EM FBT-97/1



The formation of compounds at gasification conditions, with and without internal catalysts

E Bjorkman C Larsson


OF THS oocvmn m mam® 29-04

NUTEK, 1 17 86 Stockholm, Tel 08-744 95 00 (Fran 1 november 1992 nytt nummer: 08-681 91 00) RaDDorterna kan bestallas fran Studsviksbiblioteket, 61 1 82 Nykopinq. Tel 0155-22 10 90, 22 10 00, Fax 0155-26 30 44 DISCLAIMER

Portions of this document may be illegible in electronic image products. Images are produced from the best available original document. 1997-05-20

Narings- och teknikutvecklingsverket

Titel: The formation of nitrogen compounds at gasification conditions, with and without internal catalysts

Forfattare: Bjdrkman, E. and Larsson, C. TPS Termiska Processer AB, 611 82 Nykoping


Rapportnummer: FBT-97/1

Projektledare: Eva Bjorkman

Projektnummer: A5551-1 (3)

Projekthandlaggare Rolf Ingman pa NUTEK:

Postadress Besoksadress Telefon Fax Internet e-post 117 86 Stockholm Liljeholmsvagen 32 08-681 91 00 08-19 68 26 http://www.nutek.se [email protected] TPS Termiska Processer AB TPS-96/48 Bjdrkman, E. Larsson, C. 1996-10-28



The objective of the project is to substantially decrease the emissions of nitrogen oxides from biomass IGCC by minimising the conversion of fuel nitrogen to and in the gasification process. To reach the objective, the influence from different parameters on the formation of ammonia and hydrogen cyanide from fuel nitrogen has been studied. The parameters studied are the bed material, different functional forms for the nitrogen and the concentration. The results so far indicate that it is possible by simple methods, to influence the formation of NH3 and HCN by for example modifying the bed material, but that the functional form of the nitrogen is of minor importance.

Approved by: TPS Termiska Processer AB TPS-96/48 Bjdrkman, E. Larsson.C. 1996-10-28


The dominating nitrogen compounds that can be found in the fuel gases produced from gasification processes, are ammonia and hydrogen cyanide. These compounds will be converted to nitrogen oxides in the gas turbine combustor in the IGCC process. The NOX emissions after the gas turbine combustor can be reduced by conventional flue gas cleaning, i.e. SCR (Selective Catalytic Reduction), but it is advantageous if this can be avoided. The decomposition of ammonia already in the fuel gas is desirable, but are difficult as most of the active materials are deactivated by the gas. One example are dolomite which decompose ammonia in dry gas but not with steam present (1). It would be preferable if the fuel nitrogen could be converted to N2 directly in the gasifier.

Depending on the fuel, the relation between ammonia and hydrogen cyanide differs. Small amounts of hydrogen cyanide produced during low temperature gasification of biomass (2) (compared to ). The conversion of fuel nitrogen to NH3 has been found to be 17-24% for peat, 10-20 % for bark and 15-17 % for coal, from with slow heating rate. During the same conditions the conversion of fuel nitrogen to HCN was 5- 11% for coal and considerably less from peat and bark (3).

Differences in the conversion of fuel nitrogen to HCN and NH3 have generally been explained in terms of nitrogen functionality (4,5). HCN is expected to be the principal nitrogen-containing pyrolysis product when the tuel-N is present in pyridinic or pyrrolic structures (aromatic structures), and NH3 when the fuel-N is present in amino groups. It has also been suggested that as the fuel rank decreases the number of amino groups in the fuel increases and therefore, the conversion of fuel n-^ogen to NH* gams importance. Since the thermal stability of amino groups is lower than that of aromatic nitrogen compounds, NH3 should be released before i iCN, nevertheless experiments have shown that HCN is formed before NH3 (6). Fuel oxygen seems to play an important role in the fuel nitrogen chemistry, especially phenolic OH groups in the structure increases the conversion of nitrogen to NH3(7,8).

To catalyse the formation of N2 from fuel nitrogen by internal catalysis is highly desirable. Internal catalysis of solid phase combustion/gasification reactions, mostly on coal, has been studied previously. Some of the more recent results are presented in Table 1.

Table 1 Summary of recent literature in the catalytic gasification/combustion field. catalyst fuel process effect mineral matter coal combustion no effect on NO formation (9) FeOOH coal gasification increased solids reactivity (10,11) Ca, K coal gasification increased solids reactivity (12) Ca, Sr, Ba carbon gasification increased solids reactivity (13) FeSCU wood gasification increased solids reactivity (14) Ca synth.char oxidation increased solids reactivity (15) CaO coal oxidation increased solids reactivity (16) mineral matter coal combustion no effect (17) Ni-aluminate biomass pyrolysis fast reaction and catalytic reforming (18) Na coal gasification increased solids reactivity (19) Fe coal pyrolysis reduced NH3 production (20)

As can be seen from the table, there is only one report on the effect of internal catalysis on the formation of nitrogen products. In that one, a FeCI3 was precipitated on to brown coal. The authors achieved a 50-60% conversion of the fuel nitrogen to N2 when the coal was pyrolysed in helium. Without the but else with the same experimental conditions, the conversion to N2 is less than 3%. The explanation given in TPS Termiska Processer AB TPS-96/48 Bjorkman, E. Larsson, C. 1996-10-28

the publication is that at 900°C the iron is completely reduced and present as ultra fine iron particles (20-50 nm), which are responsible for the efficient conversion to nitrogen.

The objective of this proposed project is to substantially decrease emissions of nitrogen oxides from biomass IGCC by maximising the conversion of fuel-nitrogen to nitrogen gas in the gasification process. To reach the objective, the influence from different parameters on the formation of ammonia and hydrogen cyanide from fuel nitrogen have been studied. The studied parameters are the bed material, different functional forms for nitrogen, and the oxygen concentration.


The experiments have been performed with a fixed bed quarts reactor placed in a small electric furnace, see Figure 1. To achieve comparable residence times, the same bed height was used in the experiments. The fuel used was a mixture of and the desired nitrogen source (aniline, , propyl ) on 1:1 volume base. The model fuel was injected into the bed in the reactor with a GC capillary. The fuel feeding rate was 3.5 ml/h. A mixture of N2 and 2% O2 was used as a carrier gas (flow rate 11/min, which gives a air/fuel ratio of approximately 0.2). The product gas was analysed on FTIR, the studied components were NH3, HCN, NO, CO, CO2, CH4, C2H4. The investigated temperature interval was between 750-900°C. The experimental set-up made it possible to study the influence of different variables such as nitrogen source, oxygen concentration, temperature, bed material and air/fuel ratio.

bed material • quarts wool »^^ \ .—quarts filter !

... c r , . . , * —to FTIR capillary for fuel injection ' \ electric furnace

flow direction

Figure 1 The experimental set-up

The bed material used was dolomite, quarts sand, AI2O5, Ni-catalyst, olivine sand, CaO and CaO doped with 10 mole-% K, Na or Fe. When doping was required, the doping material was added as a solution and then dried according to following scheme;

Step 1: 10°C/min to 150°C, total time 120 minutes. Step 2: 5°C/min to 300°C, total time 240 minutes. Step 3: 2°C/min to 800°C, total time 240 minutes Step 4: 25°C/min to 25°C, total time 60 minutes

then the catalyst was crushed and sieved to fraction 0.7 to 1.4 mm. The CaO based bed materials were used in smaller amounts (1 gr. instead of 9 gr.), but they were mixed with quarts sand to achieve comparable the bed heights. TPS Termiska Processer AB TPS-96/48 Bjdrkman, E. Larsson, C. 1996-10-28


To get a more understandable report, it has been divided into four different parts, first is the study of the effect of doping a basic bed material, secondly the influence of the nature of the nitrogen, third effect of the oxygen concentration and finally a mapping study of different catalyst. Since there are four different small subchapters we have put together the presentation of the results with the discussion.

3.1 Doping the bed material

It is known from previous studies that oxides can have catalytic effects. As an example, MgO is a basic oxide and it can abstract a hydrogen from at temperatures around 800°C, which produces a hydrogen radical and a (21). In the present work CaO and doped CaO was used as the bed material. With the aim to increase the basicity of the bed material, and we used potassium (K) and (Na) as the doping agent. Furthermore, for comparisons reasons, we also doped CaO with iron (Fe). The results showed that the amount of formed NH3, HCN and NO were not sensitive to the basicity of the bed material but iron had a clear catalytic effect, see Figure 2.

CaO CaO + Na

ppm 1000

Figure 2 The amount of gaseous nitrogen compounds in fuel gas produced from aniline. Bed material was doped CaO TPS Termiska Processer AB TPS-96/48 Bjorkman, E. Larsson, C. 1996-10-28

3.2 The nitrogen source

In this study we have used three different nitrogen containing , namely pyridine, aniline and propyl amine, see Figure 3.

aniline pyridine propyl amine

Figure 3 The chemical structure of aniline, pyridine and propyl amine.

The three model compounds were chosen since they have different carbon-nitrogen bonds. In pyridine the carbon-nitrogen bond is aromatic, in aniline is an aliphatic bond attached to aromatic ring, and in propyl amine aliphatic. The difference between the bonds and the impact of can be explained with the following:

In organic chemistry the thermal decomposition reactions are called homolytic cleavage reactions. In these reactions, the chemical bond is broken in the way that the two electrons in the bond are shared by the two produced molecular fragments, i e two radicals are produced. The ease at which a bond can participate in homolytic cleavage reaction can be correlated to its' bond energy, i e the weaker the bond the more easily it brakes. A sLudy of the literature (22) and comparison with the carb.,, analogues gives the bond eneiyies in Table 2. A comparison of the energies given in the table gives that carbon-nitrogen bonds are more easily broken than carbon-carbon bonds. Furthermore, when one of the radical produced is a phenyl radical, as for aniline and toluene, the bond energy is lowered considerably, while the carbon-hydrogen bond is strong.

Table 2 Bond energies in KJ/mole C-C bonds bond C-N bonds bond C-H bonds bond energy energy energy

520* 400* 427

CH 264 =200 3 -CH 335 -NH 260 CH 427 * estimated from the bond length "estimated by comparison with the bond energies from the other entries in the table

Based on the discussion above, we assumed that aniline and propyl amine should produce mainly ammonia and pyridine should produce hydrogen cyanide, but the result did not conform with our assumption. Figure 4 shows that aniline, over a bed of quarts sand, produced more HCN than NH3. Furthermore, pyridine seemed to be more reactive than aniline since the jump in the ammonia curve that can be seen in Figure 4 occurs at lower temperatures for pyridine than for aniline. TPS Termiska Processer AB TPS-96/48 Bjorkman, E. Larsson, C. 1996-10-28

Aniline Pyridne



Figure 1 ': he formation of NH,, HCN ai id NO over quarts sand.

Aniline Pyridne

1103 1000- 900 f 800- 700- V T-90CTC 600- \ 500- \ i ppm 400- 300 200- W0- ./ 0- • •••=••*»»» M=rl I'l I A A-A A A-i

750 800 Propyl amine

Figure 5 The formation of NH3, HCN and NO over iron doped CaO TPS Termiska Processer AB TPS-96/48 Bjorkman, E. Larsson, C. 1996-10-28

The same investigation, but with iron doped calcium oxide as the bed material shows that pyridine and aniline behave in a similar way, while propyl amine differed, see Figure 5.

In Figure 6 is the results from the different nitrogen sources summarised. The figure shows that the main nitrogen gaseous compound produced was ammonia, independently of the origin of the fuel nitrogen. When comparing the heights of the bars in each group of two, it can be noticed that the trend for pyridine (grey) and aniline (white) surprisingly similar. This indicates that the reactivity of aniline and pyridine are similar although the different bond energies, see Table 2. Another explanation could be that secondary reactions on the bed material is of importance, but it is contradicted by the fact that for the propyl amine (black bars) the trends differed.

qietts GO+Fe 3000- NO 2500





0 GO qierts GO+Fe

Figure 6 The concentrations of NH3, HCN and NO on the product gas for different nitrogen sources. Bar to left represents the concentration measured at 750°C, bar to right the concentrations at 900°C.

Summarising, the chemical nature of the carbon-nitrogen bond seems to have little or negligible effect on the formation of ammonia and hydrogen cyanide in gasification reactions.

3.3 The oxygen concentration

Since the formation of HCN and NH3 did not seems to be governed by the chemical nature of the fuel nitrogen, some other factor must influence the formation of gaseous nitrogen compounds. Our preliminary results indicates that one factor can be the oxygen concentration, see Table 3. The table shows that the ratio HCN/NH3 was dependent on the oxygen concentration. The more reduced the , the more HCN was produced. The bed material was also of importance, but same trend could be TPS Termiska Processer AB TPS-96/48 Bjorkman, E. Larsson, C. 1996-10-28

detected (lower oxygen concentration gave more HCN) for both quarts sand and iron doped calcium oxide.

Table 3 The ratio of HCN/NH, at different O, concentrations and bed material, aniline was used as the nitrogen source. quarts sand CaO + Fe

O2(%) HCN/NH3 02(%) HCN/NHs 10 0.62 10 0 5 0.58 5 0.02 2 1.45 2 0.09 1 1.32 1 0.30

3.4 Different catalysts

Finally, we will present the results from a mapping investigation of different bed materials, the results are presented in Figure 7.

An explanation of the figure is maybe needed. In the figure is the concentration of ammonia, hydrogen cyanide and nitrogen oxide in the gas plotted versus the total concentration of gaseous carbon species, which can be regarded as a measurement of the carbon conversion. Dots to the lower right in the figures represent a catalyst which have a high carbon conversion but produces low amount of the studied nitrogen compound. The figures may be difficult to interpret, but they should only be used to identify outlayers.

The nickel catalyst is a most effective v.dtalyst followed hy the iron doped calcium ovide. accoridng to Figure 7. They produced low amount of ammonia, hydrogon cyanide and nitrogen oxide, but still had a high carbon conversion. Other resuits are that quarts sand and olivine sand produced more hydrogen cyanide than the other bed materials. Dolomite seemed to produce a lot of ammonia but this can be due to the fact that steam inhibits the ammonia decomposition on dolomite (1)

Since TPS has a lot of experience in tar cracking on dolomite (23, 24), the dolomite results may be explained further. The apparent low carbon conversion for dolomite in Figure 7 is most probably due to the fact that dolomite do not crack the lower aromatics, such as , as effec'J-'tj as the nickel catalyst. The apparent high ammonia formation, is probably due the high cracking ability for dolomite on the parent nitrogen source (aniline) but a low decomposition property for ammonia in the wet gas (1)-


To conclude, these results can be summarised with: • the chemical nature of the fuel bound nitrogen is probably of minor importance for the formation of either ammonia or hydrogen cyanide • the Ni catalyst and iron doped CaO reduced the amount of ammonia and hydrogen cyanide produced during pyrolysis/gasification • quarts and olivine sand enhanced the HCN emission • aniline and pyridine showed, unexpectedly, similar behaviour and produced similar amounts of NH3 and HCN • propyl amine was more reactive then aniline and pyridine • low oxygen concentrations increased the ratio of HCN/NH3 TPS Termiska Processer AB TPS-96/48 Bjorkman, E. Larsson, C. 1996-10-28

CaO+Fe 3000 - CaO + K CaO+Na 2500 - V CaO m quarts sand 2000 - E olivine sand a. dolom ite 1500 - AI2O5 N i catalyst 1000 -

500 -

10000 15000 20000 25000 30000 35000 40000 45000 50000

CO2+CO + CH4+2*C2H4(ppm)

900 -


700 -

a. a. soo - 400 - o 300 -

200 -

100 -

10000 15000 20000 25000 30000 35000 40000 45000 50000 CO2+CO + CH4+2*C2H4(ppm)

200 - 180 - 160 - 140 - 120 - 100 - 80 - 60 - 40 - 20 - 0 - 20000 25000 30000 35000 40000 45000 50000 4+2*C2H4 (ppm)

Figure 7 The concentration of NH3, HCN and NO respectively as function of the total carbon concentration on the product gas for different bed materials. TPS Termiska Processer AB TPS-96/48 10 Bjorkman, E. Larsson, C. 1996-10-28


The project is planned to continue with theoretical studies and kinetic/thermodynamic calculations describing fuel nitrogen and ammonia chemistry in gasification process and in fuel gas. For the most active catalyst there will be continuos test with specific N- compounds and direct measurements of N- conversion. We plan also to develop the experimental method further so that conversion of N-species to N2 can be analysed.


(1) Bjorkman, E., Sjostrom, K. Energy & Fuel, 5 (1991) 753-760 (2) Kurkela, E., Stalberg, P., Fuel Processing Tecnology, 31 (1992) 23-32 (3) Leppalahti, J., Fuel, 74 (1995) 1363. (4) Nelson. P.F., Kelly, M.D., Wornat, M.J., Fuel, 70 (1991) 403. (5) Boardman, R.D., Smoot, L.D., 1989 Symp. on Stationary Combustion Nitrogen Oxide Control, EPRI (1989) 6B-1. (6) Wojtowicz, M.A., Pels, J.R., Moujlin, J.A. Fuel Process. Technol. 34 (1993) 1. (7) Hamalainen, J.P., Aho, M.J., Tummavuori J.L Fuel, 73 (1994) 1894. (8) Hamalainen, J.P., Aho, M.J. Fuel, 74 (1995) 1922. (9) Gonzalez de Andres, A., Thomas, K.M., Fuel73 (1994) 635 (10) Asami, K., Osuka, Y., Ind.Eng.Chem.Ftes. 32 (1993) 1631 (11) Tanaka. S., U-emura, T., Ishika, K., Nagayoshi, K., Ikenaga, N., Ohme, H : Sui-^i. T.. Energy & F-jfJs 9 (1995) 45 i 12i Abotni, C.M.K.., Bota, K.8., DOE/PC/89760--T11 (13) Suzuki, T., Ohme, H., Watanbe, Y., Energy & Fuels 8 (1994) 649 (14) Ponder, G.R., Richards, G.N., Energy & Fuels 8 (1994) 705 (15) Gopalakrishnan, R., Fullwood, M.J., Bartholomew, C.H., Energy & Fuels 8 (1994) 984 (16) Cope, R.F., Arrington, C.B., Hecker, W.C., Energy & Fuels 8 (1994) 1095 (17) Menedez, R., Alvarez, D., Fuertes, A.B., Hamburg, G., Vleeskens, J., Energy & Fuels 8 (1994) 1008 (18) Arauzo, Radlcin, D., Piskorz, J., Scott, D.S., Energy & Fuels 8 (1994) 1192 (19) Gokarn, A.N., Muhlen, H.J., Fuel74 (1995) 124 (20) Ohtsuka, Y., Hiroshi, M., Watanabe, T., Asami, K. Fue/73 (1994) 1093 (21) Ito, T., Wang, J.X., Lin, C.H., Lunsford, J.H. J. Amer. Chem. Soc. 107 (1985) 5062 (22) Breitmaier, E., Jung, G., Organische Chemie I; Georg Thieme Stuttgart (1986) pp 582-585 (23) Alden, H., Espenas, B.G., Rensfelt, E. in Research in Thermochemical Biomass Conversion ed. Bridgewater, A.V., Kuester, J.L., Elsevier Applied Science, New York, (1988) pp 975-986. (24) Alden, H., Bjorkman, E., Carlsson, M., Waldheim, L in "Advances in Thermochemical Biomass Conversion, ed. Bridgewater Blackie Academic & Professional, Glasgow (1992) pp 216-232. Termiska Processer AB


Bjorkman, E., and Larsson, C. TPS Termiska ProcesserAB, 611 82 Nykoping