Development of Porosity in Pittsburgh No. 8 Coal Char As Investigated

Apr 23, 1998 - Contrast-matching small-angle neutron scattering (CM-SANS) can decouple closed porosity from open porosity in materials. This technique...
33 downloads 13 Views 65KB Size
542

Energy & Fuels 1998, 12, 542-546

Development of Porosity in Pittsburgh No. 8 Coal Char As Investigated by Contrast-Matching Small-Angle Neutron Scattering and Gas Adsorption Techniques M. Mirari Antxustegi,† Peter J. Hall,*,† and Joseph M. Calo‡ Department of Pure and Applied Chemistry, University of Strathclyde, 295 Cathedral Street, Glasgow G1 1XL, Scotland, and Division of Engineering, Box D, Brown University, Providence, Rhode Island Received August 26, 1997. Revised Manuscript Received February 12, 1998

Contrast-matching small-angle neutron scattering (CM-SANS) can decouple closed porosity from open porosity in materials. This technique has been applied to Pittsburgh No. 8 coal char, which is known to be nonporous. It is shown that there is broad agreement between CM-SANS and gas adsorption techniques. Scattering in the CM samples was due to ash and internal fluctuations in the neutron scattering density and contributed 5% to the total scattering in dry samples.

Introduction The development of porosity in carbonaceous materials is important to a number of practical applications, and there is a large and detailed literature associated with this subject. The number of experimental techniques on which this literature is based is surprisingly limited. Gas adsorption is by far the most common technique followed by intrusion methods and particle scattering.1 Small-angle scattering (SAS) from porous materials is a powerful technique for the examination of porous solids. The technique is becoming increasingly important as a porosity characterization technique and has been applied to a range of materials such as aerogels,2 glasses,3 polymers,4 and carbons.5 Most SAS studies of porous carbon have used X-ray photons as the scattered objects. There are a number of theories for the interpretation of SAXS data to yield a variety of structural data ranging from pore size distributions6 through mass and surface fractal dimensions7,8 to the radii of gyration for pores of varying geometries.9 Although most SAS investigations of porous materials have used X-rays, there are a number of important differences between the use of X-rays and the use of †

University of Strathclyde. Brown University. (1) Lowell, S.; Shields, J. E. Powder Surface Area and Porosity; Chapman and Hall: London, 1984. (2) Schaefer, D. W.; Olivier, B. J.; Ashley, C.; Beaucage, G.; Richter, D.; Fargo, B.; Frick, B.; Fischer, D. A. J. Non-Cryst. Solids 1994, 172, 647-655. (3) Agamalian, M.; Drake, J. M.; Sinha, S. K.; Axa J. D. Phys. Rev. E 1997, 55, 3021-3027. (4) Hall, P. J.; Ruiz Machado, W.; Gascon Galan, D.; Barrientos Barria, E. L.; Sherrington, D. C. Faraday Trans. 1996, 92 (14), 26072610. (5) Mondragon, F.; Quintero, G.; Jaramillo, J.; Calo, J. M.; Ruiz, W.; Hall, P. J. J. Mater. Sci. 1997, 32, 1455-1459. (6) Sheu, E. Y. Phys. Rev. A 1992, 45 (4), 2428-2438. (7) Schmidt, P. W. J. Appl. Crystallogr.. 1991, 24, 414. (8) Teixeira, J. J. Appl. Crystallogr. 1988, 21, 781. (9) Hjelm, R. P. J. Appl. Crystallogr. 1985, 18, 452-460. ‡

neutrons in small-angle scattering experiments. Primarily, the scattering cross section for X-rays increases with atomic number. Neutron scattering cross sections have no simple systematic variation with atomic number.10 Neutron scattering cross sections can vary greatly between isotopes of the same element. This has two important consequences for the analysis of complex materials such as coal chars. Although contrast matching is possible in SAXS,11 it is much easier for SANS using appropriate isotopic substitution. Also, a major problem in the use of SAXS to investigate the pore structure of coal chars is that scattering from ash can make a large contribution to the overall scattering. This is because the principal elements in the ash (e.g., Fe, Si, Al, and O) have higher atomic weights than carbon and consequently scatter X-rays more intensely. In general, for SAXS experiments on coal chars there will be three contributions to scattering: open porosity, closed porosity, and mineral matter. As mentioned, there is no systematic variation of neutron scattering length, b, with atomic number. For example,10 bcarbon ) 6.6 fm, boxygen ) 5.8 fm, bsilicon ) 4.1 fm, baluminum ) 3.3 fm, and biron ) 9.5 fm. Consequently, there may be an “averaging” in scattering lengths of the ash material, and this may not be such a problem. One objective here is to determine the extent of scattering from mineral matter in SANS experiments by using a coal char with little or no closed porosity. Easy contrast matching gives SANS some unique features in the analysis of porous materials in that it can decouple closed porosity from open porosity. Also, SANS can give information concerning porosity in solvent-swollen materials. With regard to decoupling open and closed porosity in materials, it is well established that in SAXS experi(10) Koester, L. Springer Tracts Mod. Phys. 1977, 80, 1. (11) Hua, D. W.; D’Souza, J. V.; Schmidt, P. W.; Smith, D. Characterization of Porous Solids III. Stud. Surf. Sci. Catal. 1994, 87, 255.

S0887-0624(97)00153-9 CCC: $15.00 © 1998 American Chemical Society Published on Web 04/23/1998

Porosity in Coal Char

ments, and SANS experiments on dry samples, that scattering is from both closed and open porosity.12 In SANS experiments it is possible to fill open porosity with a liquid that has the same neutron scattering cross section as the solid matrix. In these “contrast-matching” experiments, scattering from open porosity is eliminated and, in a two-phase approximation, scattering is from closed porosity. Contrast-matching SANS (CM-SANS) has been applied by Hall et al.4 to investigate the presence of closed porosity in a series of styrene-divinylbenzene resins produced with varying amounts of porogen during polymerization (a porogen is a chemically inert substance added during polymerization to produce porosity within polymers). Contrast matching with a (nonswelling) mixture of ethanol and [2H6]-ethanol reduced scattered intensities in all of the resins, since the contrast-matching liquid eliminated scattering from open porosity. Following contrast matching there was still residual scattering that Hall et al.4 interpreted as scattering from closed porosity. They demonstrated that resins produced with small amounts of porogen had high levels of closed porosity and were therefore ineffective as ion-exchange resins. An example of SANS to investigate the pore structure of a solvent-swollen material was demonstrated by Hall et al.13 who performed SANS on a series of styrenedivinylbenzene resins swollen to different extents with a variety of solvents. For these experiments, the scattering contrast between the solid matrix and the solvent was increased by the use of deuterated solvents. The presence of closed porosity is important to a number of carbons, and the increase of surface area during the production of active carbons may involve the opening of closed porosity. Therefore, the technique of CM-SANS and its ability to decouple open and closed porosity have considerable potential to investigate the development of porosity in carbons. To begin these investigations in a systematic way, the objective is to use CM-SANS on a carbon of known low porosity, to investigate the porosity developed during gasification, and to compare the SANS results with those of BET N2 adsorption experiments. From high-pressure mercury porosimetry experiments on the Pittsburgh char,14 it has been determined that its density is 1.67 g cm-3, which comes close to representing the true “skeletal” carbon density. This corresponds to a neutron scattering density of 5.6 × 1010 cm-2. [2H8]-toluene also has a scattering density of 5.6 × 1010 cm-2, which therefore makes it suitable as a neutron contrast-matching material for the Pittsburgh No. 8 coke without the need for isotopic mixing. Modeling of Scattering Data The analysis SANS data has been discussed in detail by Higgins and Benoıˆt.15 In this discussion the notation of Sheu6 will be used with slight modification. For an

Energy & Fuels, Vol. 12, No. 3, 1998 543

isotropic system the variation of scattering intensity with q for a polydisperse system is given by

∫P(q, R)N0f(R) dR

I(q) )

I(q) is equal to the differential scattering cross section (dΣ/dΩ). P(q, R) is the intraparticle structure factor, f(R) is the normalized size distribution function, R is the size of the scattering objects, and N0 is the number of scattering objects. q is the scattering wave vector defined by

q)

4π sin(θ) λ

(2)

where λ is the neutron wavelength and θ is half of the scattering angle. For a particle of size R, P(q, R) can be written as

P(q, R) ) bv2G(q, R)[Vp(R)]2

(3)

Vp(R) is the volume of the particle and G(q, R) is the scattering kernel. bv is the contrast factor per unit volume between the solid material and the (empty) pore. Guinier16 has defined a normalized intraparticle scattering factor 〈P ˜ (q)〉, defined as

〈P ˜ (q)〉 )

∫G(q, R)Vp2f(R) dR ∫Vp2f(R) dR

(4)

such that eq 1 can be written as

I(q) ) A〈P ˜ (q)〉

(5)

In eq 5, A is a term that contains information about the concentration of scattering particles and will be treated as a fitting parameter in the model. One model for interpreting scattering data from porous materials that has been widely used is a combined fractal and polydisperse sphere. The fractal part of the scattering dominates at low q and accounts for scattering from meso- and macroporosity (this porosity with widths of >20 Å, corresponding to q ≈ 0.05 Å-1). The polydisperse spheres dominate at high q and account for scattering from microporosity. For surface fractals Schmidt et al.7 and Teixeira8 have shown that

I(q) ) I0Γ(5 - D){sin[π(D - 1)/2]}q-(6-D)

(6)

I0 is a constant, and Γ(5 - D) is a γ function. D is the surface fractal dimension. For spherical particles G(q, R) takes the following form:16

G(q, R) ) (12) Mahajan, O. P. In Coal Structure; Meyers, R. A., Ed.; Academic Press: New York, 1982. (13) Hall, P. J.; Gascon Galan, D.; Ruiz Machado, W.; Mondragon, F.; Barrientos Barria, E.; Sherrington, D. C.; Calo, J. M. Faraday Trans. 1997, 93 (3), 463-466. (14) Hall, P. J.; Calo, J. M. Abstr. Pap.sAm. Chem. Soc. 1990, 35 (3), 705-712. (15) Higgins, J. S.; Benoıˆt, H. C. Polymers and Neutron Scattering; Oxford Science Publications: Oxford, 1994.

(1)

[

]

3j1(qR) qR

2

(7)

j1 is the first-order Bessel function. The sphere radii have a Schultz distribution6 (16) Guinier, A.; Fournet, G. In Small Angle Neutron Scattering; Walker, C. B., Kudowitch, K. L., Eds.; Wiley: New York, 1955; pp 1923.

544 Energy & Fuels, Vol. 12, No. 3, 1998

f(R) )

[ ]

z+1 R2 Γ(z + 1) R0

z+1

[

exp

Antxustegi et al.

]

-(z + 1)R R0

(8)

R0 is the mean sphere radius, and z is a variance parameter that characterizes the polydispersity. Sheu6 has given an explicit statement for 〈P ˜ (q)〉 for spheres with a Schultz size distribution as

〈P ˜ (q)〉 ) 8πR(z+1){R-(z+1) - (4 + R2)-(z+1)/2 cos(ζ1) + (z + 1)(z + 2)[R-(z+3) + (4 + R2)-(z+3)/2 cos(ζ3)] 2(z + 1)(4 + R2)-(z+2)/2 sin(ζ2)} (9) where

R≡

z+1 qR0

(10)

and

ξi ) (z + i) tan-1(2/R)

(11)

Equations 5, 6, and 9 enable scattering data to be easily fitted to the fractal/polydisperse sphere model treating D (eq 6), A (eq 5), R0, and z (eq 9) as fitting parameters. A Marquardt fitting routine was used.17 One of the most useful parameters for the analysis of porous materials is the Porod invariant (PI), defined as

PI )

∫q2I(q) dq

(12)

PI is related to the void fraction of the material under investigation, φ, by15

2π2bv2φ(1 - φ) )

1 V

∫0∞I(q)q2 dq

(13)

V is the sample volume under investigation. The relationship between φ and the interfacial scattering surface area depends on the assumption of pore geometry and size distribution. In general this is complex especially if, as is generally the case, the system is polydisperse. Normally, PI values are not formally calculated from SANS data because of deviations from q-4 behavior due to the fractal nature of the surfaces involved. Also, the integral should be evaluated between 0 and infinity. However, the q range of most instruments means that the maximum in scattering cannot be detected at the lowest q values available. Therefore, PI values calculated from SANS data are under-representations of the true values. Nevertheless, PI values give a useful comparison of how the void fraction of materials changes following treatment. Experimental Section BET surface areas were obtained from adsorption of nitrogen at 77 K using a Quantasorb apparatus over the range 0.01 < P/P0 < 0.95. For all of the chars investigated here, the surface area determined from CO2 adsorption at 273 K was the same as the surface area determined from 77 K N2 adsorption isotherms. (17) Press, W. H.; Flannery, B. P.; Teukolsky, S. A.; Vetterling, W. T. Numerical Recipes; Cambridge University Press: Cambridge, 1988.

Figure 1. Nitrogen adsorption isotherms at 77 K for Pittsburgh No. 8 coal char gasified to different levels of burnoff in air at 673 K. The adsorption for the ungasified char is shown on an expanded scale in the inset. The SANS was performed at the Intense Pulsed Neutron Source (IPNS) at the Argonne National Laboratory at the small-angle diffractometer (SAD).18 The sample holders were made of Suprasil with a path length of 0.2 cm. The scattering data were corrected for the scattering from the sample holder and other instrumental backgrounds. Normalization for the sample thickness and transmission was made, and the data were scaled to yield absolute intensities. A detailed description of the SAD instrument and its calibration have been given by Thiyagarajan et al.18 Pittsburgh No. 8 coal was selected from the Argonne Premium Coal sample program.19 The original coal contained 9.2 wt % ash. The principal constituents of the ash were 25.2 wt % Al2O3, 19.5 wt % Fe2O3, and 45.9 wt % SiO2. The yield of char following pyrolysis was 14%, which implies that the coal char contains 14 wt % ash. Great care was taken to avoid contact with air during the pyrolysis procedure. The coal was heated under a slight positive pressure of nitrogen at 10 K min-1 to 1273 K with a heat soak time of 1 h. The resulting char was ground to between 60 and 100 Tyler mesh. For SANS on the contrast-matched samples, the char was mixed with excess deuterated toluene and placed in an ultrasonic bath for 4 h. Activation of the char was by air with a flow rate of 70 mL min-1 in a tube furnace at 673 K such that there were no mass-transfer limitations of oxygen to the surface of the char.

Results and Discussion Figure 1 shows the 77 K nitrogen adsorption isotherms. For clarity, the isotherm for the ungasified char is shown on an expanded scale. The surface area of the ungasified char was 8 m2 g-1, which suggests no significant open microporosity, or a small amount of macroporosity. This has been previously observed by Hall and Calo14 and is typical of chars from coking coals. The isotherm is of type 2 according to the BDDT20 classification, which is typical for nonporous materials or materials that have pore systems with significant amounts of meso- or macroporosity. As discussed, the gas adsorption gives no indication of the possible existence of closed porosity or whether porosity development proceeds by opening porosity or by developing new porosity. The pore size distributions from the adsorp(18) Thiyagarajan, P.; Epperson, J. E.; Crawford, R. K.; Carpenter, J. E.; Klippert, T. E.; Wozniak, D. G. J. Appl. Crystallogr. 1997, 30, 280-293. (19) Vorres, K. S. Energy Fuels 1990, 4, 420-426. (20) Brunauer, S.; Deming, L. S.; Deming, W. S.; Teller, E. J. Am. Chem. Soc. 1940, 62, 1723.

Porosity in Coal Char

Figure 2. Pore size distributions obtained from the 77 K nitrogen adsorption data of Figure 1 derived according to the Roberts method.20

Figure 3. Small-angle neutron scattering from unactivated Pittsburgh No. 8 coal char: dry (a); contrast-matched by mixing with deuterated toluene (b); the difference (c) between scattering curves a and b.

tion isotherms calculated by the Roberts21 method are shown in Figure 2. The unactivated carbon has a fairly broad distribution peaking at around 25 Å. At 0.8% burnoff, there is a broad distribution of large porosity and the development of considerable microporosity. The 4.5% burnoff sample shows that the distribution has shifted dramatically to the smaller pores of