Journal of Cleaner Production 156 (2017) 226e234

Contents lists available at ScienceDirect

Journal of Cleaner Production

journal homepage: www.elsevier.com/locate/jclepro

Recycling of rare earths from lamp phosphor waste: Enhanced

dissolution of LaPO4:Ce3,Tb3 by mechanical activation
Steff Van Loy a, Koen Binnemans b, Tom Van Gerven a, *
a
KU Leuven, Department of Chemical Engineering, Celestijnenlaan 200F, B-3001 Heverlee, Belgium
b
KU Leuven, Department of Chemistry, Celestijnenlaan 200F, B-3001 Heverlee, Belgium

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

Article history: With the promoted use of compact uorescent lamps and the increasing amount of stockpiled compact
Received 21 December 2016 uorescent lamp waste, lamp phosphors are becoming an interesting secondary source for critical metals
Received in revised form (yttrium, europium and terbium). This paper explores on the mechanism of a new potential hydro-
14 March 2017
metallurgical route to improve the recovery efciency of rare-earth elements from the green phosphor
Accepted 23 March 2017
Available online 27 March 2017
LaPO4:Ce3,Tb3 using mechanical activation as a pretreatment step prior to leaching. By applying an
intense frictional action, the leaching yields of rare-earth elements were enhanced from 0.9% to 81% at
room temperature, as a consequence of the change in activation energy. The shrinking core model was
Keywords:
Rare earth elements
employed for modelling the leaching kinetics of the rare-earth elements and to calculate the decreasing
Phosphors apparent activation energy of the unmilled and milled samples (from 68 kJ mol1 to 1.4 kJ mol1). This
Fluorescent lamps difference can be explained by the physicochemical changes, including structural decomposition, specic
Resource recovery surface area increase and particle size reduction, which were related to the corresponding leaching
Ball-milling pattern. The optimized mechanical activation procedure was successfully applied to lamp phosphor
Urban mining waste, containing a mix of different phosphors. After sequential removal of the halophosphate phosphor
and the red YOX phosphor, 99.0%, 87.3% and 86.3% of La, Ce and Tb present in the LaPO4:Ce3,Tb3
phosphor could be dissolved. These observations provide more insight in the mechanical activation
process and may contribute to a more sustainable alternative route to the rare-earth element recycling
industry.
2017 Elsevier Ltd. All rights reserved.

1. Introduction recycling of these metals from end-of-life products has several

The use of compact uorescent lamps (CFLs) has been widely
adopted over the last decades, as CFLs are 70% more energy-
efcient and have 10 times longer lifetimes than incandescent
lamps, despite the expected rapidly growing market share (15%
p.a.) of light emitting diodes (LEDs) (Bonneville Power
Administration, 2015; Lim et al., 2013; McKinsey and Company,
2012; NorthWestern Energy, 2015). Phosphors for uorescent
lamps and CFLs are an important application of rare-earth elements
(REEs). Regardless of the decreasing demand of REEs (Eu and Tb) for
phosphors by two third since 2011, the general demand for REEs is
expected to grow by 6% (Roskill, 2015). Since REEs account for more
than 20 wt% within lamp phosphor waste (Table 1), including three
of the most critical REEs (yttrium, europium and terbium),
* Corresponding author.
E-mail address: tom.vangerven@kuleuven.be (T. Van Gerven).
advantages over primary supply. Besides the presence of REEs in
concentrated form in comparison to the minimum industrial grade
of rare-earth ores (0.06e2.0 wt%), recycling reduces the environ-
mental impact, landlling problems and co-production of the un-
wanted REEs, resolving the so-called balance problem (Binnemans
and Jones, 2014; Machacek et al., 2015; Tan et al., 2016a).
Lamp phosphors consist of a mix of a red (YOX), green (LAP or
CAT), and blue (BAM) phosphor (Table 2). Besides the REE-bearing
phosphors, lamp phosphor waste can consist of up to 50% hal-
ophosphate phosphor (HALO) which has low intrinsic value and
contains no REEs (Binnemans et al., 2013; Dupont and Binnemans,
2015). A lot of research has been conducted both by industry and
academic research groups to efciently recycle the rare earths from
lamp phosphor waste (Binnemans and Jones, 2014; Binnemans
et al., 2013; Tunsu et al., 2015; Wu et al., 2014). Most studies
focus on the recovery of yttrium (Y) and europium (Eu) from the
red phosphor YOX, because these are the elements that are the
easiest to recover and they represent the highest value in the

http://dx.doi.org/10.1016/j.jclepro.2017.03.160
0959-6526/ 2017 Elsevier Ltd. All rights reserved.
 S. Van Loy et al. / Journal of Cleaner Production 156 (2017) 226e234 227

(Bal
a
z, 2008; Boldyrev, 2006). Recent studies have applied me-
Abbreviations chanical activation for recycling of REEs from the YOX and LAP lamp
phosphors (Tan et al., 2016a; Zhang and Saito, 1998; Zhang et al.,
BPR ball-to-powder ratio 2000). The mechanical activation method investigated by Tan
CFL compact uorescent lamp et al. (2016b) in combination with a leaching process has suc-
D crystallite size ceeded in increasing the dissolution of REEs (Y, Eu and Tb) from
3 lattice strain uorescent lamp waste.
LED light emitting diode Despite the number of publications on the recovery of REEs from
MA mechanical activation lamp phosphors, studies using MA are scarce and lacking infor-
PS particle size mation on the effect of MA on individual phosphors. By milling the
S:L solid-to-liquid (ratio) complete lamp phosphor waste (including HALO and YOX), no
SSA specic surface area denite conclusions can be drawn on the effect of MA for the
wt% percentage weight total phosphors which are difcult to leach (such as LAP). The scope of
this paper is to conduct a systematic study of the effect of MA on the
reactivity of LAP phosphor, with emphasis on the physicochemical
changes and the milling parameters, to reveal the activation
mechanism. In the present study, the potential use of the deter-
Table 1 mined optimal MA parameters for recycling rare-earth metals from
Composition of lamp phosphor waste, analyzed by XRF. lamp phosphor waste was discussed.
Element wt% Element wt%
2. Experimental
Al 11.05 Mn 0.54
Ba 2.10 Na 7.14
Ca 21.42 Px 10.64 2.1. Materials
Ce 3.05 Sb 0.36
Cl 0.50 Si 10.89 The lamp phosphor LAP (La0.57Ce0.27Tb0.16PO4, 99%) was pur-
Eu 1.30 Sr 2.47 chased from Nichia (Japan). Lamp phosphor waste was kindly
La 3.94 Tb 1.28
provided by OSRAM (Munich, Germany). Nitric and hydrochloric
Mg 0.88 Y 19.62
acid solutions were prepared using ultrapure water (Milli-Q, Mil-
lipore; 18.2 MU cm1 at 25  C), concentrated nitric acid (65 wt%,
Chem-Lab) and concentrated hydrochloric acid (37 wt%, Merck).
phosphor waste. Recycling of the green phosphors (LAP or CAT) is
The ICP standards (Fluka, 1000 mg mL1) and lithium metaborate
also very interesting due to the high concentration of the critical
(99.9%) were purchased from Sigma-Aldrich (Diegem, Belgium).
and expensive terbium (Tb). However, the green phosphors are
much more difcult to dissolve than YOX, and often high concen-
2.2. Equipment and characterization
trations of mineral acids in combination with high temperatures
(e.g. 18 M H2SO4, 120e230  C) or cracking with molten sodium
Quantitative analysis of the leachates was performed by
hydroxide or sodium carbonate is required (Porob et al., 2012; Wu
Inductively Coupled Plasma-Optical Emission Spectrometry (ICP-
et al., 2014). These methods are not attractive due to the waste
OES, Optima 8300, Perkin Elmer) for concentrations higher than
generation, the use of corrosive chemicals and the high energy
10 ppb and by Inductively Coupled Plasma-Mass Spectrometry
consumption. Therefore, effective methods for the recovery of REEs
(ICP-MS, 7700X, Agilent) for concentrations lower than 10 ppb. The
from the green phosphors, with a low energy consumption and
concentrations of REEs in the LAP lamp phosphor were determined
high recovery rates are highly desirable.
by quantitative analysis of the solutions (in triplicate) after LAP was
Mechanical activation (MA) by ball-milling has been used to
fully dissolved in a 37% HCl solution via microwave-assisted
enhance the acid digestion of various minerals that are hardly
digestion (Monowave 300, Anton Paar). For the analysis of the
soluble, including monazite, which has a similar structure to that of
waste phosphor, a sample of 100 mg (in triplicate) was dissolved by
LAP (Abdel-Rehim, 2002; Achimovi
cova and Bala


z, 2005; Kim et al.,
fusion (1000  C) with lithium metaborate (500 mg). The melt was
2009). This technique has also been applied for the recovery of
dissolved in 50 mL of diluted nitric acid (3% HNO3). XRF analysis
various metals from secondary waste streams (Ou et al., 2015; Tan
was performed using an X-ray uorescence spectrometer (PW
and Li, 2015). Typically MA refers to the application of shear, impact
2400, Philips, Netherlands). The particle size distribution of the
and compression forces that induce a modication in the physi-
activated samples was characterized by laser particle size analyzer
cochemical properties of the solids, enhancing their reactivity
(Mastersizer 3000, Malvern, UK) using the laser diffraction method
in liquid mode. The specic surface area was determined by a
surface area analyzer (NOVA 2200e, Quantachrome, USA)
Table 2 at 196  C. XRD data was recorded using an X-ray powder
Overview of the approximate lamp phosphor waste composition found in lamp
diffractometer (D2 phaser, Bruker, Germany). Amorphous content,
phosphor waste (Binnemans et al., 2013).
crystal size and strain of the material was calculated by total Riet-
Type Name Content Economic veld renement analysis with the TOPAS software (Bruker, Ger-
phosphor (wt%)a value
many). The powder morphology was investigated by scanning
White Ca4.86Mn0.10Sb0.04Sr0.004(PO4)3Cl0.10F0.90 40e50 Low electron microscopy (XL30, Philips, Netherlands).
(HALO)
Red Y2O3:Eu3 (YOX) 20 High
Green LaPO4:Ce3,Tb3 (LAP) 6e7 High 2.3. Mechanical activation of lamp phosphors
CeMgAl11O19:Tb3 (CAT) 6e7 High
Blue BaMgAl10O17:Eu2 (BAM) 5 Low The mechanical activation of the phosphors was conducted
a
Besides the presence of phosphors, the lamp waste consists of ne glass parti- using a planetary ball-mill (P-7 premium, Fritsch, Germany) in air.
cles (up to 20 wt%), and Al2O3. Phosphor samples were mixed with zirconia balls ( 1, 3 or
228 S. Van Loy et al. / Journal of Cleaner Production 156 (2017) 226e234

10 mm) with a ball-to-powder ratio (BPR) of 50:1 (wt%) and put

into a zirconia bowl (80 mL inner volume, 47 mm). The ball mill
vessels were loaded with different ball sizes using a xed lling
volume of 32%. Samples were milled for various time periods
(15e60 min) at different rotational speeds (100e800 rpm) with
intermediate cooling periods (10 min) for each 15 min of milling
time to avoid accumulated heat. Ethylene glycol (1 wt%) was added
as surface-active substance (lubricant) to prevent deposition of the
sample on the inner wall of the bowl.

2.4. Acid leaching process

Leaching experiments were performed in 4 mL vials. The sam-

ples (0.1 g) were mixed with the leaching agent (4 N HNO3 or 4N
HCl) with a solid-to-liquid ratio (S:L) of 50 g/L and shaken
(200 rpm) at 25  C for 120 min. Afterwards, the solution was l-
trated using a polyethylene terephthalate syringe lter (PET,
0.45 mm) or a polyvinylidene uoride membrane lter (PVDF,
0.45 mm) if recovery of the solid residue is required. Solid residues Fig. 1. Leaching yield (%) of REEs (by 4 N HNO3, 120 min) from LAP lamp phosphor as a
were washed with Milli-Q water (50 mL) and dried at 105  C for function of the rotational speed (100e800 rpm) of the planetary ball mill for (60 min of
24 h prior to further characterization and analysis. The leaching milling time) for different grinding media sizes ( 1, 3 and 10 mm).
yield (%) is calculated by eqn (1):

Amount of metal in the leachate optimal conditions identied in the previous experiment (1 mm
% (1) grinding balls, 600 rpm). Fig. 2 shows the leaching yield over time,
Total amount of metal in the sample
obtained for two series of experiments; with and without MA of the
The expressed leaching yield represents the average yield of the phosphor. Note the break in the Y-axis. The recovery of REEs from
REEs in the phosphor as leaching percentages are similar for all unmilled LAP is low due to the lower dissolution rate of the large
REEs (La, Ce and Tb). crystalline particles. On the other hand, when the particles are
milled prior to leaching, the leaching yield of mechanically acti-
3. Results and discussion vated LAP (81%) was almost two orders of magnitude higher than
the value observed for a non-activated sample (0.9%) after 240 min
3.1. Effect of mechanical activation prior to leaching of leaching time. The leaching of mechanically activated LAP shows
two regimes: an initial rapid increase in the REE concentration
The milling conditions are of prime importance for evaluation of (during the rst minutes), followed by a gradually decreasing
the different reactions and physicochemical changes that may leaching rate. This pattern can be explained by the initial dissolu-
occur. Therefore, two crucial milling parameters (ball size and tion of the mechanically activated particles which contain a higher
milling speed) were investigated and their effect on the leaching number of defects at the outer surface. As the leaching progresses,
yield of REEs from LAP was analyzed. Fig. 1 shows the leaching yield towards the inner core of the particles, the number of defects
measured for different ball sizes (1e10 mm) and varying rotational decrease, resulting in a decreasing leaching rate. This illustrates
speeds (100e800 rpm). The results indicate that the use of a milling that the use of a combined MA and acid treatment is really
operation has a signicant effect on the leaching yield of LAP. The
use of 1 mm grinding balls and a rotational speed of 600 rpm
resulted in the dissolution of almost 80% of the REEs, which is the
highest yield obtained at room temperature in this study. No large
differences can be observed between the results of 1 mm or 3 mm
grinding balls. The decrease in the leaching yield after increase in
the rotational speed beyond a threshold value, may be explained by
the complex ball motion in the grinding vial. With increasing
milling speed, the milling pattern changes from cascading to cat-
aracting up to centrifugation, which could affect the grinding ef-
ciency (Burmeister and Kwade, 2013). The results show that a
smaller ball size is more benecial for increasing the reactivity of
the lamp phosphor powder which is in contrast to previous reports
where large grinding balls were used (Hiroshi et al., 2001; Tan et al.,
2016a). In general, one can expect that larger grinding balls are
more effective since a larger mass will result in a higher impact
transfer. However, it was also previously observed that the use of
smaller grinding balls could be benecial due to the intense (shear-
induced) frictional action (Suryanarayana, 2001). This promotes the
physicochemical changes of the solids, resulting in more reactive
particles. All further experiments have been conducted with 1 mm
grinding balls and at a rotational speed of 600 rpm, since this gave Fig. 2. Changes in the leaching yield (%) of REEs (by 4N HNO3) from unmilled and
the highest leaching yield. milled ( 1mm grinding balls, rotational speed 600 rpm, 60 min) LAP phosphor as a
The effect of the leaching time was also investigated for the function of leaching time. Note the break in the Y-axis.
 S. Van Loy et al. / Journal of Cleaner Production 156 (2017) 226e234 229

signicant on the recovery process. This positive effect could be a
consequence of the decrease in activation energy of the leaching
process.
3.2. Kinetic modelling of the leaching process
difference in leaching model between milled and unmilled sample
is in line with observation described in the literature (Tan et al.,
2016a), who stated that the controlling step changed from chemi-
cally controlled to diffusion controlled due to the milling operation.
The apparent activation energy was calculated by the Arrhenius
equation (Simha et al., 2016a, 2016b) (eqn (4)):

To investigate the effect of the leaching kinetics, the leaching A expEa =RT ka (4)
process was investigated, using the shrinking core model
The apparent activation energy for leaching of REEs was calcu-
(Levenspiel, 1999). The use of the model is justied as the lamp
lated to be 68 kJ mol1 for the unmilled and 1.4 kJ mol1 for the
phosphors are a non-porous material (Tan et al., 2016a; Verma
milled sample (rst stage) (Fig. 4). Both values t the observed
et al., 2013). Data obtained from the leaching of milled and un-
mechanism limiting step, as in general, chemically-controlled re-
milled LAP at varying temperatures (ranging from 40  C to 80  C)
actions have an Ea > 40 kJ mol1, while diffusion-controlled re-
was used for this kinetic study (Fig. 3). The diffusion-controlled
actions have a lower Ea (<40 kJ mol1) (Abdel-Aal and Rashad,
(eqn (2)) and chemical-controlled model (eqn (3)) were both
2004; Uar, 2009).
tested for the leaching of REEs:
1
1  1  x3 ka t (2) 3.3. Physicochemical changes by mechanical activation

2
As explained in the previous section, the decrease in activation
1  31  x3 21  x ka t (3) energy and consequential improvement in reactivity by prolonged
milling can be attributed to the accumulated energy, stored in the
Here x represents the leached fraction of REEs at time t (min),
particles, in the form of defects or other structural modications.
with ka the apparent reaction rate constant (min1). Fig. 3 shows
Mechanical treatment of the solids leads to the introduction of
the effect of temperature on the leaching yield of the unmilled
linear defects, e.g. ion and atom vacancies, interstitial ions and
(bottom) and milled (top) samples. The results show that temper-
dislocations besides a change of specic surface area (SSA) and
ature has a large effect on the leaching of the unmilled sample, with
particle size (PS). A change in the bond angles and broken bonds
the leaching yield increasing from 0.74% at 40  C to 11.0% at 80  C
will result in an increase in the amorphous content (Boldyrev,
within a period of 40 min. The effect is less pronounced with the
2006). The leaching yield of REEs from LAP phosphor as a func-
milled sample, with the leaching yield increasing from 80.9% to
tion of milling time is shown in Fig. 5. Previous experiments were
87.0% by the increase in temperature from 40 to 80  C. The
conducted using nitric acid as leaching agent, but could also be
diminished effect of the temperature after mechanical activation is
replaced by hydrochloric acid as a cheaper alternative. Both acids
consistent with previous reports (Tang et al., 2010; Zhao et al.,
showed similar leaching patterns, however a higher leaching yield
2009). The unmilled sample could be tted with the chemically-
was obtained by the use of nitric acid. The duration of the milling
controlled model as the leaching was controlled by the chemical
process had a positive inuence on the leaching yield, with an in-
reaction at the surface of the core of the unreacted particle. On the
crease in leaching yield from 42% to 79% from 15 to 60 min using
contrary, the milled sample could not the tted with the same
4 N HNO3. The largest increase in leaching yield was observed from
chemically-controlled model. The initial leaching of the milled
0 to 15 min of milling time. An extended milling period, longer than
sample (Fig. 3) is a very fast process in which more than 70% is
30 min, represents a marginal increment in leaching yield due to
dissolved within 20 min (at 40  C). Therefore, the process is
the possibly increasing difculty to modify the crystal structure.
diffusion-controlled during the rst leaching period (the rst
Therefore, it can be recommended to mill for a shorter period and
20 min). Afterwards, there is a shift to chemically-controlled due to
remill the solid residue after leaching, to improve the overall ef-
the consumption of all mechanically activated particles. The
ciency of the system.
The fraction of amorphous phase can be calculated from the

Fig. 3. Effect of leaching temperature on the leaching yield (%) of unmilled (bottom)
and milled (top) LAP phosphor with 4 N nitric acid ( 1 mm grinding balls, 600 rpm,
60 min). Note the break in the Y-axis. Fig. 4. Arhennius plot for unmilled and milled LAP phosphor.
230 S. Van Loy et al. / Journal of Cleaner Production 156 (2017) 226e234
Fig. 5. Leaching yield of REEs (%) from LAP versus milling time by HNO3 (4 N) or HCl
(4 N) at room temperature for 120 min for different activation times ( 1 mm
grinding balls, rotational speed 600 rpm).
decrease of the integrated intensity of the diffraction lines, ob-
tained by using X-ray powder diffraction. The amorphous (A) and
crystalline content (X), can be described according to Ohlberg and
Strickler (1962):
 
U0 Ix
X * *100 ; A 100  X (5)
I0 I0
where I0 and Ix refer to the integrated intensities of the diffraction
lines of the initial and ground samples, while U0 and Ux are the
background signals of the initial and ground samples. It is assumed
that the unmilled LAP is fully crystalline and can be used as the 0%
amorphous content benchmark. The effect of the mechanical acti-
vation on the crystallinity of the LAP phosphor as a function of time
was clear from the X-ray diffraction pattern (Fig. 6). It is observed
that a prolonged milling operation caused a decrease in the peak
intensity and broadening of the main peaks as well as disappear-
ance of peaks at higher angle, as a result of the increase in amor-
phous content. No formation of new crystalline phases was
observed. Fig. 6 also shows that the majority of the amorphous
phase was generated during the early stages of the milling process
Fig. 6. Changes in the XRD pattern of the LAP phosphor for different activation times
( 1 mm grinding balls, rotational speed 600 rpm).
Table 3
Overview of the calculated amorphous content, crystallite size and strain of the
LAP phosphor under different activation times.
Milling time Amorphous content Crystallite size Lattice strain
(min) (%) (nm) (%)
0 0 614.6 14.6 3.2 0.1
15 63.9 1.3 50.1 1.3 34.2 3.4
30 79.6 2.2 18.0 0.9 53.3 7.4
45 82.0 1.8 15.1 0.3 61.1 4.7
60 82.9 1.2 13.4 0.2 77.8 5.0
(within 15 min). Further increase in the amorphous phase was less
pronounced after 30, 45 and 60 min. This observation was deter-
mined from the quantitative results of the amorphous content,
along with the change in crystallite size (D) and lattice strain (3 )
calculated by the method of Rietveld structure renement (Table 3).
A sharp increment in the amorphous content was observed in the
early stages of the milling process, after which a stable value was
reached. After 60 min of milling, the amorphous content was 83%,
which implies that about 17% of the material remained crystalline.
This can also be concluded from the X-ray pattern which still
showed the presence of low intensity peaks after 60 min of milling.
A decrease in D and an increase in 3 was noticed after milling for
60 min. In contrast with a minimum value for D, 3 in the crystal
lattice kept increasing with extended milling (after 30 min). This
implies that the stress in the crystal lattice keeps increasing. The
results indicate that the crystal structure of the LAP phosphor was
transformed to a distorted state, with the distortion mostly located


at the particle surface (core/shell structure) (Bala z et al., 2013). It is
proposed that the concentration of defects increases and D de-
creases from the center to the surface of the particles. The more
crystalline content is located in the center of the particle. With high
impact energy, when using bigger grinding balls (10 mm), MA can
cause the generation of rare-earth oxides along with changes in the
crystal structure as determined by Tan et al. (2016b). However, with
the use of smaller grinding media, the possibility of conversion to
rare-earth oxides is limited. Therefore it is postulated that the in-
crease in reactivity is correlated only with the changes in the crystal
lattice.
An increase in the specic SSA and reduction in the PS can
effectively inuence the mass transfer of the leaching process,
especially in case of diffusion-controlled processes. In general,
milling operations are associated with a decrease in PS. However,
opposite trends are observed in Fig. 7. In the early stages of the
Fig. 7. Changes in mean numerical particle size (-, left Y-axis) and specic surface
area (C, right Y-axis) of the LAP phosphor under different activation times ( 1 mm
grinding balls, rotational speed 600 rpm).
 S. Van Loy et al. / Journal of Cleaner Production 156 (2017) 226e234 231

Fig. 8. SEM images of non-activated and activated LAP phosphor: A) non-activated 2500, B) activated 25000, 15 min, C) activated 25000, 45 min, D) activated 25000, 60 min.

milling process (rst 15 min) a sharp decrease in numerical PS is phosphor (HALO) and the REE-bearing phosphors, up to 20% ne
found, accompanied by an increase in the SSA. However, prolonged glass particles (cullet) and small amounts of alumina. The pres-
milling caused a gradual increase in the mean numerical PS from ence of other components could alter the mechanical activation
0.30 mm to 0.39 mm, after respectively 30 and 60 min of milling. The efciency of the LAP phosphor. Therefore, a full recycling process
SSA values changed from 13.6 m2/g to 7.8 m2/g in that same time was adopted from the schemes proposed by Tunsu et al. (2015)
interval. A clear inverse correlation is observed between the PS and
the SSA. After 15 min, the rate of re-welding or agglomeration of
particles exceeds the particle breakage rate, which cannot be
avoided at intensive grinding conditions (Bal a

z, 2000). For longer
grinding periods, agglomeration of the particles was more pro-
nounced. This was supported by SEM images (Fig. 8), which showed
the increasing degree of agglomeration and particle growth of the
ne particles. In general, agglomeration will negatively affect the
mass transfer of the leaching process, resulting in a decreasing
leaching rate.
There is a clear correlation between the determined deterio-
rated crystal structure, reected in the change in amorphous con-
tent, D and 3 , and the leaching rate of activated LAP. It can be
observed 3 is strongly correlated to the leaching yield (Fig. 9). The
agglomeration could explain the decreasing correlation at longer
grinding times as an increase in PS limits the mass transfer.
Consequentially, the dissolution of REEs from LAP is not only
associated with the SSA (and correlated PS), but to a much larger
extent to the mechanical deformation of the crystal structure.

3.4. Full waste recycling process

Fig. 9. Correlation of the lattice strain and the leaching yield of LAP using 4N HNO3
Lamp phosphor waste contains, besides the halophosphate with different activation times ( 1 mm grinding balls, rotational speed 600 rpm).
232 S. Van Loy et al. / Journal of Cleaner Production 156 (2017) 226e234

and by Wu et al. (2014), comprising the selective removal of

Fig. 10. Proposed owsheet for the recovery of REEs from uorescent lamp waste. The
process comprises the selective leaching of impurity metals and REEs with hydro-
chloric acid solutions by taking advantage of the differences in their leaching kinetics,
and mechanical activation of LAP prior to leaching.
phosphors by exploiting the differences in solubility and leaching
kinetics. An alternative process has been developed by Dupont and
Binnemans (2015) using an ionic liquid leaching system which
allows the selective dissolution of YOX without dissolving the
other phosphors (HALO, LAP and BAM), to overcome the problem
of co-dissolution and this approach could improve the overall ef-
ciency of the process. Selective dissolution will reduce the
number of separation stages needed in subsequent solvent
extraction. An overview of the proposed ow sheet is shown in
Fig. 10. The feedstock is dry phosphor powder, containing small
glass particles and alumina, from which the mercury has been
removed by thermal treatment.
In order to remove the HALO phosphor, and other impurity
metals such as alumina, a rst leaching step was carried out for
5 min at room temperature with a 2 N HCl solution and a S:L ratio of
100 g/L. Almost 95% of the Ca, contained in the HALO phosphor,
could successfully be removed (Table 4). However, leaching of
HALO phosphor also lead to the co-dissolution of Y (4.2%) and Eu
(6.6%).
Next, the resulting residue was leached again to efciently
dissolve the REEs (Y and Eu) present in the red phosphor YOX.
Leaching was carried out for 60 min at elevated temperature
(70  C) using a 2 N HCl solution with a S:L ratio of 150 g/L. Hence
more than 95% of Y and 85% of Eu could be selectively dissolved.
As only little amounts of Ce, La and Tb (<1%) from the green
phosphor (LAP) dissolve alongside Eu and Y, separation by solvent
extraction is easier to achieve in comparison to a total dissolution
step in which all REEs are dissolved. In the third stage, the Tb-rich
residue is mechanically activated at optimal conditions deter-
mined in Section 3.1 ( 1 mm grinding balls, 600 rpm, 60 min)
and can be leached using 4 N HCl (50 g/L) at room temperature,
during which 99.0%, 87.3% and 86.3% of respectively La, Ce and Tb
was dissolved at room temperature. Leaching of the Tb-rich res-
idue without MA, resulted in a low recovery (<1%) of La, Ce and
Tb. These recovery results of the mechanically activated material
are better than the earlier described results (79%, using 4 N HNO3
for 120 min) on the pure LAP. This indicates that the presence of
the small glass particles (cullet) positively enhances the mechan-
ical activation process of LAP. The grinding of the cullet will create
very ne glass material (<1 mm), which is assumed to act itself as a
ne grinding medium. As a consequence, the frictional action is
increased during the milling, enhancing the structural defects in
the crystal lattice of the LAP. The presence of cullet in the waste is,
therefore, benecial for the overall mechanical activation process.
The presence of other components in the waste was also seen to
overcome deposition of material on the walls of the milling ves-
sels. Almost full recovery of Y and La could be achieved with this
process, the lower leaching percentages of Eu, Ce and Tb can be
explained by the presence of REEs in the remaining low-soluble
residue including CAT and BAM. These phosphors containing mi-
nor amounts of REEs, could be recovered by alkaline fusion using
sodium hydroxide (800  C) or sodium carbonate (1000  C) (Zhang
et al., 2013).

Table 4
Leaching rates of rare earth in three steps acid leaching tests (%).

Element First step acid leaching Second step acid leaching Third step acid leaching without MA Third step acid leaching with MA Total leaching rates (with MA)

Ca 94.6 4.5 2.5 0.1 0.1 0.1 0.8 0.1 97.9 4.7
Y 4.2 0.1 95.7 1.9 0.1 0.1 0.1 0.1 100.0 2.1
Eu 6.6 0.1 88.5 1.4 0.5 0.1 1.0 0.1 96.1 1.6
La 0.4 0.1 0.3 0.1 0.1 0.1 99.0 1.1 99.7 1.3
Ce 0.4 0.2 0.5 0.1 0.2 0.1 87.3 1.5 88.2 1.8
Tb 0.6 0.2 0.3 0.1 0.2 0.1 86.3 1.4 87.2 1.7
 S. Van Loy et al. / Journal of Cleaner Production 156 (2017) 226e234
4. Conclusions
This study investigated the use of MA as pretreatment for
leaching of REEs from lamp phosphor waste, with emphasis on LAP
phosphor. Both the activation mechanism and the effect of milling
parameters were explored. The main ndings of this research can
be grouped into following four points:
1. The MA process was promoted by the use of smaller grinding
media as a consequence of the intense frictional action. The
rotational speed was proven to be a crucial parameter and
strongly related to the milling media. By milling, the leaching
yield could be enhanced to 81%, which is almost two orders of
magnitude higher than the value obtained for the inactivated
sample (0.9%) using 4 N HNO3 at room temperature for 240 min.
2. MA resulted in a decrease of the activation energy of the
leaching process, as a result of the changes in the crystal
structure, particle size and specic surface area. The shrinking
core model suggested a change in the process mechanism from
a chemically- to a diffusion-controlled process, reected in the
lowering of the activation energy from 68 kJ mol1 to
1.4 kJ mol1.
3. The activation time (i.e., milling time) had a positive impact on
the leaching of REEs, and inuenced physicochemical changes
such as amorphization, crystallite size decrease, strain increase,
particle size reduction and specic surface area increase. It is
postulated that the structural decomposition is the main
mechanism for the improved leachability of REEs from LAP.
With extended milling, the rate of physicochemical changes
diminishes which is reected in the recovery rates. Therefore, it
is suggested to perform milling in stages with shorter milling
and leaching periods, in which the unaffected residue is
remilled.
4. The MA process has proven to be a useful method for recovery of
the REEs from lamp waste phosphor, after removal of the HALO
and YOX phosphor, resulting in a concentrated stream of La, Ce
and Tb in which respectively 99.0%, 87.3% and 86.3% could be
leached from the phosphor. The cullet, present in the waste, has
an additional benet to the process due to the increasing fric-
tional action.
It was conrmed that the MA pretreatment described in this
paper lead to a major improvement in the leaching efciency of
REEs from LaPO4:Ce3,Tb3. The process has major advantages over
the current industrial process, using 18 M H2SO4 (120e230  C) or
molten alkali (NaOH or NaCO3), as the dissolution of REEs could be
achieved under much more moderate conditions (4 N HNO3, 25  C).
Besides the advantages of the lower energy requirement, acid
consumption and reduced waste water, this process is efcient and
simple. It has huge potential for replacing to a large extent the
primary production of Y, Eu and Tb, for usage in phosphors for light
emitting diodes, security markers and magnets. This work does not
only provide insight into the mechanism of mechanical activation
of lamp phosphors but also contributes to a more environmentally
friendly and energy-efcient industrial process for the full valori-
zation of the REEs from lamp phosphor waste.
Acknowledgments
This research is supported by KU Leuven (projects GOA/13/008
and IOF-KP RARE3). SVL thanks the IWT Flanders (1S23516N) for a
PhD fellowship. The authors would like to thank Tobias Hertel for
the help with XRD Rietveld renement analysis.
233
References
Abdel-Aal, E.A., Rashad, M.M., 2004. Kinetic study on the leaching of spent nickel
oxide catalyst with sulfuric acid. Hydrometallurgy 74, 189e194. http://
dx.doi.org/10.1016/j.hydromet.2004.03.005.
Abdel-Rehim, A.M., 2002. An innovative method for processing Egyptian monazite.
Hydrometallurgy 67, 9e17. http://dx.doi.org/10.1016/S0304-386X(02)00134-2.
Achimovi
cova, M., Bala

z, P., 2005. Inuence of mechanical activation on selectivity
of acid leaching of arsenopyrite. Hydrometallurgy 77, 3e7. http://dx.doi.org/
10.1016/j.hydromet.2004.09.008.
Bal
a
z, P., 2008. Mechanochemistry in Nanoscience and Minerals Engineering.
Springer-Verlag, Berlin Heidelberg.
Bal
a
z, P., 2000. Extractive Metallurgy of Activated Minerals. Elsevier, Amsterdam.
Bal
a
z, P., Achimovi
cova, M., Bal
a
z, M., Billik, P., Cherkezova-Zheleva, Z., Criado, J.M.,
Delogu, F., Dutkov a, E., Gaffet, E., Gotor, F.J., Kumar, R., Mitov, I., Rojac, T.,
Senna, M., Streletskii, A., Wieczorek-Ciurowa, K., 2013. Hallmarks of mecha-
nochemistry: from nanoparticles to technology. Chem. Soc. Rev. 42, 7571e7637.
http://dx.doi.org/10.1039/c3cs35468g.
Binnemans, K., Jones, P.T., 2014. Perspectives for the recovery of rare earths from
end-of-life uorescent lamps. J. Rare Earths 32, 195e200. http://dx.doi.org/
10.1016/S1002-0721(14)60051-X.
Binnemans, K., Jones, P.T., Blanpain, B., Van Gerven, T., Yang, Y., Walton, A.,
Buchert, M., 2013. Recycling of rare earths: a critical review. J. Clean. Prod. 51,
1e22. http://dx.doi.org/10.1016/j.jclepro.2012.12.037.
Boldyrev, V.V., 2006. Mechanochemistry and mechanical activation of solids. Russ.
Chem. Rev. 75, 177e189. http://dx.doi.org/10.1007/BF01557732.
Bonneville Power Administration, 2015. LED Market Intelligence Report. https://
www.bpa.gov/ee/utility/research-archive/documents/momentum-savings-
resources/led_market_intelligence_report.pdf.
Burmeister, C.F., Kwade, A., 2013. Process engineering with planetary ball mills.
Chem. Soc. Rev. 42, 7660e7667. http://dx.doi.org/10.1039/c3cs35455e.
Dupont, D., Binnemans, K., 2015. Rare-earth recycling using a functionalized ionic
liquid for the selective dissolution and revalorization of Y2O3:Eu3 from lamp
phosphor waste. Green Chem. 17, 856e868. http://dx.doi.org/10.1039/
C4GC02107J.
Hiroshi, M., Lee, J., Nakagawa, T., Kano, J., Saito, F., 2001. Estimation of extraction
rate of yttrium from uorescent powder by ball milling. Mater. Trans. 42,
2460e2464.
Kim, W., Bae, I., Chae, S., Shin, H., 2009. Mechanochemical decomposition of
monazite to assist the extraction of rare earth elements. J. Alloys Compd. 486,
610e614. http://dx.doi.org/10.1016/j.jallcom.2009.07.015.
Levenspiel, O., 1999. Chemical Reaction Engineering. John Wiley & Sons., New York
http://dx.doi.org/10.1021/ie990488g.
Lim, S.R., Kang, D., Ogunseitan, O.A., Schoenung, J.M., 2013. Potential environmental
impacts from the metals in incandescent, compact uorescent lamp (CFL), and
light-emitting diode (LED) bulbs. Environ. Sci. Technol. 47, 1040e1047. http://
dx.doi.org/10.1021/es302886m.
Machacek, E., Richter, J.L., Habib, K., Klossek, P., 2015. Recycling of rare earths from
uorescent lamps: value analysis of closing-the-loop under demand and supply
uncertainties. Resour. Conserv. Recycl 104, 76e93. http://dx.doi.org/10.1016/
j.resconrec.2015.09.005.
McKinsey & Company, 2012. Lighting the Way: Perspectives on the Global Lighting
Market. https://www.mckinsey.de/les/Lighting_the_way_Perspectives_on_
global_lighting_market_2012.pdf.
NorthWestern Energy, 2015. NorthWestern Energy CFL Lighting Market Study.
http://www.northwesternenergy.com/docs/default-source/documents/
defaultsupply/plan15/volume2/nweclightingmarketstudy.
Ohlberg, S.M., Strickler, D.W., 1962. Determination of percent crystallinity of partly
devitried glass by X-ray diffraction. J. Am. Ceram. Soc. 45, 170e171.
Ou, Z., Li, J., Wang, Z., 2015. Application of mechanochemistry to metal recovery
from second-hand resources: a technical overview. Environ. Sci. Process. Im-
pacts 17, 1522e1530. http://dx.doi.org/10.1039/c5em00211g.
Porob, D.G., Alok, S.M., Kumar, N.P., Chandran, R.G., Comanzo, H.A., 2012. Rare Earth
Recovery From Fluorescent Material And Associated Method. United States
Patent. US 8,137,645 B2.
Roskill, 2015. Rare Earth: Market Outlook to 2020, fteenth ed. London.
Simha, P., Mathew, M., Ganesapillai, M., 2016a. Empirical modeling of drying ki-
netics and microwave assisted extraction of bioactive compounds from Ada-
thoda vasica and Cymbopogon citratus. Alex. Eng. J. 55, 141e150. http://
dx.doi.org/10.1016/j.aej.2015.12.020.
Simha, P., Ramanathan, A., Thawani, B., Jain, P., Hussain, S., Ganesapillai, M., 2016b.
Coal y ash for the recovery of nitrogenous compounds from wastewater:
parametric considerations and system design. Arab. J. Chem. http://dx.doi.org/
10.1016/j.arabjc.2016.11.013.
Suryanarayana, C., 2001. Mechanical alloying and milling. Prog. Mater. Sci. 46,
1e184.
Tan, Q., Deng, C., Li, J., 2016a. Innovative application of mechanical activation for
rare earth elements recovering: process optimization and mechanism explo-
ration. Sci. Rep. 6, 1e10. http://dx.doi.org/10.1038/srep19961.
Tan, Q., Deng, C., Li, J., 2016b. Enhanced recovery of rare earth elements from waste
phosphors by mechanical activation. J. Clean. Prod. 142, 2187e2191. http://
dx.doi.org/10.1016/j.jclepro.2016.11.062.
Tan, Q., Li, J., 2015. Recycling metals from wastes: a novel application of mecha-
nochemistry. Environ. Sci. Technol. 49, 5849e5861. http://dx.doi.org/10.1021/
234 S. Van Loy et al. / Journal of Cleaner Production 156 (2017) 226e234
es506016w.
Tang, A., Su, L., Li, C., Wei, W., 2010. Effect of mechanical activation on acid-leaching
of kaolin residue. Appl. Clay Sci. 48, 269e299. http://dx.doi.org/10.1016/
j.clay.2010.01.019.
Tunsu, C., Petranikova, M., Gergori c, M., Ekberg, C., Retegan, T., 2015. Reclaiming
rare earth elements from end-of-life products: a review of the perspectives for
urban mining using hydrometallurgical unit operations. Hydrometallurgy 156,
239e258. http://dx.doi.org/10.1016/j.hydromet.2015.06.007.
Uar, G., 2009. Kinetics of sphalerite dissolution by sodium chlorate in hydrochloric
acid. Hydrometallurgy 95, 39e43. http://dx.doi.org/10.1016/j.hydromet.2008.
04.008.
Verma, H.R., Sahu, S.K., Pandey, B.D., Mankhand, T.R., Materials, A., 2013. Kinetics of
hydrometallurgical extraction of rare earth metals from waste phosphor. Int. J.
Res. Eng. Technol. 2, 251e255.
Wu, Y., Yin, X., Zhang, Q., Wang, W., Mu, X., 2014. The recycling of rare earths from
waste tricolor phosphors in uorescent lamps: a review of processes and
technologies. Resour. Conserv. Recycl. 88, 21e31. http://dx.doi.org/10.1016/
j.resconrec.2014.04.007.
Zhang, Q., Lu, J., Saito, F., 2000. Selective Extraction of Y en Eu by non-thermal acid
leaching of uorescent powder activated by mechanochemical treatment using
a planetary mill. Shigen - - Sozai 116, 137e140.
Zhang, Q., Saito, F., 1998. Non-thermal extraction of rare earth elements from
uorescent powder by means of its mechanochemical treatment. Shigen - -
Sozai 114, 253e257.
Zhang, S.G., Yang, M., Liu, H., Pan, D.A., Tian, J.J., 2013. Recovery of waste rare earth
uorescent powders by two steps acid leaching. Rare Met. 32, 609e615. http://
dx.doi.org/10.1007/s12598-013-0170-6.
Zhao, Z., Zhang, Y., Chen, X., Chen, A., Huo, G., 2009. Effect of mechanical activation
on the leaching kinetics of pyrrhotite. Hydrometallurgy 99, 105e108. http://
dx.doi.org/10.1016/j.hydromet.2009.06.002.