See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/293330298

Recovery of gold from waste electrical and electronic equipment (WEEE) using
ammonium persulfate

Article  in  Waste Management · February 2016

DOI: 10.1016/j.wasman.2016.01.043

CITATIONS READS

20 356

3 authors, including:

Andrea Alzate María Esperanza López

University of Antioquia University of Antioquia
5 PUBLICATIONS   20 CITATIONS    52 PUBLICATIONS   131 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Elaboración de un cemento óseo con una adición de biovidrio View project

Desarrollo de un nuevo proceso de lixiviación selectiva de bajo impacto ambiental con el ion persulfato (S2O82-) para la extracción de oro y neodimio desde residuos
eléctricos y electrónicos View project

All content following this page was uploaded by María Esperanza López on 14 October 2018.

The user has requested enhancement of the downloaded file.

 Waste Management 57 (2016) 113–120

Contents lists available at ScienceDirect

Waste Management
journal homepage: www.elsevier.com/locate/wasman

Recovery of gold from waste electrical and electronic equipment (WEEE)

using ammonium persulfate
Andrea Alzate a,b,⇑, Maria Esperanza López a, Claudia Serna a
a
GIPIMME Research Group, Department of Materials Engineering, University of Antioquia, CL 67 53-108, Medellin, Colombia
b
Ingeniería, Suministros y Montajes S.A.S, INSUMON S.A.S, CL 36 36-9, Medellín, Colombia

a r t i c l e i n f o a b s t r a c t

Article history: This paper presents a novel methodology to recover gold from waste electrical and electronic equipment
Received 3 November 2015 (WEEE) using ammonium persulfate ((NH4)2S2O8). Gold was recovered as a fine coating using substrate
Revised 28 January 2016 oxidation without shredding or grinding process. The WEEE sample was characterized giving values of
Accepted 30 January 2016
Au: 1.05 g/kg, Fe: 86.00 g/kg, Ni: 73.64 g/kg, Cu: 26.65 g/kg. The effect of (NH4)2S2O8 concentration
Available online 6 February 2016
(0.22–1.10 M), oxygen (0.0–1.4 L/min) and L/S ratio (10–30 mL/g) on the main responses (substrate oxi-
dation and Au recovery) was investigated implementing response surface methodology with numerical
Keywords:
optimization. A quadratic model was developed and quantities greater than 98% of Au were recovered.
Gold recovery
Waste electrical and electronic equipment
The findings presented suggest that, optimized quantities of ammonium persulfate in aqueous highly
(WEEE) oxygenated media could be used to extract superficial gold from WEEE.
Ammonium persulfate Ó 2016 Elsevier Ltd. All rights reserved.
Response surface methodology
Recycling

1. Introduction
Non polluting methodologies of valuable metals recovery from
secondary sources have been a current topic in several studies
because of their low environmental impact. Many of these studies
have focused on recovering gold avoiding the hazardous implica-
tions of mineral extraction such as water pollution, deforestation
and health repercussions. Nowadays, one of the main secondary
sources to recover precious (Au, Ag, Pd) and base metals (Cu, Ni,
Fe) is waste electrical and electronic equipment (WEEE) (Yazici
and Deveci, 2014). The rapid growth in the manufacturing of tech-
nological devices has generated large quantities of WEEE. The
annual WEEE global growth has been estimated at 8.8% (2004–
2011) and 17.6% (2011–2016) calculating for 2016 a global volume
of 93.5 millions of tons (Akcil et al., 2015; Yu et al., 2014). These
values expose the adequate final disposal of WEEE as one of the
main challenges in WEEE management due to waste of valuable
metals (Hadi et al., 2013) and the environmental problems associ-
ated with conventional disposal and recycling methods.
Conventional disposal methods of WEEE such as landfill, incin-
eration and hydrometallurgical recycling techniques generates soil
and water pollutants and harmful substances that are released into
⇑ Corresponding author at: GIPIMME Research Group, University of Antioquia, CL
67 53-108, Medellin, Colombia.
E-mail address: andreaalzatenaranjo@gmail.com (A. Alzate).
the air. On the first hand, landfill and incineration produce toxic
substances that comprehend Hg, Pd, Cd, dioxins, furans and heavy
metals vapors (Yazici and Deveci, 2013). However, developing
countries still use landfill and incineration causing public health
and environmental risks. Risks of landfill leachate production and
air, soil, water or underwater pollution, make these disposal pro-
cess unsuitable (Hadi et al., 2015a). On the other hand, hydromet-
allurgical recycling techniques with strong acids (HCl, HNO3,
H2SO4/H2O2) or oxidative reagents (cyanide, thiourea, halide,
nitrate and iodide) have been suggested by a variety of authors
to process WEEE and reach the recovery of gold (Birloaga et al.,
2014; Petter et al., 2014; Shibayama et al., 2013). Methods of gold
recovery from WEEE with hydrometallurgical techniques focus on
primary physical separation (shredding, grinding) and secondary
leaching of total metallic fraction (Bas et al., 2014). The metallic
fraction (precious and base metals) is obtained after shredding
and grinding, which are recognized by dust pollution generation
(Naseri Joda and Rashchi, 2012). In addition, grinding fraction is
dissolved using strong acids or oxidative reagents (Zhang et al.,
2012) and the resulting solution is separated through time-
consuming chemical process (cementation, solvent extraction, pre-
cipitation or coagulation) with the aim to recover the metal of
interest from the solution (Syed, 2012). Hydrometallurgy exten-
sively use cyanide, thiourea, halides and some strong acids. These
agents are recognized by its toxic potential, low chemical stability
and environmental impact (Tuncuk et al., 2012).

http://dx.doi.org/10.1016/j.wasman.2016.01.043
0956-053X/Ó 2016 Elsevier Ltd. All rights reserved.
114 A. Alzate et al. / Waste Management 57 (2016) 113–120

Nowadays, strategies on WEEE management have been adopted agents due to the lack of research on its use and optimization in
to mitigate health and environmental problems. The strategies recovering gold from WEEE through substrate oxidation. In addi-
contemplate the development of tools as Material Flow Analysis tion, the produced persulfate ions (S2O2
8 ) are not absorbed or
(MFA), Multi Criteria Analysis (MCA) and Life Cycle Assessment bio accumulated in the soil after the process (Hernandez, 2005)
(LCA) (Kiddee et al., 2013). In develop countries, MFA and MCA and the generated by-products (sulfates) have not a negative effect
have been successful applied to estimate the generation of WEEE on the environment (Syed, 2006). Comparing with potassium per-
and take environmental decisions to solve multi-criteria problems sulfate, sodium persulfate and cupric chloride, (NH4)2S2O8 has a
on disposal, while LCA has been used in several studies to evaluate greater leaching power to oxidize base metals than potassium
the environmental impacts of WEEE (Wäger et al., 2011). Studies and sodium persulfate (Babu et al., 2002) and is operationally safer
on LCA concluded that, compare with landfill or incineration, recy- and less toxic than CuCl2. The toxicity reduction responds to the
cling techniques are more appropriate to manage WEEE (Kiddee absence of Cl2 (g) during the reaction.
et al., 2013). Nevertheless, any recycling process perform without In brief, the system of methods applied was carried out without
environmental care may produce a highest impact on soil, air, shredding or grinding stages which reduced secondary dust pollu-
water and humans. Developing suitable and optimal recycling tion, reaction time and contaminant by-products. This study aimed
methodologies capable of avoiding pollution, reducing reagents at maximizing gold recovery from WEEE with a novel methodology
toxicity and time-consuming reactions has become one of the most that includes the partial oxidation of metal substrate with ammo-
important topics in WEEE management research. nium persulfate in oxygenated media and the numerical optimiza-
Hence, in recent years, some methodologies that attempt to use tion of the most significant parameters.
environmentally friendly agents and the optimization of gold and
base metals recovery from WEEE have been suggested (Barbieri
2. Experimental
et al., 2010; Ha et al., 2014; Syed, 2006). For instance, alternative
agents to extract non-leaching gold that include potassium persul-
2.1. Materials and reagents
fate (K2S2O8) (Syed, 2006) and cupric chloride (CuCl2) (Barbieri
et al., 2010) were studied to recover gold from WEEE. These agents
Intel Celeron–Pentium electronic processor scrap from end-of-
were used to oxidize and leach the metal substrate (Ni, Fe, and Cu)
life computers supplied by a local recycling company was the sam-
where gold was superficially associated as coating (Barbieri et al.,
ple used (Fig. 1). A total amount of 50 processors were selected by
2010). The partial leaching of the substrate permitted gold recov-
reference, shape, weight, superficial distribution of gold and man-
ery in a solid particulate state (Barbieri et al., 2010; Syed, 2006).
ufacturer to ensure sample homogeneity and statistical signifi-
Due to the elimination of gold leaching, it was possible to reduce
cance. Intel Celeron and Intel Pentium processors of 5  5 cm2
reaction time avoiding purification stages and achieving the 98%
and an average weight of 8.97 g (Fig. 1) were used for chemical
in Au recovery with minimum formation of contaminant by-
characterization, substrate oxidation and gold recovery tests with-
products or total agent regeneration (Barbieri et al., 2010; Syed,
out shredding or grinding stages.
2006). Redox potential for the production of (SO2
4 ) ions in aqueous
A sample of 35 g was used to determine the amount of gold and
potassium persulfate was estimated in 2.01 V (Huang et al., 2002)
metal substrate (Fe, Ni, Cu) by chemical digestion using aqua-regia
and 0.48 V for the production of (CuCl2)
ions in acid solution
(Lee et al., 2011; Petter et al., 2014) followed by microwave plasma
(Lundström et al., 2009). These estimations exposed the ability of
– atomic emission spectroscopy (MP-AES, AGILENT 4100) (Table 1).
K2S2O8 and CuCl2 to oxidize Ni, Fe and Cu and release gold from
After chemical characterization, aqueous commercial grade ammo-
the substrate. Despites its great advantages, persulfate and cupric
nium persulfate (P98% (NH4)2S2O8) with a water solubility of
chloride oxidative systems to recover non-leaching gold from
850 g/L at 25 °C (Hernandez, 2005) was the selected environmental
WEEE have not been extensively investigated.
reagent used to produce sulfate ions (SO
4 ), which partially oxi-
Selective persulfate oxidation proposed by Syed (2006) cat-
dized the metal substrate breaking the Au–Cu–Ni–Fe bond and
alyzed with oxygen can be optimized in order to maximize the
allowing gold to be extracted in its original non-leaching state.
recovery of gold from WEEE reducing parameters like agent con-
sumption and reaction time (Birloaga et al., 2013; Hadi et al.,
2015b; Jordão et al., 2016). A way to optimize this process is to
use response surface methodology (RSM), a series of statistical
techniques. RSM has been adopted in several studies to find opti-
mal region of operation into an experimental design space
(Montgomery, 2013). This methodology was reported in WEEE
management to determine the greater amount of gold leaching
evaluating the incidence of thiosulfate, cooper and ammonia con-
centration (Ha et al., 2014). Besides, the extraction of Cu, Fe, Ni,
Ag and Pd from waste printed circuit boards (WPCBs) was studied
adopting RSM to establish the conditions that maximize metals
extraction using H2SO4–CuSO4–NaCl solutions (Yazici and Deveci,
2013).
In this study, an environmentally friendly methodology to
extract non-leaching gold by partial substrate oxidation from
WEEE using ammonium persulfate (NH4)2S2O8 was developed.
The incidence of (NH4)2S2O8 concentration, oxygen and liquid/solid
ratio on the recovery of gold was analyzed using response surface
methodology with numerical optimization. The effects of (NH4)2S2- Fig. 1. Processor scrap sample (ref. Intel Celeron–Intel Pentium).
O8 concentration (0.22–1.10 M), oxygen (0.0–1.4 L/min) and liq-
uid/solid ratio (10–30 mL/g) over the gold recovery were studied
in five different levels through a central composite design (CCD). An oxygen tank (99.9% O2) with flow control was used to deliver
(NH4)2S2O8 was selected over other environmentally friendly O2 and catalyze the oxidative reactions.
 A. Alzate et al. / Waste Management 57 (2016) 113–120
Table 1
Chemical composition of the processor scrap.
Element Fe Ni Cu Ag
Content (g/kg) 86.00 73.64 26.65 <0.001
2.2. Substrate oxidation and recovery of gold
Partial oxidation of substrate (Fe, Ni, Cu) was performed using
different concentration (0.22–1.10 M) of ammonium persulfate
(P98% (NH4)2S2O8) with a liquid solid rate variation between 10
and 30 mL/g. The solutions were prepared in deionized water by
heating at boiling point and reagent incorporation. Mechanical
stirring was implemented to reach complete dissolution and speci-
ation of persulfate (S2O8)2
in the interest oxidative anion (SO2

4 ).
Eq. (1) was calculated using HSC Chemistry software (2002).
1 þ
S2 O2

8 þ H2 O ! O2 þ 2SO2
0
4 þ 2H ðDGð95 CÞ ¼
121:29 kJÞ
2 model, the quadratic effect of the factors on the recovery of gold
Different solutions were transferred to glass reactors (400 mL
nominal capacity) where oxygen (99.9%O2) was added in different
flow rates (0.0–1.4 L/min) to maximize the oxidation of the sub-
strate and reduce reaction time. The solutions were heating at a
constant temperature of 95 °C (Syed, 2006) and the selected
amount of processor scrap samples were put into glass reactors
for a period of 10 min. During the reaction period, iron, nickel
and copper were partially oxidized and gold was released from
the substrate. The resulting solutions were analyzed by atomic
absorption spectroscopy (AAS, THERMO S4). Substrate oxidation
was determined using metallurgical balances between initial con-
centration of base metals in the scrap and final concentration into
the solution. The soluble by-products of the oxidative reactions
and the equilibrium constants were predicted using Medusa
software (2010) and HSC Chemistry software (2002).
Breaking Fe–Ni–Cu–Au bond was achieved by partial substrate
oxidation and gold was removed in its original solid state using
micro scale pressure washing with a subsequent process of filtra-
tion. Au recovery was calculated based on the initial gold content
in the scrap and the kinetics was determined measuring the time
dependent release of gold at predetermine intervals (2, 4, 6, 10
and 14 min) under the influence of oxidative parameters. Recovery
of Au was optimized using response surface methodology and a
second order mathematical model was developed. The chemical
composition and morphology of recovered gold was analyzed by
scanning electron microscopy (SEM/EDX, JEOL JSM-6490LV).
2.3. Sequential experimental design with numerical optimization
Study of the substrate oxidation was performed in previous
investigations adopting a full factorial 23 experimental design
(Montgomery, 2013). Combinations of factors and levels were done
by duplication to establish the experimental error. The selected
factors were A: Concentration of (NH4)2S2O8, B: Oxygen and C: Liq-
uid/solid ratio in two different levels (low (
1) and high (+1))
(Table 2).
Table 2
Experimental design for substrate oxidation, factors whit coded and actual levels.
Factors Coded levels
(
1)
A: [(NH4)2S2O8] (M) 0.44
B: Oxygen (L/min) 0.2
C: Liquid/solid ratio (mL/g) 15
115
Sixteen experiments were randomly performed giving as result
two main responses (substrate oxidation and Au recovery) that
Au were statistically analyzed using Design Expert software (2015).
1.05 The significance of the effects and its interactions over responses
were evaluated using analysis of variance (ANOVA) where p-
values were generated to prove the null hypothesis with a confi-
dence level of 95% (a = 0.05). Two empirical first order equations
were developed to evaluate the significance of the factors (A, B,
C) on the substrate oxidation and Au recovery. The equations were
established considering the regression linear model of three factors
with fixed effects (Eq. (2)) (Montgomery, 2013, 2011).
y ¼ b0 þ b1 x1 þ b2 x2 þ b3 x3 þ b12 x1 x2 þ b13 x1 x3 þ b23 x2 x3
þ b123 x1 x2 x3 ð2Þ
Eq. (2) was used to predict the behavior of the response (y)
related with the calculated linear (b1, b2, b3) and interaction (b12,
b13, b23, b123) coefficients and the symbols for each A, B, and C inde-
pendent factor (x1, x2 and x3). In order to adjust the first order
ð1Þ
was analyzed. Axial (
a, +a) and center (0) points were included
in the operation region and Au recovery was optimized using a
central composite design (CCD) with response surface methodol-
ogy (RSM) (Jordão et al., 2016; Rath et al., 2012). The effects of
the selected factors; A, B, C were investigated with an increase
from two to five levels (lowest (
2), low (
1), center (0), high
(+1) and highest (+2)) (Table 3).
The amount of runs for CCD was giving by (2k + 2k + N0) where k
is the number of factors, 2k is the starting factorial design, and N0 is
the number of experiments at the center. A total number of 3 con-
trollable factors, 23 factorial design with one replica and two repli-
cates in the center were used for a total of twenty-five
experimental runs. The response was statistically analyzed and
numerical optimization was established using Design Expert
software (2015). Linear and quadratic effects and their interaction
significance were evaluated by analysis of variance (ANOVA) with a
confidence level of 95% (a = 0.05). A quadratic polynomial equation
was developed to evaluate the significance of the factors (A, B, C) in
the Au recovery. The equation was established for a second order
polynomial model (Eq. (3)) (Montgomery, 2013; Bas et al., 2007;
Walpole et al., 2012). Where b11, b22 and b33 are the quadratic
coefficients.
y ¼ b0 þ b1 x1 þ b2 x2 þ b3 x3 þ b11 x21 þ b22 x22 þ b33 x23 þ b12 x1 x2
þ b13 x1 x3 þ b23 x2 x3 ð3Þ
3. Results and discussion
3.1. Substrate oxidation
Oxidation of substrate from the sample was run adopting all the
possible combination between factors and real levels. The experi-
ment design layout is shown in Table 4 where the individual per-
centage (%) for Cu, Ni and Fe oxidation is presented as secondary
responses. The secondary responses permitted calculating the total
substrate oxidation. In Table 4, it is possible to observe that when
Table 3
Experimental design for Au recovery, factors whit coded and actual levels.
Factors Coded levels
(+1)
a = (
2) (
1) (0) (+1) +a = (+2)
0.88 A: [(NH4)2S2O8] (M) 0.22 0.44 0.66 0.88 1.10
1.0 B: Oxygen (L/min) 0.0 0.2 0.6 1.0 1.4
25 C: Liquid/solid ratio (mL/g) 10 15 20 25 30
116 A. Alzate et al. / Waste Management 57 (2016) 113–120

Table 4
Experimental design layout for substrate oxidation.

Exp Factors with actual levels Substrate oxidation and Au recovery over 10 min
N0 A B C %Cu %Ni %Fe Substrate oxidation (%) Au recovery (%)
[(NH4)2S2O8] O2 L/S ratio
(M) (L/min) (mL/g)
1–2 0.44 0.2 15 1.74 ± 0.06 31.70 ± 1.55 6.42 ± 1.46 14.38 ± 0.33 73.61 ± 1.20
3–4 0.88 0.2 15 1.94 ± 0.25 10.80 ± 0.36 17.94 ± 1.09 12.37 ± 0.04 81.69 ± 1.95
5–6 0.44 1.0 15 4.21 ± 0.04 24.50 ± 1.78 74.41 ± 3.46 46.75 ± 2.14 88.81 ± 0.37
7–8 0.88 1.0 15 1.72 ± 0.09 4.82 ± 0.42 4.55 ± 0.28 4.62 ± 0.59 88.72 ± 1.15
9–10 0.44 0.2 25 2.42 ± 0.16 31.50 ± 0.94 9.73 ± 0.38 17.69 ± 0.05 86.82 ± 1.57
11–12 0.88 0.2 25 3.66 ± 0.18 42.33 ± 2.18 7.86 ± 0.81 19.72 ± 0.16 94.61 ± 0.63
13–14 0.44 1.0 25 2.79 ± 0.30 40.07 ± 0.17 43.03 ± 2.55 38.29 ± 1.03 88.85 ± 0.28
15–16 0.88 1.0 25 5.00 ± 0.24 44.49 ± 1.05 3.37 ± 0.30 19.56 ± 0.06 99.43 ± 0.18

the substrate was oxidized and leached at 19.72% (Exp 11–12) and
19.56% (Exp 15–16) Au recoveries of 94.61% and 99.43% were
respectively achieved. The significance of factors and interactions
for substrate oxidation response was calculated performing
ANOVA for the factorial model (Table 5). All the studied linear fac-
tors had p-values < 0.001, indicating the statistical significance of
each selected factor in the substrate oxidation. Moreover, interac-
tion between (NH4)2S2O8 concentration – Oxygen and (NH4)2S2O8
concentration – L/S ratio influenced directly the oxidation while
the interaction between oxygen – L/S ratio (p-value = 0.1196 > a)
had not effect in the response. The calculated standard deviation
for the model was 1.24%.
The regression model that describes the substrate oxidation
from processor scrap with (NH4)2S2O8 was established calculating
the coefficients for each factor and developing a first order equa-
tion (Eq. (4)).
(FeSO4)+ and (NiSO4 (a)) (Medusa software, 2010). The equilibrium
constant for this sulfates were established in (Log K(95°C) = 3.44)
and (Log K(95°C) = 3.06) respectively (HSC Chemistry software,
2002). The coefficient of determination (R2) was estimated in
0.99 proving that the model explains the 99% of the variability in
the response.
Oxidation and leaching of the substrate were achieved by the
diffusion of (SO2

4 ) and (O2) into the pines of the processors pene-
trating the solid sample from welding points to the core conformed
by Fe–Ni–Cu (Fig. 2a). Oxidative dissolution of substrate took place
breaking the Fe–Ni–Cu–Au bond without leaching of Au from the
sample. Release of solid particulate gold was achieved after remov-
ing the sample from the reactor and applying a micro scale pres-
sure washing (Fig. 2b).
3.2. Recovery of gold – first order model

Substrate oxidation ¼ 21:46
7:61A þ 5:63B þ 2:14C A linear model for the release of solid gold from the sample as a

7:62AB þ 3:43AC
0:53BC response of the substrate oxidation was analyzed with the 23 fac-
torial design. An empirical first order equation was developed
þ 2:42ABC ð4Þ
(Eq. (5)).
Eq. (4) shows that B and C were the most effective linear factors,
Au Recovery ¼ 89:61 þ 3:30A þ 3:63B þ 4:61C
0:67AB
while A was the less effective linear factor. Positive sign of B and C
indicates that the increment of the oxygen and L/S ratio positively þ 1:30AC
1:92BC þ 1:37ABC ð5Þ
influenced the oxidation and leaching of the substrate. This could
All the linear factors and the interaction AC and ABC had a pos-
be explained by the preferential formation of soluble oxides
itive influence on the Au recovery. The calculated standard devia-
include CuFeO2 (Medusa software, 2010) with an equilibrium con-
tion of the data was 5.52%. The linear model equation sets out to
stant greater than zero (Log K(95°C) = 4.45) (HSC Chemistry
obtain high values of Au recovery through coefficients of determi-
software, 2002). The negative sign of A indicates a negative linear
nation close to 100%. However, the estimated coefficient of deter-
influence in the response. However, the interaction between
mination was (R2 = 0.72) and the adjusted determination was
[(NH4)2S2O8] – L/S ratio (AC) has a positive influence. This influence
(Adj-R2 = 0.54). With these results the model explains the 72% of
demonstrates that (NH4)2S2O8 needs a careful balance between
the variability. The low values of correlation suggested possible
water added and quantity of scrap to enable the speciation of
quadratic effects. Hence, the model was adjusted to the curvature
(SO2
2

4 ) and to oxidize the substrate. The (SO4 ) ions leached the
with the aim to optimize the Au recovery.
metallic nickel and iron by formation of soluble sulfates includes

3.3. Recovery of gold – second order model

Table 5
Analysis of variance (ANOVA) for substrate oxidation. Optimal release of solid gold by substrate oxidation was
Source Sum of Degree of Mean p-value achieved running the twenty-five experiments of the central com-
squares freedom squares posite design (Fig. 3a–c).
Model 2719.61 7 388.52 <0.0001 The experiment design layout with the actual levels of each fac-
A – [(NH4)2S2O8] 925.38 1 925.38 <0.0001 tor is shown in Table 6. The standard deviation of the response (%
B – Oxygen 507.83 1 507.83 <0.0001 Au recovery) was calculated to be 62.20%. The maximum recovery
C – L/S Ratio 73.53 1 73.53 <0.0001 of gold (99.80%) was reached in the axial point (+2, 0, 0) where
AB 926.59 1 926.59 <0.0001
AC 188.24 1 188.24 <0.0001
reagent concentration was increased at 1.10 M and oxygen was
BC 4.35 1 4.35 0.1196 added in the central level of 0.6 L/min with an L/S ratio of 20 mL/
ABC 93.70 1 93.70 <0.0001 g. In Table 7, the ANOVA for the model is shown. All quadratic
Residual 16.79 11 1.53 effects were statistically significant with p-values of <0.0001 at a
Pure error 12.55 10 1.26
confidence level of 95%. This results confirm the curvature associ-
Corrected total 2736.41 18
ated to the model.
 A. Alzate et al. / Waste Management 57 (2016) 113–120 117

Fig. 2. Substrate oxidation, (a) sample before oxidation, (b) sample after 10 min of oxidation and micro scale pressure washing.

Fig. 3. Au recovered, (a) after filtration, (b) after drying – measure scale: above in inches (in), below in centimeters (cm), (c) after drying – focal length: 55 mm.

Table 6 Table 7
Experimental design layout for Au recovery. Analysis of variance (ANOVA) for Au recovery – surface quadratic model.

Exp Aleatory Factors with actual levels Au recovery Source Sum of squares df Mean square p-value
over 10 min
A – [(NH4)2S2O8] 625.36 1 625.36 <0.0001
N0 N0 A B C Au recovery (%)
B – Oxygen 665.39 1 665.39 <0.0001
[(NH4)2S2O8] (M) O2 L/S ratio
C – L/S ratio 410.94 1 410.94 <0.0001
(L/min) (mL/g)
AB 3.62 1 3.62 0.4011
1 17 0.44 0.2 15 74.45 AC 0.077 1 0.077 0.9014
2 19 0.44 0.2 15 72.76 BC 158.70 1 158.70 <0.0001
3 20 0.88 0.2 15 83.07 A2 187.44 1 187.44 <0.0001
4 3 0.88 0.2 15 80.31 B2 235.43 1 235.43 <0.0001
5 25 0.44 1.0 15 89.07 C2 166.06 1 166.06 <0.0001
6 11 0.44 1.0 15 88.54 Residual 72.69 15 4.85
7 4 0.88 1.0 15 98.53 Pure error 8.74 10 0.87
8 8 0.88 1.0 15 97.90 cor total 2269.99 24
9 23 0.44 0.2 25 87.93
10 10 0.44 0.2 25 85.71
11 12 0.88 0.2 25 94.16
12 9 0.88 0.2 25 95.05 was reached in 20 min. The time dependent released of gold from
13 13 0.44 1.0 25 88.65 the substrate was evaluated at 2, 4, 6, 8, 10 and 14 min. The kinet-
14 2 0.44 1.0 25 89.04 ics for the recovery of gold by substrate oxidation in the axial point
15 7 0.88 1.0 25 98.92 (
2, 0, 0), central point (0, 0, 0) and factorial points (+1, +1, +1) (
1,
16 6 0.88 1.0 25 99.30
17 1 0.22 0.6 20 74.09

1,
1) is presented in Fig. 4.
18 22 1.10 0.6 20 99.80 Surface plots were generated for the combination of three fac-
19 18 0.66 0.0 20 72.98 tors with actual values where the third factor was fixed at the cen-
20 15 0.66 1.4 20 97.91 tral level. Fig. 5a and b shows a wide red region in the middle of the
21 5 0.66 0.6 10 76.38
operation surface where recoveries of gold were greater than 98%.
22 24 0.66 0.6 30 98.97
23 14 0.66 0.6 20 99.39 This can be achieved under combined (NH4)2S2O8 concentration
24 16 0.66 0.6 20 98.86 ranging from 0.66 to 1.10 M with Oxygen flow from 0.5 to 1.4 L/
25 21 0.66 0.6 20 99.26 min and L/S ratio from 20 and 30 mL/g. The linear interaction
between AB (p-value = 0.4011) and AC (p-value = 0.9014) did not
present a significant effect on the response. This could be explained
Experiments at central levels (Exp 23–25) showed high Au because of the preferential substrate oxidation, the interaction of
recoveries (P98.86%) minimizing (NH4)2S2O8 concentration at the quadratic factors and the loss of linearity in the model. Fig. 5c
0.66 M and reducing the time of the substrate oxidation reaction shows that increasing oxygen and L/S ratio up to the central levels
to 10 min. These results can be compared with those presented with a fixed reagent concentration value of 0.66 M permitted to
by Syed (2006) using potassium persulfate where the Au recovery recover quantities of gold greater than 98.86%.
118 A. Alzate et al. / Waste Management 57 (2016) 113–120
Fig. 4. Kinetics of Au recovery at axial, center and factorial points.
The quadratic polynomial equation for Au recovery (Eq. (6)) was
developed in terms of coded factors. All the linear factors and the
interaction AB and AC had a positive influence on the response
while increasing the linear levels for B and C had a negative influ-
ence on the response as the BC interaction showed. The Predicted
R-Squared of 0.86 was in reasonable agreement with the adjusted
R-Squared of 0.95 with a difference 60.20, and the coefficient of
determination (R2 = 0.97) proved that the 97% of the variability in
the response has been explained by the model.
Au Recovery ¼ 98:77 þ 5:10A þ 5:27B þ 4:14C þ 0:48AB
þ 0:069AC
3:15BC
3:11A2
3:48B2
2:92C2
ð6Þ
Numerical optimization was carried out using Design Expert
software (2015) with the goal to maximize the recovery of gold.
Lower limit for Au recovery was established at 98% and the upper
Fig. 5. Surface plots of Au recovery (a) A: [(NH4)2S2O8] and B: oxygen with C = 20 mL/g, (b) A: [(NH4)2S2O8] and C: L/S ratio whit B = 0.6 L/min, (c) B: oxygen and C: L/S ratio
with A = 0.66 M.
limit at 100%. The values of concentration and oxygen were mini-
mized while the L/S ratio was fixed in its original range. Optimum
values of (NH4)2S2O8 concentration (0.65 M), Oxygen (0.58 L/min)
and L/S ratio (24 mL/g) were calculated to reach Au recoveries of
100%. Fig. 6 shows the contour plot for the prediction of maximum
recovery of gold in the optimum conditions.
The Au recovered was analyzed by scanning electron micro-
scopy (SEM/EDX) in order to evaluate chemical composition and
morphology. The results are presented in Fig. 7a and b. Pure gold
was removed from the substrate (Fe–Ni–Cu) as a fine coating
(Fig. 7a) without proof of corrosion on the Au surface. This finding
can be compared with the obtained by Barbieri et al. (2010) using
CuCl2. EDX was performed in different Au samples with replication
in five different points. Fig. 7b shows the grade of purity for Au
removed at the optimized levels. Higher peaks in Fig. 7b corre-
spond to the main energetic levels of gold that are presented at
2.12 keV and 9.71 keV.
4. Conclusions
Gold was recovered from waste electrical and electronic equip-
ment (WEEE) by a novel methodology of substrate optimized oxi-
dation with commercial grade ammonium persulfate ((NH4)2S2O8).
The parameters that determined the substrate oxidation were
studied through 23 full factorial experimental design. The reagent
concentration used in this design, oxygen flow and L/S ratio, were
statistically significant with p-values < 0.0001 while the interac-
tion oxygen-L/S ratio had not effect on the response. Oxidation of
Cu, Ni and Fe of 19.56% was obtained with (NH4)2S2O8 concentra-
tion of 0.88 M, Oxygen of 1 L/min, L/S ratio of 25 mL/g and reaction
 A. Alzate et al. / Waste Management 57 (2016) 113–120
Fig. 6. Contour plot for the numerical optimization of Au recovery.
Fig. 7. SEM analysis for Au recovery (a) fine coating of Au, (b) EDX spectrum for Au recovered.
time of 10 min. In that time, the oxidative partial dissolution of
substrate took place breaking the Fe–Ni–Cu–Au bond, and non-
leaching gold was recovered in a 99.43%. The development of a first
order equation for the prediction of Au recovery explains partially
the variability associated to the model. In order to maximize the Au
recovery a second order model was carried out using response
surface methodology. With RSM was possible to minimize the
(NH4)2S2O8 concentration (0.66 M), Oxygen (0.6 L/min) and L/S
ratio (20 mL/g) and Au recoveries greater than 98.86% were
achieved. Numerical optimization generated lower values for the
factors (NH4)2S2O8 concentration (0.65 M), Oxygen (0.58 L/min)
and L/S ratio (24 mL/g) in which gold could be recovery in a pre-
dicted value of 100%. The methodology for Au recovery was imple-
mented without grinding or shredding which had a positive impact
in the reduction of total reaction time. This indicates that the
recovery could be optimized allowing a most effective environ-
mental process. The findings presented in this paper suggest that
optimum levels of ammonium persulfate concentration, oxygen
flow and liquid/solid ratio can be used as a novel environmental
methodology to recover gold from WEEE.
Acknowledgements
The authors would like to sincerely acknowledge the committee
for research development of Antioquia University (CODI) (Project
119
No. PRG-2014-779), the national fund for financing of science,
technology and innovation (FRANCISCO JOSE DE CALDAS) (Project
No. FP44842-591-2014) and the University of Antioquia sustain-
ability strategy (2014–2015) for the financial support to success-
fully develop this research. The authors would also like to thank
the Colombian companies LITO S.A.S and INSUMON S.A.S for pro-
viding samples, staff and facilities for the experimental tests.
References
Akcil, A., Erust, C., Gahan, C.S., Ozgun, M., Sahin, M., Tuncuk, A., 2015. Precious metal
recovery from waste printed circuit boards using cyanide and non-cyanide
lixiviants – a review. Waste Manage. (New York, N.Y.) 45, 258–271. http://dx.
doi.org/10.1016/j.wasman.2015.01.017.
Babu, M.N., Sahu, K.K., Pandey, B.D., 2002. Zinc recovery from sphalerite concentrate
by direct oxidative leaching with ammonium, sodium and potassium
persulphates. Hydrometallurgy 64, 119–129. http://dx.doi.org/10.1016/S0304-
386X(02)00030-0.
Barbieri, L., Giovanardi, R., Lancellotti, I., Michelazzi, M., 2010. A new
environmentally friendly process for the recovery of gold from electronic
waste. Environ. Chem. Lett. 8, 171–178. http://dx.doi.org/10.1007/s10311-009-
0205-2.
Bas, D., Boyacı, I.H., 2007. Modeling and optimization I: Usability of response
surface methodology. J. Food Eng. 78, 836–845. http://dx.doi.org/10.1016/j.
jfoodeng.2005.11.024.
Bas, A.D., Deveci, H., Yazici, E.Y., 2014. Treatment of manufacturing scrap TV boards
by nitric acid leaching. Sep. Purif. Technol. 130, 151–159. http://dx.doi.org/
10.1016/j.seppur.2014.04.008.
120 A. Alzate et al. / Waste Management 57 (2016) 113–120
Birloaga, I., Coman, V., Kopacek, B., Vegliò, F., 2014. An advanced study on the
hydrometallurgical processing of waste computer printed circuit boards to
extract their valuable content of metals. Waste Manage. 34, 2581–2586. http://
dx.doi.org/10.1016/j.wasman.2014.08.028.
Birloaga, I., De Michelis, I., Ferella, F., Buzatu, M., Vegliò, F., 2013. Study on the
influence of various factors in the hydrometallurgical processing of waste
printed circuit boards for copper and gold recovery. Waste Manage. 33, 935–
941. http://dx.doi.org/10.1016/j.wasman.2013.01.003.
Design-Expert, 2015. Design-Expert, Version 9.0.4.1. Stat-Ease Inc., 2021 East
Hennepin Ave, Suite 480, Minneapolis, MN 55413.
Ha, V.H., Lee, J., Huynh, T.H., Jeong, J., Pandey, B.D., 2014. Optimizing the
thiosulphate leaching of gold from printed circuit boards of discarded mobile
phone. Hydrometallurgy 149, 118–126. http://dx.doi.org/10.1016/j.
hydromet.2014.07.007.
Hadi, P., Gao, P., Barford, J.P., McKay, G., 2013. Novel application of the nonmetallic
fraction of the recycled printed circuit boards as a toxic heavy metal adsorbent.
J. Hazard. Mater. 252–253, 166–170. http://dx.doi.org/10.1016/j.
jhazmat.2013.02.037.
Hadi, P., Ning, C., Ouyang, W., Xu, M., Lin, C.S.K., McKay, G., 2015a. Toward
environmentally-benign utilization of nonmetallic fraction of waste printed
circuit boards as modifier and precursor. Waste Manage. 35, 236–246. http://dx.
doi.org/10.1016/j.wasman.2014.09.020.
Hadi, P., Xu, M., Lin, C.S.K., Hui, C.-W., McKay, G., 2015b. Waste printed circuit board
recycling techniques and product utilization. J. Hazard. Mater. 283, 234–243.
http://dx.doi.org/10.1016/j.jhazmat.2014.09.032.
Hernandez, O., 2005. Persulfates. UNEP Publications SIDS Initial Assessment Report
for SIAM 20, pp. 1–192.
HSC Chemistry, 2002. Chemical Reaction and Equilibrium Software V. 6.1. Outotec
Research Oy.
Huang, K.C., Couttenye, R.a., Hoag, G.E., 2002. Kinetics of heat-assisted persulfate
oxidation of methyl tert-butyl ether (MTBE). Chemosphere 49, 413–420. http://
dx.doi.org/10.1016/S0045-6535(02)00330-2.
Jordão, H., Sousa, A.J., Carvalho, M.T., 2016. Optimization of wet shaking table
process using response surface methodology applied to the separation of copper
and aluminum from the fine fraction of shredder ELVs. Waste Manage. 48, 366–
373. http://dx.doi.org/10.1016/j.wasman.2015.10.006.
Kiddee, P., Naidu, R., Wong, M.H., 2013. Electronic waste management approaches:
an overview. Waste Manage. 33, 1237–1250. http://dx.doi.org/10.1016/j.
wasman.2013.01.006.
Lee, C.-H., Tang, L.-W., Popuri, S.R., 2011. A study on the recycling of scrap
integrated circuits by leaching. Waste Manage. Res.: J. Int. Solid Wastes Pub.
Cleansing Assoc. ISWA 29, 677–685. http://dx.doi.org/10.1177/
0734242X10380995.
Lundström, M., Aromaa, J., Forsén, O., 2009. Redox potential characteristics of cupric
chloride solutions. Hydrometallurgy 95, 285–289. http://dx.doi.org/10.1016/j.
hydromet.2008.06.009.
Medusa, 2010. Software for Chemical Equilibrium Diagrams. Royal Institute of
Technology, Stockholm, Sweden.
View publication stats
Montgomery, D., 2013. Design and Analysis of Experiments, second edition. Limusa
Wiley, Mexico.
Montgomery, D., 2011. Engineering Statistics. Arizona State University: John Wiley
& Sons Inc..
Naseri Joda, N., Rashchi, F., 2012. Recovery of ultra fine grained silver and copper
from PC board scraps. Sep. Purif. Technol. 92, 36–42. http://dx.doi.org/10.1016/
j.seppur.2012.03.022.
Petter, P.M.H., Veit, H.M., Bernardes, a.M., 2014. Evaluation of gold and silver
leaching from printed circuit board of cellphones. Waste Manage. 34, 475–482.
http://dx.doi.org/10.1016/j.wasman.2013.10.032.
Rath, S.S., Nayak, P., Mukherjee, P.S., Roy Chaudhury, G., Mishra, B.K., 2012.
Treatment of electronic waste to recover metal values using thermal plasma
coupled with acid leaching – a response surface modeling approach. Waste
Manage. (New York, N.Y.) 32, 575–583. http://dx.doi.org/10.1016/j.
wasman.2011.11.001.
Shibayama, A., Tongamp, W., Altansukh, B., Haga, K., Hosoi, A., 2013. Electronic
waste treatment: Part 1. Autoclave oxidation-leaching using pyrite waste from
mine tailing. Hydrometallurgy 137, 92–100. http://dx.doi.org/10.1016/j.
hydromet.2013.05.007.
Syed, S., 2012. Recovery of gold from secondary sources—a review.
Hydrometallurgy 115–116, 30–51. http://dx.doi.org/10.1016/j.hydromet.2011.
12.012.
Syed, S., 2006. A green technology for recovery of gold from non-metallic secondary
sources. Hydrometallurgy 82, 48–53. http://dx.doi.org/10.1016/j.hydromet.2006.
01.004.
Tuncuk, a., Stazi, V., Akcil, a., Yazici, E.Y., Deveci, H., 2012. Aqueous metal recovery
techniques from e-scrap: hydrometallurgy in recycling. Miner. Eng. 25, 28–37.
http://dx.doi.org/10.1016/j.mineng.2011.09.019.
Wäger, P.A., Hischier, R., Eugster, M., 2011. Environmental impacts of the Swiss
collection and recovery systems for waste electrical and electronic equipment
(WEEE): a follow-up. Sci. Tot. Environ. 409, 1746–1756. http://dx.doi.org/
10.1016/j.scitotenv.2011.01.050.
Walpole, R., Myers, R., Myers, S., Ye, K., 2012. Probability & statistics for engineers &
scientists, 9th ed. Prentice Hall.
Yazici, E.Y., Deveci, H., 2014. Ferric sulphate leaching of metals from waste printed
circuit boards. Int. J. Miner. Process. 133, 39–45. http://dx.doi.org/10.1016/j.
minpro.2014.09.015.
Yazici, E.Y., Deveci, H., 2013. Extraction of metals from waste printed circuit boards
(WPCBs) in H2SO4–CuSO4–NaCl solutions. Hydrometallurgy 139, 30–38. http://
dx.doi.org/10.1016/j.hydromet.2013.06.018.
Yu, L., He, W., Li, G., Huang, J., Zhu, H., 2014. The development of WEEE management
and effects of the fund policy for subsidizing WEEE treating in China. Waste
Manage. 34, 1705–1714. http://dx.doi.org/10.1016/j.wasman.2014.05.012.
Zhang, Y., Liu, S., Xie, H., Zeng, X., Li, J., 2012. Current status on leaching precious
metals from waste printed circuit boards. Proc. Environ. Sci. 16, 560–568.
http://dx.doi.org/10.1016/j.proenv.2012.10.077.