COMPATIBILITY OF SELECI'ED CERAMICS WITH STEAM-METHANE REFO-R ENVIRONMENTS'

James R Keiser and Michael Howell Oak Ridge National Laboratory

P.O. Box 2008 Oak Ridge, TN 37831-6156

Joseph J. Williams$ and Robert k Rosenbera Stone & Webster Engineering Corporation

245 Summer Street Boston, MA 02107

ABSTRACT

Conventional steam reforming of methane to synthesis gas (CO and HJ has a conversion efficiency of about 85%. Replacement of metal tubes in the reformer with ceramic tubes offers the potential for operation at temperatures high enough to increase the efficiency to 98 to 99%. However, the two candidate ceramic materials being given strongest consideration, sintered alpha silicon carbide and silicon carbide particulate-strengthened alumina, have been shown to react with components of the reformer environment. The extent of degradation as a function of steam partial pressure and exposure time has been studied, and the results suggest limits under which these structural ceramics can be used in advanced steam-methane reformers.

Kevwords: silicon carbide, silicon carbide particulate-strengthened alumina, steam-methane reformer, corrosion, oxidation, steam corrosion

INTRODUCTION

Studies sponsored by the US. Department of Energy (DOE), Office of Industrial Technologies, indicate that some high-temperature processes in chemical, petrochemical, and other industries could achieve significantly greater efficiencies if improved materials permitted operation at higher temperatures, higher pressures, and/or in more corrosive environments." Consequently, the high-pressure heat exchange system (HiPHES) program was established by DOE to demonstrate the advantages of higher temperatures and/or pressures in industrial-sized

Research sponsored by the US. Department of Energy, Assistant Secretary for Energy Efficiency and Renewable Energy, Office of Industrial Technologies, Industrial Energy Efficiency Division and Materials for Advanced Industrial Heat Exchanger Program, under contract DE-AC05-%OR22464 with Lockheed Martin Energy Research

t Now with Raytheon Engineers and Constructors, Cambridge, MA 02142. 3 Retired from Stone & Webster Engineering Corporation.

carp.

''The subnutted manuscript has been authored by a contractor ofthe u s gOVanment under contract NO DE-ACOS- 96oR22464 Accordmgly, the U S Government retams a nonexclmlve, myalty-fiee license to publlsh or repro& the published fom o f h s cmtnbutlon, or allow othm to do for U S Govrnmt purposes **

~ -__ ~- - ~ ~ --_ ~

systems. 'Ibo of these projects invohe the assessment of materials for heat exchangers in hazardous, industrial waste incinerators. In another HiPHES project, Stone & Webster Engineering Corporation (SWEC) of Boston, Massachwtts, is developing a high-temperature, high-pressure steam-methane reformer. SWEC proposes to use ceramic tubes in the reformer for containment of the reactant and produd gases because metallic tubes would not have sufficient strength to contain the pressure differential across the wall at the proposed operating temperature. Metal tubes would, however, be used within the ceramic tubes to provide a parallel counter flow path for the product gas as shown schematically in Figure 1. The proposed commercial-sized reformer would contain more than 600 ceramic tubes that would have an outside diameter of about 8.9 an (3.5 in.) and a length, based on design constraints, of about 9 m (30 ft). 'Ihe length of the tube could likely be reduced if a ceramioto-metal joining technique were identilied that would produce a joint capable of operation at higher temperatures and pressures. Although evaluation of joining techniques is an hportant aspect of this study, it is not included in this paper. Rather, the paper focuses on the effects of steam partial pressure and time on the corrosion of the two primary ceramic materials, sintered a-Sic and Sicstrengthened alumina.

BACKGROUND

The oxidation behavior of silicon carbide in oxygen and oxygen-steam mixtures has been the subject of many studies (see, for example, re€ 4). However, comparatively little has been reported on the oxidation behavior of alumina-matrix ceramic composites. The extensive studiesui6 of the reaction of silicon carbide in oxygen or moist air have shown that, under atmospheric conditions, this material oxidizes according to one of the reactions:

and

Sic + 3 H,O(g) = SiO, + 3 Hdg) + CO(g). (2)

The silica layer that forms is generally continuous and adherent and causes further oxidation of the silicon carbide to occur at a progressively decreasing rate. The longtime corrosion of silicon carbide has been shown to follow a parabolic relationship with time.

K however, the oxygen partial pressure is especially low, generally less than l@= atm, another reaction is expected to occur at a significant rate:'

Sic + O,(g) = SiO(g) + CO(g). (3)

Rather than a solid, adherent oxide being formed on the silicon carbide, the volatile corrosion product, SiO, is formed. Formation of this product means that less or no protective solid corrosion product will form on the silicon carbide surface so that further oxidation may not be limited by diffusion through an ever-increasingly thick surface layer of silica. As a result, oxidation of silicon carbide, under these circumstances, can occur at a linear rate resulting in more rapid loss of material from the silicon carbide.

It has also been shown* that in a high partial pressure water vapor, a volatile corrosion product may be formed by one or both of the reactions:

SiO, + 2 H,O(g) = Si(OH),(g). (4)

and

2 SiO, + 3 H,O(g) =F Si,O(OH),(g). (5)

Another studyg found that several atmospheres pressure of water vapor can result in the formation of a volatile species by reaction 4. A situation in which rapid weight change was observed1o that was also likely due to volatilization of the corrosion product occurred in studies conducted at the British Petroleum (BP) Sunbury facility. On the basis of this study, personnel at the BP America facility indicated the corrosion rate of silicon carbide was affected by the steam partial pressure. Tests conducted at 850°C indicated that the corrosion rate increased dramatically for steam partial pressures above about 6 atm.

.

A study conducted by Opila'' established that sodium Contamination in alumina components of the experimental system can cause oxidation rates of silicon and silicon carbide to increase, apparently because of the formation of less protective sodium aluminosilicate scales. Although it was not induded in the study cited, it should be expected that oxidation of alumina strengthened by silicon carbide particulates might be affected by sodium contamination in a similar manner. Care was taken to minimize such contamination in the ORNL tests.

In the early stages of this HiPHES program, the oxidation behavior in moist gas at 1 atm was determined for a number of ceramic materials that were candidates for heat exchanger tubes." This report showed that, at 1260°C in a gas mixture of Ha CO, COB and CH4 at atmospheric pressure and containing 0.295 atm water vapor, continuous, adherent oxide layers formed on the various silicon carbides, while a mullite layer was identitied on a silioon carbide/alumina-matrix composite. The second stage of this study invohred the exposure of the more promising ceramic materials in a series of tests conducted by BP America. These tests were intended to determine the thermal stability of these materials over a temperature range of 593" C (1100°F) to 1260" C (2300" F), at pressures up to 40.8 atm (600 psia), and for times of as long as 500 h. Because of limitations in the experimental system, the maximum temperature of these tests was 1038°C (1900°F), the maximum pressure was 20.4 atm (300 psia), and only one test ran for 500 4 the remainder ran for about 100 h. Most of the samples from these tests were examined at Oak Ridge National Laboratory (ORNL), and the results were presented in a previous publication.u Some analyses were also conducted by BP America, and their results were presented in a report that had restricted circulation."

MATERIALS AND SPECIMENS

Based on results of previous studies that were conducted to identi@ the most promising ceramic materials,'2 the tests described in this report were conducted using a silicon carbide ceramic, Hexoloy SA made by The Carborundum Company"), and a silicon carbide particulate-strengthened alumina ceramic composite manufactured by DuPont Lanxide Compositescz, using their patented directed metal oxidation process. The Carborundum product is pressureless-sinteredy alpha silicon carbide that is approximately 99% pure and has a density of about 3.10 g/cm3 (theoretical density is 3.210 g/cm3). The DuPont Lamides Composites product contains roughly 45 to 50 vol % silicon carbide particles, 10-15% unreacted aluminum alloy, and the balance alumina. One problem encountered with this latter material is seepage of molten aluminum alloy to the sample surface during elevated-temperature tests. In an effort to prevent molten alloy from migrating to the surface, one of the h a l steps taken by the manufacturer during sample preparation is to coat the sample surface with an aluminosilicate barrier coating. This has proven to be effective in containing the aluminum alloy.

For the corrosion tests, flexure bar samples, 3 x 4 x 50 mm, were used. These flexure bars were weighed before and after exposure, and then four-point flexure tests were conducted at the exposure temperature. Selected specimen cross sections were examined using light and electron microscopy as required.

EXPERIMENTAL, CONDITIONS

The test system used for these studies was initially constructed for static atmosphere tests at temperatures as high as 1260°C (2300°F) and pressures up to 20.4 atm (300 psia). Because a need arose for the tests to be conducted in a dynamic atmosphere, a system was designed and constructed to permit a premixed gas to be combined with a measured amount of steam to simulate the atmosphere in any section of a steam reformer.

For these studies, a mixed gas with composition 18.4% CO, 5.4% CO, 0.4% CH, and balance H, was routed though a mass flow controller to regulate the amount of gas being fed into the system. A pneumatically driven metering pump was used to control the amount of high-purity distilled water that was introduced into the evaporator portion of the test system. The amounts of mixed gas and distilled water supplied to the system were manuaIIy adjusted to give the desired proportion of water and gas. This test mixture was fed individually to the ceramic tubes used as test samples and as sample holders.

The ceramic tubes were spm-a€iy constructed to fit this test system. The closed-one-end tubes were 8.9 cm

The Carborundum Company, Structural Ceramics Division, Niagara Falls, NY 14302 ('1 DuPont Lanxide Composites, In&, Newark, DE 19714

(35 in.) in outside diameter, had a length ranging from 96.5 to 121.9 an (38 to 48 in.) and had an expanded section at the open end to provide a sealing surface and a means to hold the tube. niiS expanded section accx>mmodated a stainless steel flange such that the pressurized gas mixture could be fed to and removed from the tubes at pressures as high as 20.4 atm. Effluent gas from the ceramic tubes was conducted to a pneumatically driven back pressure control valve which was controlled through a feedback loop that incorporated a pressure transducer. Water in the effluent gas was condensed by passing the gas through an aircooled mil. This water was collected and measured, and the volume of the remaining effluent gas was determined by passing the gas through a calibrated flow meter. Within each ceramic tube the flexure bars were positioned horizontally in side holes in a recrystallized alumina tube that acted as a support such that they were exposed to the gases flowing from the bottom to the top of the tube.

Tests have been conducted over a wide range of conditions; temperatures have covered the range of 816 to 1 W C (1500 to 2300°F) while steam partial pressures have ranged from 2.04 atm (30 psia) to 6.12 atm (90 psia). The total system pressure was U.6 atm (200 psia) for all but one test which was conducted at a slightly lower pressure. Most of the tests conducted lasted for 100 h, although one test lasted for 500 h. Testing program wil l be continued with exposure times up to 6OOO h planned.

RESULTS AND DISCUSSION

Figure 2 shows the measured weight changes as a function of steam partial pressure for 100-h exposures conducted at 1177°C (2150°F) with a total pressure of 13.6 atm (200 psia). The calculated oxygen partial pressures for these tests are shown in Table 1. These results indicate that an increase in the steam partial pressure had little or no effect on the weight change of the silicon carbide particulate-strengthened alumina, but the same increase in pressure resulted in significant increases in weight gains by the silicon carbide samples. This effect of steam partial pressure on weight gain is reflected in the scale thicknesses measured on the samples and shown in Figures 3 and 4. The micrographs shown in Figure 3 for silicon carbide show that the scale thickness increased with the steam partial pressure. Figure 4 shows micrographs of silicon carbide particulate-strengthened alumina samples after exposure at various steam partial pressures. There is some slight variation in scale thickness with steam partial pressure, but the bamer coating also accounts for a portion of the scale.

In an effort to determine the effect of longer exposure on the corrosion rate of these materials, a 500-h exposure was conducted with a nominal steam partial pressure of 3.4 atm (SO psia). Because of experimental difficulties during the earlier 100-h test, no results were obtained for the 3.4 atm (SO psia) steam partial pressure test condition, but estimates were obtained for those rates by extrapolation of the data in Figure 2. Using those data points and the ones obtained from the 500-h exposure, the results shown in Figure 5 were produced. These results appear to indicate a decreasing rate of weight gain with exposure time for the silicon carbide particulate strengthened alumina but a nearly linear relationship for the silicon carbide. Plotting the weight gain of the alumina versus the square root of exposure time (see Figure 6) indicates that, for the Limited data adable , this ceramic composite actually may oxidize at a rate slower than predicted by a parabolic relationship.

If one takes into account the weight change results that have been observed and the calculated oxygen partial pressures (shown in Table 1) resulting from the high steam concentrations, the oxidation reactions that are most likely to be occurring on the surface of the silicon carbide sample can be predicted. Because the oxidation rate, as indicated by weight change, of the silicon carbide appears to be linear, rather than parabolic, there could be some mechanism operating to remove some corrosion product. Spallation must always be considered, but the sample appearance, both macro- and microscopically, did not indicate that is occurring at a significant rate. The formation of silicon monoxide, SiO, at a significant rate is considered unlikely because of the relatively high oxygen partial pressure. Consequently, it appears likely that one of the silicon hydroxide compounds shown as the products in Eqs. (4) and (5) forms, along with the conventional oxidation product SiOb This would be consistent with a dependence on steam partial pressure.

In order to better explain the weight change behavior of these materials, especially the silicon carbide, additional testbg is needed. In particular, much longer duration tests are needed to provide more information about the various compounds that may be forming as well as to indicate whether or not scale spallation should be a concern.

CONCLUSIONS

For the test conditions used, these studies indicate that the oxidation rate of silicon carbide, as indicated by weight gain, is a direct function of the steam partial pressure and appears to be linear with time. These observations are consistent with the formation of some volatile silicon hydroxide products. Weight changes in the silicon carbide partidate-strengthened alumina showed little or no dependence on the steam partial pressure and oxidized parabolically. Additional, longer tests are needed to determine if the silica layer on the silicon carbide will remain intact and sewe to limit the rate of oxidation.

ACKNOWLEDGMENTS

Other sponsors contriiuting to the overall program included Gas Research Institute, BP America, and Stone & Webster Engineering cofporation. The authors thank P.F. Tortorelli, LG. Wright, and J.R DiStefano for technical review of this manuscript and I(. Spence for technical editing.

1.

2

3.

4.

5.

6.

7.

a

9.

10.

11.

1 2

13.

14.

REFERENCES

"Assessment of High Pressure Heat Exchange Systems Technology - Vol. I Technology Assessment and Research and Development Needs," DOEYCW40716-2 (Vol. l), U.S. DOE, October 1987.

"Assessment of High Pressure Heat Exchange Systems Technology - Vol. I1 Appendices," DOE/CE/40716-2 (VoL Z), U.S. DOE, October 1987.

"Development of a High Pressure Heat Exchange System (HiPHES) for Reforming of Methane," DOE/ID/12797-1, U.S. DOE, May 1990.

N. S. Jacobson, Journal of the American Ceramics Society 76,l (1993): p. 3.

LUJ.T. Ogbuji and E.J. Opila, Journal of the Electrochemical Society 142,3 (1995): p. 925.

J.k Costello and RE Tressler, Journal of the American Ceramics Society, 69,3 (1986): p. 674.

KE Kim and D.W. Readey, "Active Oxidation of SIC in Low Dew-Point Hydrogen Above 1400eC,' Ceramic Transactions, Vol. 2; Silicon Carbide '87, American Ceramics Society, Westerville, OH (1987): p. 301.

EL Brady, Journal of Physical Chemistly, 57 (1953): p. 706.

A. Hashimoto, Geochimica et Cosmochimica Acta, 56 (1992): p. 511.

RC. Ruhl, BP Research, Cleveland, Ohio, private communication to J.1. Federer and J. R. Keiser, Martin Marietta Energy Systems, Oak Ridge National Laboratory, Oak Ridge, TN and R.A Rosenberg, Stone & Webster Engineering Corporation, Boston, MA, December 11,1990.

EJ. Opila, Journal of the American Ceramics Society, 77,3(1994): p. 730.

J.I. Federer, H.E Kim, and A.J. Moorhead, "Corrosion of Sic and Oxide-Composite Ceramics by a Simulated Steam-Reformer Atmosphere," Oak Ridge National Laboratory, ORNUIU-11828, September 1991.

J.R. Keiser, J.I. Federer, J.J. Williams, and R.A. Rosenberg, "Evaluation of Ceramic Materials for an Advanced Steam-Methane Reformer," CORROSIONIp3, paper no. 239, (Houston, Tx: National Association of Corrosion Engineers, 1993).

AY. Sane, "Corrosion Tests on Ceramic Materials," BP America confidential report, December 1992.

Table 1. Calculated oxygen partial pressure for the gas mixtures used for corrosion tests. For the tests, a mixed gas with a composition of 18.4% CO, 5.4% COB 0.4% CH,, and balance & was combined with steam to give the steam concentrations shown.

t rt t Figure 1. Schematic of heat exchanger tube for Stone & Webster‘s high-pressure, high-temperature steam-methane reforming system.

-“...o”-. 1.2-

1.0-

0.8 - 0.6 - 0.4 -

.+”.” ..... ” ..... 0 0.2 - o.--------

0.0 1 I 1

0 2 4 6 8

STEAM PARTIAL PRESSURE (atm)

Figure 2. Weight change versus steam partial pressure for samples exposed at 1177°C and 13.6 atm total pressure.

.

PP(H20) = 30 psia

PP(H20) = 70 psia

ry(H2O) = 90 psia Figure 3. Micrographs showing effect of steam partial pressure on the corrosion product formed on silicon carbide.

PP(H20) = 30 pia

PP(H20) = 7Opsia

PP(H20) = 90 psia Figure 4. Micrographs showing effect of steam partial pressure on the surface layer on silicon carbide particulate-strengthened alumina.

I - , L

4 n cy

Sic

..-----SO o,-----*----- 160 260 360 460 560 6 IO

EXPOSURE TIME (h)

Figure 5. Weight change versus exposure time for samples exposed at 1177°C to a total pressure of 13.6 atm and a steam partial pressure of 3.4 atm. Data for 100-h exposure extrapolated from 100-h results for 2.04 and 4.76 atm.

4 Sic n

cy

w 0 z 4 I 0 I- S c3 :

3-

2-

1 -

"."...+."" Sic-strengthened A1203

0 5 10 15 20 25

SQUARE ROOT OF EXPOSURE TIME (h 1 '2)

Figure 6. Weight change versus the square root of exposure time for samples exposed at 1177°C to a total presswe of 13.6 atm and a steam partial pressure of 3.4 atm. Data for 100-h exposure extrapolated from 100-h results for 204 and 4.76 atm.

DISCLAIMER

This report was prepared as an account of work sponsored by an agency of the United States Government. Neither the United States Government nor any agency thereof, nor any of their employees, makes any warranty, express or implied, or assumes any legal liability or responsi- bility for the accuracy, completeness, or usefulness of any information, apparatus, product, or process disclosed, or represents that its use would not infringe privately owned rights. Refer- ence herein to any specific commercial product, process, or service by trade name, trademark, manufacturer, or otherwise does not necessarily constitute or imply its endorsement, recom- mendation, or favoring by the United States Government or any agency thereof. The views and opinions of authors expressed herein do not necessarily state or reflect those of the United States Government or any agency thereof.