54 Ultramicroscopy 34 (1990) 54-59 North-Holland

Characterization of filamentous carbon on Ni/MgO catalysts by high-resolution electron microscopy

D a v i d J. Smi th , M.R . M c C a r t n e y

Center for Solid State Science and Department of Physics, Arizona State University, Tempe, A Z 85287, USA

E. T r a c z a a n d T. Borowieck i b

a Central Laboratory, I, Department of Chemistry, Maria Curie Sktodowska University, 20-031 Lublin, Poland

Received 21 January 1990; at Editorial Office 28 May 1990

High-resolution electron microscopy has been used to characterize the filamentous carbon deposits which are obtained during the steam-reforming of n-butane, as catalyzed by Ni supported on MgO. Lower reaction temperatures ( - 400-500 o C) resulted in partially graphitic filamentous fibers, with small crystals, identified as predominantly Ni, usually located at the fiber tip. Higher reaction temperatures ( - 650-700 ° C) gave carbon tubes or shells with a more highly developed graphitic nature, while the microcrystals at the ends of the tubes, identified as mainly NiO, were typically encapsulated by several graphitic sheets. Overall, these results are further confirmation that the morphology of the carbon deposits, and their growth mechanism, is independent of the type of catalyst and the particular catalytic reaction.

1. Introduction

The growth of filamentous carbon deposits is well known to occur during the metal-catalyzed decomposition of hydrocarbons such as acetylene [1], and their occurrence in nuclear reactors has also been reported [2]. The carbonaceous residue has a morphology that is strongly influenced by the particular reaction temperature. Baker and colleagues [3,4], and later Audier [5,6], have done much work attempting to classify and to under- stand the filamentous growth mechanisms. In the present study, we have applied the technique of high-resolution electron microscopy (HREM) to the characterization of the carbon deposits formed over a range of temperatures during the steam-re- forming of n-butane, using MgO-supported Ni as the active catalyst. This support has been reported to reduce the amount of coking normally associ- ated with hydrocarbon steam-reforming [7,8]. Careful off-line measurements of electron micro- graphs using optical diffraction and computer image processing were needed to positively iden- tify the microcrystals associated with the whisker

growth. Further details of the catalysis studies will be published elsewhere [9].

2. Experimental details

The initial N i / M g O catalysts were prepared by wet impregnation and chemical deposition, fol- lowed by reduction in hydrogen. Samples suitable for electron microscopy were obtained following the Ni-catalyzed steam-reforming of n-butane according to the reaction:

C4H10 + H 2 0 ~ nCO + y H 2 + uC.

Reaction temperatures ranged f rom 4 0 0 ° C to 700 o C. Holey carbon films were used to support the reaction products during examination in the electron microscope. Most observations were made with a JEM 4000EX HREM, operated at 400 keV, with typical magnifications of 200000 × and 500 000 × . Care was taken to operate the micro- scope under standardized imaging conditions, in particular with the same objective lens current for all exposures of experimental images in order to

0304-3991/90/$03.50 © 1990 - Elsevier Science Publishers B.V. (North-Holland)

D.J. Smith et al. / Filamentous carbon and Ni / MgO catalysts 55

minimize random systematic errors associated with recording of the electron micrographs which might otherwise affect the magnification calibration. Measurements of the lattice fringes present in the micrographs were mostly made from light-optical diffraction patterns recorded with a standard opti- cal bench. Typical accuracies of the spacing mea- surements were - 1%-2% in areas of the micro- graphs containing well defined fringes. An off-line computer with image-processing capabilities was also used for checking the fringe spacings in some cases, but effectively the same results were ob- tained.

3. Results

A typical low-magnification electron micro- graph of the Ni/MgO catalyst after it had been

Fig. 1. Typical low-magnification image of Ni /MgO catalyst showing filamentous carbon deposits formed during reaction at 400 o C. Note microcrystals (arrowed) at the ends of the carbon

filaments.

used at a reaction temperature of 400 o C is shown in fig. 1. The characteristic filamentous mor- phology of the carbon deposit is clearly evident, and several examples (arrowed) show the micro- crystals often present at the tips of the filaments. The diameters of the carbon filaments are ap- proximately the same size as the attached metal particles with sizes for these samples of the order of 10 to 70 rim.

Higher-magnification imaging was necessary in order to identify the nature of the microcrystals. Three typical examples, recorded with the corre- sponding carbon filaments overlapping the carbon support film to avoid mechanical instability caused by the incident electron beam, are shown in figs. 2a-c. All of these micrographs show crystalline particles at the ends of partially graphitic whiskers. Optical diffraction patterns indicate that these particles are primarily Ni although small crystal- lites of NiO can be unambiguously identified at the various positions labelled A-E. Fig. 3 shows a much smaller particle ( - 10 nm across) which is entirely NiO. Finally, fig. 4 shows another particle where the Ni{ l l l} planes are well aligned with some of the planes of the carbon filament.

Information about the filamentous deposits is also available from the high-resolution images. In general, the carbon filaments are observed to be comprised of disordered layers of carbon whose separation is approximately that of the basal planes of crystalline graphite (0.34 nm). The fila- ments seem to have the same density across their diameters and therefore show no evidence for "hollow" cores. Moreover, the carbon sheets dis- play a pronounced tendency to develop contours closely matching those of the exit surface of the catalyst particle at the metal-carbon interface (see figs. 2-4). It appears that, as further layers of carbon are added, these contours relax and, away from the particle, the sheets are oriented so that they are approximately perpendicular to the growth direction. This alignment is significant in view of the growth mechanism which has been proposed for these filaments [4-6,9].

A typical image of the Ni/MgO catalyst fol- lowing reaction a t a temperature of 650°C is shown in fig. 5. It was again observed, as for the lower-temperature reaction specimens, that the di-

56 D. J. Smith et al. / Filamentous carbon and Ni / MgO cata.fysts

Fig. 2. clysc alline clystz

articles present at the ends of carbon filaments following reaction at 400°C. (a) Part+ :le of ki te at A. (b) Particle of Ni with NiO crystallites at B and C. (c) Particle of Ni with MO at D and E.

NiO

ameters of the carbon filaments closely matched those of the attached catalyst particles. These fila- ments, however, were invariably more crystalline and had a distinctly tubular shape. This tubular nature of the filaments is confirmed in high-mag- nification images, such as fig. 6 which shows a small particle apparently encapsulated by the whisker. The more highly developed crystallinity of the carbon filament is also obvious, particularly towards the bottom and left of this particular image.

Another interesting example is shown in fig. 7. In this case, the (lengthy) carbon whisker has bent back upon itself so that the end of the tube and a portion of its length are visible in the same image. The enhanced crystallinity and tubular nature of

the carbon filament relative to those shown above is readily apparent, Optical diffraction was neces- sary to establish that the crystal located at the end of the filament was partly nickel (labelled A) and partly nickel oxide (labelled B). However, the ex- posed end of the particle could not be identified because it was in an orientation which did not give rise to lattice fringes in the image. Measurements of other particles not shown here indicated that nickel oxide was considerably more prevalent than nickel in the higher-temperature reaction prod- ucts.

Finally, fig. 8 shows a carbon filament without a terminating metal particle which appears to have bent over such that the electron beam is aligned along its axis. The graphitic layers have developed

D.J. Smith et at,. / Filamentous carbon and Ni / MgO catalysts 57

Fig. 3. Partially enclosed small particle of NiO at end of carbon filament.

so that, away from the (missing) metal particle, the roughly circumferential sheets were oriented approximately parallel to the growth direction. From this and other images it appears that these carbon filaments are formed from shells of graphitic layers which encapsulate the filaments

Fig. 5. Typical low-magnification image of Ni /MgO catalyst showing the development of filamentous carbon deposits with more pronounced crystalfinity and tubular nature following

reaction at 650. C.

but are not continuous along their lengths, show- ing evidence of having originated as elongated cones. These cones apparently form at the gas- metal-carbon interface and are stacked inside each other (see figs. 6-8) with less dense, non-graphitic material at their centers. This amorphous material is associated with the interior portions of the exit surface of the catalyst particle in regions relatively distant from the gas-metal interface. It can be concluded that this distinctive morphology of the tubes does not depend on the presence of a cone-- shaped particle, as reported for the growth of filamentous carbon by iron-alloy-catalyzed dispro- portionation of carbon monoxide [6], since it was observed for a variety of metal particle shapes.

Fig. 4. Particle of Ni at end of carbon filament showing alignment of Ni{ l l l} planes with 0.34 nm graphitic planes in

the growth direction of carbon filament.

4. Discussion and condusions

An overview of our results for the MgO-sup- ported Ni catalyst confirms that the morphology

58 D.J. Smith et al. / Filamentous carbon and Ni / MgO catalysts

of the filamentous carbon deposits is consistent with the results reported previously for different catalyst-support systems and decomposition reac- tions. For example, at lower reaction temperatures ( - 400-500 ° C), the carbon filament is only par- tially graphitic whereas, at higher temperatures ( - 6 0 0 - 7 0 0 ° ) , crystaninity is much more pro- nounced and the carbon deposit is more tube- (or shell-)like in its character. In the former, the carbon sheets developed contiguously with the exit surface of the catalytic particle with successive sheets forming roughly turbostratic layers. In con- trast, the graphitic layers of the filaments formed during 6500C reduction were aligned with the tides rather than the exit surface of the particles so that they formed encapsulating shells with less dense mat~a ls in their interiors. Moreover, with the benefit of higher-resolution observation, under carefully calibrated conditions, it was possible to determine that the identity of the small crystals located at the ends of the whiskers changed from being primarily Ni in the lower-temperature reac- tion products to mainly NiO in the higher-temper-

Fig. 7. High-resolution image showing parts of a lengthy graphitic tube and the terminating particle which is partly Ni

(at A) and partly NiO (at B).

Fig. 6. Small particle encapsulated in long graphitic carbon tube.

ature products. In the latter samples, it sometimes appeared that the NiO particles were wholly en- capsulated by several graphitic sheets, in which case it might be anticipated that the catalytic activity would be reduced by lack of access to the active catalyst particle, It should be noted that no evidence was found for lattice spacings corre- sponding to those of Ni3C, This carbide was re- ported to be a metastable intermediate phase in the formation of carbon filaments in other cata- lyric reactions [10,11].

The different morphologies observed for the graphitic sheets as a function of temperature is significant in terms of the kinetics and diffusion rates appropriate to the various proposed growth mechanisms and will be a topic of future research. Several examples were observed, such as the fila- ment shown in fig. 4, where the uppermost lattice planes of the filamentous carbon were closely aligned with the lattice planes of the adjacent

D.J. Smith et al. / Filamentous carbon and Ni / MgO catalysts 59

should be n o t e d tha t ou r resul ts p rov ide fur ther suppo r t for the behef , he ld b y o ther researchers , tha t the m o r p h o l o g y of f i l amentous ca rbon growth, when it occurs , is effect ively i n d e p e n d e n t of the c a t a l y s t - s u p p o r t sys tem and the pa r t i cu l a r reac- t ion.

Acknowledgements

This w o r k has been pa r t i a l l y s u p p o r t e d b y N S F G r a n t DMR-8514583 .

Fig. 8. A graphitic whisker viewed partially end-on which shows well developed crystallinity around the circumference of

the tube.

ca ta lys t par t ic le . This pa r t i cu l a r resul t is con- s is tent wi th the g rowth m e c h a n i s m p r o p o s e d b y Baker [4] whereby ca rbon is supposed to dif fuse th rough and a r o u n d the ca ta lys t par t ic le and then depos i t u p o n the under ly ing carbon . F ina l ly , i t

References

[1] R.T.K. Baker, M.A. Barber, F.S. Feates, P.S. Harris and R.J. Waite, J. Catal. 26 (1972) 51.

[2] M.R. Everett, D.V. Kinsey and E. Romberg, in: Chem- istry and Physics of Carbon, Vol. 3 (Dekker, New York, 1968) p. 289.

[3] R.T.K. Baker and P.S. Harris, Chemistry and Physics of Carbon, Vol. 14 (Dekker, New York, 1978) pp. 83.

[4] R.T.K. Baker, Catal. Rev. Sci. Eng. 19 (1979) 161. [5] M. Audier, A. Oberlin, M. Oberlin, M. Coulon and L.

Bonnetain, Carbon 19 (1981) 217. [6] M. Audier, A. Oberlin and M. Coulon, J. Cryst. Growth

55 (1981) 549. [7] T. Borowieeki, Appl. Catal., 31 (1987) 207. [8] E. Tracz, R. Scholz and T. Borowieeki, J. Catal., in press. [9] A.J.H.M. Kock, P.K. de Bokx, E. Boellaard, W. Klop and

J.W. Gens, J. Catal. 96 (1985) 468. [10] P.K. de Bokx, A.J.H.M. Kock, E. Boellaard, W. Klop and

J.W. Gens, J. Catal. 96 (1985) 454. [11] E. Boellaard, P.K. de Bokx, A.J.H.M. Kock and J.W.

Geus, J. Catal. 96 (1985) 481.