (i), 𝑚𝑖(𝑠𝑜𝑙𝑛 ) (mole L-1) is the concentration of solute at equilibrium,
𝑚𝑣𝑎𝑐𝑎𝑛𝑡 𝑠𝑖𝑡𝑒𝑠 (mole L-1) is the concentration of the sites that is not occupied by any
species (i). The equation can be transformed into a linear form: 1 𝑞𝑒 = 1 𝑞𝑚𝐾𝐿 1 𝐶𝑒 + 1 𝑞𝑚
Where, 𝑞𝑒 is the concentration of solute adsorbed on the surface in mg g-1, 𝐶𝑒 is the
adsorption in mg g-1. From this equation the 𝑞𝑚 value will be given and thus the 𝐾𝐿 can be calculated (García-Gómez, et al, 2016).
Results and discussion
Observations of material behaviour
Physical differences between the two slags could be observed by visual inspection. The Al-reduced material tends to be more brittle and break more easily during the crushing and grinding than the Si-reduced. This is probably due to the formation of SiO2 in the latter, which is a more stable lattice structure and resistant to physical
interactions. In the slag, there were four major components with different observable properties, the slag itself, green crystals, dark brown/black crystals, and steel/iron particles. The main fraction in the mixture was the slag itself. A visual inspection of an arbitrary subsample of the crushed material contained a mass of approximately 70-90% slag, a maximum of 1% steel/iron and the rest at equal fractions. The mix of the material after the crushing seemed to be homogeneous. This applies to both of the batches in the study.
Macro metals and anions
The water solution in equilibrium with the Al-reduced slag is very rich in dissolved elements (table 1). According to the acid titration, it has an approximate acid
neutralizing capacity of 20-21 meqv g-1 when lowering the pH from 11.6 to 0.2 (figure 2). The neutralization is mainly caused by the dissolution of slag components, rather than surface exchange of hydronium ions. As the CaO comes into contact with water, either on the waste site or in the experiment, the formed hydroxide constitutes the main source for the neutralizing process. In addition carbonatization will occur if the slag is in contact with the air during the ageing process.
Table 1. Aqueous concentrations of principal components after leaching with DI for 24 hrs in L/S 10. Al red. mg L-1 (pH) Si red. mg L-1 (pH) Al 12.2 11.5 36.5 11.0 Ca 208.8 11.5 137.2 11.0 K 2.0 11.5 1.1 11.0 Na 7.1 11.5 13.7 11.0 F- 3.7 11.5 9.3 11.0 Cl- 137.3 11.5 54.0 11.0 NO3- 9.9 11.5 5.1 11.0 SO42- 12.1 11.5 13.8 11.0
Saturation
The saturation indices presented in table 2 indicate oversaturation when they exceed 0. The observed oversaturation is of course due to the species being dissolved from the slag. This is particularly important for calcium oxide/hydroxide/carbonates and aluminium hydroxides. It can also be seen that lime remains undersaturated, either because there is no lime in these rather fresh slag samples or that it has formed Ca(OH)2 when CaO comes in contact with water. The table also indicates that there
is a possibility that the distribution of the studied metals might be affected by the formation of carbonate phases. This is particularly evident for the calcium carbonates at the high pH but also for PbCO3, CdCO3 and ZnCO3. It is however impossible to
determine from the examination of saturation if precipitation takes place on the surface of the adsorbents or in the solution. An initial surface adsorption would, however, be likely because of the impact of the surface potential as loci of reaction. Table 2. Saturation indices at equilibrium with DI, where a positive number,
highlighted in red, indicates that the system is oversaturated.
AL Si Mineral Sat. index Sat. index Al(OH)3 (am) -2.662 -2.187 Al(OH)3 (Soil) -0.152 0.323 Al2O3(s) -3.187 -2.235 Aragonite 3.17 2.989 Boehmite -0.377 0.099 CaCO3xH2O(s) 1.915 1.734 Calcite 3.313 3.133 Diaspore 1.328 1.804 Fluorite -3.108 -2.493 Gibbsite (C) 0.398 0.873 Gypsum -5.063 -5.185 KCl(s) -7.067 -7.732 Lime -16.297 -16.477 Vaterite 2.747 2.567 Acid capacity
The material reaches equilibrium after circa 4 hrs, since the difference between 4 hrs and 24 hrs is very small, as seen in figure 2. The buffer capacity is reached between 20 and 40% added HCl seen in figure 3. During the titration with concentration of 40% and 60% HCl, H2S is suspected to form as indicated by the smell of rotten eggs.
Considering the nature of the reduced materials the possible reaction is; 2 HCl + CaS CaCl2 + H2S; since lime is used to reduce the amount of sulphur in the slag.
The specific capacity is 0.02 g mol-1 with a standard deviation of 1.5*10-5 with n=2, between the different time experiment.
Figure 2. Batch titrations with pH as a function the HCL to H2O ratio.
Figure 3. pH after 4 hours equilibration as a function of initial pH after addition of acid.
Time dependent adsorption
As shown in figures 3 and 4, most of the metals have been adsorbed already after 24 minutes. With that in mind and the initial pH dependencies (table 3), the apparent equilibrium for the metal adsorption would be established within this time frame. With a longer contact time, the slags adsorbs only slightly more metal, such as Ba, Cr, V, Mo, and Li (17%; 19%; 7%; 14% and 14%, respectively) (figure 4). Hence, the materials are not porous enough to allow for an increased capacity by extending the contact time. It is also noted that vanadium is released to the aqueous phase (figure 5). The release of vanadium from the Si-reduced slag is mainly related to the acid neutralization, where the system approaches pH 12, making the surface of the Si-reduced slag be slightly positive and thus bleeding out the metals to approach an
-2 0 2 4 6 8 10 12 14 0 20 40 60 80 100 120 pH % HCl in H20:HCl
pH as a function of acid titration
4h 23h 0 2 4 6 8 10 12 14 0 1 2 3 4 5 6 7 8 9 F inal pH aft er 4h Initial pH
Equilibrium pH
Al SIequilibrium with the hydroxides in the aqueous phase. A possible formation at high pH is VO3OH2-.
Figure 4. Adsorption (%) versus contact time for Al-reduced slag, at initial pH 7.
Figure 5. Adsorption (%) versus contact time for Si-reduced slag, at initial pH 7. Adsorption at different initial pH
The adsorption efficiency of the materials when exposed to the 10 mg L-1 metal mix is shown in table 3. Here, it can be seen that a higher initial pH in the solution will have a negative impact on the adsorption. A clear example is the lithium adsorption, that reaches 26% and 21%, respectively, for Al- and Si-reduced slag at an initial pH of 2. When the initial pH approaches 8, lithium starts to leach from the material instead. For most of the metals, the steel slag had better adsorption properties at a lower initial pH, as seen in table 3. The high pH buffering capacity of the material, as seen in figure 3, makes it very difficult to maintain a constant pre-determined pH in the adsorption experiments. Since the surface of the slag mainly contains CaO and Ca(OH)2, the buffer capacity is related to the content of O- and OH-groups from
0 20 40 60 80 100 120 0 50 100 150 200 250 Ad so rp tio n (%) Time (min)
Adsorption to Al-slag at initial pH 7
BariumZinc Cadmium Vanadium Coppar Nickel Chromium Lead Molybdenum Lithium -200 -150 -100 -50 0 50 100 150 0 50 100 150 200 250 Ad so rp tio n (%) Time (min)Adsorption to Si-slag at initial pH 7
Barium Zinc Cadmium Vanadium Coppar Nickel Lead Molybdenum Lithium Chromiumthese solid species. Hence, the material is self-buffering within a pH region where there is a high adsorbent efficiency for cations.
Table 3. The average adsorption of cations expressed as % of the inventory and the relative standard deviation in italic for different initial pH (n=2 for each slag type and initial pH). Metal Al (%) (%RDS) Initial pH/Final 2/12 4/12 6/12 8/12 Lithium 26 5,6 31 16,8 60 8,2 -10 -144 Barium 75 3,5 69 2,5 73 2,3 61 3,2 Zinc 100 0,0 100 0,0 100 0,0 100 0,0 Cadmium 100 0,0 100 0,1 100 0,0 100 0,0 Vanadium 79 2,4 52 5,8 69 3,6 62 11,9 Copper 99 0,1 100 0,1 100 0,0 100 0,0 Nickel 100 0,0 100 0,0 100 0,0 100 0,0 Lead 100 0,2 99 0,0 99 0,0 99 0,3 Molybdenum 30 13,8 29 16,1 30 5,7 -9 -97,9 Chromium 48 15,1 46 17,5 47 5,9 49 4,3 Si (%) (%RSD) Initial pH/Final 2/12 4/12 6/12 8/12 Lithium 21 0,8 20 1,4 11 15,7 -19 -122 Barium 80 0,2 80 1,6 85 0,2 82 3,8 Zinc 100 0,0 100 0,0 100 0,1 100 0,0 Cadmium 100 0,0 100 0.1 100 0.0 99.6 0.0 Vanadium 63 1.7 -129 -10.4 -41 -9.6 -76 -11.4 Copper 100 0.0 100 0.0 100 0.0 100 0.2 Nickel 100 0.0 100 0.0 100 0.0 100 0.1 Lead 100 0.0 99 0.2 100 0.0 100 0.4 Molybdenum 6 32.5 9 22.0 3 0.0 -40 69.4 Chromium 14 6.4 15 14.3 12 19.4 -5 -46.4 Adsorption modelling
In figures A-J, respectively, adsorption of the adsorbate has been plotted versus the total accumulated adsorbate in the aqueous phase. Two different mechanisms of adsorption can be identified, lead-like and zinc-like. Among the metals, Cr, Cu, Li, Mo, and V, seemed to follow a lead-like behaviour, with no maximum adsorption reached under the experimental conditions. For the other metals, there is an evident maximum adsorption.
In table 4. the values for Qm, KL, KF and 1/n are presented. For zinc they should be
considered as less certain, since they are modelled with only 3 data points because of 100% adsorption. According to the Freundlich model, the majority of the surface species are formed by chemical bonds, since most of the 1/n ratios are less than one. If the LOD had been used for modelling, the information about the adsorption kinetics would have been lost. The only metal that has a possible physical bond to
the surface is molybdenum, which is plausible considering its predominance as molybdate ions. The theoretical adsorption maximum capacity for the material is estimated by using the Langmuir model, where Qm gives the theoretical value. The
results suggest that lead is adsorbed to the material, by formation of a chemical bond. Due to saturation conditions, it is more likely that it forms a precipitate (PbCO3). It is, however, possible that both mechanisms are operating but
indistinguishable when using these methods of interpretation. Disregarding the precise mechanism, the results show that non or very little of the added lead ions is left in the aqueous solution.
Since almost every tested metal forms a chemical bond with the surface the aqueous solutions pH is important. The reason behind this is that it determines how the
surface of the material will behave, with consideration of charge and composition, resulting in, a lower pH increases the binding affinity for the metals to the surface. An important finding is that the self-buffering of the slags and the kinetics would sustain a high cation adsorption as long as their capacities are not exceeded. Both slags would therefore be serious candidates for use as adsorbents in water purification, as long as metal release from them is acceptable.
Table 4. Isotherm regression between Langmuir and Freundlich model.
Langmuir model Freundlich model
Qm (mg g-1) KL R2 KF 1/n R2 Zn Al 0.272 -4.775 0.484 0.279 0.046 0.217 Zn Si 0.128 207.947 0.838 0.112 0.054 0.758 Cd Al 0.272 47.017 0.964 0.268 0.186 0.744 Cd Si 0.114 329.634 0.803 0.105 0.041 0.838 V Al 0.324 0.189 0.944 0.047 0.722 0.973 V Si 0.602 0.049 0.992 0.034 0.726 0.999 Ba Al 0.204 0.221 0.993 0.038 0.541 0.979 Ba Si 0.119 0.374 0.909 0.043 0.284 0.970 Cu Al 0.546 118.232 0.784 0.515 0.118 0.988 Cu Si 0.496 1.460 0.952 1.888 0.168 0.970 Mo Al 0.158 0.029 0.759 0.003 0.948 0.832 Mo Si 0.198 0.011 0.973 0.002 1.185 0.992 Ni Al 0.232 49.248 0.926 0.196 0.185 0.896 Ni Si 0.089 329.169 0.089 0.087 0.040 0.789 Cr Al 0.068 0.147 0.752 0.014 0.407 0.906 Cr Si 0.067 0.030 0.982 0.003 0.666 0.996 Li Al 0.118 0.021 0.971 0.003 0.747 0.984 Li Si 0.106 0.021 0.988 0.003 0.716 0.992
Figure A. Adsorption (mg g-1) of lithium versus accumulated metal loading (mg) in the aqueous phase.
Figure B. Adsorption (mg g-1) of cadmium versus accumulated metal loading (mg) in the aqueous phase.
Figure C. Adsorption (mg g-1) of zinc versus accumulated metal loading (mg) in the aqueous phase. 0 0.02 0.04 0.06 0.08 0.1 0 0.5 1 1.5 2 2.5 3 3.5 mg g -1 acc. mg aq
Lithium
Al Si 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0 0.5 1 1.5 2 2.5 3 3.5 mg g -1 acc. mg aqCadmium
Al Si 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0 0.5 1 1.5 2 2.5 3 3.5 mg g -1 acc. mg aqZinc
Al SiFigure D. Adsorption (mg g-1) of vanadium versus accumulated metal loading (mg) in the aqueous phase.
Figure E. Adsorption (mg g-1) of barium versus accumulated metal loading (mg) in the aqueous phase.
Figure F. Adsorption (mg g-1) of nickel versus accumulated metal loading (mg) in the aqueous phase. 0 0.1 0.2 0.3 0.4 0.5 0 0.5 1 1.5 2 2.5 3 mg g -1 acc. mg aq
Vanadium
Al Si 0 0.05 0.1 0.15 0.2 0 0.5 1 1.5 2 2.5 3 mg g -1 acc. mg aqBarium
Al Si 0 0.05 0.1 0.15 0.2 0.25 0.3 0 0.5 1 1.5 2 2.5 3 mg g -1 acc. mg aqNickel
Al SiFigure G. Adsorption (mg g-1) of copper versus accumulated metal loading (mg) in the aqueous phase.
Figure H. Adsorption (mg g-1) of lead versus accumulated metal loading (mg) in the aqueous phase.
Figure I. Adsorption (mg g-1) of molybdenum versus accumulated metal loading (mg) in the aqueous phase.
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0 0.5 1 1.5 2 2.5 3 3.5 mg g -1 acc. mg aq
Copper
Al Si 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 0 0.5 1 1.5 2 2.5 3 3.5 mg g -1 acc. mg aqLead
Al Si 0 0.05 0.1 0.15 0.2 0.25 0 0.5 1 1.5 2 2.5 3 3.5 mg g -1 acc. mg aqMolybdenum
Al SiFigure J. Adsorption (mg g-1) of chromium versus accumulated metal loading (mg) in the aqueous phase.
Surface capacity
In table 4. The theoretical maximum adsorption capacity (Qm) according to the
Langmuir model has been used to calculate the corresponding general adsorption capacity for the slags surface. Here it is assumed that; i) the surface area of the particles is related to mesh size of the sieve (0.9 and 2. mm); and ii) the mass of 1 grain weighs approximately 0.01 g.
Table 5. Adsorption capacity for the metals in respective slag type. mmol/m2 (0.9mm) mmol/m2 (2.0 mm) Average mmol/m2 Zn Al 16.34 3.31 9.82 Zn Si 2.96 1.56 2.26 Cd Al 3.66 1.92 2.79 Cd Si 1.54 0.81 1.18 V Al 9.64 5.07 7.36 V Si 17.89 9.40 13.64 Ba Al 2.25 1.18 1.71 Ba Si 1.31 0.69 1.00 Cu Al 13.00 6.83 9.92 Cu Si 11.81 6.21 9.01 Mo Al 2.49 1.31 1.90 Mo Si 3.12 1.64 2.38 Ni Al 5.98 3.14 4.56 Ni Si 2.28 1.20 1.74 0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0 0.5 1 1.5 2 2.5 3 mg g -1 acc. mg aq
Chromium
Al SiCr Al 1.99 1.05 1.52
Cr Si 1.95 1.02 1.49
Li Al 25.65 13.48 19.57
Li Si 23.14 12.17 17.66
Theoretical values versus experimental
The theoretical maximum adsorption capacities shown in table 4 are consistent with those obtained in the experiments (table 6). The values in bold in table 6 indicate that no saturated endpoint was reached, and the highest value measured was used as a proxy for the capacity. Some of the obtained data indicated a higher
concentration than the theoretical maxima. This is most likely caused by the alkaline pH at equilibrium induced precipitation of hydroxides and carbonates, rather than adsorption. To overcome the problem with alkaline pH at equilibrium, addition of acid is needed but then the slag itself will start to dissolve, as seen in figure 2 and 3, and there is a risk for metal release. An alternative explanation is that since it was
impossible to calculate saturation endpoints for some of the metals, the
mathematical modelling suffered from too little data in order to allow for a correct estimate of a theoretical value.
Table 6. Obtained and theoretical adsorption capacities, bold numbers indicate that saturation was not reached for the obtained data.
Obtained max adsorption (mg g-1) Theoretical Qm (mg g-1) Zn Al 0.30 0.27 Zn Si 0.15 0.12 Cd Al 0.26 0.27 Cd Si 0.12 0.11 V Al 0.44 0.32 V Si 0.31 0.60 Ba Al 0.15 0.20 Ba Si 0.10 0.11 Cu Al 0.70 0.54 Cu Si 0.53 0.49 Mo Al 0.20 0.15 Mo Si 0.19 0.19 Ni Al 0.26 0.23 Ni Si 0.11 0.08 Cr Al 0.08 0.06 Cr Si 0.05 0.06 Li Al 0.09 0.11 Li Si 0.07 0.10
Conclusions
- The retention capacities of the slags for cations is high and for a mixture of Zn, Cd, V, Ba, Cu, Mo, Ni, Cr, and Li it varies from 0.1 to 0.60 mg g-1, for the selected grain size.
- To get the maximal adsorption capacities, the initial pH should be 2 for the metals included in this study.
- Concerning time dependency, the longer a solution had contact with the material, the greater the adsorption but an apparent equilibrium was reached within 24 minutes.
- The metal retention mechanism is mainly chemical in character but for elements that form stable hydroxide, carbonate and/or mixed phases precipitation increases the overall retention efficiency.
Acknowledgements
I would like to thank Dr. Stefan Karlsson, Dr. Viktor Sjöberg, Dr. Liem Nguyen, Dr Michaela Zeiner, M.D. Frida Fart, and Joakim Jansson, Tova Alsätra, and Elli Viljanen for providing guidance and expertise through out this project, and Gunnar Ruist at the company Outukumpu Stainless for providing the slag samples.
References
Jernkontoret (2012). Stålindustrin gör mer än stål: Handbok för restprodukter, Jernkontoret teknikområde 55, ISBN 978-91-977783-2-9
Drever, J. I, (1997) The Geochemistry of natural waters: Surface and Groundwater Environments, University of Wyoming, ch. 4-5, 16
Berg, I. (2015). Validation of MP-AES at the Quantification of Trace Metals in Heavy Matrices with Comparison of Performance to ICP-MS, University of Örebro,
Karlsson, S., Sjöberg, V., Ogar, A., (2015), Comparison of MP AES and ICP-MS for analysis of principal and selected trace elements in nitric acid digests of sunflower (Helianthus annuus), Talanta, 135, 124-132
Wang Y., Baker L. A., Helmeczi E., Brindle I. D., (2017) Determination of metals in base metal ores using the Agilent MP-AES; 5991-8120EN
Repo. E., Warchoł. J.K., Westholm. L. J., Sillanpää. M., (2015), Steel slag as a low-cost sorbent for metal removal in the presence of chelating agents, Journal of Industrial and Engineering Chemistry, 27, 115-125.
Liberto. P., Chávez M. L., Abatal. M., (2011), Adsorption of heavy metals in acid to alkaline environments by montmorillonite and Ca-montmorillonite, Chemical
Engineering Journal, 171,3, 1276-1286
Nriagu, J. O., Kemp, A. L. W., Wong, H. T. K., Harper, N. (1979) Sedimentary record of heavy metal pollution in Lake Erie. Geochim. Cosmocim. Acta. 36, 454-470
Stumm, W., Baccini, P., (1978) Man-made per-turbation of lakes. In A. Lerman (Ed.), Lakes-Chemistry, Geology, Physics (pp. 91-126) New york: Springer-Verlag.
Yasemin. B., Zeki. T., (2007) Removal of heavy metals from aqueous solutions by sawdust adsorption, Journal of Environmental Science, 19, 160-166
Kamala, C., Chu, K., Chary, N., Pandey, P., Ramesh, S., Sastry, A.,& Sekhar, K.(2005). Removal of arsenic (III) from aqueous solutions using fresh and immobilized plant biomass. Water Research, 39, 2815–2826.
García-Gómez, C., Rivera-Huerta M. L., Almazán-García F., Martín-Domínguez A., Romero-Soto I. C., Burboa-Charis V. A., Gortáres-Moroyoqui P. (2016).
Electrocoagulated Metal Hydroxide Sludge for Fluoride and Arsenic Removal in Aqueous Solution: Characterization, Kinetic, and Equilibrium Studies. Water Air Soil
Bayesteh, H., Mirghasemi, A.A., (2015) Numerical simulation of porosity and tortuosity effect on the permeability in clay: Microstructural approach. Soils and
Foundations 55, 5 1158-1170
Miaosi, L., Rong, C., Azadeh, N., Liyun, G., Xiwang, Z., Wei, S., (2015) Periodic-table-style" paper device for monitoring heavy metals in water. Analytical chemistry 87, 5 2555-2559
