Ind. Eng. Chem. Res. 1999, 38, 1589-1595
1589
SEPARATIONS A Deep Desulfurization Process for Light Oil by Photosensitized Oxidation Using a Triplet Photosensitizer and Hydrogen Peroxide in an Oil/Water Two-Phase Liquid-Liquid Extraction System Yasuhiro Shiraishi,† Hiroyuki Hara,† Takayuki Hirai,*,† and Isao Komasawa†,‡ Department of Chemical Science and Engineering, Graduate School of Engineering Science, and Research Center for Photoenergetics of Organic Materials, Osaka University, Toyonaka, Osaka 560-8531, Japan
A deep desulfurization process for light oil has been investigated, on the basis of photosensitized oxidation using a triplet photosensitizer and hydrogen peroxide in an oil/water two-phase liquidliquid extraction system. Dibenzothiophene (DBT) in tetradecane was effectively desulfurized. This was achieved via the indirect photoexcitation of DBT, using benzophenone (BZP) under photoirradiation conditions of wavelength λ > 280 nm, and the effective oxidation of the excited DBT by H2O2, suppressing the photodecomposition of H2O2. The desulfurization of commercial light oil was also achieved in this process by the addition of both BZP and H2O2, and the sulfur content was reduced from 0.2 to 0.05 wt % by 48 h of photoirradiation, to meet with the new regulation in Japan. Although C5 and C6 alkyl-substituted DBTs were found to be the most difficult components to be desulfurized, they too were desulfurized effectively in the presence of BZP and H2O2. The photodecomposed sulfur-containing compounds removed into the water phase were separated successfully, using column adsorption with aluminum oxide as the adsorbent. The H2O2 solution recovered is also reusable for the desulfurization process. Introduction A novel desulfurization process for benzothiophene (BT) and dibenzothiophene (DBT), based on photochemical reaction and liquid-liquid extraction using an organic/water two-phase system, has been proposed in previous papers.1,2 Both BT and DBT, when dissolved in tetradecane, were found to be photodecomposed by the use of a high-pressure mercury lamp, and the resulting sulfur-containing compounds were removed into the water phase. The removal of DBT from light oils, however, was depressed markedly by the presence of naphthalene, owing to the triplet energy transfer from the photoexcited DBT to the ground-state naphthalene. To cope with this adverse effect of naphthalene, the effect of the addition of the triplet photosensitizer (benzophenone: BZP) or hydrogen peroxide was then studied with respect to the desulfurization of light oil.3 The desulfurization was not improved by the addition of BZP alone to the light oil phase, but the addition of H2O2 into the water phase was effective for the desulfurization of DBT, since this acts as a weak oxidizing agent for the photoexcited DBT. H2O2 is, however, known to be photodecomposed by the absorption of the short-wavelength UV light (λ < 280 nm), which is necessary for the direct photoexcitation of DBT. * To whom correspondence should be addressed. E-mail:
[email protected]. Fax: +81-6-6850-6273. Tel.: +816-6850-6272. † Department of Chemical Science and Engineering, Graduate School of Engineering Science. ‡ Research Center for Photoenergetics of Organic Materials.
In the present work, the effect of the addition of BZP for the excitation of DBT, in the presence of H2O2 under the irradiation of long-wavelength light (λ > 280 nm), on the removal of DBT from tetradecane and the resulting desulfurization of sulfur-containing compounds from several light oils was investigated. It is reported that alkyl-substituted BTs and DBTs have different reactivities in the hydrodesulfurization process and that 4-methyl-DBT and 4,6-dimethyl-DBT are the most difficult compounds to be desulfurized.4 The sulfurcontaining compounds remaining in light oil, following desulfurization according to the proposed method, were determined by GC-AED analysis, to reveal the reactivity of each compound. A separation process for the desulfurized sulfur-containing compounds, from the water phase, using solid adsorbents was also examined. Experimental Section The full procedure was described in the previous paper,1,3 and only a brief description is presented here. Feedstocks, such as commercial light oil (CLO) containing ≈0.17 wt % of sulfur, meeting with the former regulation in Japan (0.2 wt %), and straight-run light gas oil (LGO), supplied from Cosmo Petroleum Institute, were used. Relevant properties of these materials are summarized in Table 1. Organic solutions, such as the above feedstocks or n-tetradecane containing dibenzothiophene (DBT), to which benzophenone (BZP) was added, were mixed vigorously with water or hydrogen peroxide aqueous solution at the volume ratio 100/300 mL/mL, using a magnetic stirrer, and were photoirradiated by the immersion of a high-pressure mercury
10.1021/ie980651s CCC: $18.00 © 1999 American Chemical Society Published on Web 03/11/1999
1590 Ind. Eng. Chem. Res., Vol. 38, No. 4, 1999 Table 1. Properties and Composition of Light Oils commercial light oil (CLO)
straight-run light gas oil (LGO)
0.8313 15.8 83.2 0.1731 77.9
0.8548 12.7 85.2 1.434 75.4
17.5 4.6
14.9 9.7
density (288 K) (g/mL) hydrogen (wt %) carbon (wt %) sulfur (wt %) saturated fraction (vol %) aromaticsa (vol %) one-ring two-ring
a Quantities of three-ring and more than three-ring aromatics are trace ( 280 nm).
Figure 1. Effect of the irradiative wavelength on the time-course variation in the concentrations of (a) H2O2 in the water phase and (b) DBT in the tetradecane phase, in tetradecane/H2O2 (10%) aqueous solution two-phase system (λ > 280 nm).
lamp (300 W, Eikohsha Co., Ltd., Osaka), and with air bubbling (500 mL/min) at atmospheric pressure. A Pyrex glass filter was used to give a light wavelength rejection of less than 280 nm. The temperature of the solutions, during photoirradiation, was about 323 K. The sulfur-containing compounds in the light oil were analyzed by a gas chromatograph-atomic emission detector (GC-AED, Hewlett-Packard 6890, equipped with AED G2350A) and according to the procedure of Tajima et al.5 The concentrations of H2O2 in the water phase and hydroperoxide in light oils were determined by a titrimetric analysis using sodium thiosulfate and potassium iodide as an indicator.6 Aluminum oxide (surface area ) 107.6 m2/g) and activated carbon adsorbents were used, following the activation treatment, to separate the resulting sulfur-containing compounds from the water phase. Results and Discussion Effect of Benzophenone and Hydrogen Peroxide on the Desulfurization of DBT from Tetradecane. Figure 1 shows the effect of the irradiative wavelength on the decomposition of H2O2 and the desulfurization of DBT from tetradecane. Without the filter, DBT was successfully decomposed and removed into the water phase following 10 h of photoirradiation, as reported in the previous paper.3 The H2O2 was also photodecomposed significantly, since H2O2 absorbs the light at wavelengths of λ < 280 nm.7 Use of the filter, however,
markedly reduced the rate of decomposition of H2O2. The quantity of the H2O2 decomposed was only 10% as compared to the case of that without filter, and this was comparable with that for zero irradiation where only the thermal decomposition is likely to occur.8 The desulfurization of DBT was also suppressed considerably by the use of the filter, since the quantity of photoexcited DBT was decreased. These results indicate therefore that an alternative method for the indirect excitation of DBT, effected by the irradiation of longwavelength light (λ > 280 nm), is required. Benzophenone (BZP), as a triplet photosensitizer, is photoexcited at wavelengths of less than 380 nm. The triplet energies of BZP and DBT are estimated as being 287 and 285 kJ/mol, respectively.9 The energy transfer from photoexcited BZP to DBT, therefore, occurs exothermically, and as a result, the indirect excitation of DBT occurs satisfactorily, under the irradiation at wavelength λ > 280 nm, consequent to the addition of BZP.3 Figure 2 shows the effect of the addition of BZP both on the desulfurization of DBT and on the decomposition of H2O2. The rate of desulfurization of DBT is accelerated by the addition of BZP or H2O2 when added singly, but was much enhanced by the addition of BZP and H2O2 together. The rate of decomposition of H2O2 was not affected by the presence of BZP, and likewise the rate of decomposition of BZP was also not affected by the presence of H2O2. With both BZP and H2O2, the amount of H2O2 consumed for every mole of DBT following 2 h of photoirradiation was estimated as 15 mol, whereas 707 mol of H2O2 was decomposed in the case where a whole wavelength of light was irradiated without BZP. Relationship among the Concentrations of DBT, BZP, and Hydrogen Peroxide in the Desulfurization of DBT from Tetradecane. Figure 3 shows the effect of the addition of BZP and/or H2O2 on the rate of desulfurization of DBT from tetradecane at the condition of [DBT]initial ) 11 mM, corresponding to a 0.05 wt % sulfur concentration. With the addition of BZP alone, the concentration of DBT decreased markedly with
Ind. Eng. Chem. Res., Vol. 38, No. 4, 1999 1591
Figure 3. Time-course variation in the concentration of DBT in tetradecane with the addition of (a) BZP alone, (b) H2O2 alone, and (c) both BZP and H2O2.
increasing BZP concentration in the range, [BZP] ) 0.5-2.7 mM. The desulfurization rate however becomes almost saturated at [BZP] > 2.7 mM. With H2O2 alone being added, the rate of desulfurization was accelerated with increasing [H2O2], but with the rate being rather less than that for the case with BZP alone being added. With the addition of both BZP (11 mM) and H2O2, the rate of desulfurization was considerably enhanced, and almost all the DBT was removed following 8 h of photoirradiation. In this case, the greater addition of H2O2 at more than 10% however hardly affected the rate of desulfurization obtained. The relationship between the differing concentrations of BZP, H2O2, and DBT on the rate of desulfurization of DBT from tetradecane was studied. Here, the rate was defined as the difference between the concentrations of DBT in tetradecane at the times of 0 and 6 h, as rDBT ) ([DBT]0h - [DBT]6h)/6. These results are shown in Figure 4. With BZP alone, the desulfurization rate of DBT increased linearly with increasing [BZP] up to approximately 2 mM, and the rate became saturated at [BZP] > 11 mM, as shown in Figure 4a. This behavior was unaffected by the air bubbling rate, and thus, the solution appears to be saturated with oxygen. The differences of triplet energy (∆ET) between DBT and BZP in nonpolar solvents is only 2 kJ/mol. When the ∆ET value is less than a few kJ/mol, the reverse triplet-energy transfer occurs.10,11 Yamaji et al.12 have reported the formation of a triplet equilibrium in the DBT-BZP system in acetonitrile solution. Therefore, the reverse-energy transfer from excited DBT to
Figure 4. Effect of the addition of (a) BZP alone, (b) H2O2 alone, and (c) both BZP and H2O2 on the desulfurization rate of DBT from tetradecane, rDBT, with respect to the initial concentration of DBT.
ground-state BZP can occur, as shown by eq 1: f
3
BZP* + DBT a BZP + 3DBT* b
(1)
The triplet equilibrium between DBT and BZP is formed at the excited states, and the corresponding equilibrium constant can be represented as in eq 2:
( )
Keq ) exp
∆ET [3DBT*][BZP] ) 3 RT [ BZP*][DBT]
(2)
where R is the ideal gas constant and T is the temperature (K). For the present experimental conditions, the Keq value was determined as 2.11 at 323 K. The excess addition of BZP, therefore, reduces the formation of 3DBT*, and approximately 11 mM of BZP seems sufficient to increase the desulfurization rate. The desulfurization rate was increased by increasing [DBT]initial, since the equilibrium, as shown by eq 1, shifts to the right with increasing [DBT]initial. The increase in the desulfurization rate with [DBT]initial was minor at [BZP] > 11 mM, thus indicating that the quantity of O2 is less than that of 3DBT* in this high BZP concentration region.
1592 Ind. Eng. Chem. Res., Vol. 38, No. 4, 1999
Figure 5. Comparison of the desulfurization rate of DBT, its alkyl-substituted derivatives, and BT from tetradecane.
The effect of the addition of H2O2 alone is shown in Figure 4b. Here, the desulfurization rate increased only slightly with increasing [H2O2]. H2O2 acts as an oxidizing agent against photoexcited DBT,3 and therefore, the addition of H2O2 alone is ineffective under the present irradiation condition (λ > 280 nm), for which the direct excitation of DBT hardly occurs. With both BZP and H2O2 added together, as shown in Figure 4c, the rate increased 2-fold or 5-fold compared to that for the cases with only BZP or H2O2 alone additions. The rate obtained with both BZP and H2O2 present was also enhanced with increasing [DBT]initial. This is because the number of excited molecules of DBT, which can be oxidized effectively by H2O2 molecules, was increased with increasing [DBT]initial. In the combined system, the desulfurization rate was almost saturated at [H2O2] > 10%, thus indicating that the quantity of the H2O2 molecules was in sufficient excess, as compared to that of the excited DBT molecules.3 Thus, the proposed process is likely to be effective for the desulfurization of light oils containing large quantities of DBTs. Photoreactivity of Methyl-Substituted DBTs and Benzothiophene, and the Effect of Naphthalene on the Desulfurization of DBT. Methyl-substituted DBTs and benzothiophenes (BTs) are the main components of the sulfur-containing compounds in light oil, and the former are reported to be most difficult to be desulfurized, according to the hydrodesulfurization method.4,5 The photoreactivity of 4-methyl-DBT and 4,6dimethyl-DBT was, therefore, also studied. The reactivity of these substituted DBTs was found to be higher than that of DBT, and the order is 4,6-dimethyl-DBT > 4-methyl-DBT > DBT, as shown in Figure 5. This is the same order as that obtained for the case where the whole wavelength region from the mercury lamp was used.1 The methyl substituents cause the expansion of the electronic cloud on aromatic moiety. The energy transfer occurs via the collision and the association between 3BZP* and ground-state DBTs,13 and thus methyl-substituted DBTs strongly associate with 3BZP*, resulting in the higher desulfurization yield. The photoreactivity of BT was found to be higher than that of DBTs. The electron density on unsaturated 2- and 3-positions of BT is high, and therefore the energy transfer from 3BZP* to BT occurs effectively; as a result, desulfurization of BT proceeds more effectively. The excited DBT is deactivated by two-ring aromatic compounds, such as naphthalene (NP), which are contained in light oil in a large quantity.1,3 Figure 6 shows the effect of BZP and H2O2 on the desulfurization of
Figure 6. Effect of the addition of BZP and H2O2 on the timecourse variation in the concentrations of (a) naphthalene (NP) and (b) DBT in tetradecane.
DBT in the presence of NP. When using the filter (λ > 280 nm), both BZP or H2O2 when added separately were found to be ineffective for the desulfurization of DBT. This is because of the energy transfer from photoexcited DBT to the ground-state NP.3 With both BZP and H2O2 together, the desulfurization rate was increased to a level which was comparable to that in the case of H2O2 alone without a filter. As was shown in Figure 3a, the desulfurization is not enhanced by the addition of BZP in excess of 2.7 mM in the absence of NP. In the presence of NP, the rate was, in contrast, accelerated by increasing [BZP], as shown in Figure 6. Since NP is also excited by the energy transfer from BZP,14 the competitive energy transfer thus occurs, as follows: 3
BZP* + DBT f BZP + 3DBT* 3
BZP* + NP f BZP + 3NP*
(3) (4)
Thus, the addition of a large quantity of BZP is effective for the formation of 3DBT* in the presence of NP. In addition, under the irradiation conditions of λ > 280 nm, the degradation of NP is also suppressed, owing to a reduction in the direct excitation of NP. Desulfurization of Light Oils Using a Triplet Photosensitizer and Hydrogen Peroxide. Figure 7 shows the effect of the addition of BZP and/or H2O2 when applied to the desulfurization of light oils. As shown in Figure 7a, for commercial light oil (CLO), with neither BZP nor H2O2 present (condition (i)), only 10% of the sulfur was removed following 60 h of photoirradiation. Also, for either BZP or H2O2 alone ((ii) or (iii)), only 20% or 50% respectively of the sulfur was removed. With the addition of both 108 mM BZP and 30% H2O2 (iv), the desulfurization was accelerated very effectively, and the sulfur content was reduced following 48 h of photoirradiation to a value of less than 0.05 wt %. On the other hand, desulfurization of straight-run light gas oil (LGO) was found to be well-affected by the addition of H2O2 alone (iii), and any effect of the BZP did not appear to be very obvious (iv), as shown in Figure 7b. From the data shown in Table 1, LGO contains a 2-fold greater quantity of two-ring aromatic compounds than does CLO. Thus, the energy transfer from BZP to DBTs is hardly effective at [BZP] < 108 mM. The proposed
Ind. Eng. Chem. Res., Vol. 38, No. 4, 1999 1593 Table 2. Quantities of Benzothiophenes (BTs) and Dibenzothiophenes (DBTs) Having Different Carbon Number Substituents in the Feed Commercial Light Oil (CLO) and the Oil after 60 h of Photoirradiation of Long Wavelength Light (λ > 280 nm) [Conditions (i)-(iv) Are Identical to Those in Figure 7] desulfurized CLO (condition) BZP, H2O2
Figure 7. Effect of the addition of BZP and H2O2 on the desulfurization of (a) commercial light oil (CLO) and (b) straightrun light gas oil (LGO).
process is therefore effective for the desulfurization of the light oils containing low level two-ring aromatic components. The desulfurization of aromatic-rich light oils such as LGO and light cycle oil are difficult according to the proposed process. It was, however, found to be achieved effectively by employing photoirradiation of the whole wavelength region from the mercury lamp in an oil/hydrogen peroxide aqueous solution two-phase system.15 To establish the reactivities of the various sulfurcontaining compounds, each compound remaining in the light oil, following desulfurization, was determined. A third phase (emulsion) appeared during desulfurization,1 and this was demulsified by adding 30 mL of 1 M HCl aqueous solution. The recovered fraction of the light oil was dependent on the desulfurization conditions, as shown in Figure 7; the fractions being respectively for conditions (i) 95%, (ii) 93%, (iii) 87%, and (iv) 86%. The resulting light oils were analyzed by GC-AED. Table 2 lists the quantities of the individual sulfur-containing compounds remaining in desulfurized CLO, following 60 h of photoirradiation, for the various conditions according to Figure 6. The results show that more than 60% of the BTs were removed in the absence of both BZP and H2O2,2 and that the BTs were further reduced to a concentration level of less than 5% by the addition of both BZP and H2O2. Under this latter condition (iv: 108 mM BZP and 30% H2O2), more refractory DBTs were also reduced effectively to less than 30%. Figure 8 shows the remaining percentage values for BTs and DBTs, having different carbon number substituents, for desulfurized CLO. The residual percentage of BTs was decreased effectively by the addition of BZP and H2O2, although the residual percentage increased with increasing the carbon number of the substituent. With H2O2 alone, the C0-C4 DBTs were effectively desulfurized, but the desulfurization was not so easy for the C5 and C6 DBTs, thus making deep desulfurization of CLO difficult. By employing both BZP and
feed CLO
(i) 0 mM, 0%
(ii) 108 mM, 0%
(iii) 0 mM, 30%
(iv) 108 mM, 30%
species
(wt %)
(wt %)
(wt %)
(wt %)
(wt %)
