Refinery streams are typically desulfurized by the Claus process wherein elemental sulfur is produced by reacting hydrogen sulfide and sulfur dioxide in the presence of a catalyst. The Claus system uses a combustion chamber which, at 982°C to 1538°C (1800°F to 2800 ^β F) , converts 50 to 70% of sulfur contained in the feed gas into elemental sulfur. Sulfur is condensed by cooling the reaction gas to a temperature below the dew point of sulfur after which the remaining gas is heated and further reacted over a catalyst. Normally, the gas passes through at least two such Claus catalyst stages.

The different stages of the process may be represented by the following equations: H ₂ S + 3/2 0 ₂ → S0 ₂ + H ₂ 0

2H ₂ S + S0 ₂ -→ 3S + 2H ₂ 0 The overall reaction is:

3H ₂ S + 30 ₂ -» 3S + 3H ₂ 0

The final Claus exhaust gas still contains small amounts of H_S, S0 ₂ , CS , carbon oxysulfide, CO, and elemental sulfur in the form of a vapor or mist. The exhaust gas can be subjected to post-combustion to convert substantially everything to SO and then further purified by Claus after-treatments. An after-treatment process which oxidizes all sulfur compounds into SO-, is disclosed in U.S. Patent No. 3,764,665 which disclosed a process for removing sulfur oxides from gas mixtures with a solid acceptor for sulfur oxides wherein the solid acceptor is regenerated with a steam-diluted reducing gas and the regeneration off-gas is fed to a Claus sulfur recovery process. The regeneration off-gas is cooled to condense the water vapor contained therein, the cooled off-gas is contacted with a sulfur dioxide-selective liquid absorbent, and the liquid absorbent is passed to a buffer zone and then to a stripping zone wherein the absorbed S0_ is recovered from

the liquid absorbent and is supplied to the sulfur recovery process. By operating in this manner, fluctuations in the sulfur dioxide concentration of the regeneration off-gas are leveled-out and a relatively concentrated sulfur dioxide stream is supplied to the sulfur recovery process at a relatively constant rate.

Although the process of U.S. Patent No. 3,764,665 supplies relatively concentrated sulfur dioxide to the sulfur recovery process at a substantially constant rate, the off-gas must be cooled and the liquid absorbent must be transferred to a buffer zone before the absorbed SO can be stripped. Therefore, what is needed is a simpler process whereby these steps are eliminated and energy costs reduced. U.S. Patent No. 5,229,091 relates to a process for extracting sulfur from a gas containing hydrogen sulfide and sulfur oxides.

This invention is directed to a process for removing low concentrations of sulfur from a gas stream. In accordance with a broad aspect of the present invention there is provided a multi-bed process of removing sulfur compounds from a gas stream including sulfur oxides and/or other combustible sulfur-containing compounds. The process comprises the steps of combusting the other combustible sulfur-compounds when present in the gas stream with air or oxygen to convert such sulfur-containing compounds to sulfur oxide, and form a sulfur oxide enriched gas stream. The sulfur oxide enriched gas stream is contacted with first and second serially connected solid adsorbent beds of the multi-bed system for adsorbing therein the sulfur oxides in the form of inorganic sulfates or sulfur oxides or combinations thereof. A third adsorbent bed is contacted with a reducing gas stream to regenerate the third adsorbent bed by reducing the retained inorganic sulfates or sulfur oxides or combinations thereof to hydrogen sulfide and/or sulfur dioxide, and thereby form a

hydrogen sulfide and/or sulfur dioxide bearing stream. After regeneration and if necessary, the third adsorbent bed is placed in a standby mode until time for realigning the beds. The first and second serially connected beds are contacted with the sulfur oxide stream until adsorption distribution in the first and second beds indicates a need for realignment. Then the feeds to each of the beds are realigned to place the second and third beds in series with the sulfur oxide and/or sulfur dioxide enriched stream being fed to the second bed and to place the first bed in a regenerative mode. The foregoing steps are repeated whereby each of the three beds are sequentially placed in an initial bed mode in the series, a regenerative mode, and a second bed mode in the series. In accordance with a specified aspect of the invention, the process further comprises recovering sulfur from the hydrogen sulfide and/or sulfur dioxide bearing stream.

In accordance with another broad aspect of the present invention there is provided a multi-bed system for removing sulfur compounds from a gas stream which includes sulfur oxides and/or other combustible sulfur-containing compounds comprising means for combusting the other combustible sulfur-compounds when present in the gas stream with air or oxygen to convert the sulfur-containing compounds to sulfur oxide, and thus form a sulfur oxide enriched gas stream; and means for contacting the sulfur oxide enriched gas stream with first and second serially connected solid adsorbent beds of the multi-bed system for adsorbing therein the sulfur oxides in the form of inorganic sulfates or sulfur oxides or combinations thereof. The system also includes means for contacting the third adsorbent bed with a reducing gas stream to regenerate the third adsorbent bed by reducing the retained inorganic sulfates and/or sulfur oxides to hydrogen sulfide and/or sulfur dioxide, and thereby form a hydrogen sulfide and/or sulfur dioxide

bearing stream. The system further provides means for placing the third adsorbent in a standby mode if necessary after regeneration until time for realigning the beds. Means are also included for realigning the feeds to each of the beds to place the second and third beds in series with the sulfur oxide and/or sulfur dioxide enriched stream being fed to the second bed, and to place the first bed in a regenerative mode. The realigning means sequentially places each one of the three beds as an initial bed mode in the series, a regenerative mode, and a second bed mode in the series.

In another specific aspect, there is provided means for recovering sulfur from the hydrogen sulfide and/or sulfur dioxide bearing stream. Thus the present invention relates to a process for desulfurizing a gas stream, such as Claus tailgas. A set of three beds is used, with two of the three beds in series to form an adsorption train, while the third is being regenerated. Bed switching typically occurs when the first of the two beds in series is nearly saturated with SOx. This approach allows for higher utilization of the solid sorbent capacity. In addition, spare sorbent capacity is always available in case of unit upsets, and the condition of the sorbent is easily monitored. Fig. 1 is a schematic flow diagram of a three-bed system in accordance with the present invention for recovering sulfur from a gas stream including SOx and/or combustible sulfur-containing compounds;

Fig. 2 is a graphic representation of a steep adsorption front case wherein the first serially connected bed is almost totally utilized before frontal break¬ through; and

Fig. 3 is a graphic representation of a shallow adsorption front case wherein the beds are preferably switched before saturation of the first bed, and before frontal break-through of the second bed.

ith reference to Fig. 1, a hydrogen sulfide containing exhaust gas stream 10 comprising hydrogen sulfide from an elemental sulfur recovery unit 11, for example a Claus plant, is fed to a relatively small combustion furnace 12. An air stream 14 and a fuel gas stream 16 are also fed to the combustion furnace 12 at rates sufficient to maintain the temperature in the furnace 12 within the range of 482°C to 760°C (900°F to l,400 ^β F), preferably from 593"C to 704°C (1,100°F to 1,300°F), to completely convert hydrogen sulfide to sulfur oxides (SOx) . A sulfur oxide enriched gas stream from the furnace 12 is fed by a line 19 through a valve 20, a line 21 to a first solid adsorbent bed 22. The outlet 23 of the first adsorbent bed 20 is connected through a valve 31, a line 24, to the inlet of a second solid adsorbent bed 25. The output 26 of the second bed 25 is connected through a valve 27, lines 28, 29 to a stack 30. Thus, the first 22 and second 25 beds are connected in series between the furnace 12 and the stack 30, and adsorb therein the sulfur oxides from the sulfur oxide enriched stream 19. The sulfur oxides are adsorbed in the form of inorganic sulfates or sulfur oxides or combinations thereof, and the sulfur oxide depleted stream is passed to the stack 30 by line 26, valve 27 and line 28. The third solid adsorbent bed 32 is initially in a regeneration mode while the first and second beds are in an adsorbent mode. During regeneration, a stream of reducing gas from a source 33 travels by a line 34, a valve 35 and a line 36 to the third bed to reduce the retained inorganic sulfates or sulfur oxides or combinations thereof to hydrogen sulfide and/or sulfur dioxide and thus form a hydrogen sulfide and/or sulfur dioxide bearing stream which is returned to the Claus plant 11 by a line 37, a valve 38, and a line 39. Since regeneration of the third bed 32 will be completed before the first and second beds have adsorbed

a desired amount of sulfur oxides, the third bed is preferably placed on standby by closing valves 35 and 38. When adsorption distribution in the first and second beds requires that the first bed be removed from the series, the feeds to each of the beds are realigned such that the second bed 25 and the third bed 32 are placed in series between line 19 and the stack by first closing valves 20 and 31, and opening valve 40 to disconnect the first bed 22 from the second bed 25 and to connect the sulfur oxide enriched stream from line 19 to the inlet of the second bed 25. At the same time valve 41 is opened to serially connect the third bed to the second bed and valve 42 is opened to provide a path from the outlet line 37 of the third bed to the line 29 to the stack 30. As the second and third beds 25, 32 are being connected in series, the first bed 22 is placed in a regeneration mode by opening valves 43, 44 to provide a path for the reducing gas from the source 33, the valve 43, the line 22, the first bed 22, the line 23, the valve 44 and the line 39 to the Claus plant 11. The reducing gas then regenerates the first bed 22, and forms a hydrogen sulfide and/or sulfur dioxide stream which is sent to the Claus plant 11 via line 39. When the first bed is regenerated, the first bed is placed on standby by closing valves 43 and 44.

When the second and third beds 25, 32 have an adsorption distribution requiring that the second bed be removed from the series, the feeds to each of the beds are realigned such that the third and first beds 32, 22 are connected in series. The valves 50, 51, 52 to provide a path for the sulfur oxide enriched gas stream 19 through valve 50, line 36, the third bed 32, valve 51, line 53 , to the inlet of the first bed 22, and the valve 52, the line 29, to the stack 30. At the same time, the second bed is placed in a regeneration mode by opening valves 55, 56 to pass reducing

gas to the second bed 25, and return the desorbed hydrogen sulfide and/or sulfur dioxide bearing stream to the burner 12. Thus, each one of the three beds is sequentially placed in an initial bed mode in the series, taken out of the series for a regeneration mode and then placed in a second bed mode in the series.

While in an adsorbent mode, each of the beds 22, 25, 32 operate at a temperature from 482°C to 760°C (900 ^β F to 1,400°F). A temperature of from 593 ^β C to 704 ^β C (1,100"? to 1,300"F) is preferred. The oxygen content of the stream 19 entering the beds in an adsorbent mode is in an amount of from 0.10 to 10 vol%, preferably from 2 to 4 vol%. It is contemplated that the combustor 12 may be operated with excess oxygen supplied by the air stream 14. In this case, air or oxygen may not have to be added to the feed stream 19 to the beds. Pressure within each bed should be maintained at a pressure of from 10.1 kPa (about 0.1 atmosphere) to 1013 kPa (about 10 atmospheres) , preferably from 152 kPa (about 1.5) to 304 kPa (about 3.0) atmospheres. Gas hourly space velocity (GHSV) should be from 500 to 20,000 hr~ , and preferably from 3,000 to 5,000 hr . An additional benefit of operating each bed during the adsorbent mode within these parameters is that any carbon monoxide therein is converted into carbon dioxide which is released into the environment.

The adsorbent in each bed can be in the form of balls, pebbles, spheres, extrudates, channeled monoliths, microspheres or pellets. This sulfur oxide-capturing adsorbent provides absorbers or acceptors which absorb, and collect, or otherwise remove sulfur oxides from the effluent gaseous stream. In one embodiment, the bed contains Mg/Al spinels.

The outlet conduits 23, 26, 37 are monitored by a sensor. A suitable sensor is a Siemens Ultramat 22P infrared analyzer. Of course, as will be understood by

those skilled in the art, other comparable analyzing equipment can be used.

Sulfur dioxide break-through occurs when a substantial increase in the concentration of sulfur dioxide occurs in one of the lines 23, 26, 37. This increase may be in the order of from about 3 ppm to about 250 ppm in less than about 2 minutes.

During regeneration of each adsorbent bed 22, 25, 32 the temperature is maintained between 482°C (900°F) to 760°C (1,400°F); and the pressure in the reactor 20 is maintained at 10.1 to 1013 kPa (about 0.10 to about 10 atmospheres), preferably 50.6 to 304 kPa (about 0.5 to about 3 atmospheres) . The reducing gas stream 33 is directed into the bed under regeneration is at a gas hourly space velocity (GHSV) of about 10 to about 1,000 hr ^_1 , preferably about 100 to about 150 hr -1. Initially, a GHSV of about 300 hr is preferred when commencing regeneration of a fixed-bed adsorbent so that a higher concentration of liberated gases can be removed from the regenerator. As regeneration proceeds, the GHSV can be reduced to about 50 as the concentration of liberated gases diminishes. Similarly, although hydrogen is the preferred reducing gas for regeneration, other hydrocarbon reducing gases can be used. These will preferably comprise C through C ₅ hydrocarbons. Substantially improved regeneration results are anticipated when water 60 is co-fed into the bed along with the hydrocarbons. The hydrogen and/or hydrocarbon stream 32 may contain 0.0 to 50% water. Syngas, a mixture of CO and hydrogen, may also be used as a reducing gas. Although the embodiments are described with the reducing gas flowing through the beds in the same direction as the sulfur oxide enriched stream flow, it is contemplated that the reducing gas can alternatively be back-flowed through the beds during regeneration. This would ensure that the last part of the bed that adsorption gases see is very active. Also, although the Fig. 1

embodiment is described with reference to Claus tailgas, the invention is useful for any combustible sulfur- containing gas compounds, e.g. flue gases such as FCC flue gas.

Computer Simulations

A computer simulation of adsorption fronts moving through fixed beds was written to explore the benefits of various process configurations. The computer model uses a shrinking core model, cast in cylindrical geometry, to described the SOx uptake by an individual particle. This model is described in reference works such as "Chemical Reactor Analysis and Design" by G.F. Froment and K. B. Bischoff (John Wiley and Sons, New York, 1979) . The parameter values for the simulations were found to fit the breakthrough results in a laboratory reactor, with some allowance for sorbent aging. The operating conditions (e.g. bed length, flow rates, feed concentration) are suitable for a full scale embodiment.

For these experiments, the operating pressure and temperature were 111 kPa (1.1 atmospheres) and 649"C

(1,200°F). Other assumptions were a feed concentration of SO- of 0.43 vol%, gas velocity of 105 cm/sec, and an uptake capacity of the sorbent at long times of 46.5 SOx (as SO ) per gram of fresh sorbent. The bed density was 0.707 g/cm_. The dimensionless parameter of effective diffusivity divided by the particle radius, and by the mass transfer coefficient had a value of 0.11. The maximum SO concentration permitted in the effluent from the system was 1 ppm. In the base case, the sorbent was distributed in two beds each having a 135 cm length. While one bed was on adsorption duty, the other was being regenerated or was on standby after regeneration. It takes a shorter period of time to regenerate than to complete an adsorption cycle. In the second case which is an embodiment of the present

invention, the same amount of sorbent was distributed in three beds, each 90 cm long.

Two-Bed System

From the time the feed was introduced to the first 135 cm bed until the time the effluent SO- concentration reached 1 ppm, was 1,118 min (18.6 hr) . At this point adsorption on this bed was stopped, and the bed was switched to regeneration; while the other (regenerated) bed was switched into adsorption service. The total SOx loading on the first bed at this point was 36.8 wt%, therefore only 79.1% of the total uptake capacity of the sorbent could be utilized in this system. Each bed would be cycled once every 37.2 hr (i.e., 2 x 18.6 hr) . This higher cycling rate results from less than 100% sorbent capacity utilization which can lead to faster sorbent deactivation.

Another disadvantage of the two-bed system of operation is that the beds are not switched until the SO, effluent concentration actually approaches the allowable limit. This makes for vulnerability to excessive emissions if a process upset should occur near the switching time. A partial solution to this problem would be to switch beds well before the effluent SO concentration approaches the emission limit. However, this would have the effect of reducing sorbent capacity utilization even further.

Still another disadvantage of the two-stage system is that if significant SO- breakthrough at the bed exit is not permitted to occur, it can be difficult to monitor the condition of the sorbent, or even know how close the sorption front is to breakthrough. Thus, the operator would not know how much spare capacity is available in the bed.

Three-Bed System

The three-bed system of the present invention solves the above-noted problems with the conventional 2-bed system. For comparison in model calculations, the same amount of sorbent was used as in the two-bed system, but was distributed in 3 beds, each 90 cm long. Two out of the three beds were always in series performing adsorption, while a third bed was on regeneration or standby following regeneration. At the time of valve switching, the first bed in the adsorption train switches to a regeneration mode, the second bed is moved to the first bed position, and the recently regenerated bed becomes the second bed in the adsorption train.

Operation of the three-bed system significantly differs from the two-bed system. As long as the length of the adsorption front within the beds is less than the length of a single reactor, nearly 100% sorbent utilization can always be attained. For this case, which is typical, the preferred time to switch reactors is when the SO_ concentration in the effluent from the first bed reaches 95-100% of the feed concentration. At this point, the first bed is essentially saturated and is no longer sorbing much SOx. The adsorption front is then contained within the front section of the second bed. In other words, the preferred valve switching point is when the adsorption front has nearly completed passage past the exit of the first bed in the series. This contrasts with the 2-bed system where the usual valve switching point is just before the adsorption front starts passage out the exit of the single bed on adsorption duty.

The velocity of the adsorption front as it moves through a bed can be calculated from a simple mass balance:

(gas velocity) (S02 molar feed cone.) (80 g/mol S03)

Front vel.= (sorbent bulk density) (sorbent uptake capacity)

For the conditions used in this simulation, the front velocity was 0.0955 cm/ in. For a 90 cm bed length, the preferred bed switching time (i.e. the time for the front to propagate one bed length) was 15.7 hr. Therefore, each bed would be regenerated every 47.1 hr (=3 x 15.7 hr) . One advantage of the three-bed system is that the sorbent capacity is fully utilized. This translates into less frequent cycling of the sorbent, which can extend sorbent lifetime. Also, more deterioration in sorbent performance can be tolerated, since the intrinsic sorbent capacity is better utilized.

Another advantage of the three-bed system is that the full adsorption front can be allowed to break through the first bed. Monitoring of the timing and shape of this breakthrough curve can furnish valuable information on the state of the sorbent and on other developments such as nonideal flow patterns in the bed.

A third advantage of the three-bed system is that, at the time of bed switching, the adsorption front is contained in the front section of the second bed. The adsorption front, defined as the distance along the bed from a point where the S02 concentration in the gas phase drops below 99% of the feed concentration to where the SO- concentration drops to 1 ppm. For the example herein, the absorption front is about 43 cm long. Thus, at the preferred switching time, this front extended from the front of the second bed to a point 43 cm from its entrance. This left 47 cm (more than half the bed) as spare capacity in case of a process upset. For instance, if the switching event had to be delayed, adsorption could have continued another 8.2 hours before 1 ppm SO started to emerge in the effluent of the second bed.

It should be noted, however, that if subsequent cycles reverted immediately to the normal 15.7 hr switching interval, the adsorption front at the time of valve switching would remain at the exit of the second bed,

leaving no spare capacity for further upsets. In order to move the location of the adsorption front (at the time of valve switching) to its preferred location near the entrance of the second bed, the next several valve switchings must be made at intervals of less than 15.7 hr. The three-bed system of the present invention provides a significant improvement in the configuration of the fixed-bed process, and economically reduces Claus tailgas effluent concentrations of sulfur and CO below the levels attained with current treatment technologies.

Fig. 2 is an example of a steep front case. The initial bed in the series is saturated to the exit of the bed. In this case the adsorption front is steep enough that it does not extend more than half-way into the second bed. This case provides almost total utilization of the initial bed, and then the initial bed is switched out of the adsorption mode and into a regeneration mode.

In a shallow adsorption front case shown in Fig. 3, which might arise with high gas flow rates and/or large particles, it is possible that the leading edge of the second bed is not yet saturated, e.g. the first bed may only be at an 80% adsorption level. However, in the second case, there still is an advantage by splitting into two beds. If gas flows slower and/or particles are smaller, then the more steep the front. Conversely, the faster the gas and/or the larger the particle, the more shallow the slope of the adsorption rate front.

Strategies for Switching Beds In the case of a very steep adsorption front, the system would switch the first bed from an adsorption mode to a regeneration mode when the effluent from the first bed reached about 95% of inlet concentration, because the first bed is almost totally saturated and there is little frontal penetration into the second bed.

In the case of a very shallow adsorption front, there may be break through at the end of the second even though the first bed is not yet saturated. Thus, there is a need to monitor the exit from the second bed. In the intermediate case, which may be the most common, the effluent from the first bed is monitored and a mathematical model, such as the one used to calculate the results of this application, is used to predict or extrapolate what the shape of the adsorption front is in the second bed. To provide a margin of safety, switching preferable occurs before the adsorption front reaches the end of the second bed, for example at 80% into the second bed.

Leading edge of the sloped adsorption front show adsorption levels at specific locations in the beds. The relatively flat trailing portion of the curve indicates that the bed(s) is saturated.

Adsorbents

Non-limiting examples of suitable solid adsorbents for use in the present invention include the porous solids, alumina, silica, silica-alumina, natural and synthetic zeolites, activated carbon, spinels, clays, and combinations thereof. Gamma alumina, chi-eta-rho alumina, delta alumina, and theta alumina are particularly useful as adsorbents and supports because of their high surface areas.

While alpha alumina and beta alumina can be used as adsorbents, they are not as effective as gamma, chi-eta- rho, delta and theta alumina. One or more oxides of other metals can also be used as adsorbents, either alone or in combination with alumina or as spinels, such as bismuth, manganese, yttrium, antimony, tin, copper, Group IA metals, Group IIA metals, rare earth metals, and combinations thereof. Magnesium aluminate spinels are particularly useful as adsorbents. These may be magnesium or aluminum

rich with magnesium aluminate spinels preferred. Lanthanum and cerium are preferred rare earth metals. Naturally occurring rare earths, such as in the form of baestenite, are also useful adsorbents. Elemental copper or copper compound adsorbents, such as copper oxide adsorbents, can also be used. The copper oxide can be cuprous oxide and/or cupric oxide. Other copper compounds can be used, such as copper (II) sulfate, copper (II) acetate, copper (II) formate, copper (II) nitrate and/or copper (II) chloride. The adsorbents can also be a blend/mixture of high density and low density materials.

Also, a metal or metal oxide may be deposited on the solid adsorbent or may be used alone. The metal or metal oxide part of the adsorbents can be supported, carried and held on a refractory support or carrier material which also provides part of the adsorbents. The support controls the attrition and surface area characteristics of the adsorbents. The support preferably has a surface area greater than about 10 π /g and most preferably from about 50 n /g to about 500 m ₂ /g for best results. Suitable supporters include, but are not limited to, silica, alumina, kaolin or other clays, diatomaceous earth, boria, and/or mullite. The support can comprise the same material as the metal or metal oxide part of the adsorbents. The adsorbents can be impregnated or otherwise coated with at least one oxidizing catalyst or promoter that promotes the removal of nitrogen oxides, the oxidation of SO to SO in the presence of oxygen, and the regeneration of the sorbent. It is believed that SO is more readily adsorbed than SO . One useful catalyst is ceria (cerium oxide) . Another useful catalyst is platinum. Other catalytic metals, both free and in combined form, preferably as an oxide form, can be used, either alone or in combination with each other or in combination with ceria and/or alumina, such as rare earth metals, metals from

Group 8 of the Periodic Table, chromium, vanadium, rhenium,

tungsten, silver and combinations thereof. The promoter can comprise the same material as the adsorbent. An even distribution of the promoter is preferred for best results and to minimize adsorbent erosion. Useful Group IA metals include lithium, sodium, potassium, rubidium, and cesium. Useful Group IIA metals include magnesium, calcium, strontium, and barium. Useful Group VIII metals are the Group VIII noble metals (the platinum family of metals) including ruthenium, rhodium, palladium, osmium, iridium, and platinum. The rare earth metals are also useful and are referred to as the lanthanides. Suitable rare earth metals include cerium, praseodymium, neodymium, samarium, europium, gadolinium, terbium, dysprosium, holmium, erbium, thulium, ytterbium, and lutetium.

Preferably, the promoter may be selected from the rare earth metals, the platinum group metals and mixtures thereof. Particularly good results are achieved when the promoter is cerium and/or platinum, with cerium giving outstanding results.

A second promoter, if present, may be selected from the metal or the metal oxide form of iron, nickel, titanium, chromium, manganese, cobalt, germanium, tin, bismuth, molybdenum, antimony, vanadium and mixtures thereof. More preferably, the second promoter is selected from iron, nickel, cobalt, manganese, tin, vanadium and mixtures thereof. Additional metals may be also incorporated into the sorbent. For example, the sorbent may include small or trace amounts of additional metals or metal oxides, such as lanthanum, iron, sodium, calcium, copper, and titanium.

The specific amounts of the promoters included in the solid sorbent, if present at all, may vary widely. Preferably, the first promoter is present in an amount between 0.001% to 20% by weight, calculated as elemental metal, of the solid sorbent, and the second promoter is

present in an amount between 0.001% to 10% by weight, calculated as elemental metal, of the solid sorbent. Preferably, the solid sorbent includes 0.1% to 20%, more preferably 0.2% to 20%, and still more preferably 0.5% to 15%, by weight of rare earth metal, calculated as elemental metal. Of course, if a platinum group metal is employed in the solid sorbent, very much reduced concentrations (e.g., in the parts per thousand to parts per million (ppm) range) are employed. If vanadium is included as the second promoter, it is preferably present in an amount of 0.01% to 7%, more preferably 0.1% to 5%, and still more preferably 0.5% to 2% by weight of vanadium, calculated as elemental metal.

The promoters may be associated with the solid sorbent using any suitable technique or combination of techniques; for example, impregnation, coprecipitation, ion-exchange and the like, well known in the art. Also, the promoters may be added during synthesis of the sorbent. Thus, the promoters may be an integral part of the solid sorbent or may be in a phase separate from the solid sorbent (e.g., deposited on the solid sorbent) or both. These metal components may be associated with the solid sorbent together or in any sequence or by the same or different association techniques. Cost considerations favor the preferred procedure in which the metal components are associated together with the sorbent. Impregnation may be carried out by contacting the sorbent with a solution, preferably an aqueous solution, of the metal salts.

It may not be necessary to wash the sorbent after certain soluble metal salts (such as nitrate, sulfate or acetate) are added. After impregnation with the metal salts, the sorbent can be dried and calcined to decompose the salts, forming an oxide in the case of a nitrate, sulfate or acetate. The above-mentioned adsorbents are discussed in U.S. Patent No. 4,692,318.

In one general aspect, the present invention may involve use of a sorbent which is represented by the following empirical formula: Mg Al 0 , where the atomic ratio of x to y ranges from 0.1 to 10, and where z is at least as required to accommodate the valances of the Mg and Al components of the sorbent. This sorbent may have the spinel structure, and may contain one or both promoters described above.

Metal-containing spinels according to the above empirical formula that are useful in the present invention include the alkaline earth metal spinels, in particular magnesium (first metal) and aluminum (second metal)- containing spinel. Other alkaline earth metal ions, such as calcium, strontium, barium and mixtures thereof, may replace all or a part of the magnesium ions. Similarly, other metal ions, such as iron, chromium, vanadium, manganese, gallium, boron, cobalt, Group IB metals. Group IV metals. Group VA metals, the platinum group metals, the rare earth metals, Te, Nb, Ta, Sc, Zn, Y, Mo, W, Tl, Re, U, Th and mixtures thereof, may replace all or a part of the aluminum ions, preferably only a part of the aluminum ions.

The metal-containing spinels useful in the present invention may be derived from conventional and well known sources. For example, these spinels may be naturally occurring or may be synthesized using techniques well known in the art. Thus, a detailed description of such techniques is not included herein. A particularly useful process for preparing the solid sorbent is presented in U.S Pat. No. 4,728,635. The Group IA, IIA, IB metals, Group IIB metals, Group IV metals, Group VA metals, Group VIA, and Group VIII metals referred to herein are those listed in the Periodic Table of the Elements in the Handbook of Chemistry and Physics (61st Edition) . Free magnesia and/or alumina (i.e., apart from the alkaline earth metal containing spinel) also may be

included in the present solid sorbent, e.g., using conventional techniques. For example, in one embodiment, the solid sorbent preferably includes about 0.1% to about 30% by weight of free magnesia (calculated as MgO) . As mentioned above, potential solid adsorbents are magnesia rich, magnesium aluminate spinels. One example of such a spinel is a commercial magnesia rich, magnesium aluminate spinel containing 0 to 100 wt.% excess magnesia, 5 to 15 wt.% cerium, and 1 to 5 wt.% vanadium. These adsorbents are substantially described in U.S. Patent Nos. 4,790,982; 4,472,267; and 4,469,589. Another particularly suitable adsorbent is a magnesium-aluminum spinel with excess magnesia (MgO) , and with added RE and other metals, particularly with 13 wt% RE, 0.5 wt% La, 13 wt% Ce0 ₂ and 1.5 wt% V in a 1.59 mm (1/16") extrudate. In general, the magnesium aluminate spinels useful in the present invention may be prepared by methods which are conventional and well known in the art.

The following examples are illustrative of sorbents suitable for use in the reactor beds of the present invention.

EXAMPLE 1 A ceria/alumina sorbent was prepared by impregnating high pore value gamma alumina (3.18 mm (1/8") extrudate from Dycat International) with a solution of 32.7 grams

Ce(NO_) _/r .6H-,0 from Aldrich Chemical Company in 45 grams of water, using an incipient wetness technique. The material was dried for three hours at 120°C (248°F) and calcined one hour at 700°C (1,292°F), in air. The composition was approximately 11% CeO ₂ /Al ₂ 0_. This material was crushed and sieved to 14/60 mesh (API) .

EXAMPLE 2 A magnesium aluminate sorbent was prepared, starting with two solutions. Solution I contained 461.5 grams magnesium nitrate, 68.6 grams of concentrated nitric acid, and 500 mis of water. Solution II contained 209.7 grams sodium aluminate, 10.7 grams sodium hydroxide, and 500 mis of water. To Solution I were added 2 liters of water, and then over a 30 minute period, Solution II. Sodium hydroxide was then added in an amount to bring the pH up to 10.7. The resulting mixture was aged for 16 hours and then filtered. The recovered solids were dried at 170°C (338°F) for 12 hours and sized to 14/60 mesh (API) . This material had a composition of about Mg-,A1 0...

EXAMPLE 3 To make a sorbent with approximately 100 ppm platinum loading, 35 grams of the magnesium aluminate from Example 2 was impregnated using an incipient wetness technique with a solution of 0.013 gram of chloroplatinic acid (37% Pt. assay) in 16 mis of water. The resulting solids were calcined in air at 450°C (810°F) for three hours and sized to 14/60 mesh (API) .

EXAMPLE 4 A sorbent with approximately 10% ceria loading on magnesium aluminate was prepared by adding a solution of 9.71 grams cerium nitrate in 16 mis of water to 35 grams of magnesium aluminate from Example 1, using an incipient wetness method. The material was then dried for three hours at 120°C (248°F), calcined in air one hour at 700°C (1,292°F), and sized to 14/60 mesh (API). To test the sorbents' ability to sorb sulfur oxides from a gas mixture simulating an incinerated Claus tail- gas, 6 grams of each material described in Examples 1-4 were loaded in an 11mm I.D. quartz reactor with a central thermowell. The reactor was placed in a radiant furnace

for rapid heating and cooling. A gas flow of 360 ml/minute with a composition of 1% sulfur dioxide, 4% oxygen, and 95% nitrogen (on a dry basis) was established through the reactor, after the desired sorption temperature was attained. Water, in the amount of 20% of the gas flow, as added by directing part of the feed gases through a saturator held at 65.6 ^β C (about 150°F) .

The sulfur dioxide content in the effluent stream was monitored with a Siemens Ultramat 22P infrared analyzer. A cold trap between the reactor and the analyzer removed most of the water on the effluent stream. Sorption experiments were terminated when the sulfur dioxide level in the effluent exceeded 250 ppm. Sulfur dioxide breakthrough was relatively sharp. In general, the analyzer detected no sulfur dioxide for the first 80-90% of the sorption period. Sulfur dioxide concentration of less than 2 ppm during this portion of the sorption was confirmed by measurements with Drager gas measurement tubes. The calculated weight percentage uptake of sulfur oxide as SO- during the sorption period is reported in the Table below.

Regeneration of the solid sorbent was accomplished by contacting it with hydrogen, which was bubbled through a saturator to obtain about 25% water vapor content. The composition of the off-gas during reductive regeneration was determined by injections on to a Hewlett-Packard 5890 gas chromatograph equipped with a thermal conductivity detector. Usually, both hydrogen sulfide and Sulfur dioxide could be detected in the off-gas, but typically one gas or the other dominated, depending on the sorbent and on operating conditions, as indicated in the following Table.

TABLE

Sorbent Temperature of Wt% Uptake Dominate Sulfur

Material Sorption and During Compound In Re-

Identity Regeneration,°C (°F ) Sorption generation Off-Ga3

Ce0 ₂ /Al ₂ 0 ₃ (Ex. 1) 538 (1,000) 4.8 H ₂ S Ceθ ₂ Al ₂ 0 ₃ 649 (1,200) 6.2 so ₂

Mg ₂ Al ₂ 0 ₅ (Ex. 2) 649 (1,200) 4.7 H ₂ S

Pt/Mg Al _j O (Ex. 3) 649 (1,200) 33.8 ^S °2 Cθ0 ₂ /Mg ₂ Al ₂ θ ₅ (EX. 4) 593 (1,100) 14.7 H ₂ S Ceθ ₂ Mg ₂ Al ₂ 0 ₅ 649 (1,200) 25.2 SO.

The uptake of SO was greater for Mg Al_0_ promoted with Pt (Ex.3) and with CeO_ (Ex.4) was higher than for Mg ₂ Al_0 ₅ alone (Ex.2). For the ceria-promoted materials of Examples 1 and 4, magnesium aluminate was a more effective sorbent than alumina, and increasing the operating temperatures from 538°C to 649°C (1000°F to 1200 ^β F) (Ex.l), and from 539°C to 649°C (1100 ^β F to 1200 ^β F) (Ex.4) increased SO sorption which shifted the dominant off-gas sulfur species from H_S to SO_.

EXAMPLE 5

The carbon monoxide oxidation activity of two sorbents was tested by flowing a mixture of 4% carbon monoxide, 4% oxygen, and 8% carbon dioxide at a flow rate of 310 ml/min over 6 grams of each material in an 11 mm I.D. quartz reactor. Carbon monoxide and carbon dioxide concentration, as a function of reactor temperature, were monitored by Beckman Model 864 infrared analyzers. With the magnesium aluminate of Example 2, carbon monoxide was half converted at 410°C (about 770°F) and substantially all converted at 460°C (860 ^β F) . With the platinum-promoted magnesium aluminate of Example 3, carbon monoxide was half converted at 266°C (about 510°F) and substantially all converted at 282°C (540°F) . With an empty reactor, there was no detectable carbon monoxide conversion for temperatures up to 649 ^β C (l,200 ^β F) .

This example demonstrates that the designated sorbents are effective in promoting the removal of carbon monoxide in the presence of oxygen.

While the invention has been described in conjunction with specific embodiments thereof, it is evident that many alternatives, modifications, and variations will be apparent to those skilled in the art in light of the foregoing description. Accordingly, it is intended to embrace all such alternatives, modification, and variations as fall within the spirit and broad scope of the appended claims.