Preparation of carbon molecular sieves, I. Two-step hydrocarbon deposition with a single hydrocarbon

C<,rb,,,r, Vol. 31. No. 6, pp. 969-976. Printed m Great Britain.

1993 Copyright

eua8-6223193 G 1993 Pergamon

56.00 + .Kl Press LIJ

PREPARATION OF CARBON MOLECULAR SIEVES, I. TWO-STEP HYDROCARBON DEPOSITION WITH A SINGLE HYDROCARBON A. L. CABRERA,~ J. E. ZEHNER, C. G. COE, T. R. GAFFNEY, T. S. FARRIS. and J.N. ARMOR* Air Products and Chemicals, (Received

19 Junuury

Inc., 7201 Hamilton Blvd., Allentown, 1993; c~ceptrd

in revised,fimn

I

PA 18195, U.S.A.

March

1993)

Abstract-A process is described for preparing a carbon molecular sieve that is suitable for the kinetic separation of gases, such as oxygen from nitrogen. ‘The process involves modifying a carbon support, having a majority of micropores with an effective pore size of about 4.5 to 20 A, using a two-step process in which the sieve is contacted with two different concentrations of a volatile carbon-containing organic compound. The concentration of the carbon-containing compound used in the first step is larger than that in the second step, so that the pore openings of the micropores of the support are narrowed successively in two distinct steps without excessively filling the micropores themselves. Key Words-Molecular

sieve. synthesis.

air separation,

1. INTRODUCTION The use of carbon molecular sieves [CMS] to separate various gases has been known for several decades. Walker[ l] described obtaining carbon mo-

lecular sieves by charring polyvinylidine chloride and vinyl chloride-vinylidine chloride copolymer; additional approaches are described in a recent article[2]. These chars have micropores useful for the separation of hydrocarbons. The sieves can be modified by combining the char with a thermosetting furan resin or lignite pitch and carbonizing the product. Air separation can be carried out over carbon molecular sieve adsorbents, by separating oxygen from air on a kinetic basis through rapid sorption of the smaller oxygen molecules relative to the slightly larger nitrogen molecules. In order to obtain separation, the adsorbent must have pore openings of about the molecular diameter of the larger gas in the mixture (nitrogen in air). This allows for rapid adsorption of the smaller component and slower diffusion of the larger component, resulting in high kinetic selectivity. The ability to control the size of the pore openings on a CMS to exacting specifications, to tenths of an angstrom in the case of air separation, is a major challenge in preparing CMS adsorbents. Since the adsorbent is a key part of the performance of the entire process unit, improved CMS adsorbents are needed to reduce the cost of air separation by pressure swing adsorption (PSA) systems. CMS

can be made from activated carbons by various post treatments, which either narrow pores “Author to whom correspondence should be addressed. +Current address: P. Universidad Catotica de Chile, Fact&ad de Fisica. Castilla 306. Simtiago 22, Chile.

hydrocarbon

decomposition,

carbon.

or shift pore size distribution to produce a material with a bimodal pore distribution having a predominance of pores smaller than 6 &3,4]. Key to the performance of these materials as adsorbents for air separation is their specific size selectivity. One molecular dimension of importance for gas sorption in a kinetic process is the Lennard-Jones kinetic diameter, which is calculated from the minimum equilibrium cross-sectional diameter of the molecute. This measure of molecular size is widely used to relate micropore diameters for many carbon and zeolitic molecular sieves. The kinetic diameter (f~) 0 of 0, is 3.46 A and 3.64 A for N,. (These dimensions are Lennard-Jones F values; Van der Waals values are 2.8 A and 3.0 A, respectively[S].) CMS used for air separation are capable of distmguishing the 0.2 A difference between 0, and N?, which is manifested in the faster adsorption of the smaller 0: molecule. Commercial CMS materials provide a precisely controlled effective pore size distribution in the 3-4 w range to within 0.2 A! Given this degree of precision exhibited by an amorphous material, it is remarkable to note that these materials are made in tonnage quantities throughout the world with excellent quality control. Besides their application to gas separation, these CMS materials are receiving increased attention as catalysts[6,7]. Munzner et al. [S] describe the preparation of carbon molecular sieves (CMS) suitable for the separation of oxygen and nitrogen. This reference describes obtaining CMS for oxygen separation by treating coke with a carbonaceous substance that deposits carbon at 600” to 9OO”C,thereby narrowing the pores present in the coke. The average pore diameter can be adjusted by changing the intensity of the treatment. Suitable carbonaceous substances that can be used in the treatment include benzene, ethylene, ethane, hexane, cyclohexane. methanol,

969

970

A.

L. CABRERA etd.

and the like. Upon pyrolysis, carbonaceous deposits are formed that modify the pore structure of the coal-derived coke, thus improving its separating ability for 0, from air. Pitch, bitumen, tar oil, or gaseous coking materials are used as binders to prepare extrudates. Bergbau-Forschung has published extensively[9-1 l] on their coal-based CMS materials. They commercialized this material on a tonnage scale in the early 1970s. These were reported to be prepared by the pyrolysis of benzene on pretreated coal-based chars. A number of other patents to Japanese companies describe the preparation of similar materials by the deposition of hydrocarbon products onto pretreated coconut charcoal. One Takeda patent[ 121 describes making a carbon molecular sieve by first condensing or polymerizing a phenol or furan resin in the presence of a carbon adsorbent, and then carbonizing the product. Mixtures of the resins can also be used. The resin-forming material is dissolved in water, methanol, benzene, or creosote oil, and the solution is used to impregnate the carbon adsorbent. Carbonizing can be carried out at 400” to 1,OOO”C in an inert gas. This operation is said to reduce the pore diameter of the carbon adsorbent. A 1984 patent to Kuraray Chemical Co.[13] described several processes for narrowing the micropores of activated carbon by precipitating soot in the micropores, and described a method to provide improved selectivity for separating nitrogen from air. The method involved using coconut shell charcoal and coal tar binder, acid washing, adding coal tar, and heating to 950” to 1,OOO”Cfor 10 to 60 minutes. The coal tar penetrates the surface of the active carbon and decomposes to grow a carbon crystahite on the inner surface of the micropore. Later, another Kuraray patent[l4] disclosed a method for making carbon molecular sieves suitable for separating oxygen and nitrogen by using carbon from coconut shells and coal tar or coal tar pitch binder to form particles that are dry distilled at 600” to 9Oo”C, washed with mineral acid and water, dried, and then impregnated with creosote, 2,3-dimethylnapthalene, 2,4-xylenol, or quinoline, and heat treated in inert gas. Both the oxygen adsorption rate and selectivity are claimed to be improved, compared to the use of simple hydrocarbons, such as benzene, pyrolyzed in the gas phase. There are numerous other multi-step synthesis patents that describe pyrolysis of carbonaceous materials onto an activated char[lS]. For example, a recent patent assigned to Kanebo Ltd. describes the preparation of a CMS from the pyrolysis of phenol formaldehyde resins, followed by treatment with starch or polyvinyl alcohol[l6]. All these processes involve the formation of a gate-keeping layer onto the pores of the carbon via the deposition of a pyrolyzable carbonaceous material[17,18]. The gatekeeping layer is a region near the pore opening that is narrowed sufficiently to allow O2 and N, to trans-

verse, and through which 0, passes significantly faster (lo-50 times) than N,. Despite numerous attempts to reproduce several of the methods described above, we could not reproducibly prepare highly effective 02-selective materials without more specific experimental data. Part of the difficulty was in applying these recipes to insufficiently defined carbon supports. The carbon deposition procedure is very much dependent on both the chemical and physical properties of the support. These carbon supports need to be clearly defined in order to successfully apply the carbon deposit that provides a kinetic barrier and imparts 0, selectivity. Earlier, Verma and Walkerr 191 studied carbon deposition via propylene to control pore size. In this manuscript, we describe the pyrolysis of isobutylene for carbon deposition, which has been optimized for various carbon hosts to produce Ozselective adsorbents[20]. We used molecular probe studies[21] along with other physical methods to evaluate the porosity and micropore distribution of a variety of carbon supports. This resulted in the development of controllable processes for preparing these materials based upon the physical properties of the starting material[3,4,20,22-241. 2. EXPERIMENTAL 2.1 Source materials Traditionally, carbons for air separation are pelleted or extruded materials to minimize pressure drop in the large beds of sorbent used in the pressure swing process[25,26]. We have worked with several sources of commercial activated carbons and CMS materials. For the work reported here, we focused on commercial sources of non 02-selective, pelleted activated carbons that were highly microporous. Takeda 4A (CMS-4A) and 5A (CMS-SA) molecular sieves are nonselective sorbents with a substantial microporosity below 10 A, as well as considerable macroporosity above 0.5 pm. A granular, highly microporous coconut shell material from Sutcliffe Speakman (#203C) was also used as a starting material. Commercial samples of CMS for air separation processes are available from several sources. The values reported for the “commercial” CMS (CCMS) described here are typical of commercial CMS (in 1990) used for producing N,; these are commonly referred to as “3A” sieves.

Isobutylene pyrolysis leads to an effective closure of the pores. Isobutylene pyrolysis was performed under a variety of conditions within a microbalance (Cahn 2000) containing the pelleted material (118” cylinders) or pellets crushed to GO/200 mesh (74-105 pm). The carbon sample (-150 mg) was suspended in a quartz basket from a quartz fiber. A type K thermocouple was inserted into a glass sleeve directly below the quartz bucket. The sample

Preparation of carbon molecular sieves-l was evacuated to IO-? torr and refilled with He to 1 atm. The sample was heated to -62S-700°C and treated with the desired level of isobutylene in NI for a fixed period of time. In some cases a second pyrolysis step with 5% isobutylene in Nz at -650°C for a different period of time was carried out. The mass of carbon deposited (-3 mg) was monitored with time. In most cases, modification of the carbon substrate was m~~nitored by measuring the amount of carbon deposited as a mass uptake (in mg). For a dilute isobutylene feed (5% in I-Ie) with 150 mg of CMS-4A, the quartz pan was modified. A hole was made in the middle of the pan to increase the gas flow through the collection of pellets. The samples were heated to 600-675°C in 100 ccimin He at a rate of IOYYmin, held in He at the temperature for I h, and exposed to 5% isobutylene/He for l-2 h. Similar procedures were used for propylene and I, I-dimeihylcyclohexane (DMCH) pyrolysis studies (see Table 2).

Adsorption measul-ements for different carbon materials were carried out in the same Cahn 2000 microbalance at room temperature. Weight uptake curves were obtained for the carbon samples before and after modification. Pellet samples weighing about IS0 mg were suspended from the microbalante using a quartz basket. The samples were then exposed to 100 ccimin He flow, and after a stable baseline was obtained, the gas was switched to O2 or N,. The weight increase was recorded until no further gain was observed. Baseline corrections due to changes in viscosity and buoyancy were also obtained using an empty quartz basket. The 0, and N2 adsorption curves were fitted with an exponential type of curve having the form: W(r) = WN(I - exp( -Kt)), where Wn is the equilibrium loading at experimental conditions and K(0,) is the effective linear driving force mass transfer coefficient for OZ. The ratio of the coefficients for O2 and N2 is a measure of the selectivity of the adsorbent.

Pore size distributions were measured by the molecular probe method[21]. Adsorption of various organic vapors by the carbon sieves over a 24-h period were measured on a McBain-Bakr adsorption balance. About 5 g of powdered sieve were suspended from quartz spring balances in individual glass tubes. The samples were thermostated to 28°C (except for n-butane studies at OCC) and exposed to various organic vapors (P/PO = 0.5). The following liquids were chosen, because they had sufficient vapor pressure at 0 or 28°C over the desired range of pore sizes and with the following densities: CSZ (3.7 A, (I .2610 g/cc); CHzCll (4.0 A) (I .3266 g/cc);

971

n-C,H,,, (4.3 A) (0.6012 g/cc); CHCI, (4.6 A) (I .4832 g/cc); isopentane (4.9 A) (0.6201 g/cc); Ccl, (6.0 A) (1.5867 g/cc). 2.5 Oxygen cupwit> Pellets or crushed sieves (-150 mg) were suspended from the Cahn microbalance in an aluminum dish. The sample was exposed to 100 ccimin of He until a stable baseline was obtained, and then the gas flow switched to pure OZ. The total mass increase due to gas adsorption was recorded. Basetine correction due to changes in viscosity or buoyancy was accounted for by control runs with a dish filled with glass beads. Uptake measurements were made at room temperature (-25°C) and 1 atm pressure (monitored for dramatic weather fluctuations). The measurements were usually repeated in order to assure that the surface was equilibrated with O2 at I atm. Calculations for oxygen capacity in cc/cc were made by multiplying the volume of O2 (using the mass value converted into cc) per mass of carbon by the bulk density (via Hg porosimetry). 3. RESULTS AND DISCUSSION 3. I Properties of curbonsand hydrourrhons A typical micropore size distribution for CMS4A. CMS-SA, and a commercial CMS (C-CMS) used for air separation is displayed in Fig. I. Sutcliffe Speakman’s #203C has a micropore profile similar to that of CMS-SA, but with a slightly larger capacity through the range shown in Fig. 1. Porosity and equilibrium O? capacity data for the commercial carbon sources are given in Table 1. We carried out carbon deposition by pyrolyzing three different hydrocarbons: propylene (4.8 A), isobutylene (5.7 A), or dimethylcyclohexane (DMCH) (>6 A) on three different carbon sieves: CMS-4A, CMS-SA, and a commercial, Oz-selective sieve (Table 2). The products of propylene pyrolysis readily fill the micropores of CMS-4A and 5A. Isobutylene pyrolysis readily penetrates CMS-5A, but has limited adsorption in CMS-4A. The pyrolysis products from isobutylene are not adsorbed significantly by a crushed commercial sieve, but the products of propylene pyrolysis result in limited carbon deposition. Under the conditions in the microbal~~nce, py-

ADSDRPT,ON (CC/G)

3.00

3.50

4.00

4.50

5.00

5.50

PORE DIAMETER, (ii,

Fig. 1. Microporosity of CMS adsorbents.

6

972

A. L. CABRERAef al, Table 1. Properties O2 capacity cclcc, 25°C

C-CMS CMS-4A CMS-SA SS #203C

of commercial carbon sorbents

Pellet density (Hg), glcca

He density, glccb

PV, cc/g=

Micropore volume, cc/gd

1.0 1.0 0.91 0.92

1.96 2.01 2.01 1.91

0.57 0.35 0.40 0.35

0.21 0.17 0.20 0.21

E s:1 7.2

aDensity measured by Hg porosimetry (Micromeritics, from vacuum to 1 atm). bDensity measured by He pycnometry (Micromeritics). CPore volume via Hg porosimetry from 1 atm to 60,000 psi Hg. d[(liHg density)-l/density (He)]; 2.6 A to -30 A.

rolysis of DMCH on CMS-SA did not impart 0, selectivity. The pyrolysis of DMCH results in lim-

ited deposition on CMSJA (to the same level of isobutylene on CMS-4A), and even smaller deposition on CMS-4A (to the same level as propylene in commercial sieve) (Table 2). These results show that the ability of various hydrocarbons upon pyrolysis to deposit carbon is related to their molecular size, relative to the pore size of the carbon support. Although the molecular size of isobutylene is larger than the pore size indicated for CMSdA, optimal carbon deposition nevertheless occurs. This may be related to optimal adsorption of this hydrocarbon at the pore mouth, where cracking may occur, or to fragments of the pyrolyzed isobutylene being more suitable for narrowing the pores on CMS-4A. This is further complicated by the difficulty in assigning precise pore sizes and molecular dimensions for reactions at such high temperatures over such a complex surface.

3.2 One-Step isobutylene CMS-4A and CMS-SA

treatment

of

Pellets of Takeda CMS-SA have a broad pore distribution from 4-5 A. As received, they are not kinetically 02-selective (i.e., N, adsorption occurs as fast as 0, adsorption). These pellets were also

treated with isobutylene under various conditions. The results of oxygen and nitrogen adsorption measurements on CMS-5A after treatment with isobutylene are listed in Table 3. CMS-SA can be made somewhat kinetically selective to 02. In order to improve the selectivity of the materials in Table 3, treatment of CMS-5A with additional isobutylene under another set of conditions eventually produced an oxygen-selective material, but with a substantial loss in capacity. Capacity is critical to the commercial use of any CMS materials, and this loss in capacity is unacceptable. Several experiments were performed on CMSdA, using 20% isobutyleneiHe at different temperatures. Selectivity is easily imparted to this sample when only 1 mg of carbon is deposited at 650°C. A summary of the results is found in Table 4. For a material like CMS-4A, isobutylene upon pyrolysis provides a desirabIe source of carbon. Less desirable source was obtained from propylene; however, desirability depends on the porosity of the host material one chooses. Although selectivity is also imparted to pellets of CMS-5A treated with isobutylene, much more carbon deposition is needed. With CMS-SA, capacity is reduced by 13%, indicating that isobutylene easily penetrates the micropores of CMS-SA, and

Table 2. Pyrolysis of hydrocarbons

Sample C-CMSb

HydrocarbonC

Wttmg) deposited

on CMS-4A and CMS-.5A

r(min)

Deposition rate (~g/min)

propylene isobutylene

1.2 0.02

36 30

33

CMS-4A

propylene isobutylene DMCHa

6.0 4.0 1.0

3 35 42

2000 111 24

CMS-5A

isobutylene DMCHa

17.8 3.6

3 36

5933 98

I

aDMCH = l,l-dimethylcyclohexane. bCommerciaI extrudates crushed to 74-105 ,um. CConditions: 1501200mesh powder first purged with 100 cclmin of He; 100 ccimin of either: 20% isobutylene/He (62S’C), or 20% propylene (SOO’C),or 50 ccimin of 3% DMCH (675°C) in He; DMCH and the gases were at room temperature; the temperatures within parentheses corresponds to the pyrolysis temperatures.

Preparation

Table 3. Pyrolysis of 20% isobutylene/He

wt(mg)

t (min)

as rec’d 650 700 C-CMSa

6.0 6.0

48 12

on CMS-5A

limin

Carbon depositionb T(“C)

973

of carbon molecular sieves-l

MN21

WO2)

1.24 3.25 2.75

0.15

1.42 0.12

02iNz

O2 capacity

Sel

(cc/g)

8 2 23

8.6 7.5 7.9 8.0

“Typical commercial CMS sold for air separation in 1990. bOn 1SOmg of sample.

selectivity is poor compared to that of commercial materials. These experiments are consistent with our current picture that an effective treatment for closing down the pores of these carbons (to impart selectivity) is with hydrocarbon-containing molecules, which upon pyrolysis produce fragments to narrow the pore mouth without filling the micropores. The effective size of the products of isobutylene pyrolysis is similar to the pores in CMS-4A (-4.2 A). Several experiments with a lower level of isobutylene (5%) were performed on CMS-4A at 600 and 65O”C, to a total weight uptake of -1 mg (1 h at 650°C), but low selectivity was obtained. Additional experiments were performed at 625 and 650°C for longer exposure times (2 h), and attractive selectivity values were then obtained (Table 5). These results indicate that the best treatment conditions are a low concentration of hydrocarbon, the lowest possible temperature, and longer exposure time. This one-step procedure on a narrow pore sieve generates material having selectivity and uptake rates that approach those of commercial O?selective materials. However, this approach used the more expensive, narrower-pore CMS-4A. It would be more attractive to use the wider-pore carbon materials (-4-10 A), which are cheaper and more readily available.

3.3 Two-step isobutylene of CMS-5A

treatment

Since isobutylene cracking produces a good O,selective CMS from CMS-QA, but not from CMS-SA, it appeared that the average micropore size of the carbon support was critical. The differ-

ence between the unselective CMS-4A and CMS5A is the average size of micropores: -4.2 A for CMS-4A and between 4-5 A for CMS-SA. We hypothesized that isobutylene is able to impart selectivity to CMS-4A because the pyrolysis of the isobutylene molecule allows one to narrow the pore openings of CMS4A carefully. These observations suggest that the average pore size of CMS-5A has to be significantly reduced before the pore entrance is necked down to impart kinetic selectivity. The reduction of pore diameter can be accomplished by carbon deposition in the interior of micropores. The deposition of carbon in the interior has to be carefully controlled so that gas diffusion is not limited or capacity diminished significantly. Using this concept, we attempted to reduce the pore diameter of CMS-SA before attempting to neck down the entrance to impart O? kinetic selectivity (Fig. 2). To accomplish this we used two distinct steps for hydrocarbon pyrolysis: In the first step, deposition occurs in the interior using relatively harsh conditions (20% isobutylene, 700°C) (Fig. 2B); then in a second step, controlled carbon deposition occurs at the pore entrance under relatively milder conditions (5% isobutylene, 6OO”C, with longer treatment times) (Fig. 2C). Independently, we also developed a two-step procedure using a larger hydrocarbon to deposit carbon to neck down the pores, without excessive pore filling, and then a second treatment with isobutylene to provide the necessary selectivity by preferential deposition at the pore mouth[23]. Experiments were performed to test this twostep concept, and the results are summarized in Table 6. High selectivity values (21) can be imparted by cracking isobutylene on CMS-SA while maintain-

Table 4. Selectivity and 0, rates for CST-4A treated with 20% isobutylene/He Carbon deposition T(“C)

Wt(mg)

f(min)

1.1 1.5 1.0

18 54 9

limin KQ)

K(Nz)

O>lN? Sel

O2 capacity (CC/P)

CMS-4A as received 650

650 700 C-CMS

1.67 0.78 2.69 2.75

0.18 0.03 0.33 0.12

-

7.7

9 24 8 23

8.0 7.4 8.0 8.0

A. L.

974

CABRERA et al.

Table 5. Pyrolysis of 5% isobutylene on CMS-4A

TW)

Exposure (h)

Carbon wtfmg)

0ziN2 Se1

K(G) ( 1imin)

625 650

2 2

1.0 0.85

I8 33

2.31 1.36

A

ing most of the 0, capacity. These conditions need to be optimized to attain acceptable O2 rates. A CMS with a selectivity of 20 and 0, rate of 1.2 min - I might be a viable air separation adsorbent. From the results reported in Table 6, it can be seen that the two-step carbon deposition can yield adsorbents with both high selectivity and high capacity. By using a one-step treatment, the selectivity was not increased beyond 10, and a 6% capacity loss occurred by the slow deposition of carbon. This suggests the pores are too large, and isobutylene pyrolysis easily fills the pores. Hence carbon deposition is high, capacity is lost, and selectivity suffers. With a mild second deposition step, we obtained a material with moderate O2 sorption rates, and good O,/Nz selectivity and capacity. Also, the two-step carbon deposition process was carried out on another commercial coconut she&based, activated carbon obtained from Sutcliffe Speakman (#203C). This adsorbent has an average pore diameter larger than 6.0 ..&,and is also unselective for oxygen. Molecular probe studies indicate a substantial amount of microporosity exists below 6 A. The 203 C coconut shell carbon was treated with 20% isobutylene in nitrogen at 700°C until 3 mg of carbon was deposited. 0, and N, gravimetric adsorption uptake for a typical deposition

B

C

Fig. 2. Depositioaof carbon onto/into a pore. A: uncoated micropore of of isobutylene of carbon via throat (T)

-5 A; B: deposition of carbon via pyrolysis into the throat (T) of the pore; C: deposition the pyrolysis of isobutylene into some of the of B, but mainly at the pore mouth (M).

are shown in Fig. 3. (The materials used for Figs. 3 and 4 are not equilibrium O,-selective; if allowed to reach equilibrium, both uptake values will approach similar levels.) Despite several attempts, it was not possible to improve the O2 selectivity significantly for the 203 C adsorbent by any one-step carbon deposition treatment. A second carbon deposition step was carried out with 5% isobutyiene in nitrogen at 650°C to deposit 3 more mg of carbon (Fig. 4).

Table 6. Two-step isobutylene pyrolysis on CMS-5A First step 20% isobutylene

Example 1

Carbonb deposit wt(mg) 5.8 11 min

2 3 4 5

3.0 3 min 3.0 3 min 2.0 I .5 min

Second Step 5% lsobutylene

(700”Qa

K(O,)’

0,/N,

I!mIn

Seld

Cape (cc/g)

1.57

10

7.6

-

-

-

1

-

I 1

Carbon’ deposit wt(mg) 0.2 1.0 h 4.8 2.2 h 2.9 1.9 h 2.1 1.0 h 3.2 1.9 h

(650°C)”

K(O?) I!mL

OzlNz Seld

Cape (cc/g)

0.64

21

8.3

1.31

10

7.8

0.95

16

8.1

2.47

6

1.23

11

8.4

Note: Initial capacity of nonselective carbon host is 8.8 cc/g. “Treatment temperature. bWeight uptake due to carbon deposition during treatment. This weight increase was measured for I50 mg of carbon host. Deposition time, in minutes, is indicated below the weight uptake entry for each exampie. ‘Mass transfer coefficient for 0, from curve fit to W(r) = Wn(1 - expf - K(O?) 1t) for 0, adsorption. dSelectivity for 0, over NZ derived from mass transfer coefficient ratios (~(O~)/~(N~)). ‘Gas capacity (Wu) of carbon adsorbent after treatment obtained from equilibrium value of O? adsorption at 1 atm and about 25°C.

975

Preparation of carbon molecular sieves-l

sions may be appropriate, as well. The size of these compounds can be varied relative to the pore size of the carbon support and tailored to the size-selective separation desired. 4. CONCLUSION

b

1

2

3

4

5

6

T (MIN)

Fig. 3. O? or Nz adsorption on SS 203C; O2 or NZ uptake in mg on SS 203C, as received at 25”C, with 150 mg carbon: the solid line corresponds to the O2 uptake.

The material had an O2 selectivity of 6, K(OJ was still somewhat fast (1.14 min-I), and the capacity decreased from 7.6 to 6.0 cc/g. Figure 4 clearly indicates a significant increase in kinetic C&/N2 selectivity for the treated adsorbent. in principle, the hydrocarbon that supplies the carbon for narrowing the pore diameter of the support can be any volatile carbon-containing organic molecule, including hydrocarbons and compounds with hetero atoms such as oxygen, nitrogen, sulfur, silicon, and the like, provided that the compound can decompose cleanly to carbon without forming other pore-plugging materials. It is important, however, for the carbon-containing compound under pyrolysis conditions to have an effective molecular dimension smaller than the majority of the pore openings in the untreated carbon support. but large enough to preclude penetration of most of the micropores after the first step. This dimension cannot be measured at a molecular level of the compound because other effects control the ability of the molecule to enter the micropores of the adsorbent; however, the dimension can be estimated empirically by calibration with a CMS of known pore dimensions as determined by using molecular probes. Isobutylene is simply one compound that is useful for this approach. Other compounds, preferably hydrocarbons. having similar minimum molecular dimen-

A two-step pyrolysis treatment with a single hydrocarbon, such as isobutylene, allows a uniform narrow-ing of pore-mouth openings with an effective size in the range of about 4.0 to 6 A. This treatment is superior to any single-step treatment, since it allows a very high percentage of the pore openings to be selectively narrowed for oxygen adsorption relative to nitrogen, without becoming so narrow that either capacity is lost, adsorption becomes too slow, or pores are blocked. We speculate this twostep process is effective because when a hydrocarbon is carbonized in the pores of the carbon. the micropores are narrowed to the point that they discriminate between 0, and Nz by size. Continued coking with a hydrocarbon in large concentration narrows all of the pores, resulting in closure of some pores (at ~3.8 A) by the time the larger pores are narrowed to 3.X-4.0 A. The use of a lower concentration of the hydrocarbon and lower pyrolysis temperature in a discreet second step allows carbon to be deposited at the entrance of the pore mouths. This occurs due to the pyrolyzed hydrocarbon’s inability to penetrate 4.0-4.3 A pores, and hence carbonization occurs at the pore mouth entrances. Through proper matching of hydrocarbon size to the particular carbon support. and employing a twostep process for deposition of the carbonaceous residue, we demonstrated that it is possible to convert less expensive, non-selective carbon sorbents into materials of potential use for the kinetic separation of O? from air. Finally, we demonstrated that by the careful characterization of the host carbon. we were able to prepare O?-selective carbons on different commercial activated carbons by a two-step synthesis. with one hydrocarbon selected for its optimal molecular size appropriate to the pore size of the host material. A~iinoH~ieci~etnerrrs--f)ur CMS work was the result of a substantial team effort by a large group of people (besides the co-authors), including S. A. Auvil, T. A. Braymer, J. D. Moyer, J. M, Schork, and S. R. Srinivasan.

REFERENCES

L. Walker. Jr.. Minerd Indusfui~s (Jan. 1966). P. L. Walker, Jr., Corbon 28, 261 (1990). 5:J. N. Armor, Procrrdings ofthe 20131Birnniul Conference on Cnrhon, Santa Barbara, CA, June 1991, 4. T. S. Farris, C. G. Coe, J. N. Armor, and J. M. Schork, U.S. Patent .5,164,355 (1992). 5. D. W. Wreck, Zeoliw tnolecrclnr sieues, p. 636. John Wiley & Sons, New York (1974). 6. H. C. Foley, Carbon molecular sieves, properties and applications in perspective, in Perspectiuas in ~z~)/e~~~lur sieve scknce (Edited by W. H. Flank and T. E. I. P.

7 0

1

2

3

4

5

6

T (YIN)

Fig. 4. 0: or Nz adsorption after two-step isobutylene pyrolysis in mp on SS 203C . . _ on SS 203C: 0, or N,_ uotake . at 25”C, as received, with 150 mg carbon that has been treated with a two-step pyrolysis of isobutylene; the solid line corresponds to the 0: uptake.

976

7. 8.

9. 10. 11.

12. 13. 14. IS. 16.

A. L.

CABRERA

Whyte, Jr.), pp. 335-360. ACS Symposium Series #368, American Chemical Society, Washington, DC (1988). J. L. Schmitt, Carbon 29, 743 (1991). H. Munzer, H. Heimbach, W. Korbacher, W. Peters, H. Juntgen, K. Knoblauch, and D. Zundorf, U.S. Patent 3,801,513 (1974). H. Jungten, Carbon 15, 273 (1977). H. Jungten, K. Knoblauch, and K. Harder, Fuel 60, 817 (1981). H. Munzner, H. Heimbach, W. Korbacher, W. Peters, H. Juntgen, K. Knoblauch, and D. Zundorf. U.S. Patent 3,%0,769 (1976). Y. Eguchi, K. Itoga, T. Sawada, and H. Nishino, Jpn. Patent Sho 49-37036 (1974). T. Ohsaki and S. Abe, U.S. Patent 4,458,022 (1984). T. Ozaki and T. Kawabe, Jpn. Patent Sho 61-17487 (1986). R. J. Grant. U.S. Patent 3.884.830 (1975). C. Marumo; E. Hayata, and N. Shiomi,‘U.S. Patent 4,933,314 (1990).

ernf.

17. S. P. Nandi and P. L. Waiker, Jr., Fuel 54, 169 (1975). 18. K. Chihara and M. Suzuki, Carbon 17, 339 (1979). 19. S. K. Verma, Carbon 29,793 (1991); also S. K. Verma and P. L. Walker, Jr., European Patent Appl. 0146909 (1985). 20. A. L. Cabrera and J. N. Armor, U.S. Patent 5,071,450 (1991). 21. C. G. Coe, J. D. Moyer, J. N. Armor, and T. R. Gaffney, manuscript submitted for publication. 22. T. R. Gaffney, T. S. Farris, A. L. Cabrera, and J. N. Armor, U.S. Patent 5,098,880 (1992). 23. J. N. Armor, T. A. Braymer, T. S. Farris, and T. R. Gaffney, U.S. Patent 5,086,033 (1992). 24. T. Golden, P. Battavio, Y. C. Chen, J. N. Armor, and T. S. Farris, U.S. Patent 5,135,548 (1992). 25. G. V. Baron, In Gas Separation Technology (Edited by E. F. Vansant and R. Dewolfs), pp. 137-148. Elsevier, Amsterdam (1990). 26. R. V. Jasra, N. V.‘Choudary, and S. G. T. Bhat, Sep. Sci. Techn. 26, 885 (1991).