Removal and recovery of mercury from used fluorescent lamp glass by pyrolysis

* Corresponding author (samanwije@aol.com) Abstract: Recovering mercury in metallic form from used fluorescent lamps is important for environmental safety and for its re-use in industry. Presently, pyrolysis has become a widely used thermal treatment technology for various wastes. Since it is performed under anaerobic conditions, the toxic metals are not oxidized and this minimizes their leaching from the treated material. This technology also reduces the risk of formation of toxic chloro-organic compounds such as dioxins. Therefore, in the study reported in this paper, pyrolysis was applied for the treatment of used fluorescent lamp glass, with special emphasis on mercury removal and recovery in metallic form. A laboratory scale pyrolyser was used for all the experiments. Pyrolysis was conducted at different temperatures using nitrogen as the carrier gas. A cold trap was attached to condense and recover metallic mercury from the flue gas. Results showed that a temperature of 600 C is sufficient to achieve nearly 100% mercury removal. Around 20% of metallic mercury was recovered in the cold trap. The rest were trapped in the activated carbon filter and acid trap, showing that the process is able to completely prevent release of mercury to the environment.


INTRODUCTION
Mercury is a hazardous material with known toxic health effects. Its presence in the environment is a persistent and growing problem due to its ability to bio-accumulate in various food chains (Jang et al., 2005). Mercury comes from a variety of sources including batteries, paints, dyes, electronic components and most significantly fluorescent lamp tubes. Based on a 1999 National Electrical Manufactures Association (NEMA) survey, an average four-foot fluorescent lamp contains about 11.6 mg of mercury, but the amount varies with the size and the brand of the lamp. Though this concentration has considerably decreased over time, the use of fluorescent lamps in homes, schools, factories, office facilities, and parking lots has increased so that the problem still persists. Fluorescent lamps are usually preferred over incandescent lamps due to their efficient energy consumption (Hildenbrand et al., 2000). Nationwide, almost 600 million mercury containing lamps are discarded each year (ALMR, 2010), while the recycling rate is below 24%. Despite the various actions of different countries to increase the recycling rate (Hilkene et al., 2005) most lamps are still incinerated or disposed with municipal solid wastes in landfill sites (ALMR, 2010). This could release higher amount of mercury to the environment (Changsuphan et al., 2003). Hence these lamps typically require special handling and treatment at the time of disposal.
Recovering metallic mercury from fluorescent lamps is important as it can be re-used in industry. The treated glass could also be recycled. Alternative treatment strategies include dry recycling, wet acid washing, aerobic thermal treatment and electrochemical treatment. The electrochemical process has been developed for gas phase mercury recovery (Paul & David, 2002). Most lamp recyclers in the United States employ the dry process, where the lamps are crushed and the mercury vapour is captured by carbon filters. The mercury containing waste is transported to centers for further treatment and recovery of mercury (Jang et al., 2005). In Japan, wet acid washing is the preferred approach. This involves the crushing and washing of glass with strong hydrofluoric acid to remove mercury, and re-washing glass with pure water to remove excess acid. The acid solution passes through a special mercury absorbent and the treated acid is then recycled for lamp washing. The absorbent is finally cemented with special chelating agents and disposed at designated sites (personal communication, 4 August 2009). However, this process is complicated and has problems related to the leaching of mercury from cemented blocks.
There is a need to develop more practical and economical approaches to treat fluorescent lamps, with emphasis on mercury recycling and safety for handling the treated sample. In this connection, anaerobic thermal decomposition or pyrolysis is currently becoming a popular approach for use with various wastes. This process has been successfully applied for pilot-scale treatment of dioxin and polychlorinated biphenyl (PCB)contaminated sediments (Hu et al., 2006). Due to the anaerobic conditions, the risk of further formation of such toxic chloro-organic compounds is minimized. In addition, toxic metals are also stabilized thus limiting their leachability (Hu et al., 2007). This research extends the application of pyrolysis for the treatment of used fluorescent lamp glass. It was postulated that this technology has special advantages for the removal and recovery of metallic mercury from fluorescent glass.

METHODS AND MATERIALS
Setting up and operation of the pyrolysis system: The pyrolysis system (ALS, Tsukuba, Japan) is shown in Figure 1. The 8 L capacity quartz chamber is equipped with a sample holder and an internal temperature sensor. The condensation section cools down and collects the mercury vapour coming from the chamber. The temperature in the cold trap is around 0 -5 o C throughout the experiments. For safety purposes, a special type of activated carbon (DAISO, Japan) column was installed to adsorb mercury vapour that may escape from the condensation unit. A 2.5% HNO 3 acid trap was also added to remove any trace mercury prior to the release of exhaust gas to the surroundings.
During a typical operation, the required temperature programmes were initially saved in the machine. A measured amount of used fluorescent lamp glass, which was crushed to below one inch, was placed inside the chamber. The chamber cap was closed and sealed by tightening the screws. The glass manifold from the chamber to the condensation tank was heated and maintained at 220 o C throughout the experiment using a ribbon heater in order to avoid mercury condensation inside these tubes. The exhaust valve was closed and the vacuum pump was switched on to reduce the pressure to -98.8 kPa [assuming that no oxygen (O 2 ) or air remains inside]. At -98.8 kPa, the inlet valve was automatically switched on to allow nitrogen (N 2 ) gas (purity 99.999%) to purge the system at an initial flow rate of 1 -2 L/min. The exhaust valve opens automatically when internal pressure reaches 3 kPa. At this point, the nitrogen flow was adjusted to the desired experimental condition. The heater was also turned on, allowing the temperature inside the chamber to increase and decrease according to the programmed instructions. During the cooling stage, the cover of the chamber was opened when the temperature reached 400 ˚C to accelerate the cooling rate. When the temperature dropped to room temperature, the chamber cap was opened and the sample was collected.
The treated sample was stored in a dark glass bottle prior to analysis. The tar, acid trap and activated carbon were also collected and stored in a refrigerator until analysis.
Mercury removal experiments: Using the above procedure, pyrolysis runs were conducted at different temperatures (300, 400, 500, 550, 600, 700 and 800 ˚C) for 30 min to identify the best condition for mercury removal. Around 75-100 g of used lamps from Kinden Cooperation, Osaka, Japan were employed in each run. Nitrogen gas flow rate was 0.2 L/min.

Mercury recovery experiments:
Since the mercury concentration of used fluorescent lamp glass was very small, experiments were first conducted using pure metallic mercury (Hg 0 ) in order to identify the suitable conditions for its recovery. Less than 0.3 g of mercury was used in each run. A series of experiments was conducted at different temperatures (300 -800 ˚C) at 0.2 -2 L/min N 2 gas flow rates. The adsorption capacity of activated carbon was also evaluated in these preliminary experiments. Final runs were conducted using approximately 2 kg of actual fluorescent lamps.

Parameter analysis
a) mercury analysis: All samples for mercury analysis were submitted to the Environmental Research Center, Tsukuba, Japan. The total mercury was analyzed according to Suzuki et al. (2004), for all liquid and solid samples.

b) metal analysis and leaching test:
Crushed fluorescent lamp glass samples were subjected to hot acid digestion for complete dissolution of heavy metals following the standard methodologies described in American Public Health Association (APHA, 1995). The heavy metal contents (except mercury) of supernatants after digestion were analyzed by ICP 757 and 575 (Nippon Jarrel Ash Co., Ltd., Kyoto, Japan). Leaching tests were conducted following the standard methods of the Ministry of Environment, Japan (Suzuki et al., 2004).

Initial mercury content of fluorescent lamp glass
The initial mercury contents of 3 representative used fluorescent lamps were found to range from 0.89-3.25 mg/kg. For this analysis, fine particles below 2 mm, which generally consists of phosphor powder, were pre-separated prior to the determination of the initial mercury content. A separate analysis of these particles revealed a very high content of mercury (1100 mg/kg), indicating that most of the mercury is concentrated in these fractions. Studies of Jang et al. (2005) have also reported very high concentrations of mercury in the phosphor powder, which is around 94.12-97.02% of total mercury present in the lamp (Jang et al., 2005). Since during the crushing process this powder readily detaches from the glass and settles down at the bottom of the crushing vessel, careful attention must be paid on their collection. As a consequence, the mercury content of lamps may vary significantly depending on the remaining phosphor powder attached to the glass. Furthermore, lamps manufactured 10-15 years ago have even higher content (4700 mg/kg) of mercury and these are also concentrated in the phosphor component ( Bostick et al., 1996).

Mercury removal efficiency during pyrolysis
Compared to other metals, mercury generally exhibits a low boiling point of around 357 ˚C. According to Edmonds et al.(1995), the vapour pressure of mercury increases drastically from 0 to 400 ˚C. Table 1 shows the results of mercury removal during pyrolysis at different temperatures. As expected, a very high mercury removal of nearly 90% was observed even at 300 ˚C (Table 1). From 400 ˚C, the removal further increased until close to 100% was achieved at 600 ˚C ( Figure 2). However, glass deformation became apparent at 700 ˚C and the glass began to melt at 800 ˚C. Therefore, the temperatures beyond 600 ˚C were considered unsuitable for actual applications.
The efficiency of mercury removal depends on the type and the brand of lamp. According to Jang et al. (2005), complete removal may be possible at 400 ˚C but in some cases, temperatures above 500 ˚C may still be insufficient. Through the system reported in the paper it was possible to achieve a maximum removal of 99.8% at 600 o C indicating that only 0.2% of mercury initially present in raw glass was detected in the treated sample. This small fraction of mercury may be considered as strongly bound to glass matrices. However, leaching test of the pyrolyzed lamp glass showed a positive result regarding the high stability of the residual mercury. The very low concentration of mercury in the supernatant

September 2011
Journal of the National Science Foundation of Sri Lanka 39 (3) (< 0.0005 mg/L) indicates that the treated glass can be disposed safely if not recycled.

Mercury recovery with pure metallic mercury
Since the mercury concentration of used fluorescent lamp glass was low, pure metallic mercury was used for the initial studies to establish suitable conditions for its recovery. It was found that the best temperature for maximum mercury removal, without changing the glass quality, was 600 ˚C. Recovery tests were therefore conducted at 550 -600 ˚C at a holding time of 60 min.
Preliminary runs showed that mercury condensation occurred inside the glass tubes between the chamber and cold trap since the temperature in this section was below 100 ˚C. To prevent such condensation, these tubes were separately heated up to 220 ˚C. After this adjustment, metallic mercury condensation was successfully channeled to the desired collection tank. Table 2 shows that almost all metallic mercury (100%) can be recovered during pyrolysis.
Generally, the mercury containing wastes are placed in a retort and heated for long hours (4-24 hours) at a temperature of above 357 ˚C (boiling point of mercury) and below 550 ˚C ( Truesdale et al., 1993). In the system reported in this paper, the total time for the removal and recovery of mercury, including the heating and cooling stages, was only around 6 hours, suggesting that it is comparable if not superior in performance relative to established methods.
For comparison, a similar experiment was also conducted in the presence of air, instead of nitrogen to simulate incineration conditions. The air flow rate was maintained at 1 L/min. Results showed that only around 10% of metallic mercury could be recovered under this condition. The rest may possibly be in the form of HgO and dihalide species such as HgCl 2 , HgF 2 , HgI 2 and HgBr 2 (Figure 3). Zhao et al. (2010) recently reported that mercury oxidation is considerably higher in the presence of oxygen than with nitrogen. In addition, mercury oxidation with air in the presence of chlorine can also produce HgCl 2 (Puchakayala et al., 2006). HgCl 2 formation decreases at temperatures above 600 ˚C. These reactions could probably explain the low recovery of mercury in its metallic form during the simulated incineration runs.

Mercury recovery from used fluorescent lamp glass
Initial tests for mercury recovery from actual fluorescent lamps were conducted using small samples (75-100 g). With this amount, a recovery of 10-25% in the cold trap was achieved. However, the formation of metallic mercury in the condensation tank was not observed. In order to scale up the recovery, a larger quantity of crushed sample (2 kg) was used. Under this condition, clear formation of metallic mercury was observed in the cold trap ( Figure 4). Nevertheless, mass balance calculation still revealed an incomplete recovery. Analysis has revealed that 31.9 mg of metallic mercury was present in the condensation tank, which corresponds to around 19% of the total mercury in the 2 kg fluorescent lamp. In this regard, it is extremely important that mercury, regardless of form, should be prevented from being released to the environment. This study has confirmed that activated carbon is well-suited for this purpose. Huang et al. (2008) reported that commercially available activated carbon absorbs 50.5% of elemental or metallic mercury. They have also mentioned that the activated carbon produced from chicken waste absorbs 68.8% of elemental Hg in the flue gas (Huang et al., 2008). In this study, the special   type of activated carbon utilized (DAISO, Japan) was confirmed to absorb a very large amount of around 13 mg mercury/g of carbon ( Figure 6). Through this application, a huge fraction (77%) of the total mercury that escaped the cold trap was trapped. Finally, the remaining 3.3% was detected in the acid trap. The distribution of mercury after treatment is summarized in Figure 5 and Table 3. These data show the superior safety feature of the system in preventing the release of highly toxic mercury to the environment during treatment.    activated carbon. Detailed work on this aspect will be considered in the future.
In addition to mercury, the fluorescent lamp glass may also contain other organic and inorganic compounds. ICP analyses also showed that both raw and pyrolyzed glass contain P, Si, Na, Al, Mg and K. However, their individual concentrations were less than 2 ppm. Chang et al. (2007) have reported that the phosphor powder has traces of rare earth metals such as La, Gd, Tb and En. In the same study, SiO 2 , Na 2 O, MgO K 2 O, PbO, and some organic compounds have also been detected (Chang et al., 2007). During pyrolysis the apparent decomposition of organic compounds was observed (Figure 7). The pyrolyzed sample acquired a black colouration. However, with such limited qualitative data, not much can be said about the fate of these compounds. This is yet another interesting topic for future investigations on the pyrolysis of fluorescent lamps. From estimates of the above data, 8000 mg of metallic mercury may be recovered per ton of used lamp glass. However, it must be emphasized that this may vary depending on the size of lamps as well as the amount of mercury present at the time of treatment etc. (Raposo et al., 2003). There is an immediate concern about the handling of the activated carbon, where the majority of the volatilized mercury is trapped. Although a second stage pyrolysis is a reasonable choice, more investigations are necessary to evaluate the conditions that would enhance their recovery in the cold trap as metallic mercury, considering that most fractions were concentrated in the Adssorbed Hg (mg/kg)