ARTICLE

pubs.acs.org/est

Thermodynamic Analysis for the Controllability of Elements in the

Recycling Process of Metals
Kenichi Nakajima*,†
†
Research Center for Material Cycles and Waste Management, National Institute for Environmental Studies, Ibaraki 305-8506, Japan

Osamu Takeda,‡ Takahiro Miki,‡ Kazuyo Matsubae,‡ and Tetsuya Nagasaka‡

‡
Graduate School of Engineering, Tohoku University, Miyagi 980-8579, Japan

ABSTRACT: This study presents the results of chemical thermodynamic

analysis on the distribution of elements in the smelting process of metallic
materials to examine the controllability of impurities in the pyrometallurgical
technique. The results of the present work can give an answer against the
frequently given question; “Which impurity element can be removable in
metallurgical process?” or “How far can the impurity level be controlled?”.
The proposed method was applied to estimate the distribution of 29 elements for
a copper converter and 26 elements for a steel-making process and shows the
distribution tendency of elements among the gas, slag, and metal phases as well as
clarifying which metals can be recovered or removed from secondary resources in
metallurgical processes. The effects of temperature, oxygen partial pressure, and
slag composition on the distribution ratio of elements were also evaluated, and
the removal limit or controllability of impurity in these two processes was
presented. This study results in thermodynamic features of various elements in the pyrometallurgical process and also shows, even
by varying process parameters such as temperature and oxygen partial pressure, no drastic improvement of removal efficiency should
be expected, except for lead and tin in copper.

1. INTRODUCTION One of the best tools to predict and understand the behavior
Material flow analysis (MFA) is an excellent tool to quantify of impurities in the recycling of EoL metal products is a knowl-
material balance in specific areas globally,1 nationally,2 regionally,3 and edge of metallurgical physical chemistry, which is a well-estab-
by industry.4 It is also useful to systematically evaluate the flows of lished academic area studying the production of primary metals
substances/materials for resource and waste and the structure of or inorganic materials from natural minerals through pretreat-
resource consumption. Representative studies for metals, such as ment, smelting, and refining with hydro- and pyrometallurgical
aluminum,3 iron,5 copper,6 and chromium,7 including recycled technologies. Using these types of technologies, it should also be
resources and metal scrap, have provided important information for possible to predict the possibility of the removal of impurity or
a better understanding of the structure of metal flows. Discussions of recovery of elements from remelted metal scrap.
the complex web of metals and their linkages (relationships of In our previous articles, extensive discussion was made on the
elemental coexistence in natural and secondary resources), however, thermodynamic behavior of impurity elements in the metal, slag,
have been insufficient in most traditional MFAs, except for a few and gas phases of the steel-making process (both for basic oxygen
studies.8
12 The studies conducted by Murakami et al.8 and furnaces, BOF, and electric arc furnaces, EAF), copper convert-
Nakamura et al.9,10 incorporated a smelting model to account for ers, lead blast furnaces, the imperial smelting process (ISP) of
metal linkages, including the recovery of silver from the lead zinc and lead,14 the remelting process of aluminum,15 and the
metallurgical process. In their extensive works, Verhoef et al.11 and remelting process of magnesium.16 In the metallurgical pro-
Reuter and Verhoef12 suggested the importance of understanding cesses, impurities in main metal product are mostly controlled by
metal linkages in natural resource processing by introducing the oxidization
reduction reactions and evaporation except for
concept of the “metal wheel”, and Schaik13 pointed out the limitation some special cases. Taking such technological status into ac-
of recycling based on thermodynamic consideration. Even these count, the distribution ratio of impurity elements between the
representative studies, however, do not provide us the quantitative slag/metal or gas/metal phases was quantitatively evaluated in
limitations of impurity removal and the recoverability of elements in
the recycling of end-of-life (EoL) metal products. In this study, our Received: December 16, 2010
aim was to develop quantitative criteria for evaluating the possibility of Accepted: April 25, 2011
removing impurities during the metal recycling process by considering Revised: April 20, 2011
all the relevant thermodynamic parameters. Published: May 11, 2011

r 2011 American Chemical Society 4929 dx.doi.org/10.1021/es104231n | Environ. Sci. Technol. 2011, 45, 4929–4936
Environmental Science & Technology ARTICLE

Figure 1. Distribution chart of elements among gas, slag, and metal phases for the metal recovery under the simulated atmosphere of the converter of
copper smelting.

this study. The most conventional operating conditions for each conventional pyrometallurgical processes, volatilization and oxidiza-
process of metal have been chosen as process parameters to tion, were selected. In each case, the distribution ratio of a given
simplify discussions. To graphically represent the results of our element M is thermodynamically obtained with following equations
evaluation, a chart named as the “element radar chart”15 plotting
MðlÞ ¼ MðgÞ ð1Þ
the distribution tendencies of the metal elements processed by
the methods previously described was proposed.
In our previous paper, the distribution ratio of impurity pM =po aM poM po γ xM
Lg=m ¼ ¼ ¼ M M ð2Þ
elements among the metal, slag, and gas phases that would occur pSolv =p o pSolv pSolv
in a smelting furnace under typical operating conditions was
calculated as a first approximation. However, as described later in n
MðlÞ þ O2 ðgÞ ¼ MOn ðslagÞ ð3Þ
this article, the removability of impurities from remelted metal by 2
evaporation or oxidation is simultaneously affected by several
thermodynamic parameters, such as the Gibbs energy change of a MOn x MOn γMOn
K3 ¼ n=2
¼ ð4Þ
the oxidation reaction, total pressure, activity coefficients of the aM ðpO2 =p°Þ γM xM ðpO2 =p°Þn=2
impurity in the metal and its oxide in slag, oxygen partial pressure,
and temperature. Activity coefficients are also dependent on the
xMOn K3 γM ðpO2 =p°Þn=2
composition of the metal and slag. When the Gibbs energy Ls=m ¼ ¼ ð5Þ
change of the reaction exhibits a large negative value or when a xM γMOn
sufficiently low activity coefficient of oxidation product is where poM, pM, pSolv, and po are the partial pressures of the pure
achieved by selecting an adequate slag composition, the impurity element M (Pa), of M dissolved in the main metal product (solvent
can be removed even with a small activity coefficient in the metal, metal) (Pa), of the solvent metal vapor (Pa), and conversion factor
which is met in the practical process of refining of primary steel (101325 Pa/atm); aM and γM are the activity and the activity
and copper (e.g., dephosphorization of molten steel). Without coefficient of M in the pure liquid standard state; xM is the mole
knowing the effect of several process parameters on the distribu- fraction of M in the solvent metal; aMOn and γMOn are the activity
tion behavior, the removal limit of impurities from the remelted and the activity coefficient of the oxidation product MOn in slag in
metal cannot be precisely known. the pure solid standard state; xMOn is the mole fraction of MOn in
In this article, the effects of process parameters, such as slag; and pO2 is the oxygen partial pressure (Pa). Symbols of l, g, and
temperature, oxygen partial pressure, and slag composition, on slag in parentheses in eqs 1 and 3 denote the state of species as in
the distribution ratio of elements in the steel-making process and liquid solvent metal, in gas phase, and in slag phase, respectively. K3
a copper converter are theoretically evaluated, and thermody- is the equilibrium constant of reaction 3 and is given from the Gibbs
namic criteria on the removal limit or the controllability of free energy change of the reaction ΔGo3 by eq 6
impurities in these two processes are presented. Discussion is  
extended to the potential effectiveness and limits of extractive ΔGo3
K3 ¼ exp
ð6Þ
metallurgical processing as a recycling method. RT
where R and T denote the gas constant and temperature (K),
2. METHODOLOGY respectively. In the oxidation reaction 3, a monocationic oxide
The thermodynamic methodology used in this study is essentially species is assumed as an oxidation product throughout this paper.
the same as that described in our previous articles.14,15 The most For example, in the case of oxidation of aluminum in the solvent
4930 dx.doi.org/10.1021/es104231n |Environ. Sci. Technol. 2011, 45, 4929–4936
Environmental Science & Technology ARTICLE

Figure 2. Distribution chart of elements among gas, slag, and metal phases for the metal recovery under the simulated atmosphere of converter of steel making.

metal, the product is AlO1.5 rather than its general stoichiometric the regular solution behavior using eq 6a
form of Al2O3. Dissociation reaction of general stoichiometric oxide
_ Mix _ XS
to monocationic oxide is represented in eq 7 ΔHM ¼ G M ¼ RT ln γM ¼ Rð1
xM Þ2 ð6aÞ
Mx Oy ðslagÞ ¼ xMOn ðn ¼ y=xÞ ð7Þ _ _
where G XS M and ΔH M are the partial molar excess Gibbs energy
Mix

The free energy change of this reaction is reasonably assumed of mixing and the partial molar enthalpy of solution of M,
zero.17
19 Therefore, ΔGo3 and K3 can be evaluated from general respectively. The parameter R is the so-called alpha-function,
thermodynamic references. which is a temperature and composition independent constant in
The distribution ratio Lg/m and Ls/m quantitatively show how the the regular solution. Because the values of γM for mercury,
vapor pressure of M is bigger than that of solvent and how the rhenium, and rhodium in molten copper are unknown, they were
concentration of oxidation product MOn in slag is higher than that assumed to be unity as a first approximation. The solubilities of
of M in the solvent metal at given condition, respectively. A greater calcium, lead, and silver in molten iron are less than 1 mol % at
distribution ratio results in easier removal of M into the gas or 1873 K, so that the activities and vapor pressures were assumed
slag phase. to be those for the pure state for these elements.
For the calculation of Lg/m and Ls/m in eqs 2 and 5, the Tungsten, boron, aluminum, chromium, gallium, iron, germanium,
parameters were referred from published thermodynamic manganese, indium, magnesium, zinc, and strontium can be trans-
data. 20
27 ferred to slag from molten copper by oxidation, and mercury can be
removed by evaporation (Figure 1). Nickel, rhenium, tin, cadmium,
3. RESULTS and lead are partially distributed to the slag phase. On the other hand,
precious metals (silver, gold, platinum, palladium, and rhodium),
3.1. Element Distribution Tendencies. Figures 1 and 2 show bismuth, selenium, tellurium, and antimony remain in the molten
the calculated Ls/m and Lg/m for remelted copper and iron, copper phase. Most of the elements distributed to the metal phase can
respectively. In Figure 1, the distribution of 28 elements is shown be removed or recovered in further processing, such as electro-
under a copper converter’s typical operating conditions, that is, at refining (electrolysis) or hydrometallurgical processes,33 and some of
1500 K and pO2 = 1  10
1 Pa.30 In Figure 2, the distribution the elements distributed to the gas phase can also be recovered in
of 26 elements is shown under conventional operating condi- further processing.34 The recoverable elements are shown as white
tions of a BOF or an EAF for steel making, that is, at 1873 K and circles in Figure 1, and the gray circles denote the typical alloying
pO2= 1.9  10
5 Pa,25,31 where pO2 is calculated from the elements for industrial copper products. Most of the alloying elements
equilibrium between the metal and the slag. In both figures, can be removed or their concentrations can be managed by control-
the initial impurity content in the copper or iron was set at ling the processing conditions.
1 mol % (xM = 0.01) where γM is constant due to the dilute More elements tend to be distributed into the slag or gas phase in
concentration range. The activity coefficient of MOn was steel, although tungsten, molybdenum, cobalt, nickel, tin, and copper
assumed to be unity as a base case. The boundary between metal remain in molten iron. There is essentially no industrial process to
and gas was assumed to be the condition that satisfies both recover elements from slag and dust in steel making, with the
log(pM/pSolv) = 1 and log(pM) =
3. 32 For the calculation of exception of zinc recovery from EAF dust of carbon steel
Ls/m and Lg/m, the activity coefficients γM were sourced from production.35 Typical alloying elements for steel products are
published literature.21,25
29 When the temperature and the denoted as gray circles in Figure 2. Deoxidizing agents for steel
concentration ranges for the published values of γM differed refining, which have a large negative distribution ratio, are shown as
from the calculated conditions, they were estimated by assuming black triangles. They are distributed into the slag phase.
4931 dx.doi.org/10.1021/es104231n |Environ. Sci. Technol. 2011, 45, 4929–4936
Environmental Science & Technology ARTICLE

Figure 3. Distribution ratio of elements between metal and slag in a copper converter.

Figure 4. Distribution ratio of elements between metal and slag in a steel-making process.

3.2. Effect of Temperature and Oxygen Partial Pressure on copper and slag. For evaluating the possibility of change of the
the Distribution Ratio. Figures 1 and 2 visually demonstrate the distributed phase, we selected tin, lead, nickel, and indium that
elements that tend to enter the metal, slag, and gas phases under have a distribution ratio ranging from 103 to 10
2. The
conventional metallurgical processing for copper and steel. How- distribution ratios Ls/m for lead decreased with increasing
ever, operating conditions vary depending on facilities, scrap grade, temperature and crossed over zero, indicating lead tends to
and other factors. Therefore, the effects of process parameters on remain in the metal phase at higher temperatures. However, the
the distribution tendencies were further investigated. effect of temperature on the distribution ratio was not remark-
Figure 3 shows the effect of temperature and oxygen partial ably large. The distribution ratio increased with increasing
pressure on the distribution ratio of some elements between molten oxygen partial pressure. For example, most tin remains in
4932 dx.doi.org/10.1021/es104231n |Environ. Sci. Technol. 2011, 45, 4929–4936
Environmental Science & Technology ARTICLE

Figure 5. Variation of the activity coefficients of FeO and MnO in CaO-Al2O3-SiO2 slag at 1873 K calculated by a regular solution model.

Figure 6. Variation of final manganese content in remelted steel scrap with

mass of CaO-Al2O3-SiO2 slag at 1873 K. The compositions of slag A and slag B
are shown in Figure 6. Initial manganese content in molten steel was 1 mol %. Figure 7. An example of the removal limit of an element (Pb) in liquid
copper by evaporation.

molten copper at pO2 = 10
2 Pa, but it tends to be removed into Gibbs energy change of eq 7 gives the maximum oxygen partial
slag at pO2 = 10 Pa. This distribution tendency indicates that a pressure at 1500 K as 8.3 Pa, above which significant oxidation
higher pO2 is favorable for the oxidative removal of impurities in loss of molten copper occurs.
a solvent metal. The oxygen partial pressure, however, should The effect of temperature and oxygen partial pressure on the
be less than the maximum pressure determined by the equilib- distribution ratio between molten iron and slag is presented in
rium of molten copper as the solvent metal with copper oxide in Figure 4 for niobium, manganese, vanadium, chromium, tungsten,
the slag, as shown in eq 7a molybdenum, nickel, tin, and copper, all of which have distribution
ratios ranging from 104 to 10
4. In all cases, the distribution ratio
2CuðlÞ þ 1 =2 O2 ðgÞ ¼ Cu2 Oðin slagÞ ð7aÞ Ls/m decreased with increasing temperature, indicating that the
4933 dx.doi.org/10.1021/es104231n |Environ. Sci. Technol. 2011, 45, 4929–4936
Environmental Science & Technology ARTICLE

Figure 8. Flow of elements associated with the copper production process. Copper, sulfur, and oxygen are the primary materials; minor elements and
gangue minerals are not included. PGM denotes platinum group metals, including Ir, Rh, Ru, Os, Pd and Pt.

alloying elements tend to remain in the molten iron phase at higher model is applicable for the entire composition range of molten
temperatures. Once again, however, the effect of temperature was CaO-SiO2-Al2O3 slag containing 5 mol % MnO and FeO. The
not remarkably large. Increasing oxygen partial pressure increased activity coefficients of MnO and FeO were also assumed to be
the distribution ratios. The oxygen partial pressure should again be constant from 0 to 5 mol % MnO and FeO.
limited to below the Fe/FeO equilibrium (molten iron as the When molten iron containing manganese reacts with CaO-
solvent metal with iron oxide in the slag) given by eq 8 SiO2-Al2O3-FeO slag, the exchange reaction of eq 9 takes place,
and manganese transfers as MnO into the slag while a corre-
FeðlÞ þ 1 =2 O2 ðgÞ ¼ FeOðin slagÞ ð8Þ sponding amount of FeO decreases as the reaction proceeds
The critical oxygen partial pressure determined by the equilibrium Mnðin FeðlÞÞ þ FeOðin slagÞ ¼ MnOðin slagÞ þ FeðlÞ ð9Þ
between pure molten Fe and pure solid FeO was calculated to be
4  10
4 Pa at 1873 K, above which significant oxidation loss of The prediction of final manganese content by the refining with
molten iron occurs. Even at a high pO2 (just below the critical value), CaO-SiO2-Al2O3-FeO slag has been made where two typical slag
copper and tin are very hard to remove by oxidation (figure 4b). compositions shown in Figure 6 (Slag A (γFeO = 1.61, γMnO =
Copper and tin are both recognized as typical harmful impurities 0.43) and Slag B (γFeO = 3.22, γMnO = 1.27)) have been selected.
(so-called tramp elements36) for steel. The results are plotted in Figure 7 as a function of the amount of
3.3. Effect of Slag Composition on the Distribution Ratio Ls/m. slag generated. Because the activity coefficients of MnO and FeO
In the previous sections, the activity coefficient of the oxidation exhibit similar composition dependencies, the final manganese
product MOn in slag was assumed to be unity as a base case. However, content of the two slags was similar. The final manganese content
in practical operations, the slag composition is selected or designed to of the steel decreases almost linearly as the amount of slag
control the distribution behavior of elements between metal and slag generated increases in the range below 120 kg-slag/t-metal.
because the activity or activity coefficient changes depending on the Generally the amount of slag generated for refining of steel is
slag composition. For this reason, quite a lot of research has been less than 100 kg-slag/t-steel, except for very special cases, so
conducted to study the activity coefficients of oxides in various kinds that slag A or B cannot lower manganese in remelted steel below
of slags.37 However, most of the studies have focused on conventional xMn = 0.005 which is the upper limit of manganese in carbon tool
impurity elements whose origins are primary raw materials, such as steel. Final manganese content will decrease as the amount of slag
iron ore, copper concentrate, coke, and lime. Thermodynamic increases, but this would lead to increased costs and energy
information on elements whose origins are secondary resources, such consumption. Therefore, in real-world operations, an optimum
as scrap, remains quite limited. slag amount is chosen both from the specification of products
Figure 5 shows the iso-activity coefficient contours of FeO and and the cost performance.
MnO in CaO-SiO2-Al2O3 ternary slag37 at 1873 K. The contours The behavior of other elements can also be predicted if the
were obtained by “quadratic formalism” based on the regular activity coefficients of the reactant and oxidant are known for a
solution model.31,38 For the calculation, it was assumed that this given slag system. The variable range of the slag/metal
4934 dx.doi.org/10.1021/es104231n |Environ. Sci. Technol. 2011, 45, 4929–4936
Environmental Science & Technology ARTICLE

distribution ratios was calculated by changing the activity coeffi- calcium treatment 40 or nonmetallurgical solutions13 should be
cient of MOn in slag from 0.1 to 10 for typical alloying elements developed for sustainable supply of high metal products.
(Cr, Co, Mn, Mo, Ni, V, W) in industrial steel products. The This is not the case for copper and lead smelting. The primary
calculated ranges are presented as bars on the distribution ratios metallurgical processing of copper and lead are generally fol-
shown in Figure 2. As the figure indicates, these elements are still lowed by an electro-refining process (electrolysis) and extraction
not controllable, even by changing the activity coefficients of process as shown in Figure 8, which enables the effective removal
oxidation products, because the distribution ratios in initial and recovery of elements from crude metals. Figure 8 shows the
conditions are apart from zero. flow of elements associated with the copper production process.
3.4. Limitations of Evaporation Refining. As indicated in Copper, sulfur, and oxygen as a primary material, minor ele-
Figures 1 and 2, some elements can be removed or controlled by ments, and gangue minerals are not included in this figure.
evaporation from molten copper or steel, but there are limits to Figure 8 shows that some valuable elements, bismuth, gold,
removal by evaporation. The vapor pressure of the objective silver, and PGMs (platinum group metals), including copper can
element decrease while that of solvent metal slightly increases as be extracted into an anode slime in EoL products such as e-waste.
the evaporation proceeds. As a result, the vapor pressure of the In fact, metals nobler than copper, including PGMs, remain in
objective element and solvent metal become closer, and preferential copper and lead after processing in converter (Figure 1). These
removal of the impurity eventually becomes impossible. As a typical noble metals can be recovered from anode slime after electrolysis
example, we calculated the removal limit for lead from molten by an extraction process such as a hydrometallurgical process, for
copper by a vacuum treatment (Figure 7).39 The assumed initial example, a solvent extraction process.33 Thus, the contamination
mole fraction of lead in molten copper was 0.01 (3.2 mass%), and does not become an essential problem for copper and lead when
the vapor pressure of lead was 249 Pa and that of copper was 0.82 Pa. scrap containing these noble metals is supplied to the ordinary
When this molten copper alloy is treated under vacuum, essentially smelter. Even if PGMs are mixed into the copper scrap from
only the lead evaporates from the molten Cu
Pb alloy and the e-waste, they can be finally recovered via anode slime (Figures 1
concentration of lead decreases. As the process proceeds, the lead and 8).
pressure decreases while that of copper slightly increases. The lowest
pressure of about 0.1 Torr (13.3 Pa) can be attained in a ’ AUTHOR INFORMATION
metallurgical vacuum facility. Hence, lead in copper will be removed
only until its partial pressure reaches the critical value of xPb = 5.3  Corresponding Author
10
4 (0.18 mass%) for molten copper at 1500 K. The major use of *Phone: þ81-29-854-2744. E-mail: nakajima.kenichi@nies.
copper is electrical applications, and it needs a 4N (99.99%) purity go.jp.
to be highly conductive. Therefore, lead should be removed from
remelted copper scrap by a different technique if the refined copper
is to be used in electrical applications. ’ ACKNOWLEDGMENT
This research was partially supported by JSPS (KAKENHI
22360387, KAKENHI 22360218). Financial supports given by
4. DISCUSSION the ISIJ Innovative Program for Advanced Technology, The Iron
The results of the present work can give an answer against the and Steel Institute of Japan, in 2008-2010 and ISIJ Research
frequently given question: “Which impurity element can be group, Automobile Recycling from Material Industry's perspec-
removable in metallurgical process?” or “How far can the tive are also gratefully acknowledged.
impurity level be controlled?”. The element radar chart, shown
in our previous article,15 indicates that removing impurities is
much more difficult from aluminum and magnesium than from ’ REFERENCES
the other metals listed. Even so, some elements are also difficult (1) Rauch, J.; Pacyna, J. Earth’s global Ag, Al, Cr, Cu, Fe, Ni, Pb, and
to remove from molten copper and iron, as shown in Figures 1 Zn cycles. Global Biochemical Cycles 2009, 23, GB2001.
and 2. Even by varying process parameters such as temperature (2) Adriaanse, A.; Bringezu, S.; Hammond, A.; Moriguchi, Y.;
and oxygen partial pressure, no drastic improvement of removal Rodenburg, E.; Rogich, D.; Schutz, H. Resource flows: the material basis
efficiency should be expected, except for lead and tin in copper of industrial economies; World Resource Institute: Washington, DC,
1997.
(Figures 3, 4, and 6). (3) Recalde, K.; Wang, J.; Graedel, T. E. Aluminum in-use stocks in
The disadvantage of steel making processing, remelting of the state of Connecticut. Resour. Conserv. Recycl. 2008, 52, 1271–1282.
aluminum, and remelting of magnesium is that the processes (4) Nakajima, K.; Yokoyama, K.; Nagasaka, T. Substance flow
cannot effectively remove impurities. Typical alloying elements for analysis of manganese associated with iron and steel flow in Japanese
steel are national strategic metals in Japan. As shown in Figure 2, economy. ISIJ Int. 2008, 48, 549–553.
not only nickel but also tungsten, molybdenum, and cobalt are (5) Wang, T.; Muller, D. B.; Graedel, T. E. Forging the anthropo-
very stable in molten iron. This thermodynamic feature strongly genic iron. Environ. Sci. Technol. 2007, 41, 5120–5129.
indicates that loss of or contamination by these uncontrollable (6) Graedel, T. E.; Bertram, M.; Kapur, A.; Reck, B.; Spatari, S.
elements could occur in conventional recycling systems of steel, Exploratory data analysis of the multilevel anthropogenic copper cycle.
Environ. Sci. Technol. 2004, 38, 1253–1261.
aluminum, and magnesium. The removal or control of impurities
(7) Jonson, J.; Schewel, L.; Graedel, T. E. The contemporary
is essential in this type of system. In order to effectively use these anthropogenic chromium cycle. Environ. Sci. Technol. 2006, 40, 7060–
elements as important alloying elements rather than contamina- 7069.
tions, it is important to develop a presorting system for scrap and (8) Murakami, S.; Yamanoi, M.; Adachi, Y.; Mogi, G.; Yamatomi, J.
accurate scrap specifications that take material composition into Material flow accounting for metals in Japan. Mater. Trans. 2004,
account. Needless to say, new metallurgical solutions such as 45, 3184–3193.

4935 dx.doi.org/10.1021/es104231n |Environ. Sci. Technol. 2011, 45, 4929–4936

Environmental Science & Technology ARTICLE

(9) Nakamura, S.; Nakajima, K.; Kondo, Y.; Nagasaka, T. The waste (34) Tamura, N.; Miura, H. Treatment of mercury-contaminated
input-output approach to material flow analysis. J. Ind. Ecol. 2007, wastes at Itomuka plant of Nomurakohsan Co. Ltd. J. MMIJ 2007,
11, 50–63. 123, 747–750.
(10) Nakamura, S.; Murakami, S.; Nakajima, E.; Nagasaka, T. Hybrid (35) Nakajima, K.; Matsubae-Yokoyama, K.; Nakamura, S.; Itoh, S.;
Input-Output Approach to Metal Production and Its Application to the Nagasaka, T. Substance flow analysis of zinc associated with iron and
Introduction of Lead-Free Solders. Environ. Sci. Technol. 2008, steel cycle in Japan, and environmental assessment of EAF dust recycling
42, 3843–3848. process. ISIJ Int. 2008, 48, 1478–1483.
(11) Verhoef, E. V.; Dijkema Gerard, P. J.; Reuter, M. A. Process (36) Katayama, H.; Mizukami, Y. Recycling of ferrous scrap. Mater.
knowledge, system dynamics, and metal ecology. J. Ind. Ecol. 2004, 8, 23–43. Jpn. 1996, 35, 1283–1289.
(12) Reuter, M. A.; Verhoef, E. V. A dynamic model for assessment (37) Chemical Properties of Molten Slags; Ban-ya, S., Hino, M., Eds.;
of the replacement of lead solders. J. Electron. Mater. 2004, 33, The Iron Steel Inst. Jpn.: Tokyo, 1991.
1567–1580. (38) Ban-ya, S.; Hino., M. Mathematical expression of slag-metal
(13) Schaik, A. V.; Reuter, M. A. Dynamic modeling of E-waste equilibria in steelmaking Process by Quadratic Formalism Based on
recycling system performance based on product design. Miner. Eng. Regular Solution Model. Tetsu to Hagane 1988, 74, 1701–1711.
2010, 23, 192–210. (39) Wang, C. P.; Liu, X. J.; Ohnuma, I.; Kainuma, R.; Ishida, K.
(14) Nakajima, K.; Takeda, O.; Miki, T.; Nagasaka, T. Evaluation Thermodynamic assessment of the Cu-Ni-Pb system. CALPHAD 2000,
method of metal resources recyclability based on thermodynamic 24, 149–167.
analysis. Mater. Trans. 2009, 50, 453–460. (40) Isawa, T.; Wakasugi, T.; Noguchi, K.; Sano, N. Thermody-
(15) Nakajima, K.; Takeda, O.; Miki, T.; Matsubae, K.; Nakamura, namics of Antimony, Arsenic, and Tin in CaO-CaF2Melts. Steel Res.
S.; Nagasaka, T. Thermodynamic Analysis of Contamination by Alloy- 1987, 58, 296–302.
ing Elements in Aluminum Recycling. Environ. Sci. Technol. 2010,
44, 5594–5600.
(16) Hiraki T.; Takeda O.; Nakajima K.; Matsubae K.; Nakamura S.;
Nagaska T. Thermodynamic criteria for the removal of impurities from
end-of-life magnesium alloys by pyro-metallurgical techniques, Sci.
Technol. Adv. Mater. (in reviewing).
(17) Rosenqvist, T. Principles of Extractive Metallourgy, 2nd ed.; Tech
Book: Herndon, VA, USA, 1983.
(18) Yazawa, A. Thermodynamic consideration of copper smelting.
Can. Metall. Quart. 1974, 12, 443–453.
(19) Yazawa, A. Distribution of various elements between copper,
matte and slag. Erzmetall 1980, 33, 377–382.
(20) Barin, I. Thermochemical Data of Pure Substances; VCH Ver-
lagsgesellschaft mbH: Weinheim, Germany, 1989.
(21) Hultgren, R. Selected thermodynamic values and phase diagrams
for copper and some of its binary alloys; International Copper Research
Association: New York, 1971.
(22) Mehta, S. R.; Richardson, F. D. Activities of Manganese Oxide
and Mixing Relationships in Silicate and Aluminate Melts. J. Iron Steel
Inst. 1965, 203, 524–528.
(23) Kawai, Y.; Shiraishi, Y. Handbook of physico-chemical properties
at high temperatures; The Iron and Steel Institute of Japan: Tokyo, 1988.
(24) Turkdogan, E. T. Physical Chemistry of High Temperature
Technology; Academic Press: New York, 1980.
(25) The Japan Society for Promotion of Science. The 19th Com-
mittee in Steelmaking: Steelmaking Data Sourcebook; Gordon and Breach
Science Publishers: New York, 1988.
(26) Zhang, X.; Han, Q.; Chen, D. Dissolution equilibrium of
magnesium vapor in liquid-iron. Metall. Trans. B 1991, 22B, 918–921.
(27) Su, X.; Tang, N.; Toguri, J. M. Thermodynamic evaluation of
the Fe-Zn system. J. Alloys Compd. 2001, 325, 129–136.
(28) Jayakumar, M. V.; Sriramamurthy, A. M.; Naidu, S. V. N.
Calculated phase diagrams of Cu-W, Ag-W and Au-W binary systems.
CALPHAD 1988, 12, 177–184.
(29) Risold, D.; Hallstedt, B.; Gauckler, L. J.; Lukas, H. L.; Fries, S. G.
Thermodynamic optimization of the Ca-Cu and Sr-Cu systems. CAL-
PHAD 1996, 20, 151–160.
(30) Kurokawa, H.; Kemori, N. Reaction mechanism in copper-
making stage of PS copper converter. Shigen to Sozai 2003, 119,
55–60.
(31) Ban-ya, S.; Hino, M. Calculation of activities of the constituents
in FetO-(CaOþMgO)-(SiO2þP2O5) Slags by regular solution model.
Tetsu to Hagane 1987, 73, 476–483.
(32) The Japan Institute of Metals. Extractive Metallurgy; The Japan
Institute of Metals: Sendai, 87, 1999.
(33) Kurokawa, H.; Asano, S.; Nakamura, S.; Nagai, H. Develop-
ment of PGMs metallurgical process. Shigen to Sozai 2004,
120, 280–283.

4936 dx.doi.org/10.1021/es104231n |Environ. Sci. Technol. 2011, 45, 4929–4936