Journal of Nanoparticle Research 6: 215–221, 2004. © 2004 Kluwer Academic Publishers. Printed in the Netherlands.

The characterization of nanostructured CVD Ni powders using transmission electron microscopy G.J.C. Carpenter and Z.S. Wronski∗ Department of Natural Resources Canada, Canada Center for Mineral and Energy Technology, Materials Technology Laboratories, 568 Booth Street, Ottawa K1A 0G1, Canada; ∗ Author for correspondence (E-mail: [email protected]) Received 24 September 2003; accepted in revised form 14 January 2004

Key words: nanopowders, carbonyl nickel, CVD Ni, nanostructured carbon, TEM, nanoparticles

Abstract Studies of Ni powders using a state-of-the-art transmission electron microscope with a field-emission electron source show that this instrument can be an invaluable tool for studying the morphology and character of nanostructured Ni powder manufactured by a commercial nickel carbonyl vapor deposition process. This study has revealed the presence of surface coatings, including NiO and turbostratic graphite on some powders, which can degrade the electrical conductivity of the powder. Due to the high curvature of hexagonal carbon (graphene) planes, and subsequent high density of defects, the turbostratic graphitic nanocarbon may exhibit new properties.

Introduction The positive electrodes in Ni-based rechargeable batteries are a composite consisting essentially of a mixture of powders of Ni and Ni(OH)2 that are pasted into a porous support structure, for example a foam of metallic Ni. The performance of these electrodes is strongly dependent on the microstructural characteristics of the constituent materials. Results of work on Ni(OH)2 have been published previously (Wronski, 1997) and a report of a study of Ni foam is in preparation. Here, we present examples of different batches of Ni metal powders, characterized by transmission electron microscope (TEM). A brief summary that highlights some of the results of these projects has been published elsewhere (Malis et al., 2002). ‘Fine’ Ni metal powder, produced by chemical vapor deposition (CVD) from Ni carbonyl, commonly referred to as the Mond process (e.g., Pauling, 1988), is used to improve the electrical conductivity of the electrode as well as to provide extra catalytic activity

for the reactions required. ‘Extra-fine’ battery-grade Ni metal powders are a more recent development, produced by the same proprietary CVD process (Pfeil, 1987). They have a branched filamentary structure that gives a very large surface-area/mass ratio that improves the conductivity of an electrode. However, the surface microstructure of the powder, which affects its conductivity, is sensitive to the operating conditions and chemistry during the CVD process. A number of techniques have been used in our laboratory to characterize a variety of batches of Ni powder (Wronski et al., 1995), produced by proprietary CVD processes. The purpose of this paper is to illustrate the point that analytical TEM has proven to be one of the most useful techniques to assess morphology, microstructure, and microchemistry in ‘extra-fine’ Ni powders manufactured under various CVD process conditions. An unexpected benefit of this study was the demonstration of the presence of a new, nanostructured form of carbon phase that was formed, often at a very high volume fraction, on the surface of Ni particulates in a commercial CVD process.

216 Experimental techniques

Results and discussion

The extra-fine grade powders typically have filament diameters of the order of hundreds of nanometers. This greatly simplifies specimen preparation so that it was merely necessary to collect a sample of powder on a holey carbon film in order to examine it in the TEM. Some additional specimens were also prepared for sectioning in a Reichart-Jung Ultracut E ultramicrotome by impregnating the powder in Spurr’s ‘Hard Formula’ epoxy resin. To give adequate dispersion, a pointed stick was wetted with resin, dipped into the powder and stirred in some resin that was contained in a moulding capsule. The capsule was placed in a vacuum to remove any entrapped air or moisture and cured at 60◦ C for 8 h. After rough trimming, final sections were cut with a nominal thickness of 40 nm. The sections were floated off onto water and collected on copper grids. To minimize charging and improve the stability under the beam, sections were given a light, indirect carbon coating using vacuum evaporation at ∼10−5 Torr. The specimens were examined at 200 kV in a Philips CM20FEG TEM/scanning transmission electron microscope (STEM) with a Schottky high intensity field-emission electron gun (FEG), equipped with, (a) a Gatan Imaging Filter model 678, which permits energy-filtered imaging as well as parallel electron energy-loss spectroscopy (PEELS), and (b) an Oxford Instruments Link thin-window energy dispersive X-ray (EDX) detector with an eXL analyzing system.

Figure 1a is a low-magnification bright-field (BF) image of a typical particle showing the branched structure characteristic of extra-fine Ni powder. Figure 1b is an annular dark-field (DF) STEM image of a thin monofilament branch at higher magnification, showing that the microstructure was composed of Ni nanocrystals that were lightly twinned and took the form of a bamboo structure in the thinner branches. Individual Ni crystals were usually free from dislocations such that the overall dislocation density was quite low. Results of X-ray diffraction analysis reported in our previous study provided the average crystal size ca. 50 nm, and a low value for the average plastic strain, e = 0.08% in this nanostructured Ni powder (Wronski et al., 1995). The ability to examine the powder in scanning mode at high spatial resolution using a microscope having a high intensity FEG electron source, which provides a finely-focused beam with relatively high beam current, was advantageous for a number of aspects of this work. Williams (1984) has pointed out that excellent resolution can be obtained in secondary electron (SE) images obtained in a TEM/STEM. This mode of operation is particularly effective when a highly intense electron source is used and has proven useful for examining surface films on certain batches of Ni powder. An example is shown in Figure 2a, from a particular sample where fine particles could be observed, growing on the surfaces of the powder. EDX analysis, using a thin-window detector, readily revealed the presence of

(a)

(b)

Figure 1. (a) Low-magnification BF TEM image of a particle of ‘Extra-fine’ Ni, showing the typical branched structure, (b) annular DF STEM image of a single filament branch at higher magnification, showing grain boundaries and twins.

217 (a)

(b)

Figure 2. (a) SE STEM image, showing the presence of particles on the surface of a filamentary Ni particle, (b) EDX spectrum showing the presence of oxygen. Table 1. Comparison of measured interplanar spacings d (nm) from a SAD pattern with calculated d-values for Ni and NiO; the innermost diffraction ring is #1 Ring #

d (meas) A

Planar indices, (hkl) Ni

d (calc) Ni

Planar indices, (hkl) NiO

d (calc) NiO

1 2 3 4 5 6 7

0.241 0.207 0.174 0.147 0.124 0.105 0.094

None 111 002 None 022 113 None

— 0.2035 0.1762 — 0.1246 0.1063 —

111 002 None 022 113 004 133 024

0.2412 0.2089 — 0.1477 0.1259 0.1044 0.0957 0.0934

oxygen in these particles, Figure 2b. The most likely source of oxygen in surface films on Ni powder was the presence of either NiO or Ni(OH)2 . Because the thin filaments of the extra-fine Ni powder could be studied in transmission mode, selected-area diffraction (SAD) patterns could be obtained showing diffraction rings typical of finegrained crystals with a range of orientations. By calibrating the electron microscope with a standard, the radii of the rings were converted into interplanar spacings. The corresponding spacings from a typical analysis are listed in Table 1, where it is shown that the diffraction rings matched well with those expected for the face-centered cubic crystal structure of NiO, with a lattice parameter 0.418 nm (JCPDS file #4-835). Because at 300 K NiO is an insulator (e.g., Gebhard, 1997), these surface layers would have the

undesirable effect in a battery electrode of reducing the conductivity of the powder. The strong contrast observable in SE images (Figure 2a) is also a consequence of the insulating character of the oxide particles, which leads to charging under the electron beam. No evidence was found for the formation of Ni hydroxide on the particle surfaces. Because the hydroxide phase is hexagonal, with cell parameters a = 0.313 and c = 0.4605 nm, a large number of lattice planes exist which can overlap the oxide reflections. However, there are several planes that have distinctive spacings ¯ for diffraction analysis, particularly (0001), (1010), ¯ and (1012). ¯ The absence of visible reflections (1011) from these planes (Table 1) showed unambiguously that if there were any hydroxide phase on the particle surfaces, it was present only at a very low level, such that it would have little effect on the properties of the powder. Thin, more uniform oxide films were also observed on some samples. The presence of such films could be deduced from a BF image in transmission mode. Figure 3 shows slightly under-focused images of two different batches of powder. Under- or over-focused images are especially useful for revealing fine-scale variations in mass-thickness by means of Fresnel contrast. The powder shown in Figure 3a was virtually free from oxide, as shown by its ‘smooth’ appearance, a fact that was confirmed using EDX spectra and SAD patterns. That image can be compared with the ‘granular’ appearance in Figure 3b, for which a significant surface oxide film, having a thickness that was non-uniform on a scale of nanometers, was present.

218

(a)

(b) Figure 3. Under-focused high-magnification images of two batches of Ni powder; (a) very clean, free from surface films, (b) with a very thin surface oxide, having an irregular thickness at the scale of nanometers.

Such images provided a rapid means for assessing the surface characteristics of extra-fine Ni powders. It was essential to obtain images in a slightly out-of-focus condition, as very thin surface layers become virtually invisible at exact focus. While the oxide layers can be very thin, this oxide phase contributed a substantial fraction of 0.2–0.3 wt% of oxygen in this particular batch of Ni powder. For most powders the total carbon content was in the range 0.2–0.4 wt%. Some powders, particularly those which exhibited a lower surface coverage with NiO, had higher carbon content, 1.25 wt% C and increased oxygen content, 0.51 wt% O. High-resolution (TEM) imaging (HREM) proved useful in studying these Ni powders. Figure 4a shows an image of a sample of powder on which a thin surface layer of graphite could be observed. Whilst the graphite could be identified as a form of carbon using thin-window EDX,

it was necessary to use HREM (Figure 4a), PEELS (Figure 4b) or electron diffraction (SAD, Figure 4c) to show that crystalline graphite was present. In PEELS spectra, for example (Figure 4b), there was a difference between the shape of the CK edge of the graphite as compared to that from evaporated amorphous carbon. In particular, there is evidence of some structure in the σ ∗ peak from the graphite, with the main peak occurring at a lower energy than that from amorphous carbon. Because the surface graphite does not exhibit the same detailed near-edge structure as highly oriented pyrolytic graphite but more like the edge from its highly disordered form produced by ball milling (e.g., Huang, 1999), it can be concluded that the spectrum is characteristic of graphite with a very high defect density. Electron diffraction (SAD) patterns are capable of giving lattice-plane spacings to an accuracy of better than ∼2%, using the Ni reflections for calibration purposes and such measurements are significantly more accurate than measurements from HREM images. SAD patterns (Figure 4c) were used in this case to show a basal plane spacing of 0.36 nm, characteristic of ‘carbon black’. The large d-spacing compared to that of graphite (0.3354 nm) indicates that the carbon planes are bonded more weakly than in graphite, which occurs when they are parallel but rotated relative to each other. Such a form of carbon is referred to as being turbostratic in nature (Kinoshita, 1988a). Thus, the carbon layer observed in this study differs from a wellresearched laminar graphite, which forms by heating an amorphous carbon thin film on Ni and Co metal foils (Derbyshire et al., 1975). The basal planes of the turbostratic graphite tended to be parallel to the curved Ni metal surface on which they were sitting and contained many dislocations to help compensate for the large degree of lattice bending that would otherwise be required, as also indicated by the EELS near-edge structure. Because electrical conductivity is relatively low along the c-axis (perpendicular to the basal planes of graphite), this graphite layer would also be effective in reducing the overall conductivity of the Ni powder. Measurements of the electrical resistivity as a function of pressure applied to a packed bed of carbon particles were reported in our previous work (Wronski et al., 1995). There is a striking similarity of the carbon layer adjacent to a Ni particle with the ordered outer layer in a carbon black particle. Indeed, numerous HREM studies have been reported giving evidence for ordering of

219 (a)

(b)

(c)

Figure 4. (a) High-magnification TEM micrograph, showing the presence of turbostratic graphitic nanocarbon on the surface of a Ni particle, (b) PEEL spectrum, illustrating the distinction between amorphous carbon and the surface graphite, (c) SAD pattern from an area that includes the surface graphite layer; the radius of the inner diffraction ring corresponds to the interplanar spacing of the basal planes.

turbostratic carbon

(a)

turbostratic carbon

carbon black

(b)

CVD nickel metal

Figure 5. Schematic illustration of the microstructure of turbostratic carbon on (a) a carbon black particle and (b) a CVD Ni powder particle.

carbon planes near the periphery of the carbon black particle (Kinoshita, 1988b). The orientation of the outer graphitic planes has been reported to be mainly parallel to the surface of the particle, whilst the core remains disordered. Likewise, the basal graphite planes

follow the curvature of the Ni metal particle. Such a turbostratic carbon phase, which exhibits substantial curvature of the graphene planes, may represent a new form of turbostratic graphitic nanocarbon. This is schematically depicted in Figure 5.

220 When sulphur was found, it appeared to be localized in a coarse form that was unlikely to have a significant impact on the electrical properties. Conclusions

Figure 6. TEM micrograph of a cross-section through a Ni particle having an unusually high C concentration, showing the presence of turbostratic graphite in layers within the particle.

An experimental batch of powder that contained an unusually high carbon concentration (7.1 wt%) showed a complex microstructure in which layers of graphite could even be present within the Ni particles. This powder was best examined in the form of ultramicrotomed sections, as shown in Figure 6, where the different layers within the particle are identified. The graphite within the particles again tended to show an ‘onion skin’ type of structure, where the basal planes were strongly curved, with an average lattice spacing ca. 0.36 nm, and a high dislocation density to accommodate the lattice bending needed to give a high degree of curvature. This powder also had relatively poor electrical conductivity properties, as we reported previously (Wronski et al., 1995). Oxygen was present in EDX spectra collected from the carbon coating. Substantial oxygen pick up has also been observed in some grades of carbon black which exhibited decreased electrical resistivity (Delhaes & Carmona, 1981). Amorphous sulphur, in the form of clumps, ∼1 µm in diameter, was also occasionally seen, attached to powder having a high carbon concentration. Sulphur could be readily detected using EDX analysis and arose from the decomposition of H2 S, which is sometimes used as a catalyst for the carbonyl CVD reaction, depending on the operating conditions.

A state-of-the-art TEM/STEM with a field-emission electron source can be an invaluable tool for studying the morphology and character of extra-fine, nanostructured Ni powder. The following techniques have proven useful for these studies: BF- and DF-STEM and SE imaging, SAE diffraction, TEM imaging, including HREM, together with the analytical techniques of thin-window EDX and PEELS. Detailed examination has revealed that certain batches of Ni powder may show the presence of surface films, including NiO and a new phase of turbostratic graphitic nanocarbon, both of which can degrade the electrical conductivity of the Ni powder. The basal planes of the turbostratic graphite tended to be curved and contained many dislocations to help compensate for the large degree of lattice bending that would otherwise be required to follow the surfaces of the powder. Acknowledgements We are pleased to acknowledge the technical assistance of the following colleagues at MTL: Marc Charest for TEM support and Glenn Williams for preparation of the ultramicrotomed sections. We are also very grateful to Dr. Peter Kalal of Inco Ltd., for supplying the Ni powders and providing expert advice on their application and electrical behavior. References Delhaes P. & F. Carmona, 1981. In: Walker P.L. and Thrower P.A. eds. Chemistry and Physics of Carbon, Vol. 17. Dekker, New York, p. 89. Derbyshire F.J., A.E.B. Presland & D.L. Trimm, 1975. Graphite formation by the solution–precipitation of carbon in cobalt, nickel and iron. Carbon 13, 111–113. Gebhard F., 1997. The Mott Metal-Insulator Transition, Heidelberg, 37; also Tschauner O., 1998. Insulator-metal transition of NiO, Paper at the 1998 March meeting of the APS, March 16–20, Los Angeles, CA, available at http://www.aps.org/BAPS98/abs/G2570002.html. Huang J.Y., 1999. HRTEM and EELS studies of defects structure and amorphous-like graphite induced by ball milling. Acta. Mater. 47, 1801–1808.

221 Kinoshita K., 1988a. Carbon – Electrochemical and Physicochemical Properties, John Wiley, New York, p. 33. Kinoshita K., 1988b. Carbon – Electrochemical and Physicochemical Properties, John Wiley, New York, pp. 25–28. Malis T., G.J.C. Carpenter, S. Dionne, G.A. Botton & M.W. Phaneuf, 2002. In: Li Z.R. ed. Industrial Applications of Electron Microscopy. Marcel Dekker, Inc., New York, pp. 213–257. Pauling L., 1988. General Chemistry. Dover Publications, New York, p. 693. Pfeil L.B., 1987. Method for the production of nickel powder. US Patent 4,673,430 June 16, 1987.

Williams D.B., 1984. Practical Analytical Electron Microscopy in Materials Science. Philips Electronic Instruments, Inc., Mahwah, New Jersey, p. 51. Wronski Z.S., J.R. Brown, G.J.C. Carpenter, E.J.-C. Cousineau, J.A. Jackman, G. McMahon, M.T. Sheheta, S. Mikhail & P.J. Kalal, 1995. CANMET Report, MTL 95-12(TR), Canada Centre for Minerals and Energy Technology, Metals Technology Laboratory, Ottawa. Wronski Z.S., G.J.C. Carpenter & D. Martineau, 1997. In: Holmes C.F. and Landgrebe A.L. eds. Electrochemical Society Proceedings, Vol. 97-18, Batteries for Portable Applications and Electric Vehicles. The Electrochemical Society, Pennington, New Jersey, pp. 804–811.