International Summer Water Resources Research School Dept. of Water Resources Engineering, Lund University Comparative study of several kinds of adsorption materials in Hg2+ adsorption performance By Johanna Norup 2013-07-19 Johanna Norup 7th Lingfeng Summer Research School Abstract This report is a result of the 7th Lingfeng Summer Research School at Xiamen University. The project has been focused on an investigation of different adsorbents efficiency to adsorb mercury gas. The different kinds of adsorbents have been compared in a bench scale system with a nitrogen-mercury gas mixture. A comparison has been made with the amount of mercury that has passed the adsorbent without being adsorbed. This project has studied the types BPL, DARCO FGD, molecular sieve (SBA-15) and modified adsorbents: PBL-CX, DARCO FGD-CX, molecular sieve-CX. The detection of the concentration was done with an Atomic Fluorescence Spectrometry, AFS. The results showed that FGD-CX was the best adsorbent in the study and adsorbed 96% of the mercury. The worst tested adsorbent was molecular sieve that only adsorbed 34% of the mercury gas. The comparison with and without modification concluded that an activated carbon modified with benzoic acid increased the ability to adsorb mercury gas. Keywords: mercury removal, mercury gas, adsorption, activated carbon, flue gas, AFS Johanna Norup 7th Lingfeng Summer Research School Table of Contents Abstract .................................................................................................................................................. 2 1 Introduction ........................................................................................................................................ 4 1.1 Aim ............................................................................................................................................... 4 2 Background ......................................................................................................................................... 5 2.1 Activated carbon .......................................................................................................................... 5 2.2 Mercury ........................................................................................................................................ 6 2.3 Mercury - a hazardous chemical ................................................................................................... 6 3 Materials and method ......................................................................................................................... 7 3.1 Experimental setup ...................................................................................................................... 7 3.2 Sampling ....................................................................................................................................... 7 3.3 Detection of mercury concentration ............................................................................................ 9 4 Results ............................................................................................................................................... 10 5 Analysis ............................................................................................................................................. 14 5.1 Limitations .................................................................................................................................. 14 5.2 Sources of error .......................................................................................................................... 14 5.3 Conclusions ................................................................................................................................ 15 6 Acknowledgement............................................................................................................................. 16 7 References ......................................................................................................................................... 17 Johanna Norup 7th Lingfeng Summer Research School 1 Introduction Mercury is one of the most hazardous chemicals that humans utilize. Mercury is emitted into the atmosphere mainly through coal combustion and waste incineration (Hall B. et al, 1991). Mercury is not degraded in any large extension in the atmosphere. Instead it accumulates in the ecosystem and will mainly affect the top consumers in the food web, for example humans (Scala F. et al, 2011). A lot of research has been made within this topic, and everyone can agree that activated carbon is a good adsorbent for mercury gas. The questions left are how it can be cheaper and more efficient to clean the gas from mercury so the emissions will reach zero. The whole ecosystem is affected by the emissions of mercury. Both plants and animals and humans will experience negative effects from the accumulation of mercury. It is higher dosage higher in the food web, which make mercury more hazardous for predators. In this project different kinds of activated carbons ability to adsorb the mercury ion Hg 2+ are studied in a bench scale testing system, which will be a simulation of a cleaning system in an incineration fume. 1.1 Aim The project’s aim is to compare different kinds of adsorbents to see how efficient they adsorb mercury gas in the ionic form, Hg2+. The different types of activated carbon that has been studied are BPL, DARCO FGD, molecular sieve (SBA-15) and modified activated carbon: PBL-CX, DARCO FGD-CX and molecular sieve-CX. Johanna Norup 7th Lingfeng Summer Research School 2 Background 2.1 Activated carbon Activated carbon is a type of carbon that has been pulverized so the surface area increases. Nonpolarized subjects that are in contact with this area will be adsorbed by the activated carbon. The activated carbon´s pore size distribution and surface area are important properties for the activated carbon to be a good adsorbent. A large surface area is good because it offers a lot of active sites for the mercury to attach to. The area can be as large as 1500m2/g carbon (Diamantopoulou I. et al, 2009). Activated carbon can be produced of any kind of carbon containing material. The carbon can be activated by treatment with water vapor, CO2, metal chlorides, phosphates or air (Elding L I., 2013). The activated carbon BPL, which is studied in this project, stands for bituminous coal-based activated carbon material that has been activated at high temperature in a steam atmosphere. It has a surface area of 995m2/g and a pore volume of 0.488cm3/g (Chen X., 2013). The next activated carbon in this study is DARCO FGD, a steam activated carbon manufactured specifically for the removal of heavy metals found in incinerator flue gas emission streams. FGD stands for Flue Gas Desulfurization. It has a surface area of 702m2/g and a pore volume of 0.535cm3/g (Chen X., 2013). The third studied adsorbent is molecular sieve (SBA-15), a material with similar distributed holes in it with uniform size. It is designed so molecules that are larger than the size of the holes will get stuck and not get through the material (Environmental Expert, 2013). Molecular sieve has a surface area of 711m2/g and a pore volume of 0.781cm3/g (Chen X., 2013). A summarized table of the physical properties for some of the adsorbents can be seen in table 1. Sorbent-CX means that the sorbent has been modified by benzoic acid. Figure 1. Benzoic acid There are many different kinds of functional groups on the surface of the activated carbon. There are four that can bind to mercury. They are carbonyl, carboxyl, lactones and phenol groups. The most efficient are carbonyl- and carboxyl groups. Half of the adsorbents in this study is modified with benzoic acid, which is a carboxyl acid (Chen X., 2013). Johanna Norup 7th Lingfeng Summer Research School Table 1. Physical properties of the adsorbents. SBA-15 is the name of the studied molecular sieve. Carbon BET specific surface area (m2/g) Pore volume (cm3/g) Aperture (nm) FGD BPL SBA-15 BPL-CX 702 995 711 739 0.535 0.488 0.781 0.380 N/A 1.96 5.29 2.03 2.2 Mercury Natural sources of mercury emissions to the atmosphere are volcanoes, forest fires and weathering from mercury containing sediment. A natural level of exposure in the atmosphere is about 1-2ng/m3. Most of the anthropogenic mercury comes from coal combustion and waste incineration. In waste incineration the mercury originates from batteries, medical products and lamps (Elding L I., 2013). The mercury concentration in the flue gas is around 0.5mg/Nm3. In many countries the emission limit is as low as 0.05 mg/ Nm3. This means that it is very important to reduce the emissions directly at the combustion (Diamantopoulou, I. et al, 2009). About 95% of the mercury in the atmosphere is elemental (Hall B. et al, 1991). But it can be oxidized with oxygen or ozone so Hg2+ ions form. These ions can be captured by particles or water droplets and follow them down to the ground when it rains. When the mercury ions come in contact with the ground they will bind to organic compounds and follow streams and rivers to lakes and seas (Elding L I., 2013). 2.3 Mercury - a hazardous chemical In high dosages mercury can damage the nervous system, since mercury can penetrate the bloodbrain barrier. It also affects the kidneys, the liver and the immune system (Sterner O., 2010). The Minamata disease tells us how dangerous methyl mercury is. Methyl mercury can be formed when anaerobic organism are in contact with inorganic mercury in aquatic systems (EPA, 2013). In the 1950’s a chemical factory near Minamata in Japan emitted mercury to the sea were it was accumulated as methyl mercury in fish and seafood. The people of Minamata, which lived on fishing, were affected by the methyl mercury that was formed (Harada, M. 1995). Methyl mercury affects the peripheral nervous system that can give the consumer motor and mental disorders. Since mercury can bind to the fat tissue in a human body, it maintains there and cannot be excreted with the urine (Swedish EPA, 2013). Johanna Norup 7th Lingfeng Summer Research School 3 Materials and method Mercury adsorption trials were performed in two steps. First the bench scale testing system was used to take samples. Then an AFS machine was used to detect the mercury concentration. 3.1 Experimental setup To compare the different activated carbons a bench scale testing system was prepared as seen in figure 2. The input was nitrogen gas (1 L/min) mixed with mercury gas, Hg2+ (500ng/min). The gas mixture was heated to 140°C to be more similar to incineration gas temperature. The adsorption reactor is in an oven where the adsorption of mercury occurs. The gas is lead into the adsorption reactor or directly to two bottles filled with KCl solution, where the mercury binds to the chloride ion so the concentration can be measured. Figure 2. Sketch over the bench scale testing system. 3.2 Sampling The procedure starts with weighting the activated carbon (30 mg mixed with 5 g quartz) and put it into the adsorption reactor. The bottles with KCl are replaced once every hour to collect five samples in total. But at first the initial concentration was measured. The gas was then lead directly to the KCl solutions without going through the adsorption reactor and samples were taken every 30 minutes. The following procedure was then done both for the initial concentration and the samples. Every sample with KCl-mercury solution was stabilized with 5% KMnO4 solution so the mercury ions will be stable. Johanna Norup 7th Lingfeng Summer Research School Figure 3. Bench scale testing system. Three new samples, 2ml each to put in 25ml test tubes, were taken from every hourly sample. This resulted in 15 new samples. All new sample was put out together with 0.5ml concentrated H2SO4, 0.25ml concentrated HNO3, one drop of 5% KMnO4 solution and 0.75ml 5% K2S2O8. Then the solution was diluted to 20ml. Three blank samples with KCl solution were also prepared in the same way to detect the natural mercury concentration or error in the machine. All the samples were then heated to 94°C for two hours. Figure 4. New samples. Johanna Norup 7th Lingfeng Summer Research School 3.3 Detection of mercury concentration After putting one drop of 10% hydroxylamine sulfate, ((NH2OH)2 ·H2SO4 ) into the samples to make them transparent, the samples are ready to be tested in the Atomic Fluorescence Spectrometry, AFS. Mercury sends out 253.7 nm wavelength of fluorescence under ultraviolet excitation. The intensity of fluorescence is proportional to the concentration of mercury. In the AFS, KBH 4 was used as reducing agent and 2% HNO3 was used as current-carrying. KBH4 react with the mercury sample and the ionic mercury turns into elemental mercury so the AFS can detect the concentration of mercury´s wavelength. Argon gas was preceded as a carrying gas for the mercury gas to the instrument for detection. The result appeared on a computer screen connected to the AFS where all the data was collected and saved, see figure 5. Figure 5. The Atomic Fluorescence Spectrometry To prepare the machine, standard solutions were tested first and a standard curve was performed. Then after the machine was cleaned with HNO3 and KBH4, the samples’ concentrations could be measured. To calculate the actual concentration, the measured mean value of the three samples was subtracted with the mean value of the three blank samples. The same procedure was then repeated for all different adsorbents in the study. Johanna Norup 7th Lingfeng Summer Research School 4 Results The different adsorbents have been compared with a blank test and then with each other. The y-axis in figure 6 and 7 shows the relation between the initial mercury concentration of the nitrogen mercury gas mixture, c0, and the mercury concentration that has been adsorbed of the current adsorbent, cx. On the x-axis the adsorption time is represented. Figure 6 shows all the studied adsorbents’ capacity to adsorb mercury over five hours. The lowest mercury concentration in the output gas is found in FGD-CX. The average ratio for FGD-CX is 96%, which means that 96% of the mercury is adsorbed when the gas has passed the FGD-CX adsorbent. The worst adsorbent in the study is molecular sieve. It has an average ratio of 34%, see table 2. 1,200 1,000 BPL Cx/Co (%) 0,800 BPL-CX FGD 0,600 FGD-CX 0,400 molecular sieve 0,200 molecular sieve-CX initial concentration 0,000 1 2 3 4 5 Adsorption time (h) Figure 6. Ratio of adsorbed mercury over time with studied adsorbents. Table 2. The average ratio of adsorbed mercury for the different activated carbons. Type Initial concentration BPL BPL-CX FGD FGD-CX Molecular sieve Molecular sieveCX Average ratio Cx/Co 0,00 0,67 0,83 0,87 0,96 0,34 0,53 Johanna Norup 7th Lingfeng Summer Research School To get the results neater, extended values can be ignored and the graph’s lines will be straighter, see figure 7. The conclusion will be the same but it is easier to see the difference between the adsorbents. These extended values may be there due to errors in the procedure, see limitations and sources of error. 1,200 Cx/Co (%) 1,000 0,800 BPL 0,600 BPL-CX FGD 0,400 FGD-CX 0,200 molecular sieve 0,000 molecular sieve-CX 1 2 3 4 5 Adsorption time (h) Figure 7. Ratio of adsorbed mercury over time with studied adsorbents with ignored extended values. The result with modified values gives the same conclusion; the best adsorbent is still FGD-CX that adsorbs 98% of the mercury with modifications according to table 3. Molecular sieve is also still the worst adsorbent in the study which only adsorbs 19% with the modifications. Table 3. The average ratio of adsorbed mercury for the different adsorbents with modified values. Type Average ratio Cx/Co Initial concentration 0,00 BPL 0,67 BPL-CX 0,83 FGD 0,90 FGD-CX 0,98 Molecular sieve 0,19 Molecular sieve-CX 0,60 Johanna Norup 7th Lingfeng Summer Research School Cx/Co (%) To see the difference in the ability to adsorb mercury with or without modification with benzoic acid, figure 8, 9 and 10 shows the same adsorbent with and without CX. This is done to see if it the modification is an efficient way to improve the adsorption for the activated carbon. BPL comparison 1,000 0,900 0,800 0,700 0,600 0,500 0,400 0,300 0,200 0,100 0,000 BPL BPL-CX 1 2 3 4 Adsobent time (h) 5 Figure 8. BPL and BPL-CX’s ability to adsorb mercury. FGD comparison 1,000 Cx/Co (%) 0,950 0,900 FGD 0,850 FGD-CX 0,800 0,750 1 2 Adsobent time3(h) Figure 9. FGD and FGD-CX’s ability to adsorb mercury. 4 Johanna Norup 7th Lingfeng Summer Research School Molecular sieve comparison 0,800 0,700 Cx/Co (%) 0,600 0,500 Molecular sieve 0,400 0,300 molecular sieve-CX 0,200 0,100 0,000 1 Adsorbent 2 time (h) 3 Figure 10. Molecular sieve and molecular sieve-CX’s ability to adsorb mercury. Johanna Norup 7th Lingfeng Summer Research School 5 Analysis As seen in table 1, BPL has larger surface area than FGD. That would indicate a good adsorbent. But in this study this is not the case. This can be because of the limitations in this project. No conclusion can be made which one of the studied that are the best. Other studies have shown that BPL is better than FGD, while this study shows that FGD is better (Chen, X. 2013). Therefore more experiments have to be done with the same procedure for a longer time to get more accurate results. A reason for the molecular sieve to be the worst adsorbent is that molecular sieve has relatively poor thermal and chemical stability at high temperatures (Lin Y.S. et al, 1994). In a comparison with the pore volume, there is also hard to find any correlations. The molecular sieve has the largest pore volume while BPL has the smallest. Since the best adsorbent in this study was FGD, this result indicates that the pore volume does not have any impact on the adsorption ability. The result makes it hard to find any correlations between the physical property of the adsorbent and the ability to adsorb mercury. The figures 8, 9 and 10 shows that the adsorption is improved if the adsorbents are modified with benzoic acid. This was expected since more effective active sites for mercury to attach to occurs with the modification. The benzoic acid is a carboxyl acid and it increases the sites that are easiest for mercury to attach to on the adsorbent. This means that the less efficient active sites; phenol and lactones groups are less on the activated carbon’s surface area. Therefore the adsorbents modified with benzoic acid could be better adsorbents, which have been confirmed in the study. 5.1 Limitations This project lasted from the 24th of June until the 19th of July. This means that there was a limited time to take all the samples and detect the concentrations. The project was delayed due to an electrical outage and the fume carport stopped working. Every adsorbent was studied for five hours, even if some could adsorb mercury for a longer time. The adsorbents’ time depending ability to adsorb mercury has not been considered in this study. It also has to be taken into consideration that the gas mixture is a prototype of a flue gas fume. The gas mixture will not look the same in reality and other trace gases in flue gas can affect the adsorbents’ ability to adsorb mercury. The bench scale testing system is only a simulation of a cleaning system in reality, which also affect the result. 5.2 Sources of error Almost everything has worked out as planned, but there are always some human factors that have to be taken into consideration. One example is that the KCl solutions once were installed before the gas was lead into the adsorption reactor, and therefore the concentration would be rather misleading. The flasks with KCl solution was once put in the wrong way, so the gas came in where it should go out, so there was some solution that was dropped. Johanna Norup 7th Lingfeng Summer Research School 5.3 Conclusions Due to this study the DARCO FGD-CX is a better adsorbent than the others studied in this project. The worst adsorbent in this study is molecular sieve. In comparison with all studied adsorbents, the ones modified with benzoic acid are better than them without. More studies have to be done to make a general conclusion about which adsorbents that has the best ability to adsorb mercury and what it depends on. Johanna Norup 7th Lingfeng Summer Research School 6 Acknowledgement The author wants to thank Professor Jinjing Luo for supervising this project. A special thank to assistant Xiaobao Chen for many valuable comments and explanations about the experiments. Many thanks to Syuan Lei Jhao for a good cooperation in this project. A last thank to the organizers at Lund University and Xiamen University for making the Lingfeng Summer Research School possible with many memorable moments and cultural experiences. Johanna Norup 7th Lingfeng Summer Research School 7 References Cai Y., Atomic Fluorescence in Environmental Analysis, 2009, Chen X., oral communication June-July 2013 Diamantopoulou, I., Skodras G., Sakellaropoulos G.P, Sorption of mercury by activated carbon in the presence of flue gas components, 2009 Elding L I, The National Encyclopedia, http://www.ne.se/lang/kvicksilver, read 2013-07-03 EPA, http://water.epa.gov/scitech/swguidance/standards/criteria/aqlife/methylmercury/factsheet.cfm, read 2013-07-08 Hall B. Schager P., Lindqvist O., Chemical reactions of mercury in combustion flue gases, 1991 Harada, M., Minamata Disease and the Mercury Pollution of the Globe, 1995 Lin Y.S., Deng S., Adsorption and desorption of sulfur dioxide on novel adsorbent for flue gas desulfurization, 1994 Ohlsson R., The National Encyclopedia, http://www.ne.se/aktivt-kol , read 2013-07-05 Scala F., Chirone R., Lancia A., Elemental mercury vapor capture by powdered activated carbon in a fluidized bed reactor, 2011 Sterner O., Chemistry, Health & Environment, 2010 Swedish EPA, http://www2.naturskyddsforeningen.se/Regional%20Office%20Files/Kretsar%20och%20l%C3%A4nsf %C3%B6rbund/Sk%C3%A5ne/Lund/F%C3%B6reningsdokument%20LNF/ESS/Fakta%20om%20kvicksil ver.pdf, read 2013-07-04 The Environmental Expert, http://www.environmental-expert.com/products/bpl-4x10-granularactivated-carbon-60487, read 2013-07-12