©) resem PIT: $0043-1354(96)00332-6 Wat Res. Nol 31. No 4 9p. 199-746 1997 °C 1907 Ete Scene Lid Allright sere Patel on Creat Brian ‘op issast $1700 000 CHARACTERISATION AND APPLICATIONS OF ACTIVATED CARBON PRODUCED FROM MORINGA OLEIFERA SEED HUSKS BY SINGLE-STEP STEAM PYROLYSIS A. MICHAEL WARHURST'®, GORDON L. MeCONNACHIE! and SIMON J. T. POLLARD" & ‘Department of Civil and Enviconmental Enginering. University of Edinburgh, King’s Buildings, West Mains Read, Edinburgh EH9 3), U.K, and 'Scotssh Environment Protection Agency, Environmental Strategy Directorate, Erskine Court, The Castle Business Park. Surling FRO ATR, UR. (Received February 1996; accepted in revised form November 1996) Abstract—The seed husks of the multipurpose tree Moringa oliera ate potently & waste procuc that ‘may be available in large quantities, and previous work has demonsrated that a microporous activated Carbon caa be prod ed fram them by arbonisation under nitrogen followed by activation in scar. This research examines the eficacy of simpler and cheaper activation proces, singlestep steam pyrolysis activation, with a view to promoting the production of low-cost activated carbon sn the developing Wor Husks were heated ina seam atmosphere to 780°C for 30 or 120:min, or 800 C for 30min. then the resulting carbons were tested to determine their iodine numbers and adsorption isotherms for phenol ‘Fnitrophenol and methylene blue. Phenol and 4nitrophenol were adsorbed rapidly by al ree carbons, ‘with 80-90% ww adsorbed in the Rrst 30 min, whereas methylene blue adsorption was sloner. Pyrolysis 1 500‘C for 30 min produced a carbon (yield 12.2% w') with an iodine number of 703 me g' phenol specie surface ares (SSA) of 629 mg a 4-ntrophenol SSA of 664 mg ' and a methylene blue SSA. ‘of 211 mg". The carbon produced at 730°C for 120:min yield 1.9% ww) bad similar properties, but the one produced at 750°C for 30 min (eld 16.6% ww) hada less developed porosity. The adsorbance characteristics of the two best carbons were superior to those produced previously by the conventional two-stage carbonisation activation, and were competitive with commercial carbons. These results ‘demonstrate tha steam pyrolysis activation of 47. eleera husks could provide a lowcost local Source ‘of high quality activated carbon inthe developing world.) 1997 Elevier Selence Lt, All ights reserved ‘Key words— activated carbon. appropriate technology. Moringa oleifra. steam pyrolysis, water eatment NOMENCLATURE. b= Langmuir adsorption coofient BET = Brunauer, Emmett and Teller method fr surface area calculation ‘equilibrium concentration of solute (mol I~) = amount of solute adsorbed on the adsorbent (mol g°9) [Langmuir monolayer coverage constant (mol Freundlich constant (mol p") Freundlich constant specific surface area (mg) Giles and Nakhwa adsorption monolayer ca- pacity from experimental isotherm (mol g INTRODUCTION In recent years there has been a focus on the use of appropriate, low-cost technology for the treatment of drinking water in the developing world (Gupta and ‘Chaudhuri, 1992; Pollard er al., 1992; Cairneross and Feachem, 1993). One major cost associated with “Author to whom all correspondence should be addressed [Fax: (01796) 446 885; Tel (01786) 457 700) 189 water treatment in these countries is the importation of water treatment chemicals and other materials. Exchange rates are frequently unfavourable, and foreign currency is scarce, making the importation of chemicals and materials a significant burden on the water treatment budget. Extreme variability in exchange rates often affects a water authority's budgeting capability, and can cause unpredictable losses and price increases, One solution to these problems is the indigenous production of treatment materials, either by local production of chemicals, or the use of alternative, locally available, treatment materials. One example of the later is use of a coagulant which is contained in the seeds of the tree Moringa oleifera (common names include Drumstick and Horseradish tree), a ‘multi-purpose, drought-tolerant tree which grows in ‘many areas, including Northern India, Aftica and the Caribbean (Morton, 1991). The seeds of this tree contain a coagulant protein (Gassenschmidt e al. 1995; Ndubigengesere etal, 1995) which can replace conventional coagulants such as aluminium salts, in both domestic (Jahn, 1988) and larger scale water70 treatment (MeConnachie er al., 1994). In addition, the seed contains a high quality oil, which has many uses (Morton, 1991). A potential waste product from the production of the coagulant and oil is the seed husk. We have previously shown that this husk can be converted into 2 high quality, microporous activated carbon (Pollard et a., 1995), that could potentially be used to replace imported carbons in applications such as removal of taste and odour compounds from water supplies. Our initial work on the production of carbon from M. oleifera husk used a two stage ccarbonisation-activation process, involving heating the husk to 485°C for 30min under nitrogen, followed by exposure to a steam/nitrogen mixture at 850°C for Smin (Pollard et al., 1995). This process required nitrogen, which might have to be imported into a developing country, and two heating stages, involving significant energy input. Reappraisal of the literature, however, showed that other workers GGergova et al, 1994) bad successfully used a single-step steam pyrolysis activation to. make carbons from a variety of materials. Preliminary tests on carbons produced from husks at a variety of temperatures demonstrated that steam pyrolysis activation produced high quality carbon (Warhurst eta, 1996). More detailed examination of carbons using nitrogen adsorption allowed selection ‘of optimum carbons based on a balance between surface area development, microporosity and product yield (Warhurst er al., 1996, 1997). Three carbons were chosen for further characterisation: $750/30, Which had been heated to 750°C then held at that temperature for 30 min; $750/120, also heated to 750°C, but held for 120 min; and $800,/30, which was hheated to 800°C and held for 30min. The yields and BET N; surface areas were as follows: $750/30, 16.6% w/w, 588 m? g-'; $750/120, 11.9% ws, 730 my? 2° $800/30, 12.2% w/w, 710-m? g!(Warhurst eral, 1997). The carbon produced using the conventional two-stage process had a similar surface area of 734 m? sg ',and.a yield of 9.6% w/w, though this carbon was acid-rinsed to remove water-soluble ash (Pollard et al, 1995). For comparison, a commercial peat-based activated carbon, Norit SA4, was found to have a surface area of 705 m? g-' (Pollard er al., 1995), ‘The aim of this work was to characterise three steam-pyrolysed carbons using molecular probes in the solution phase, so as to facilitate comparison with commercial activated carbons used in. developing ‘countries, and provide a sound basis for carbon production in the developing worl. EXPERIMENTAL DETAILS. Materials M-oleifera husks were obtained from Malawi, where they had been produced as 2 waste product from the production ‘of M. olefera sed for use in pilot and Tulsa testing of its coagulant properis. Because of the nature of the A.M, Warhurst et al. seedfhusk separation process, this husk material also Contained powdered sed, which was not ofan even particle ze, varying between a ponder and particles several rilimetres in diameter, Carbonisation-actvation of husks Carbonisation of M. olefera husks was performed in & Carbolte 12/65 tube furnace containing a purpose-made quartz tube (MultcLab, Newcasde, UC), connccted to Quick connection atone end and a ball and cup joint at the othe, allowing flow-through of « controlled gaseous atmosphere. Charges of 30 of M. oleifera husks were placed in a eustomised qusrtz combustion boat, loaded into the quartz tube and positioned in the ceatee of the furnace The temperature of the sample was monitored by @ thermocouple probe (type K probe, connected toa PL 8013 thermometer) positioned just above the sample. Steam, ienerated by a flask of water on a heating mantle, entered the quartz tube at a eate of approximately 2m (iguid water) min. A condenser was connected to the ball and up joint at the other end ofthe tube, andthe condensate vas collected in a beaker “The sample was heated at a fixed rate of approximately 20°C min” tothe appropriate temperature, and held 3 this temperature (10°C) for the time speciied, After the furnace had cooled to between 200 and 300°C, the shonixed husks were removed, allowed to coal in air For afew minutes and then weighed to determine the yield. The fatbon was then moved to a desiccator for storage ‘Carbon, atleast duplicated for each set of temperature and lime, were later mixed, homogenised and. sieved to -<250 um, dried in an oven at 110°C overnight, then stored ina desiccator over silica ge Adsorption kineries ‘The adsorption kinetics of the model adsorbates penal, 4ritrophenol and methylene blue (BDH, U-K.) on selected carbons were determined in advance of isotherm measure ‘ents. The adsorption times used were 48, 31, 24, 8, 4, 2, ‘and 0.5 h each for phenol and ¢-nitropheno, and 100,72, 48,31, 24, 8,42, Vand 0.5h for methylene blue. Carbon samples of 0.5 g were added 10 100ral of 10 mM adsorbate Solution in 280m? Erlenmeyer Masks, which were sealed with Suba-seas, then shaken at 90'sirokes min-" in a temperature controled shaker bath at 25°C. After the equied time, the samples were filtered and analysed as, described below for the adsorption isotherms, Adsorption isotherms Seven-point,single-adsorbate adsorption isotherms were constructed for phenol, nitrophenol and methylene blue, at 30, 38, 20, 15, 10, 7 and 3mM of phenol and -nirophenol, and 30, 20, 15, 10, 7, 3 and ImM of methylene biue. These. “molecular probes" represent ausorbates of distinct molecular cros-section, so providing information on the characteristic porosities of M. olifera carbons. Carbon samples of 0.5 g were added 10100 ml of adsorbate solution in 280 om? Erlenmeyer asks, sealed ‘with Suba-seals. then shaken at 90 strokes min" ina temperature controlled shaker bath for 24h at 25°C Following uptake, suspensions were vacuum-ftered ‘through Whatman No. $42 filter paper, diluted as necessary, then acidified with 1 drop of 1 MHCL per 10 ml sample Residual adsorbate concentrations were determined using a ‘ux. spectrophotometer (Shimadzu UV-160A), 270:nm for ‘Phenol, 329 nm for -ntrophenol and 665 am for methylene blue, ‘The isotherms were then analysed using the Langmuir and Freundlich models. The Langmuir model assumes that 2 monolayer ofthe adsorbate adsorbs to the earbon surface,Moringa oefora activated carbon 761 leading tothe equation: = OG T+ BG o) where g. isthe amount of solute adsorbed on the adsorbent {in mol g-), Cris the equilibrium concentration of solute (in ‘mol Qs the Langmuir monolayer coverage constant {in (mol g”!) and b is the Langmuis adsorption coeficient ‘The constants ate derived from 2 plot of 1/C. against 1g ‘The Freundlich model is purely empirical, and. ts expresied a aa Kee a ‘where the Freundlich constants Ay and 1 are derived from 4 plot of log C against log @ ‘Adsorbate-specihe apparent surface areas were derived by comversion of 0°, using the ported cross-sectional areas of the model adsorbates. The molecular crosesectionalazess used for the three adsorbates were: 5.22 = 10-m" for phenol (Pollard ea, 1995), 5.25 = 10°" ey’ for nitro Phenol (Pollard ef al. 1991) and 39% 10°" m! for methylene blue, assuming end-on adsorption (Barton, 1987). {A further estimate of the adsorbate-speciic surface area was obtained by the method of Gils and Nakhwa (1962), in which the “Knee point” on the adsorption isotherm is ‘extrapolated to the 4. axis and taken as the point of ‘monolayer coverage, This extrapolated point, tx ean then be converted to un apparent surface area as done with Q” Iodine manber determination Iodine numbers were determined using a method based on ASTM D4607-86 (ASTM, 1991b). A thre-point isotherm ‘was used, with an equilibration time of 30s, and the residual Todine concentration was determined. by Wration” with Sodium thiosulphate (BDH. UK.) with starch as an indicator. The iodine aumber was operationally defined ax the quantity of iodine (BDH. UK) adsorbed (in mg g carbon) at a residual iodine concentration of 002 N RESULTS Adsorption kineties Adsorption kinetics were determined for phenol on carbons $750/120 and $750/30, for 4-nitrophenol on carbon $800/30 and for methylene blue on carbon '$800/30. The results are shown in Fig. 1. Adsorption ff phenol and 4-nitrophenol on the carbons was rapid, with 90% wiw of the 4-nitrophenol adsorbed bby $800/30 after 48 h being already adsorbed after 05h, and 80% wiw of the phenol adsorbed by erate ret te) rm Fig. |. Adsorption kines. Fig. 2. Phenol adsorption isotherms $750/120 after 48h adsorbed in the first OS. A comparison of the phenol adsorption kinetics of ‘carbons $750)30 and 750/120 demonstrated that the 750/120 carbon adsorbed more phenol, and adsorbed it slightly faster, with 88% w/w of the final amount adsorbed in Ih as against 80% w/w on the 750,30 carbon The methylene blue uptake was. substantially slower, with only about 30% wiw of the final total adsorbed in the first OSh, and about 80% w/w adsorbed after 24. The variability of the methylene blue resulis in this experiment (and in its adsorption isotherm) makes it more difficult to identify. the precise time at which its adsorption reaches a plateau. Ic was decided that a 24h equilibration time would be used for the isotherm experiments, which would approximate equilibrium conditions for the phenol and 4nitrophenol adsorbates, though the methylene blue would not be fully equilibrated, Adsorption isotherms Phenol, 4nitrophenol and methylene blue adsorp- tion isotherms forthe three carbons are shown in Figs 2, 3 and 4, respectively. For all the adsorbates, the 5800/30 carbon showed the highest adsorptive capacity, with $750/120 slightly lower and $750/30 the lowest. The difference between the carbons was particularly marked in the adsorbance of methylene blue, where the $750/30 carbon adsorbed substan- lilly less than the others, The isotherms were analysed using the Langmuir and Freundlich models, the results of which are shown in Table 1, along with the regression Fig. 3. 4-Nitrophenol adsorption isotherms