Low pressure equilibrium between H 2 S and alkanolamine revisited

Transcription

1 Indian Journal of Chemical Technology Vol. 6, May 1999, pp Low pressure equilibrium between H S and alkanolamine revisited M V Jagushte & V V Mahajani* Department of Chemical Technology, University of Mumbai, Matunga, Mumbai , Indi a Received 4 September 1998; accepted 11 April 1999 A-bso rpti on of H S in aqueous a1kanolamine solution is of considerable commercial importance. A simple method is presented to determine the equilibrium pressure having very low loading of H S, with the help of 'sulfide' ion-selective electrode. By thi s method H S-DEA and H S-TEA VLE-data were obtained at different amine concentration and temperatures. A theoretical analysis of equilibrium between H S and aqueous a1kanolamine solution has been reviewed. A semiempiri cal model with a correlation parameter, ~ -fac t or, is introduced to predict the equilibrium between H S and alkanolamines. Absorpti on of H S into aqueous alkanolamine solutions is of considerable industrial relevance particularly in natural gas, associated gas and biogas sweetening, Claus tail gas cleaning, refinery gas sweetening etc. Many times, process engineers are concerned wi th selective removal of H S from the sour gas stream. Tertiary ami nes such as methyl diethanolamine (MDEA), and triethanolamine (TEA) are find ing popularity among gas processing engineers wh o would like to have selective removal of H S. The absorption of H S in aqueous alkanolamine solution takes place with reversible chemical reaction.... (I) Due to reversible nature of the reaction there always ex ists finite equilibrium concentration of H S. The H S laden rich amine solution from the absorber is sent to regenerator or stripper where H S is knocked off at higher tempe.ature, say 1 1O-1SoC and near atmospheric pressure. The regenerated solution is then returned back to the absorber. Thus absorber and regenerator are coupled together to remove H S. The stripper overhead gas which is very rich in H S is then processed suitably to convert it into elemental "Sul fur". Purely, from economic consideration the amine solution has not been generated completely. Thus the degree of sulfidation defined as, a = (moles of H S) (moles of amine) *For correspondence... () a is finite and can be In the range 0.0 to 0.1 depending upon the local techno-economic considerations. Due to reversible nature of the chemical reaction, the absorber is operated in counter current mode of operations. The driving f0rce, at the top of the column depends on the equilibrium pressure of H S over regenerated amine solution. This is often called as back-pressure of H S over the regenerated solution. The absorption of H S is gas film controlled and at any point in an absorber, the rate is given by, I I I - - = KG a kg a ( H )(kla)(e) (3a)... (3b) Knowledge of both gas and liquid side mass transfer coefficients, namely. kg and kl and knowledge of effective interfacial area a is essential. The enhancement factor, E, can be computed based on theories of mass transfer with chemical reaction 1.. From Eq. (3a), it is seen that at the top of the. absorber, partial pressure of H S ( PH S), in the bu lk gas (sweetened gas) being very small, the driving force (P HS - P~ S ) depertds on the equilibrium partial pressures of H S, ( P~ S)' over the regenerated solution. Thus the knowledge of P~S becomes very important to process design engineer because small error in P~ S can make considerable error in driving force, as ( P HS ), itself will be very low.

2 16 INDIAN 1. CHEM. TECHNOL., MAY 1999 Experimental technique for measuring Vapour liquid equilibrium data are such that accurate data is available in the middle range of loading and pressures.1 5 and very few equilibrium' data are reported in the low loading region, encountered in commercial operations. A general method for obtaining VLE data at low loading is to sparge inert gas into an excess of known amine solution followed by analysis of the gas fo r H S. However, this method had not provided the reliable VLE data at very low partial pressure of H S. Recently, the increasing demand for sulfur free-fuel, the stricter air pollution control legislation and the incentives for sulfur recovery have all resulted in removal pf H S at very low level, which require the better experimental technique to understand the VLE'behaviour of H S in the lean loading region. There are very few techniques reported in the literature, which measure VLE data at very low loading of H S. Amongst, Rogers et al. 6 has used FfIR spectroscopy to measure VLE data at low. 7 partial pressure of H S. Rochelle et al. have developed electrode method for the measurement of VLE using ph-silver sulfide electrode. These in vesti gators related partial pressure of H S with electrode potential difference between the ph and sulfide electrode by the equation, I log H,S=Constant+ (E H, - Es= ) But the value of constant in the above equation was calculated based on the assumption of Kent & Eisenberg model 8 as stated by Rochelle et al.' Naturally, this might introduce some deviation in the measurement as in this model (Kent & Eisenberg), the values of equilibrium constants (K, & K ) were determined by forcing computational fit on the reported data available in the literature. In the present paper, a simple method to determine equihbrium partial pressure of H S, (P~ S) over alkanolamine solution, having very low a values, with the help of 'su lfide' ion-selective electrode has been presented. This experimental technique will aid to generate the valuable equilibrium data at very low loading of H S in aqueous alkanolamine solutions. A theoretical tulalysis of equilibrium between H S and alkanolamine solution has been reviewed and a simpler semi-empirical model is presented to aid the process design engineer. - {) U f- ~L. M TO CONDUCTIVITY ME1~R,--+ H TIC) 3 ~ '-- 7 f-:.-:-:- '. 1, ~h-- ~.,.,. ' D-Lo '. \.., ' ':,..,'... II////A I~ :~: 6"' m Fi g.! - Expcrimenlal sel-up fo r Vapour Liquid Equi lihrium measurement Experimental Procedure Materials-Diethanolamine (DEA), triethanolamine (TEA) and sodium sulfide used were of analytical grade with 99.5% purity and obtained from S.D. Fine Chem. Ind., Mumbai, (India). H S gas was generated in Kipp's apparatus. Sulfuric acid was used to -generate H S from iron sulfide so that contamination due to HCI and water is avoided. Also sufficient quantity of gas is generated and purged out so that air contamination in H S is eliminated. Experimental set-up-the experimental set-up consisted of a gas bubbler with magnetic stirrer to enhance equilibrium process. The equilibrium cell is fitted with conductivity probe. The exit of the cell is connected to a glass reservoir. The gas-circulating blower takes gas from reservoi r and passes into the equilibrium cell. The pressure maintained in this system is practically near atmosphere. The entire assembly is placed into a constant temperature bath except gas circulating blower. Since temperatures are not widely different from ambient 300C, the heat loss from the blower to surrounding can safely be neglected. Fig.l shows a schematic experimental diagram for the entire set-up. Experimental Proced'1re-A known quantity of alkanolamine solution was taken in an equilibrium cell. H S gas was injected into reservoir to get desired partial pressure. The gas-circulating blower was then started. Some H S would get adsorbed into alkanolamine solution. To compensate this, additional quantity of H S was injected so that the system is near atmospheric pressure. The approach to equilibrium is monitored with the help of a conductivity probe.

3 JA GUSHTE & MAHAJANI: LOW PRESSURE EQUILIBRIUM BETWEEN HzS ALKANOLAMfNE 17 Since the reaction of HS with aqueous alkanolamine solution is ionic in nature, the concentration of ionic species remains constant after reaching equilibrium. The constant reading of conductivity probe over a few hours indicates that the equilibrium was achieved. At this stage, the gas composition was identical in the cell as well as in the reservoir. The reservoir was then isolated from the system with the help of valves. A known quantity of caustic, which is far excess than required, is added to the reservoir with the help of liquid syringe. It was then well mixed by shaking and left for about 48 h so that entire amount of HS gas is absorbed into aqueous NaOH solution. A sample was taken from amine solutioh with the help of a gas-tight syringe and introduced into caustic solution to convert it into Na S. With the help of sulfide ion-selecti ve electrode, both samples were analyzed for sulfide content and hence H S content was back calculated both in the gas phase and in the liquid phase. The sul fide ion selective electrode (ORION, USA make) was cali brated before the reading with the help of st.. ndard aqueous solutions of NaS. It was ensured that all readings were obtained in linear behaviour of the electrode. as Il , lo: 0.3.= ::t Il c:x::xx:o present work ~ Mather etai.[5] 0.0 -rrrnrrt-rrt"t't"t"t"rnrrt't't'"rt"t"t"t"t"t't"t"t"rnrrt-rrrri at / ( l-ci\) J( l()' Fi g. - Pl ot of a ~ / ( I - a) versus PH S of H S-OEA (M) system at 313 K. The reacti on between amine and H S IS in stantaneous as compared to rate of diffusion. o Results and Discussion During absorpti on of H S in an aqueous solution of alkanolamine fo llowing reactions take place and equilibria are establi shed, Dissociation of HzS.. (4a) Protonatiol1 n.famine K" = (H+)(An: H) (AmH ) Dissociation of water... (7a).. (7 b).. (4b) H0 ~ ~ H+ + OH- (8a) also, we have... (Sa) Kw = (H+) (OH- ) Amine dissociation in water (8b)... (5b) (ArnH) + HO OIl (ArnH/) + OH-.. (9a) Reaction of amine with H S K _ (AmH +) (OH- ) b- (AmH).. (9b).. (6a) and then from Eqs (7b) & (8b), K _ (AmH/ ) (HS- ) SAm - (H S) (AmH)... (6b) K _ Kw b-- Ka.. (9c)

4 1-8 INDIAN 1. CHEM. TECHNOL., MAY (10) The solubility is assumed to be established instantaneously at gas-liquid interface. Also there exist a possibility of reaction of H S with air as given below due to the pres~nce of air ions in liquid phase (Eqs 8a & 9a) (H+)= ( Ka) (AmH/) (AmH).. (13) for the sake of argument, the reaction wherein S= is formed is considered, It is being assuming that all amine can form S= (14) (1 5) Thus it is seen that during absorption of H S in aqueous alkanolamine solutions, the entire absorption process can be attributed to. reaction of H S with amine provided concentration of (OH- ) is sufficiently small so that the Eq. ( 11") can be practically ignored. Further, the dissociation of HS- to S= via Eqs (5a & 5b) be very small so that only reaction can be considered by Eqs (4a) and (4b). The above aspects have been dealt in details by da Silva and Danckwerts 9 For the sake of clarity it shall be reviewed here also. For selective absorption of H S the use of tertiary amine such as triethanolamine, (TEA), methyldiethanolamine (MDEA) is recommended. Mahajani and Joshi 10 have reviewed the kinetics of reactions between CO and alkanolamines (primary, secondary and tertiary alkanolamine). MDEA is very popular tertiary. amine for H S removal. However MDEA reacts with CO also. Authors recommended, the use of TEA for selective removal of H S. The kinetic selectivity for H S would be more in the case of TEA because CO reacts very slowly with TEA as compared to MDEA. Equilibrium with respect to all types of amines, namely, primary, secondary and tertiary shall be discussed. The experimental results would be compared with those published in case of diethanolamine system and data would be presented for triethanolamine system. The condition underwhich, there will be negligible formation of s= as compared to HS- From Eq. (5b), (s=) ( K S ) --=-- (HS-) (H+) Also from Eq. (7b),.., (1 ) where 0:. = sulfidation ratio Eq. ( 13), gives _ (mole of H S) (moles of amine) (H+)= ( K,) (0:.) (AmH)o (1- a) (AmH)o =3:0:. ( K,) (I - 0:.) Therefore,.. (16).. ( 17).. (18) However, in majonty of cases fi rst dissociation IS dominating as given in Eq. (4a) (AmH) = (1-0:. ) (AmH)o.. (19) (S=) (KS)(I - 0:.) --=-=--- (HS-) (Ka) 0:... (0) From Eqs ( 18) and (0) one can see that the ratio of K S I Ka is more important than sulfjdation ratio. By and large the loading of H S in the absorber will not exceed 0.3, (i.e. 0:. = moles of H S per moles of amine). Under such a situation the multiplier of K S I Ka in Eq. ( \8) will be 0.67 and that from Eq (0) will be.33. In the regenerated solution (lean solution) loading, a, could be as low as 0.0 and therefore the multiplier of K S I Ka could be 4 in Eqs ( 18) and 49 in Eq (0). Thus a condition can arise where,

5 JAG USHTE & MAHAJANI: LOW PRESSURE EQUILIBRJUM BETWEEN H S ALKANOLAMrNE 19 Table \-Various equilibria as a function of temperature I. MEA (monoethanolamine) pk, = exp ( / 1),. DEA (diethanolamine) pk, = T, 3. TEA ( tri ethanolamine) pk, = / T, 4. MDEA (meth yldiethanolamine) pk, = / T, 5. First dissoci ation constant for H S in water pk ls = / T log T, 6. Second di ssociation constant for H S in water pks= / T, 7. Solubility of H S in water 10g HH S = / T , 8. Di ssociation constant of water In (Kw) = / T In T , K Table -Values of S for different alkanolami nes at Ka and 33 K Amine MEA DEA MDEA TEA Temperature, (K ) (for K S and K, refer to Table I) (s=) = (50)* (KS ) «50x I0-4 (HS ) ( Ka) ( K S) --xi06 ( Ka) ( Ia)... ( Ib) Ka == (kmol. m- 3 ) Ka == (kmol. m- 3 ) Ka == (kmol. m- 3 ) Reference [Perrin 1] [Perrin 1] [Perrin 1] Ka == (kmol. m- 3 ) [Little 13 j KJ S == (kmol. m- 3 ) KS == (kmol. m- 3 ) H H S == (kmol. m- 3 Pa- I ) '[Bosch l 4 ] [Bosch l4] [Boschl4] Kw= (gmoi.kg- 1 ) [Edward 15] Various equilibria used to illustrate above condititms as well as in formulating the model at later stage, have been presented in Table I. Table presents such computations of K S / Ka for' monoethanolamine (MEA) (primary alkanolamine); diethanolamine (DEA) (secondary alkanolamine); triethanolamine (TEA) (tertiary alkanolamine) and methyl diethanolamine (MDEA) (tertiary alkanolamine) at three different temperatures. Thus it is seen that, the above inequality (Ib) is easil y sati sfi ed and hence formation of S= can be neglected without sacri ficing engineering accuracy. The condition under which there will be no significant reaction between OH - and H S While formulating a mathematical model fo r va pour-liquid equilibrium. It is essential that one has to consider reacti on between H S (solute gas) and OW as given by Eq ( II ). However, if there are in significant quantity of hydroxyl ions (OW) as compared to free alkanolamine (AmH) present in solution, then the reacti on between H S and OH- can be safely neglected without sacrificing engineering accuracy. Therefore, if (OW ) / (Am H) < 10- contributi on due to (OH-) can be neglected. Thus, so that contribution of S= can be neglected without sacrificing engineering accuracy..., ()

6 130 INDIAN J. CHEM. TECHNOL., MAY 1999 (K w) < (AmH)xlO-4 (Ka)... (3) For MEA, DEA, TEA and MDEA, Table 3 exhibits this ratio. It is thus seen that between temperature 303 to 33 K the lowest value of Kw I Ka is X 10-6 in the case of TEA. Therefore under real operating conditions where concentration of free amine could be more than 0.1 M, it is seen that inequality of Eq. (3) is easily satisfied. At the top of the countercurrently operated absorber free amine concentration being higher, say > M, in case of MEA, also this inequality holds true. As absorbent goes down the column, free amine concentration goes down and in case of MEA, the inequality may not hold true at the bottom part of the absorber. However at the bottom portion of the absorber, the effect of equilibrium partial pressure of H S in the gas phase is not worrying. As seen from Table 3, as temperature increases, Kw 1Ka also increases. Nonetheless, under real operating conditions it is seen that inequality of Eq. (3) is easily satisfied. From the forgoing discussions, it is seen that there will be negligible formation of S= and under normal conditions, the contribution of OR- to absorption of H S could be neglected. H S - aqueous alkanolamine equilibrium With the help of above assumption, H S alkanolamine equilibrium can now be presented Eq. (6b) gives, K _ (AmH/) (HS- ) SAm - (H S) (AmH) and Eq. (4b), gives, K Table 3-- Yalues of ~of different alkanolamines at 303, 313 Ka and 33 K Amine MEA DEA MDEA TEA ( :: ) x IO'(k~lIm') Temperature, K Therefore, (H S) = (H+) (HS- ) (K HS ) Eq. (7b), gives (H+) = (Ka) (AmH/) (AmH)... (4)... (5) Table 4-H S-aqueous alkanolamine equilibrium data System Partial Pressure of H S (KPa) Sulfidation Ratio (a) Partial Pressure of H S (Kpa) Sulfidation Ratio (a) Temperature = 313 K Temperature = 33 K Q.loo Concentration = M Concentration = 4 M '" J>,

7 JAGUSHTE & MAHAJANI: LOW PRESSURE EQUILIBRIUM BETWEEN H S ALKANOLAMINE 131 Table 5-Variation of ~-factor with respect to temperature and amine concentration for H S- DEA and H S-TEA systems , Amine Concentration Temperature ~-factor kmol / m 3 K DEA DEA TEA TEA , III c..,.\ii 1.0.S ::t c o 0.30 ~ CIS c.. ~0. 0 C o. 0 *,"TTT"rTTTTTT1"TTT"rrrrTTnrrTTTTT1"TTT"rTTTTTT1rrTTTTT1"TTT"rTTT~ c</(1-0'.) x 10' Fig. 4-Plot of a / ( I-a) versus PH S of H S-TEA (M) system at 3 13 K pressure through Eq. ( 10). Thus equilibrium pressure of H S over amine solution is given by ';/(1-0:.) x 10' Fig. 3-Plot of a / ( I-a) versus PH ZS of H S-DEA (M) system at 33 K. Substituting the above value of (H+) in Eq. () (HzS) = (Ka) (AmH/) (HS- ) (K ls ) (AmH) relation obtained is, a= (moles of HzS) (moles of amine)... (6) If (AmH)o is the original concentration of amine, then Eqs (6a) & (6) gives, (HzS) = (Ka) ~ (AmH)o (K ls ) (I-a)... (7) Concentration of (HzS) is also correlated with... (8) If a graph of PH S against a Z / ( I-a) is plotted, for the given concentration of amine (AmH)o, it would be straight line through the origin. Using the equilibrium data tabulated in Table. 4, Fig. to Fig. 5 exhibit such plots for H S-DEA and H S-TEA system at different conditions. For H S-DEA system, the comparison with dat" published by Mather et al. 11 is presented in Fig.. The data fit in the model quite well. At any given time in an absorber there exists many ionic species as given by Eqs (1 )-( 11) and therefore the adsorbent becomes a non-ideal solution. Atwood et al. 4 have shown that even for ionic,strength of 0.075, the mean activity coefficient for MEA, DEA and TEA reduces to approximately 0.75 from unity. In order to account for non-ideality, the factor ~ has been introduced which takes into account deviation of the system parameters from ideal situation. Thus,... (9)

8 13 INDIAN 1. CHEM. TECHNOL., MAY ; Q.. 'T is introduced to predict the equilibrium between H S and alkanolamine. Acknowledgement One of the authors (MV J) wishes to thank the G P Kane Trust for awarding research fellowship to enable this investigation. Nomenclature (AmH) (A mh)" = concentration of amine, (kmol.m- 3 ) = o ri ginal concentration of amine, (kmol.m- 3 ) = interfacial area, (m 1.m- 3 ) = enhancement Factor = solubility of H S, (kmol. m- 1.Pa- l ) ''1' Fi g ~~~rnnn"~nnnn"TTrrrnno~,,rrnnrl B.0 at 3 13 K c:j.j( 1-0() J: 10" Plot of a ' / ( I - a ) versus PH S of H1S-TEA (4M) system This fa c t o r -~ can be function of original amine concentration and degree of sulfidation and te mpe rature. Thus values of K a, K ts H H,S reported at infinite dilution can be considered and then correction may be introduced through factor-~. Vapour liquid equilibrium data of H S-DEA and H S-TEA system have been correlated with the help of Eqs (R) and (9). Table 5 shows ~-fa ct or for typical system involving aqueous solutions of DEA, and TEA-systems. It can be seen that for DEAsystem, the va ri ation in the value of ~ -fac tor was more (nearly one hal f) for change in temperature of 10K as compared to TEA-system, where the amine concentrati on was varied from M to 4M. It may imply that the temperature rather than amine. concentrati on has the pronounced effect on the parameters wh ich govern the ~-fa c t or. Conclusion An experimental tec hnique has been developed for the measurement of vapour-liquid equilibrium between H S and aqueous alkanolamine solution using ion-selective electrode at low partial pressure of H S. Usi ng this technique, VLE-data for DEA-H S and TEA-l-hS systems were measured by varying the te mpe rature and amine concentration. By considering all the equilibria in the system, a sem iempirical model is presented. A correlation parameter, ~ - fac t or, K S Ka ( pka) ~h ' (pk h) KG = equil ibrium constant for dissociation of H S in Eq. 4a, (kmol.m- 3 ) = equilibrium constant in Eq. (5b), (kmol.m- 3 ) = equilibrium constant in E. (7b), (kmol.m-.l ) =equilibrium constant in Eq. (9b), (kmol.m- 3 ) = overall mass transfer coefficient, (kmol.m-.pa- I. s- I ) = gas side mass transfer coefficient, (kmol m-.pa- l.s- 1 ) = liquid side mass transfer coefficient, ( m.s- I ) = equilibrium constant fo r reaction of amine wi th H"S in Eq. (6b), (kmol.m- 3 ) = equi li brium constant in Eq. (8b) in Table I, (gmoi.kg- 1 ) = parti al pressure of H S in!he bulk of gas phase. (Pa) =predi cted equilibrium partial pressure of H S in the bul k of gas phase, (Pa) =equi librium parti al pressure of H S. (pa) R., =rate of absorption, (kmol.rn- 3. sec- I) a =suilldation ratio, Eq. () f3 =correctioll factor. Eq. (9) All {,K ' s are defined in Table I. References Doraiswamy L K & Sharma M M, Heterogeneous Rcacciolls - Analysis. Examples and Design (Wiley Interscience, New York) Danekwerts P V, Gas-Liquid Reactions (McGraw Hil l Book Co) Lcibush A G & Shneerson A L, } Appl Chem (U 5 5 R ), 3 ( 1950) At wood M R, Arnold R C & Kindrick. Ind Eng eire"" 49 (1957) Muhlbauer H G & Monaghan P R, Oil Gas J, 55 ( 1957) Rogers E A, A I e h E J, 43 ( 1997) Rochelle G T, Tseng P C, Ho W S & Savage D W, Ind Eng e ire", Res, 7 ( 1988) 195.

9 JAGUSHTE & MAHAJANI: LOW PRESSURE EQUILlBRlUM BETWEEN H,S ALKANOLAMINE Kent R L & Eisenberg B, Hydrocarbon Processing, 55 (1976) da Silva T A & Danckwerts P V, I Chem Eng Symposium Ser, 8 (1968) Mahajani V V & Joshi J B, Gas Seper and Purifi, (1988) 50 1 I Mather. A E, Otto F D & La! D, Can 1 Chem Eng, 63 (1968) 1 Perrin D D, Dissociation Constants Of Organic Bases In Aqueous Solutions (Butterwort London), Little R J, Bos M & Knoop G J, 1 Chem Eng Data, 35 (1990), Bosch H, Solvents and Reactors for ~cid Gas Treating, Ph D Thesis, Universiteit-Twente, Edwards T J. Maurer G J, Newman J & P,ausnitz M, A I Ch E 1, 4 (1978) 966.