Waste Management 23 (2003) 879–886

www.elsevier.com/locate/wasman

Mercury speciation in fluorescent lamps by thermal release analysis

Cláudio Raposoa,*, Cláudia Carvalhinho Windmöllerb, Walter Alves Durão Júniorb
a
Nuclear Technology Development Center—CDTN/CNEN, Cidade Universitária, 30.123-970 Belo Horizonte, Minas Gerais, Brazil
b
Chemistry Department, Federal University of Minas Gerais (UFMG), Cidade Universitária 30.123-970 Belo Horizonte, Minas Gerais, Brazil

Accepted 1 April 2003

Abstract
In this work, mercury speciation in phosphorus powder matrices and soda lime glass waste from new and spent fluorescent lamp
wastes has been studied by thermo-desorption/atomic absorption spectrometry (TDAAS), X-ray diffraction (XRD), cold vapor-
atomic absorption (CV-AAS) and atomic emission spectrometry/inductively coupled plasma (ICP/AES). TDAAS results show the
presence of oxidized forms of mercury, i.e., Hg1+ and Hg2+, especially in wastes with high mercury concentration. Such forms are
mobile, and therefore represent a potential hazard waste material. Glass TD profiles of spent fluorescent lamps suggested the pre-
sence of mercury strongly linked to the matrix, which desorbs only at high temperatures.
# 2003 Elsevier Ltd. All rights reserved.

1. Introduction condensation. However, recent studies carried out in

Mercury lamps (fluorescent, compact fluorescent,
mercury vapor, sodium vapor and metal multi-vapors
and mixed) use mercury as a vital component for its
functioning. Mercury concentration in these lamps var-
ies considerably depending on the manufacturer, lamp
type and year of manufacturing. The amount of mer-
cury in a 40 W/T121 standard fluorescent lamp has
decreased over the past years from 41 mg in the early
1990s to 21 mg after 1997 (USEPA, 1997). Once the
lifetimes of these lamps have ended, they generate a
hazardous waste, which must be properly disposed of
(USEPA, 1999; Raposo, 2001, in press).
Several technologies to treat lamp waste, called Best
Demonstrated Available Technology—BDAT, such as
thermal and chemical treatment and stabilization, have
been evaluated by the United States Environmental
Protection Agency (USEPA). Recent patent literature
research (Raposo and Carvalho, in press) has shown
that there is a world trend towards thermal treatment
for recovering mercury from waste by vaporization and
* Corresponding author. Tel./Fax: +55-21-313-4993403.
E-mail addresses: raposoc@urano.cdtn.br (C. Raposo), claucw@
dedalus.lcc.ufmg.br (C. Carvalhinho Windmoller).
1
The most commonly used fluorescent lamp is the tubular type 40
W/T12, length of 122 cm (4-foot) and 3,8 cm of diameter (1.1=2 inch).
0956-053X/03/$ - see front matter # 2003 Elsevier Ltd. All rights reserved.
doi:10.1016/S0956-053X(03)00089-8
Brazil (Raposo and Roeser, 2001) show that mercury
lamps are thrown in the garbage in almost all cases, and
also that only a few, about 3%, are recycled by means
of chemical treatment.
Data of the National Electrical Manufacturers Asso-
ciation—NEMA (USEPA, 1997) in a work carried out
along with Sylvania, a lamp manufacturer, pointed out
that the mercury contained in an up-to-date fluorescent
lamp is probably 0.2% (0.042 mg) elemental mercury
(vapor phase) and the other 99.8% (20.958 mg) divalent
mercury form incorporated into the phosphorus powder
(Table 1). USEPA indeed insists on emphasizing that
the speciation of mercury contained in fluorescent lamps
is a controversial and complex subject.
Controversies are documented in Peer Review Report
(USEPA, 1998a) on Mercury Model application and
validation, which was designed to estimate mercury
emission from fluorescent lamps disposal. The most
significant doubts concern the speciation of mercury
contained in fluorescent lamps.
One of the reviewers noted that the concentration of
vapor phase elemental mercury in these lamps when
they are not electrified is limited by the vapor pressure
of the metal and thus cannot exceed saturation. Lastly,
mercury is added to these lamps to generated UV
radiation by electrical stimulation of elemental mercury
atoms in the gas contained in the lamp. Besides, it is
880 C. Raposo et al. / Waste Management 23 (2003) 879–886
Table 1
Mercury content of fluorescent lamps
Year Mercury (mg/T12 lamp)
Elemental Divalent Total
Pre-1992 0.082 40.918 41.000
1992–1996 0.060 29.940 30.000
1997–2007a 0.042 20.958 21.000
Source: USEPA, 1997
a
Mercury Emissions Model (The modeling period begins in 1992
and ends in 2007).
also possible that mercurous oxide might be used for its
spontaneous disproportionation to elemental and mer-
curic forms, and the greater ease of dispensing small
amounts per lamp using mercuric oxide (USEPA, 1997).
The mercury minimal content in fluorescent lamp has
been determined from NEMA’s demonstration that an
insufficient level of mercury will result in the premature
failure of a fluorescent lamp. This phenomenon is called
‘‘mercury starvation’’ within the industry. Individual
F40T12 lamps rated for 20,000 h of life need 10 mg of
mercury (0.7 ml) to ensure that the life-limiting mechan-
ism of such lamps is electrode emission mix depletion
and not mercury starvation. Variations in the mechan-
ical process used to insert the mercury into a fluorescent
lamp generally require that the average amount be set to
no lower than 15 mg, so that the minimum amount in
an individual lamp is at least 10 mg and lifetime is not
shortened due to insufficient mercury (USEPA, 1998b).
Another relevant aspect is the interaction between
mercury and other fluorescent lamp components. Sev-
eral research studies demonstrated a strong interaction
of mercury with Na present in glass (Doughty et al.,
1995; Thaler et al., 1995; Dang et al., 1999) and also
with alkali metal carbonates (CaCO3, BaCO3 and
SrCO3), and Sn from the tin oxide (SnO2) which coats
the electrodes (Dang et al., 1999).
In this work the thermo-desorption technique was
used to differentiate the species of mercury found in
mercury lamp wastes, which will certainly be relevant in
studies concerning their final disposal and recycling.
Characterization of lamp phosphorus powder and glass
matrices and the quantitative analysis of the total mer-
cury were also carried out.
The mercury thermo-desorption technique of solid
samples has been used since 1904 to determine total
mercury in ores from cinnabar mines and also in geo-
chemical prospecting (Koksoy et al., 1967; Aston and
Riley, 1972; Henry et al., 1972; Lidums 1972). Nowa-
days, the utilization of this technique has been revived
mainly as an important tool in studies of mercury spe-
ciation in solid matrices, since different species of mer-
cury desorb at different temperature ranges (Goleb,
1971; Windmöller et al., 1996; Biester and Nehrke,
1997; Biester et al., 2000).
2. Materials and methods
The samples for thermo-desorption study and char-
acterization were collected from 20–40 W fluorescent
lamp tubes, new and spent with different diameters from
assorted manufacturers. The lamps selected were initi-
ally cleaned and carefully imploded under a hood ade-
quate for mercury vapor exhaustion. Phosphorus
powder was manually collected by scraping with a
stainless steel spatula.
2.1. Characterization of fluorescent lamp glass and
phosphorus powder
In order to determine the approximate composition of
glass and phosphorus powder from fluorescent lamp,
the following techniques were used: atomic absorption
spectrometry (GBC Avanta AAS 932) for Na2O, K2O,
B2O5, CaO, MgO and Al2O3, atomic emission spectro-
metry/inductively coupled plasma (Spectroflame FMV
05 ICP/AES) for PbO and gravimetry for SiO2.
Elemental composition of phosphorus powder was
determined by colorimetry (P), potentiometry (F and
CI) and ICP/AES (the other elements). X-ray Diffrac-
tion (Rigaku-Geigerflex) was used to identify the crys-
talline phases.
2.2. Hg thermo-desorption
2.2.1. Qualitative analysis by thermo-desorption/atomic
absorption spectrometry (TDAAS)
The samples were analyzed by thermo-desorption
followed by atomic absorption spectrometry as descri-
bed in literature (Windmöller et al., 1996). The samples
were heated at a rate of 33
C/min from room tem-
perature to 570
C. A constant nitrogen flow (200 ml/
min) passing through the sample (masses varying up to
40 mg) took thermo-desorbed vapors to the detection
cell of an atomic absorption spectrophotometer CG-
Analytical (model GBC 380). The temperature over the
sample was monitored by a thermocouple. In this way,
one obtains registration of extinction (absorbance units)
as a function of the temperature over the sample, here-
inafter called thermograms or TD profiles. These pro-
files give information on the interaction force of
mercury with the matrix, and can be compared with
standard mercury profiles (spiked samples).
2.2.2. Preparation and analysis of spiked samples
Four samples were prepared by spiking the mercury
species standards, i.e., Hg0, Hg2Cl2, HgCl2 and HgO,
manufactured by Riedel-de Haën/Fluka, 99.5% (Ger-
many) to phosphorus powder from samples previously
submitted to thermal treatment for taking away the
existing mercury with a concentration of 5,000 ppm in
order to simulate commercial mercury lamps. The
 C. Raposo et al. / Waste Management 23 (2003) 879–886 881

samples spiked with mercury were homogenized by Table 2

maceration in an agate crucible. Thermograms were Approximated chemical composition of the main elements of
phosphorus powder used in a 40 W/T12 fluorescent lamp
obtained at different times after spiking to investigate the
possible effects or matrix interactions with the species Element Mass% Element mg.g1 Element mg.g1
added (7, 14, 30 and 45 days after the initial doping).
Ca (37.91 0.11) Al (580 30) Co (41)
TD profiles did not show any significant change. P (18.12 0.36) Ba (310 15) La (354)
F (2.95 0.14) Mg (143 2) Nd (313)
2.2.3. Glass samples tests Mn (0.86 0.02) Sr (125 7) Ce (121)
Cl (0.72 0.04) Ni (90 4) Sm (60.3)
Two glass samples from fluorescent lamps were Sb (0.61 0.01) Pb (38 3) Dy (30.3)
tested: one from a spent lamp and another from a Fe (0.19 0.01) Cr (10 1) Er (30.3)
new lamp (both with 200 mg mass). The phosphorus Cd (0.18 0.01) Zn (71) Ho (20.3)

powder was carefully removed from the glass before
thermo-desorption analysis. The analytical procedure
used was similar to the one used for spiked samples
and phosphorus powder matrix. The samples were
submitted to heating to 800
C (apparatus maximum
temperature).
2.3. Determination of total mercury in the phosphorus
powder matrix and glass samples
Total mercury determinations were carried out by
using a cold vapor-atomic absorption (CV-AAS) appa-
ratus manufactured by Perkin Elmer, model FIMS 400.
Phosphorus powders matrix and pulverized glass sam-
ples, in aliquots of 0.5 g, were solubilized in HNO3 (2:1)
and diluted to 1,000 ml. Hg determinations were per-
formed according to Standard Methods for the Exami-
nation of Water and Wastewater procedures (American
Public Health Association et al., 1995). All analyses
were made in duplicate.
3. Results and discussion
3.1. Determination of main chemical elemental analysis
of phosphorus powder matrix
A 40 W standard fluorescent lamp typically contains
from 4 to 6 g of phosphorus powder, which represents
approximately 2% of its total mass. This powder is
mainly calcium phosphate, being fluorapatite
Ca5(PO4)3F the predominant crystalline phase as
revealed by XRD analysis. Other minor phases were
also identified: CaCO3—calcite; CaHPO4—monetite,
CaHPO4.2H2O—brushite, TiO2—rutile/anatase and
b-Ca3(PO4)2—whitlockite. The approximate chemical
composition of the main elements of this powder,
shown in Table 2, is very close to the chemical compo-
sition of the apatite standard manufactured by Astimex
Scientific Limited and used as a reference in micro-
analysis studies (Astimex, 2001). Other elements in
minor concentration such as Cu, Mn, Pb, Sb, Sn and
rare earths were also present in this powder (Raposo et
al., 2000). In general, a combination of fluorescent
minerals and rare earths is used in phosphorus powder
Analysis methods: Colorimetry (P); Potentiometry (F and Cl), ICP/AES
(other elements); oxygen not analyzed by direct method. Chemical composi-
tion of synthetic apatite standard—Astimex (2001): O—38.07%; F—3,77%;
P—18.42%; Ca—39.74%.
to achieve a good range of color response. Phosphorus
powder is originally mercury free. This metal is added
during lamp manufacturing.
3.2. Analysis of the spiked samples
Thermograms of phosphorus powder samples spiked
with the mercury standards Hg0, Hg2Cl2, HgCl2, HgO
are shown in Fig. 1. The following features can be
observed:
(a) Hg0—the desorption of this mercury phase
occurs at low temperatures ranging from room
temperature (RT) up to 180
C (maximum at
150
C). The peak has sharp conformation (lep-
tokurtic form);
(b) Hg2Cl2—an intense peak appears from 100 to
250
C, with a maximum at 225
C and a
shoulder in the interval of 250 up to 350
C of
HgCl2 formed due to the decomposition of some
Hg2Cl2 . These two peaks are probably related to
the two decomposition steps, according to Eqs.
(1) and (2):
Hg2 Cl2 ! Hg0 þ HgCl2 ð1Þ
HgCl2 ! Hg0 þ Cl2 ð2Þ
This was confirmed by the TD profile of the sam-
ple spiked with HgCl2, which showed similar des-
orption temperature range of the HgCl2 shoulder.
(c) HgCl2—presents a broad peak (platikurtic form)
between 150 and 340
C, (maximum at approxi-
mately 275
C);
(d) HgO—thermo-desorption in the temperature
interval from 250 to 500
C, with a maximum at
approximately 400
C. The peak conformation
also shows a planikurtic shape at the interval
from 380 to 430
C.
882 C. Raposo et al. / Waste Management 23 (2003) 879–886
Fig. 1. TD profiles of spiked mercury standards in phosphorus
powder matrices of spent fluorescent lamps.
The general order of temperature desorption observed
in this work was:
Hg0 < Hg2 Cl2 < Hgcl2 < HgO
3.3. Analysis of the phosphorus powder samples of new
and spent lamps
The main characteristics of the samples submitted to
thermo-desorption tests are reported in Table 3.2 Ana-
lytical results of total mercury contents exhibited in this
work show that the phosphorus powder matrix (pre-
dominant crystalline phase—fluorapatite) is the hazar-
dous material responsible for polluting spent fluorescent
lamp waste. Total mercury concentrations in the phos-
phorus powder matrix from selected samples varied
from 300 30 mg.g1 to 13,300  1,300 mg.g1. Higher
concentrations were observed in spent lamps, and the
minimum in new ones, which was expected because the
interactions of mercury with the matrix increases with
the lamp use time (Doughty et al., 1995; Thaler et al.,
1995). Considering that there must not exist variations
in mercury content between a new to a spent lamp, the
difference pointed results from mercury in spent lamps
is essentially adsorbed into the phosphorus powder
matrix, and to a minor extension, into the glass, and
possibly into the other components (emissive dust, fila-
ments, electrodes), while in new lamps, the phosphorus
powder matrix is not the preferential site for mercury
and the glass do not show interaction with this metal.
This is expected since mercury is introduced during the
lamp manufacturing process as vapor phase elemental
mercury. Thermo-desorption results are also summar-
ized in Table 3. They demonstrate the correlation
2
In Table 3, the term main refers to the signal with the largest area
in relation to the TD profile, while secondary refers to the one with the
smallest area.
between the concentration and the mercury species
contained in the phosphorus powder matrix.
From the TD profiles, the main features were:
(i) Hg0—was the dominant mercury species detected
in LF01, LF03, LF04, LF08-LF10 (spent
lamps—Table 3, Fig. 2a) and LF11, LF13-LF17
(new lamps—Table 3, Figs. 2b and 2d);
(ii) Hg1+—was detected in LF01, LF03-LF07, LF10
(spent lamps), LF11, LF13, LF16-LF17 (new
lamps) and as the main species in LF05-LF07
(spent lamps—Fig. 2c);
(iii) Hg0+Hg1+—both metallic and a partially oxi-
dized form were released as the main species from
LF04 (spent lamp) and LF11, LF13 (new
lamps—Fig. 2d);
(iv) Hg 2+—occurred in LF01, LF02-LF04, LF06,
LF09 (spent lamps), LF12, LF14 (new lamps)
and as the main species in LF02 (Fig. 2e) and
LF12 samples, curiously one of them was a new
lamp sample (LF12), which would be expected to
have little mercury in the oxidized state;
(v) metallic mercury and oxidized forms (Hg0, Hg1+
and Hg2+) were released from samples LF01,
LF03, LF04 and LF06. In the LF03 profile
(Fig. 2f), it can also be observed a narrow peak at
410
C. This desorption temperature is coin-
cident with maximum temperature in the TD
profile of the sample spiked with HgO. More-
over, the configuration of this profile is different
from all the profiles studied, probably due to
some additive present in this lamp and absent in
the others.
The highest mercury concentrations were observed in
samples in which Hg1+ (LF05 and LF06) or Hg2+
(LF02) was the main mercury species. Part of Hg1+ and
Hg2+ species might be present as chlorides, since the
TD profiles obtained were similar to those of Hg2Cl2
and HgCl2 standards. Concerning divalent mercury,
some research studies suggest the formation of HgO due
to the reaction of Hg0 with free oxygen generated by the
decomposition of the alkaline metal oxides that are
volatilized in the electrical discharge zone of the lamp
(Foust et al., 1999; Klinedinst et al., 2001). On compar-
ing the thermograms in Fig. 2 with that of the HgO
spiked matrix (Fig. 1), one observes that only the LF02
sample (Fig. 2e) has a signal similar to that relative to
oxide. Nowadays, several patents have been deposited
in the USA for the use of organic complexing agents
(ascorbic acid—C6H6O6 and sodium glutamate—
C6H11NaO7) that may be incorporated in the phospho-
rus powder and/or cementing base matrix during lamp
manufacture aiming to reduce mercury leaching in
Toxicity Characteristic Leaching Procedure (TCLP), or
by occasion of the waste final disposal (Haitko et al.,
 C. Raposo et al. / Waste Management 23 (2003) 879–886 883

Table 3
Characteristic of samples submitted to thermo-desorption tests and results

Sample Description Hg concentration Mercury species

(mg.g1)
Main Secondary

Phosphorus powder of spent fluorescent lamps

LF01 F40T12/LDP 4,000400 Hg0 Hg1+; Hg2+
LF02 F40T12/LDP 8,900900 Hg2+
LF03 F40T12/LD/U 3,500400 Hg0 Hg1+; Hg2+
LF04 F40T10/ELD 6,300600 Hg0; Hg1+ Hg2+
LF05 F40T12/LDE 6,700700 Hg1+
LF06 F40T12/LDE 13,3001,300 Hg1+ Hg0 ; Hg2+
LF07 F40T12/ELD 3,600400 Hg1+ Hg0
LF08 F40T12/SLD, Duramax 1,430130 Hg0
LF09 F40T10/LDE 2,100200 Hg0 Hg2+
LF10 F20T12/SLD, Duramax 1,700150 Hg0 Hg1+

Phosphorus powder of new fluorescent lamps

LF11 F40T12/SLD, Duramax 750 80 Hg0; Hg1+
LF12 F40T10/LDE 330 30 Hg2+
LF13 F40T12/SLD, Duramax 760 80 Hg0; Hg1+
LF14 F40T12/SLD, Duramax 800 50 Hg0 Hg2+
LF15 F40T12/SLD, Duramax 750 60 Hg0
LF16 F40T10/LDE 300 30 Hg0 Hg1+
LF17 F20T12/SLD, Duramax 950 40 Hg0 Hg1+
Soda lime glass of fluorescent lamps
Sample Description Hg concentration (ng.g1) Mercury species
1 Spent lamp 4,300400 Possibly Hg1+; Hg2+
2 New lamp 18 2 –

Analysis method: Cold vapor-atomic absorption (CV-AAS).

1998; Foust et al., 1998; Woodwart et al., 1999).
Another alternative under study is the introduction of
an anti-oxidizing agent in the lamp structure, such as
iron (Foust et al., 1999), manganese compounds, copper
and also halogen compounds (NaBr and NaCl) (Kline-
dinst et al., 2001), precisely to avoid the oxidation of the
metallic mercury into more soluble forms. In the present
work, the possibility of incorporation of these com-
plexing and/or anti-oxidizing agents is discarded at least
in spent lamps, since they were manufactured before the
patents (last 2 years). On the other hand, even con-
sidering the possibility of this technological innovation
in new lamps, the thermo-desorption profile should evi-
dence this, i.e., oxidized species should not appear.
The presence of Hg1+ and Hg2+, as verified, is rather
worrisome, because these species, more soluble than
metallic species, could fail many lamps in TCLP tests. It
must be noted that the North American regulatory limit
in TCLP tests is 0.2 mg.l1 (USEPA, 1999, 2000). The
results of mercury concentration in the phosphorus
powder matrix shown in Table 3, firstly adjusted for
phosphorus powder mass contained in a 40-W fluor-
escent lamp tube (4–6 g) and, secondly for the total
mass of each lamp (average of 275 g), reveal that the
mercury content exceeds this regulatory limit, as for
example in waste represented by samples: LF02, LF04,
LF05 and LF06. The wastes thus characterized are
considered hazardous and must receive special treat-
ment. The results also show a strong tendency of
occurrence of high mercury concentrations in condi-
tions in which the oxidized form prevails, be it either
Hg1+ isolatedly, Hg2+ or (Hg1++Hg2+).
The interaction of mercury with the phosphorus
powder matrix from spent lamp samples was evinced in
our TDAAS study and the results showed that the spe-
cies of mercury in spent lamp waste are Hg0, Hg1+ and
Hg2+. Hg0 and Hg1+ predominate upon the divalent
species, differently from the data presented in USEPA’s
Report (1997), which suggests the divalent type as pre-
dominant in this material.
3.4. Analysis of glass samples
The glass used in fluorescent lamps, the soda type, has
approximately the following composition: SiO2
(69.2 1.4)%, B2O3 (16.5 0.5)%, CaO (4.2 0.2)%,
Na2O (3.1 0.2)%; Al2O3 (2.7 0.1)%, K2O (1.8 0.1)%,
MgO (1.4 0.1)% and PbO < 0.1% (Raposo et al., 2000).
Fig. 3 shows the TD profile of the spent lamp glass
matrix with a mercury concentration of about
(4,300 400) ng.g1. Mercury desorption was recorded
in an interval ranging from 240 to 800
C, maximum
peak at 618
C. This temperature interval is not com-
parable to any known specific mercury standard, which
884 C. Raposo et al. / Waste Management 23 (2003) 879–886
Fig. 2. TD profiles of phosphorus powder samples of spent and new fluorescent lamps.
Fig. 3. TD profile of glass sample of a spent fluorescent lamp.
suggests that there is a type of interaction between
mercury and the glass components. Dang et al. (1999)
suggested that most of the Hg species strongly chemi-
sorb onto Na present in the glass or form intermetallic
compounds with this element, i.e., mercury in Hg0 state.
In turn, Doughty et al. (1995) postulated that electro-
chemical reactions occur inside the lamp, among them
mercury oxidation due to plasma formation. Oxidized
mercury can diffuse into the glass matrix. The thermo-
gram shown in Fig. 3 seems to agree with Doughty et
al.’s hypothesis because the signal only appears after
200
C, which is a characteristic of oxidized species. The
appearance of oxidized species (Hg1+ and Hg2+) in the
phosphorus powder matrix, according to what is shown
in this paper, represents then a strong argument for the
linking of these species and, not Hg0, to glass.
In the case of the new lamp glass matrix, mercury
concentration of about (18  2) ng.g1, the absorbance
signal was practically zero for a sample mass equal to
the one used for the spent lamp test (200 mg). This
shows that the mercury interaction is proportional to
the lamp use time, which was more precisely determined
as its square root by Thaler et al. (1995), and that it
depends on the physicochemical processes that occur
inside it (Dang et al., 1999).
4. Conclusions
The analytical results presented in this paper show
that the phosphorus powder matrix is the hazardous
 C. Raposo et al. / Waste Management 23 (2003) 879–886
and polluting material in the waste of spent fluorescent
lamps. Mercury concentrations in this type of matrix
are much higher than that in new lamps of the same
kind, in some cases 40-fold. The basic difference is that
mercury in spent lamps is adsorbed primarily into the
phosphorus matrix and, to a lesser extend, into glass
and other components. In new lamps, mercury pre-
dominates as a vapor phase element.
The interaction between mercury and the phosphorus
matrix is evidenced in TDAAS studies, which revealed
the presence of both Hg0, the species initially added, as
well as the Hg1+ and Hg2+ species. The species found
most commonly and in larger quantity were Hg0 and
Hg1+, however there were cases in which the pre-
dominance of Hg2+ was absolute. In contrast to data
from the USEPA Report (1997) which suggest that the
divalent species predominates in this material, our
results show that the mercury species Hg0, Hg1+ and
Hg2+ occur in spent lamps, with the predominance of
Hg0 and Hg1+ over the divalent species. The thermo-
grams also showed that mercury could be released at
very high temperatures, over 400
C.
The desorption temperature of mercury of glass sam-
ples of spent lamps reaches levels much higher than
those of any known specific mercury standard, reaching
even 800
C. Even considering that glass wastes of
fluorescent lamps are not classified as hazardous or
semi-inert in leaching/solubilization tests, special atten-
tion must be paid to these results in case this material is
recycled/reused, for high temperature conditions may
release mercury in substantial quantities, depending on
the volume of material used in the process.
The Hg+1 and Hg+2 species can form compounds
more soluble than Hg0 and therefore have a greater
impact on the environment, since they can be trans-
ported via the aqueous phase more easily, oxidized
and methylated. In terms of treatment of fluorescent
lamp waste (spent and/or new lamps), either thermal
or any other, it can be concluded that the ideal mer-
cury recovery conditions are far from being the same
for different kinds of lamps, since these lamps contain
not only different species, but also different quantities
of mercury.
Desorption temperatures must be taken into con-
sideration during the thermal processing of lamp wastes.
References
American Public Health Association et al.,1 995. Standard Methods
for the Examination of Water and Wastewater, nineteenth ed.
Washington, DC (3112-Hg B, 3500-Cd B e 3500-Pb B).
Astimex Scientific Limited, 2001. Mineral Standard Mount—Serial 99-
045. MINM 25-53. Available at: http://www.proscitech.com.
Aston, S.R., Riley, J.P., 1972. The determination of mercury in rocks
and sediments. Anal. Chim. Acta 59, 349–354.
Biester, H., Nehrke, G., 1997. Quantification of mercury in soils and
885
sediments—acid digestion versus pyrolysis. Fresenius J. Anal.
Chem. 358, 446–452.
Biester, H., Gosar, M., Covelli, S., 2000. Mercury speciation in sedi-
ments affected by dumped mining residues in the drainage area of
the Idrija Mercury Mine. Slovenia. Environ. Sci. Technol. 34, 3330–
3336.
Dang, T.A., Frisk, T.A., Grossman, M.W., Peters, C.H., 1999. Iden-
tification of mercury reaction sites in fluorescent lamps. J. Electro-
chem. Soc. 146 (10), 3896–3902.
Doughty, D.A., Wilson, R.H., Thaler, E.G., 1995. Mercury-glass
interactions in fluorescent lamps. J. Electrochem. Soc. 142 (10),
3542–3550.
Foust, D.F., Akins, W.W., Arsena, V.J., Doliac, D.J., Haitko, D.A.,
Jansma, J.B., 1998. Method for Incorporating Anti-oxidants Within
a Fluorescent Lamp’s Base Shell. US Patent 5,821,682.
Foust, D.F., Haitko, D.A., Dietrich, D.K., 1999. Control of Leach-
able Mercury in Fluorescent Lamps by Iron Addition. US Patent
5,998,927.
Goleb, J.A., 1971. The use of permanent mercury sources for spectro-
scopic studies of varying amounts of mercury, Appl. Spectrosc. 25,
94–96.
Haitko, D.A., Foust, D.F., Dietrich, D.K., 1998. Antioxidant Control
of Leachable Mercury in Fluorescent Lamps. US Patent 5,754,002.
Henry, H.G., Stever, K.R., Heady, H.H., 1972. Determination of
mercury in low-grade ores. Appl. Spectrosc. 26 (2), 288–293.
Klinedinst, K.A., Shinn, D.B., Fowler, R.A., 2001. Mercury Vapor
Discharge Lamp Containing Means for Reducing Leachable Mer-
cury. US Patent 6,169,362.
Koksoy, M., Bradshaw, P.M.D., Tooms, J.S., 1967. Notes on the
determination of mercury in geological samples. Trans. Inst. Mining
Met. Sect. B, 76, b121-b124..
Lidums, V., 1972. Determination of mercury in small quantities by
direct combustion combined with cold vapor-atomic absorption.
Chem. Scripta 2, 159–163.
Raposo, C. Classificação dos resı́duos gerados por lâmpadas de mer-
cúrio, quanto aos riscos potenciais ao meio ambiente e à saúde
pública. Revista Quı́mica Nova, Brasil, (Código 195/00, in press).
Raposo, C., Carvalho, C.F. Recycling processes for mercury lamps—
The state of the art. Revista Quı́mica Nova. Brasil (Código 135/00,
in press).
Raposo, C., Brito, W., Menezes, M.A.B.C., Palmieri, H.E.L., Leonel,
L.V., Oliveira, R.C., 2000. Caracterização quı́mica dos principais
constituintes de lâmpadas de mercúrio com vistas ao controle
ambiental. In: V ENAN (Encontro Nacional de Energia Nuclear),
Rio de Janeiro. p. 145.
Raposo, C., Roeser, M.H., 2001. Contamination of the environment
by the current disposal methods of mercury-containing lamps in the
State of Minas Gerais, Brazil. Waste Management 21 (7), 661–670.
Raposo, C., 2001 Contaminação ambiental provocada pelo descarte
não-controlado de lâmpadas de mercúrio no Brasil. Ouro Preto,
Brasil. Tese de Doutorado. Doctorate thesis, Departamento de
Geologia, Universidade Federal de Ouro Preto, Brazil, p. 194.
Thaler, E.G., Wilson, R.H., Doughty, D.A., 1995. Measurement of
mercury bound in the glass envelope during operation in fluorescent
lamps. J. Electrochem. Soc. 142 (6), 1968–1970.
USEPA, 1997. Mercury Emissions from the Disposal of Fluorescent
Lamps. Final Report. Washington: 1 v. Available at: http://www.
epa.gov/epaoswer/hazwaste/id/merc-emi/merc-pgs/merc-rpt.pdf.
USEPA, 1998a. Peer Review of the USEPA Analytical Model: Mer-
cury Emissions from the Disposal of Fluorescent Lamps. Comment
response document. Comment No. 3-8. Available at: http://www.
epa.gov/epaoswer/hazwaste/id/merc-emi/merc-pgs/peerrev.pdf.
USEPA, 1998b. Comments Related to Industry Source Reduction
Efforts. Response to comments document. Final rule on hazardous
waste lamps. (DCN FLEP-00156). 1 v. Available at: http://www.
epa.gov/epaoswer/hazwaste/id/merc-emi/merc-pgs/vol-2.
USEPA, 1999. Hazardous Waste Management System; Modification
886 C. Raposo et al. / Waste Management 23 (2003) 879–886

of the Hazardous Waste Program; Hazardous Waste Lamps; (40
CFR Parts 260, 261, 264, etc.), Final Rule, Federal Register/vol. 64,
No. 128, pp. 36466–36490.
USEPA, 2000. Background Information on Mercury Sources and
Regulations. Available at: http://www.epa.gov/grtlakes/.
Windmöller, C.C., Wilken, R.D., Jardim, W.F., 1996. Mercury
speciation in contaminated soils by thermal release analysis. Water,
Air and Soil Pollution 90, 399–416.
Woodwart, D.R., McGuire, S.D., Hellebrekers, W., 1999. TCLP
Compliant Fluorescent Lamps. US Patent 5,898,265.