石 油 学 会 誌 Sekiyu Gakkaishi, 36, (6), 479-484 (1993) 479

[Regular Paper] Hydrorefining of Shale Oil (Part 4) Pretreatment for Removal

Toshiaki HISAMITSU†1)*, Kazuyuki GOMYO†1), and Fumio MARUYAMA

Central Research Lab., Nikko Kyodo Oil Co., Ltd., Niizo-minami 3, Toda, Saitama 335

(Received February 8, 1993)

With the objective of selecting suitable reaction conditions and catalysts for a guard reactor used in shale oil hydrorefining for arsenic removal, two kinds of Mo-Ni-Al2O3 catalysts and γ-alumina have been evaluated. It was confirmed that most of the arsenic present in shale oil is removed under such relatively

mild reaction conditions as 300℃ and 2.0 LHSV, using conventional Mo-Ni catalysts for hydrorefining. All arsenic removed from the shale oil deposited on the catalyst, and no arsenic was detected in the effluent gas from the reactor. The arsenic deposition profiles in the catalyst pellets vary depending on the reactivity and diffusibility of arsenic compounds. Suitable catalysts for the upper layer in the guard reactor should be provided with medium initial activity for arsenic removal, with larger pore diameter, and with the shape of pellets with a greater ratio of external surface area to volume. The metal contents of the catalyst should be increased commensurate with the expected amount of arsenic deposition.

1. Introduction In one of our preceeding paper1) we have reported that hydrodenitrogenation (HDN) activ- Shale oil is rated one of the most promising ity of the catalyst substantially decreased with time hydrocarbon alternatives to petroleum because of on stream, and this deactivation was ascribable to its huge amounts of reserves as well as its relatively the arsenic compounds in the shale oil. This rich content comparing with other became evident when a Colorado shale oil, which alternatives such as tar sands oil and liquefied coal. contains 27ppm of arsenic was hydrorefined. Shale oil contains, however, unstable olefinic Curtin2) has disclosed that some part of arsenic compounds, organic compounds, and compounds present in shale oil can be precipitated arsenic compounds in greater amounts than by non-catalytic thermal treatment at temperatures conventional petroleum crude oil. The unstable above 316℃. However, it may be more practical olefinic compounds and the organic nitrogen to remove the arsenic by depositing it on the compounds have deteriorating effects on storage adsorbent or catalyst in a pretreatment reactor stability. The organic nitrogen compounds are before charging the shale oil to succeeding also poisonous to the catalysts used in the catalytic processes. Sullivan et al.3) used alumina downstream processes. pellets for a guard reactor placed upstream of a Hydrorefining is a potential means to remove hydrorefining reactor in order to remove the these harmful compounds and to upgrade the shale arsenic and compounds in a Colorado shale oil to the so-called synthetic crude oil which can be oil. Cottingham and co-worker4) have reported further processed in the conventional petroleum that substantial part of arsenic compounds in a refineries. However, the presence of arsenic com- Colorado shale oil was removed by hydrorefining pounds and unstable olefinic compounds cause over a molybdenum-cobalt catalyst under such hydrorefining more difficult, since the arsenic relatively mild reaction conditions as 0.5 LHSV at compounds poison hydrorefining catalysts, and 316℃ or 1.0 LHSV at 347℃ under 70kg/cm2. the carbonaceous deposition brought about by the Young5) has disclosed that 95% of arsenic in a unstable olefinic compounds fouls both process Colorado shale oil was removed by hydrorefining equipment and catalyst. over a 5% Mo-33% Ni-Al2O3 catalytst at 3.5kg/cm2, 4.2 LHSV, and 343℃.

* To whom correspondence should be addressed. This work has been done with the objective of †1) (Present address) Mizushima Refinery, Nikko Kyodo Oil selecting suitable catalysts for a guard reactor of Co., Ltd., 2-1 Ushio-dori, Kurashiki, Okayama 712 hydrorefining processes for shale oil. Two kinds

石 油 学 会 誌 Sekiyu Gakkaishi, Vol. 36, No. 6, 1993 480 of molybdenum-nickel catalysts and a γ-alumina 2.3. Hydrorefining support have been evaluated for their activity for Two types of high-pressure unit were used in the arsenic removal. hydrorefining experiments. A microunit, The other objective is to observe the fate of the equipped with a reactor of 10mm in inner arsenic compounds removed from shale oil. The diameter, 300mm in length, and 10ml catalyst arsenic distribution in the catalyst bed and in the loading capacity, was used in Experiment-1 catalyst pellets was investigated after hydrorefin- (evaluation of arsenic removal activity). The ing the shale oil. reaction temperature is defined as the average of three temperatures measured at top, middle, and 2. Experimental bottom of the catalyst bed. A bench unit, equipped with a reactor of 25mm 2.1. Feed Oil in inner diameter, 1,200mm in length, and 100ml The feed oil used here is a vacuum gas oil catalyst loading capacity, was used in Experiment- fraction separated from a Colorado shale oil under 2 (study on arsenic distribution in the catalyst bed reduced pressure. The boiling temperature range and catalyst pellets). The reaction temperature is is from 330 to 550℃ at atmospheric pressure. the average of five temperatures measured along The feed oil properties are as follows: the catalyst bed height at regular intervals. The

Arsenic: 27wtppm, Density: 0.9244g/ml at γ-alumina support was placed in the preheater Nitrogen: 1.14wt%, 15℃, section. The effluent gas from the high-pressure : 0.45wt%, Viscosity: 15.93 cSt at gas-liquid separator was washed by passing Carbon: 85.2wt%, 50℃, through a dilute nitric acid solution of magnesium Hydrodgen: 12.0wt%, Pour point: 27.5℃, nitrate and a dilute aqueous solution of sodium Conradson carbon hydroxide successively, to trap the gaseous arsenic residue: 0.31wt%. products. 2.2. Catalysts The experiments were conducted under the 2.2.1. γ-Alumina Support following conditions: A pseudobohemite alumina was kneaded with dilute nitric acid, then extruded into a cylindrical Experiment-1 Experiment-2 form, dried at 130℃, calcined at 600℃,and broken into pieces about 2 to 5mm in length. The pellet diameter, its specific surface area and pore volume are 0.8mm, 210m2/g and 0.63ml/g, respectively. 2.2.2. Catalyst-A Molybdenum and nickel were impregnated on After Experiment-2 was completed, both the γ- the above γ-alumina support using aqueous alumina support in the preheater section and solutions of ammonium heptamolybdate and catalyst was discharged from the reactor, and nickel nitrate and calcined at 500℃. The catalyst separated into five equal parts along the bed contains 12wt% Mo and 3wt% Ni. height, and washed with toluene using a Soxhlet's 2.2.3. Catalyst-B extractor. While kneading the same pseudobohemite 2.4. Aanlysis alumina used for the above γ-alumina support, a Elemental analysis of the liquid samples was dilute nitric acid, an aqueous solution of am- carried out using the following apparatus: ni- monium heptamolybdate, and an ammoniacal trogen content by chemiluminescence with a solution of nickel carbonate were added in Dohman model DN-10; sulfur content by burning succession. The dough was extruded into a the sample in an flow in a ceramic tube cylindrical form of 1.0mm in diameter, dried at according to JIS K 2541; carbon and hydrogen 130℃, calcined at 500℃, and broken into pieces contents by a CHN-analyzer with a Perkin-Elmer about 2 to 5mm in length. The catalyst contains model 240-C; and arsenic by f lameless atomic 5wt% Mo and 33wt% Ni. The specific surface adsorption, a Hitachi model 180-80, after burning area and pore volume are 280m2/g and 0.56ml/g, the sample in a bomb. respectively. Elemental analysis of the used alumina and Prior to use, the catalysts were sulfided in-situ catalysts was done using the following apparatus: using a gas oil spiked with 1.0vol% carbon aluminum content by ICP with a Seiko Denshi disulfide at temperatures increasing from 150 to Kogyo model SPS-1100; carbon content by electric 300℃. titration with a Kokusai Denki model Coulomat- ic-C, after burning the sample in a furnace;

石 油 学 会 誌 Sekiyu Gakkaishi, Vol. 36, No. 6, 1993 481

arsenic, nitrogen, and sulfur contents by the same pounds removed over the alumina may be the methods used for the liquid samples. "easy -to-remove" types such as amides and/or The arsenic distribution in a cross-section of the nitriles. used catalyst pellets was measured by X-ray It is generally known that a wide variety of microanalysis (XMA) using a Hitachi model organic nitrogen compounds such as amines, X-500. amides, pyrroles, pyridines, quinolines, nitriles, etc. is present in shale oil, and that these 3. Results and Discussion compounds differ from each other in their basicity as well as in reactivity. According to Ford et al.7), 3.1. Experiment-1 the amide-type nitrogen compounds are pref- Catalyst-A (12% Mo-3% Ni) is a typical hydro- erentially hydrogenated under mild reaction con- refining catalyst for processing heavy oils. And ditions. Skala et al.8) have also reported that Catalyst-B (5% Mo-33% Ni) was prepared accord- lauronitrile was hydrodenitrogenated at 280℃. ing to the patent5) which claimed to provide with a 3.2. Experiment-2 greater arsenic deposition capacity. 3.2. 1. Liquid Product The results of Experiment-1 are summarized in Experiment-2 was conducted at 300℃, 40kg/ Table 1. The arsenic content decreases from 27 to cm2 for 126h over Catalyst-B. The reaction 15wtppm over the γ-alumina at 300℃. How- temperature of 300℃ was decided based on a ever, it almost levels off at higher temperatures. compromise between more arsenic removal and Meanwhile, the arsenic content decreases to 2- less coke formation. The coking propensity of 3wtppm at 300℃ and close to zero at 360℃ with shale oil will be discussed in detail in a succeeding either of the catalysts. Accordingly, the reaction paper9). temperature around 300℃ is sufficient to achieve Table 2 shows that, throughout the experiment, above-90% arsenic removal under the reaction arsenic, nitrogen and sulfur removals are almost conditions employed here. In addition, it seems constant at 90%, 3.5%, and 44%, respectively. that about a half of the arsenic compounds in the Thus, most of the arsenic is removed under mild shale oil is of the type removable without a catalyst reaction conditions under which the HDN reac- and that the other half can be removed only with a tion hardly takes place. Here, as the reaction catalyst. This is in accord with the results of pressure decreases from 80kg/cm2 in Experiment- Curtin2). He has reported that some part of 1 to 40kg/cm2 in Experiment-2, the arsenic and arsenic compounds in shale oil could be removed sulfur contents remain unvaried, but the nitrogen by a noncatalytic thermal treatment. content increases slightly from 0.98 to 1.10wt%. In contrast with the arsenic compounds, the 3.2.2. Used Catalyst nitrogen compounds seem considerably more (1) Arsenic refractory. Even with Catalyst-A, having the As shown in Experiment-1, both Catalysts-A and highest activity, nitrogen removal is only 17% at B, 12% Mo-3% Ni-Al2O3 and 5% Mo-33% Ni-Al2O3, 300℃ and 53% at 360℃. It is interesting to note, exhibited equally high activities in spite of the however, that even over the γ-alumina, 8-9% of significant difference in their metal composition. the nitrogen is removed at 300℃, although no This may indicate that a good initial activity for additional increase is attained by increasing the arsenic removal is obtained by metal loading of temperature up to 360℃. The nitrogen com- only several percent. On the other hand, we have reported in the preceeding paper1) that the

Table 1 Results of Experiment-1 maximum arsenic deposition capacity of 8% Mo-3% Ni-Al2O3 catalyst was 6-7% based on the weight of fresh catalyst, and that the arsenic

Table 2 Results of Experiment-2

Reaction temperature: 300℃, LHSV: 2.0h-1. Reaction pressure: 80kg/cm2, LHSV: 2.0h-1. Reaction pressure: 40kg/cm2, Catalyst-B.

石 油 学 会 誌 Sekiyu Gakkaishi, Vol. 36, No. 6, 1993 482

Table 3 Elemental Analysis of Alumina and Catalyst Used in Experiment-2

poisoned the catalysts presumably by expelling the sulfur atoms, which are constituents of the active sites, from the surface of metal sulfides. In addition, as will be described later, one arsenic atom seems to expel two sulfur atoms from the metal sulfides. This may indicate that the optimum metal content of a catalyst should be decided with due regard to the expected amount of arsenic depo- sition. Here, Catalyst-B with a significantly greater nickel content was tested in an attempt to find out a way to improving the arsenic deposition capacity. As shown in Table 3, the arsenic Fig. 1 Arsenic Distribution in Cross-section of Cata- content of Catalyst-B decreases steeply from lyst Pellets 1.90wt% for the first layer to 0.32wt% for the second layer: then it decreases gradually down to 0.05wt% for the fifth layer. Although the efficacy catalyst pellets. Pazos et al.11) have pointed out of the increased nickel content against the arsenic that the higher the reactivity of metal compounds, deposition tolerance could not be ascertained the greater is the ratio of the chemical reaction rate because of insufficient time on stream, the large to the diffusion rate; hence the metals are more difference in the arsenic content between the first readily deposited on the outer edge of pellets. and second layers, in addition to the high initial Ware and co-worker12) have also reported that activity, may imply that this kind of countermeas- strongly diffusion-limited compounds resulted in ure is a promising approach. a steep metal gradient at the edge of the pellets, and Arsenic distribution profiles in the cross-section less diffusion-limited compounds resulted in of the catalyst pellets are shown in Fig. 1. The relatively uniform metal deposition across the arsenic concentration decreases from the outer pellets. Thus, the metal distribution profile surface toward the center of the pellet in the first varies depending on the reactivity and diffusibility layer where the arsenic content is 1.90wt%. On of the metal compounds involved. the other hand, the arsenic distributes rather Meanwhile, Jeong et al.13) have suggested the uniformly in the pellet from the second layer where presence of both organic and inorganic arsenic the arsenic content is 0.32wt%. compounds, including As2O5 and phenylarsenic Numerous studies have been reported on acid, in a Green River shale oil. They have also hydrodemetalation reactions of vanadium and attributed the fairly constant arsenic levels in the nickel compounds present in petroleum residue. four fractions, separated from the shale oil Sato et al.10) have reported that there are two according to their boiling point ranges, to the dominant patterns of metal distribution in used presence of a uniform mixture of organoarsenic catalyst pellets. One is a uniform distribution, compounds with varying structural complexity. which is typical of nickel deposition on large pore Judging from the different arsenic distribution catalyst pellets. And the other is a U-shaped profiles between the first and second layer catalyst distribution with the edge maximum, which is pellets, the arsenic compounds in the shale oil typical of vanadium deposition on small pore seem to be grouped roughly into two groups based

石 油 学 会 誌 Sekiyu Gakkaishi, Vol. 36, No. 6, 1993 483 on their reactivity. One is an "easy-to-remove" the form of molybdenum and nickel sulfides. type, most of which is removed in the first layer and The sulfur content increases from 9.28wt% for the results in the U-shaped profile. The other is a first layer to 11.2wt% for the third layer; then it "hard -to-remove" type that reacts slowly and stays constant thereafter. This is in line with the results in the uniform profile in the second layer. results observed in the preceeding paper1), where it The significantly greater arsenic deposition in the was pointed out that some part of the arsenic first layer than the total arsenic deposition in the deposited on the catalyst interacted with molyb- other four layers may indicate that the "easy-to- denum sulfide and/or nickel sulfide by its remove" type accounts for the predominant part of substitution of the catalytically active sulfur atoms the arsenic compounds in the shale oil. There- on the metal sulfides. The corrected sulfur fore, catalysts appropriate for the upper layer contents for the five catalyst layers, based on the should be provided with medium activity, with assumption that one arsenic atom substitutes for larger pore diameter, and with the shape of pellets two sulfur atoms, are nearly the same within the with a greater ratio of external surface area to range of 10.7 to 11.3wt%. volume in order to reduce the diffusional Atomic ratios of sulfur (S/C) and nitrogen (N/C) limitations. to carbon for the alumina, and N/C for the catalyst A small amount of arsenic also deposits on are greater than those for the feed oil. This may alumina. And it increases along the bed height imply that the heteroatomic compounds convert to with increasing temperature in the preheater. coke more easily than the hycfocarbon compounds The amount of arsenic contained in the alumina, in the feed oil. Furimsky16) has reported that the catalyst, liquid products, and gaseous products nitrogen content of coke deposited on the catalyst correspond respectively to 6%, 75%, 10% and used for hydrotreatment of carbazol was signif- <0.001% of the total arsenic (1.25g) brought into icantly greater than the sulfur concentration in the the reactor throughout the experiment. The case of benzothiophene. He17) has also attributed missing 9% may be accounted for by the loss of fine the enhancement of coke formation by basic powder rich in arsenic from the outer surface of nitrogen compounds to the interaction with Lewis pellets during the handling of the alumina and the acid sites on the catalyst surface via the unpaired catalyst. electron on nitrogen and to the ability of the sites to (2) Carbon, Sulfur, and Nitrogen remove proton from the catalyst surface. The alumina contains more carbon than the It is also interesting to note that both S/C and N/ catalyst. This agrees with the observation of C for the alumina are greater in the upper layers, Teran et al.14). They have reported that a greater while N/C for the catalyst stays constant through- extent of coke deposited on alumina supports than out the layers. This might have something to do on catalysts when a heavy gas oil, derived from tar with the fact that the coke deposition on the sands bitumen by hydrothermal cracking, was alumina support was decreased by impregnating hydrorefined over them. It should be noted, molybdenum on it, as reported by Teran et al.14). however, that the carbon content of the alumina increases only slightly from the first to the fifth 4. Conclusions layer in spite of the fact that the temperature increases from about 150℃ at the first layer to The present study on hydrorefining of shale oil 300℃ at the fifth layer. The other point to be led us to the following conclusions: noted is that the carbon content of the catalyst 1) Most of the arsenic present in the shale oil is decreases from the top to the bottom layer, in removed by using conventional Mo-Ni-Al2O3 contrast to the observation that it generally catalysts under such relatively mild reaction increases from the inlet to the outlet when used for conditions as 300℃ and 2.0 LHSV. hydrorefining of straight-run heavy petroleum 2) The arsenic removed from the shale oil deposits oils15). Such carbon distribution profiles in the on the catalyst and no arsenic was detected in the alumina and catalyst bed may be attributable to the effluent gas from the reactor. thermally unstable characteristics of shale oil. 3) A suitable catalyst for the upper layer in the This coking propensity calls, particularly from guard reactor should be provided with medium industrial consideration, special attention to deal- initial activity for arsenic removal, with larger pore ing with such problems as fouling of equipment diameter, and with the shape of pellets with a and plugging of the catalyst bed. The next greater ratio of external surface area to volume. paper9) will report the results on of 4) The metal content of the catalyst should be thermally unstable olefinic compounds. decided taking into consideration that each arsenic Most of the sulfur contained in the catalyst is in atom expels two sulfur atoms from the active sites

石 油 学 会 誌 Sekiyu Gakkaishi, Vol. 36, No. 6, 1993 484 on the metal sulfides. 6) Inoue, Y., Gomyo, K., Hisamitsu, T., Ozaki, H., Sekiyu Gakkaishi, 28, (5), 374 (1985). 7) Ford, C. D., Holmes, S. A., Tompson, L. F., Latham, D. Acknowledgment R., Anal. Chem., 53, 831 (1981). This work is a part of the study carried out at the 8) Skala, D. U., Saban, M. D., Jovanovic, J. A., Ind. Eng. Research Association for Petroleum Alternatives Chem., Prod. Res. Dev., 27, 1186 (1988). Development (RAPAD). The authors are grate- 9) Hisamitsu, T., Gomyo, K., Maruyama, F, Sekiyu ful to RAPAD and Nikko Kyodo Oil Co., Ltd. for Gakkaishi., 36, (6), 485 (1993). their permission to publish this paper. 10) Sato, M., Takayama, N., Kurita, S., Kwan, T., Nihon Kagaku Zasshi, 92, 834 (1971). 11) Pazos, J. M., Gonzalex, J. C., Salazar-Guillen, A. J., Ind. References Eng. Chem., Process Des. Dev., 22, 653 (1983). 12) Ware, R. A., Wei, J. J., J. Catal., 93, 122 (1985). 1) Hisamitsu, T., Gomyo, K., Maruyama, F., Ozaki, H., 13) Jeong, K. M., Montagna, J. C., ACS Div. Fuel Chem., 29, Sekiyu Gakkaishi, 30, (6), 404 (1987). (3), 307 (1984). 2) Curtin, D. J., U.S. Pat. US 4 029 571. 14) Teran, M., Furimsky, E., Parsons, B. I., Fuel Process. Tech., 3) Sullivan, R. F., Stangeland, B. E., Frumkin, H. A., Prep. 2, 45 (1979). No. 25-78, API Refining Dep. 43rd Midyear Meeting, 15) Kwan, T., Sato, M., Nihon Kagaku Zasshi, 91, 1103 (1970). 16) Furimsky, E., Ind. Eng. Chem., Prod. Res. Dev., 17, (4), 329 (1978). 4) Cottingham, P. L., Carpenter, H. C., Ind. Eng. Chem., (1978) Process Des. Dev., 6, 212 (1967). 17) Furimsky, E., Erdoel Kohle, Erdgas, Petrochem. 5) Young, D. A., U.S. Pat. US 4 046 674. Brennst-Chem., 35, (10), 455 (1982).

要 旨

シェール オイルの水素化精 製 (第4報)

脱 ヒ素前処理

久 光 俊 昭 †1), 後 明 和 幸 †1), 丸 山 文 夫

(株)日 鉱 共 石 中 央 研 究 所, 335 埼 玉 県 戸 田市 新 曽南3 †1) (現在)(株)日鉱 共 石 水 島製 油 所 試 験 研 究 室, 712岡 山 県倉 敷 市 潮 通2-1

シ ェー ル オ イル を水 素化 精 製 す る 際 の 脱 ヒ素 前 処 理 反 応 器 に れ な か っ た。 ヒ素 の沈 積 量 は 触 媒 床 上 部 で 多 く, 下 方 に行 くに 適 した反 応 条 件 と触 媒 を選 定 す る た め, 2種 類 の モ リブ デ ン- 従 っ て減 少 した。 一方, 触 媒 床 最 上 層 か ら取 り出 した 触 媒 で は ニ ッケ ル-ア ル ミナ触 媒 と1種 類 の γ-アル ミナ の 脱 ヒ素 活 性 を ヒ素 が粒 子 外 表 面 か ら 内部 に 向 か って 減 少 して い るが, 2層 目 評価 した 。 そ の結 果, 石 油 精 製 に使 用 され て い る 通 常 の モ リブ か ら の触 媒 で は粒 子 の外 部 か ら内 部 まで 比 較 的 均 一 に分 布 して デ ン-ニ ッケ ル-ア ル ミナ 触 媒 を使 用 し, 300℃, 2.0 LHSVの い た。 こ の よ うな分 布 の 変化 は, ヒ素化 合 物 の 反応 速 度 と拡 散 比 較 的 温和 な 条件 に て 水素 化 処 理 す る こ と に よ り, 原 料 油 中 に 速 度 の違 い に よ っ て生 じる と考 え られ る 。 した が って, 前 処 理 含 まれ る ヒ素 の90%以 上 を除 去 で き る こ とが 判 明 した。 反 応 器 の上 部 に は適 度 な 活性 を有 し, 細 孔 径 お よ び粒 子 の 外 表 さ らに, 上 記 触 媒 の 一 つ を使 用 して シ ェー ル オ イ ルの 水 素 化 面 積/体 積 比 の 大 き な触 媒 が 適 して い る 。 また, 予 想 され る 運 精 製 を130時 間 行 っ た が, そ の 間 で の 触 媒 活 性 は安 定 して い 転 終 了 時 で の ヒ素 の蓄 積 量 に比 例 して, 触 媒 上 の 金属 担 持 量 を た 。使 用 後 の反 応 器 内部 に お け る ヒ素 の分 布 を調 べ た 結 果, 除 増 やす べ きで あ る。 去 され た ヒ素 の ほ とん どが触 媒 上 に あ り, 排 ガス 中 に は検 出 さ

Keywords Shale oil, Pretreatment, Hydrorefining, Heterogeneous , Arsenic removal

石 油 学 会 誌 Sekiyu Gakkaishi, Vol. 36, No. 6, 1993