Analysis of WC grain growth during sintering using electron backscatter diffraction and image analysis

https://doi.org/10.1016/j.ijrmhm.2007.10.004Get rights and content

Abstract

The WC carbide grain size is important for the technological properties of cemented carbide cutting tools. In the present work the WC carbide grain size distribution is determined after milling and sintering for 0.25, 1 and 8 h at 1430 °C.

The WC grain size distribution, both for the powder after milling and the sintered specimens, is determined by two different methods, i.e. image analysis on scanning electron microscopy (SEM) images and electron backscatter diffraction (EBSD). It should be noted that in this work the 2D grain size distribution is considered.

The EBSD analysis clearly shows that the special Σ2 boundaries are present in the powder and that their fraction decreases during sintering and particularly during the early stages. When the Σ2 boundaries are omitted in the EBSD analysis the results of the grain size measurements for the two methods agree quite well.

Introduction

Cemented carbides are made by liquid-phase-sintering during which the grain size increases through coarsening. In so-called normal grain growth there is a gradual increase in grain size but no drastic change in the shape of the grain size distribution function except for its increase in average grain size. Abnormal grain growth, i.e. the excessive growth of some grains leading to a few very large grains, is often observed in cemented carbides. The abnormal grains have a negative influence on the properties and thus it is very important to learn how to avoid abnormal grain growth. It is also important to understand grain growth in general since the average grain size is important for the properties. Thus many studies have been performed on grain growth in cemented carbides.

However, it is difficult to measure the grain size accurately and many different methods have been used. Often image analysis based on SEM images is used, but a drawback is that the resolution makes it hard to detect small grains and the method is very time-consuming. EBSD can be used to obtain better spatial resolution, since the images are constructed from crystallographic information. A problem with both methods is that abnormally large grains can be missed since the magnification is adjusted to see the most frequently occurring grain size. It should also be observed that the evaluations are performed on 2D cross-sections and in this work no attempts are done to transform the 2D grain size distribution to 3D.

Kumar et al. [1] recently used EBSD to characterize grain boundaries during sintering of WC–Co cemented carbides. In particular they observed so-called Σ2 boundaries in WC. Such boundaries exhibit low interfacial energy and they are usually not wetted by Co during sintering [2], [3], [4]. Lay and Loubradou [5] suggested that these grain boundaries were present already in the powder. Also Kim et al. [6] confirmed that the Σ2 grain boundaries present in liquid-phase-sintered WC–Co alloys originate from the powder. However, although they occur frequently in the powder directly after milling, they are largely annihilated during sintering [1].

At this point a word of caution is needed. The CSL (coincident-site-lattice) concept only refers to the relative orientation of two grains and does not take into account the orientation of the grain boundary itself. Usually the two special cases of twist and tilt boundary are considered but in general the grain boundary between two crystals of a given orientation relationship could have any orientation. One expects the low-energy only for some grain boundary orientations. For the case of Σ2 boundaries in WC the twist boundary represents the lowest energy [2]. Such a boundary is created by cutting the crystal along a (101¯0) plane and rotate one half 90° around its normal.

The purpose of the present work is to investigate if EBSD is suitable for determination of the grain size distribution function during grain growth. The WC grain size distribution will be obtained after sintering using both image analysis based on SEM images and EBSD analysis. The similarities and differences between the methods are discussed. Especially the influence of the Σ2 boundaries are of interest since they are hard to observe using conventional image analysis.

The present study is part of an attempt to model the WC carbide grain growth. The model [7] is based on 2D nucleation of growth and shrinkage ledges. The time evolution of the size distribution function is obtained by solving the non-steady-state Langer–Schwartz equation numerically. Also experiments are performed on a series of WC–10 wt%Co powders with different grain size for a number of time and temperature sintering cycles. The comparison with experimental data is critical when validating the model and therefore it is important to have accurate experimental data. It is also essential to understand the differences between data obtained by the two techniques.

Section snippets

Materials

A powder with a composition of WC–10 wt%Co was produced using raw material from H.C. Starck with a nominal WC grain size of 0.9 μm according to the Fisher sub sieve sizer (FSSS). The powder was milled for 40 h and heated with a rate of 10 °C/min to 1430 °C, sintered for a number of different holding times (15 min, 1 and 8 h) and furnace cooled. The grain size was determined both for the sintered specimens and for the powder before sintering after removal of pressing agent and Co.

As shown by Wang et

Results and discussion

Fig. 1, Fig. 2, Fig. 3 show the evolution of the microstructure during sintering by means of optical microscope and EBSD images. The results of the grain size measurements from the image and EBSD analysis are shown in Table 2, Table 3 and are also presented as cumulative particle size distributions, shown in Fig. 4, Fig. 5, Fig. 6, Fig. 7. The EBSD results are presented both as “EBSD incl. Σ2”, where the Σ2 boundaries are treated as ordinary grain boundaries and “EBSD excl. Σ2”, where the Σ2

Conclusions

A good agreement between image analysis and EBSD for WC grain size measurements is obtained if the Σ2 grain boundaries are considered as special boundaries which are omitted from the grain detection routine. An interesting observation for both techniques is that the WC grain size distribution is poly-disperse in the powder before milling and Gaussian after sintering.

Despite the fact that the amount of Σ2 boundaries decrease rapidly during the initial stages of sintering we believe that they

Acknowledgements

This is a part of a project financed by the Brinell Centre Inorganic Interfacial Engineering (BRIIE), supported by the Swedish Agency for Innovation Systems (VINNOVA), AB Sandvik Coromant, Seco Tools AB and Atlas Copco Secoroc AB. The authors would like to thank professor G. Wahnström for stimulating discussions. Dr. Bo Jansson at Seco Tools AB and Dr. Göran Stenberg at Atlas Copco Secoroc AB are also acknowledged.

References (9)

There are more references available in the full text version of this article.

Cited by (0)

View full text