The following article appears in the journal JOM, 50 (9) (1998), pp. 58-64.

The Key Role of Impurities in Ancient Damascus Steel Blades

CONTENTS
INTRODUCTION DAMASCUS STEEL Reproducing Wootz Damascus Blades THE SWORDS Zschokke Swords Four Wootz Damascus Blades THE LADDER PATTERN THE ROLE OF IMPURITIES IN BAND FORMATION EXISTING DAMASCUS BLADES QUALITY WHY THE ART WAS LOST ACKNOWLEDGEMENTS References

The art of producing the famous 16-18th century Damascus steel blades found in many museums was lost long ago. Recently, however, research has established strong evidence supporting the theory that the distinct surface patterns on these blades result from a carbide-banding phenomenon produced by the microsegregation of minor amounts of carbide-forming elements present in the wootz ingots from which the blades were forged. Further, it is likely that wootz Damascus blades with damascene patterns may have been produced only from wootz ingots supplied from those regions of India having appropriate impurity-containing ore deposits.

INTRODUCTION

This article is concerned with the second type of Damascus steel, sometimes called oriental Damascus. The most common examples of these steels are swords and daggers, although examples of body armor are also known. The name Damascus apparently originated with these steels. The steel itself was produced not in Damascus, but in India and became known in English literature in the early 19th century³ as wootz steel, as it is referred to here. Detailed pictures of many such wootz Damascus swords are presented in Figiel's book,⁴ and the metallurgy of these blades is discussed in Smith's book.⁵

Unfortunately, the technique of producing wootz Damascus steel blades is a lost art. The date of the last blades produced with the highest-quality damascene patterns is uncertain, but is probably around 1750; it is unlikely that blades displaying low-quality damascene patterns were produced later than the early 19th century. Debate has persisted in the metallurgy community over the past 200 years as to how these blades were made and why the surface pattern appeared.^6-8 Research efforts over the years have claimed the discovery of methods to reproduce wootz Damascus steel blades,^9-12 but all of these methods suffer from the same problem—modern bladesmiths have been unable to use the methods to reproduce the blades. The successful reproduction of wootz Damascus blades requires that blades be produced that match the chemical composition, possess the characteristic damascene surface pattern, and possess the same internal microstructure that causes the surface pattern.

DAMASCUS STEEL

a

b

Figure 1. (a) A reconstructed wootz Damascus blade showing the Damascene surface pattern containing a combined Mohammed ladder and rose pattern. (b) A longitudinal section of the same blade showing the bands of cementite particles responsible for the surface pattern.

Both the internal microstructure and the chemical composition of these steels were well established early in this century.^11,13 The internal microstructure of a wootz Damascus blade possessing a high-quality damascene surface pattern is a unique metallurgical microstructure.⁸ It consists of bands of small (generally around 6 mm diameter) particles of Fe₃C (cementite) clustered along the band centerline. The bands have a characteristic spacing in the 30-70 mm range and are contained in a steel matrix. The structure of the steel matrix varies depending on how the smith heat-treated the blade, but it is generally found to be pearlite. The bands lie parallel to the forging plane of the blades. By manipulating the angle of the blade surface relative to the plane of the bands, the smith can produce a variety of convoluted patterns of intersection of the bands with the blade surface. With polishing and etching, the Fe₃C particles cause the bands to appear white and the steel matrix nearly black; thus, the surface pattern is created.

Reproducing Wootz Damascus Blades

A detailed picture description of the production process for this blade has recently been published.¹⁴ In addition, the technique has been fully described in the literature,^15-17 and it has been shown that blades possessing high-quality damascene patterns can be repeatedly produced utilizing the technique. The technique is, in essence, a simple reproduction of the general method described by the earlier researchers. A small steel ingot of the correct composition (Fe + 1.5C) is produced in a closed crucible and is then forged to a blade shape. However, some key factors are now specified. These include the time/temperature record of the ingot preparation, the temperature of the forging operations, and the type and composition level of impurity elements in the Fe + 1.5C steel. It appears that the most important factor is the type of impurity elements in the steel ingot. Recent work^17-18 has shown that bands of clustered Fe₃C particles can be produced in the blades by the addition of very small amounts (0.03% or less) of one or more carbide-forming elements, such as V, Mo, Cr, Mn, and Nb. The elements vanadium and molybdenum appear to be the most effective elements in causing the band formation to occur. An obvious question raised by these results is, are these elements also present at low levels in the 16-18th century wootz Damascus blades?

THE SWORDS

Figure 2. Macrophotographs of Zschokke sword blades.

This article presents the results of a study of these four samples. Also, four additional wootz Damascus blades, all thought to be a few hundred years old, have been acquired and are included. Hence, all of the blades studied here are more than two centuries old and were presumably made from wootz steel. These blades are referred to as genuine wootz Damascus blades to differentiate them from the reconstructed wootz Damascus blades made by the technique developed by the authors.

Zschokke Swords

Pieces were cut from one end of each of the samples with a thin diamond saw. A 2 cm length was cut for chemical-analysis studies, and an 8 mm length sample was used for microstructure analysis. The chemical analyses were done using emission spectroscopy on a calibrated machine at Nucor Steel Corporation. Table I presents the chemical analyses, along with the values reported by Zschokke. Agreement between the analyses done by Zschokke in 1924 and the present data is reasonably good.

Micrographs of surface and transverse sections of the remaining three swords are shown in Figure 3. The micrographs of the surfaces are, in effect, taper sections through the bands seen on the micrographs of the section views, and, as expected, the widths of the bands are expanded in the surface views.

a	b	c
d	e	f
Figure 3. Micrographs of Zschokke blades showing (a) the surface of blade 7, (b) a transverse section of blade 7, (c) the surface of blade 9, (d) a longitudinal section of blade 9, (e) the surface of blade 10, and (f) a transverse section of bade 10.

Rockwell C hardness data were taken along the centerline of the transverse sections of all four swords in order to more fully characterize them. A large variation in hardness was found and is presented in Table II. The hardness correlated with the matrix microstructure. The matrix structure of the blades underwent a transition from pearlite at the thin tip to a divorced eutectoid ferrite + cementite at the fat end (thickness = 3-4 mm). These structures are consistent with recent kinetic studies of the eutectoid reaction in hypereutectoid steels.^19-20 The studies show that in two-phase (austenite + Fe₃C) steels, the divorced eutectoid transformation (DET) dominates at slow cooling rates and the pearlite reaction dominates at higher cooling rates; the DET is favored as the density of the Fe₃C particles in the transforming austenite increases. Hence, the matrix microstructures indicate that the blades were air-cooled with pearlite dominating near the faster cooling cutting edge. The dominance of the DET matrix structure in swords 7 and 10 probably results from the higher amount of interband Fe₃C present in these swords.

In swords 7 and 10, the particles are dominantly plate-shaped with the thin direction aligned in the forging plane of the sword blades. Consequently, the area of the particles on the sword face is generally larger than on the sections. The standard deviation of the data was consistently in the range of 20-25%, so that differences in the areas on the three surfaces are problematic, whereas, the differences in minimum and maximum diameters are significant. For blades 7 and 10, the maximum/minimum aspect ratio of the particles averages around three on both transverse and longitudinal sections and around two on the sword faces. The ratios are slightly less for blade 9, reflecting the more globular shape of the particles and the observation that the oblong particles do not have their broad face well aligned in the forging plane, as they do on blades 7 and 10.

Four Wootz Damascus Blades

a

b

Figure 4. (a) The surface of a kard blade showing the emery-paper scratches and the burn mark made by the emission spectrograph analysis. (b) The region near the burn mark after refinishing.

After polishing, the effects of emission-spectrometer analysis are no longer apparent (Figure 4b). The original damascene pattern has been restored, and, even knowing where the crater had been, it is very difficult to detect its presence by visual examination. The damascene pattern in this blade is well formed, but requires a small magnification to be clearly seen, because of the combination of the small size of the cementite particles and the large amount of interband Fe₃C particles in this blade. Digital micrographs of the surface of the kard blade were examined with the same image-analyzing technique used for the Zschokke blades. The results presented at the bottom of Table III illustrate the significantly smaller area of the Fe₃C particles in the kard blade versus the Zschokke blades.

THE LADDER PATTERN

Figure 5. (a) A longitudinal section view after a notch cut across blade-(b) Distortion of the carbide bands by forging flow. (c) A schematic of the blade surface showing band spacing after forging flow.

A qualitative illustration of the mechanism of ladder formation using the notch-cutting technique is presented in Figure 5. For a given band spacing in the blade, the band spacing on the blade surface, S, is controlled by the angle of the band planes with the blade surface, shown as a on Figure 5. When a is increased, the band spacing on the surface decreases. It is the systematic variation of this angle a that results in the oscillation of the spacing of the bands on the blade surface and causes the characteristic wavy damascene pattern. Experiments show that the waviness of the pattern can be dramatically increased by simply using rounded or peening hammer heads to increase variations in a. After cutting a groove across a nearly finished blade, subsequent forging causes the metal at the groove base to fill the groove cavity. The direction of this forging flow is shown by the vertical arrow of Figure 5a. This flow deforms the bands, causing the local a angles at the blade surface to increase at the notch wall locations. The resulting decrease of the surface spacing S at those two locations is shown in Figure 5c. If the grooves are wide, the narrowed band spacing appears along the two walls of the groove and not the bottom, depending on the depth of the groove. The details of the pattern are a complex function of the width and depth of the groove cavity and the shape of the groove cavity.

Experiments have been carried out on the reconstructed wootz Damascus blades in which the ladder and rose pattern were produced by both the groove-cutting and groove-forging techniques. The patterns in the blade of Figure 1 were made with the groove-cutting technique, and detailed photographs of the process have recently been published (Figure 6a).¹⁴ These patterns may be compared to similar ladder/rose patterns made by the die-forging technique (Figure 6b). The circular pattern in Figure 6b (called the rose pattern on ancient blades) was made with a hollow cylindrical die, while the pattern in Figure 6a was made by removing metal with a specially shaped solid drill. In the case of the die-forged patterns, the ridges produced by the upsetting action of the die were removed with a belt grinder prior to additional forging.

A comparison of the ladder patterns produced by grinding versus forging reveals nearly identical features (Figure 6). Figiel points out that there is a large variation in the pattern in the bands of the several examples presented in his book.⁴ Hence, this study is only able to conclude that the ancient smiths produced the ladder patterns by making parallel grooves across the surface of nearly finished blades, either by forging or cutting/grinding.

THE ROLE OF IMPURITIES IN BAND FORMATION

a

b

Figure 6. The ladder and rose pattern produced by (a) grooves cut into the surface of the nearly finished blade and (b) grooves forged into the surface of the nearly finished blade.

It is well established^25-28 that the ferrite/pearlite banding of hypoeutectoid steels results from microsegregation of the X element in Fe-C-X alloys, where X is generally manganese, phosphorus, or an alloy addition. For the example X = P, it is established that the microsegregation of phosphorus to the interdendritic regions (IRs) causes ferrite to nucleate preferentially in the IRs. If the cooling rate is slow enough, the ferrite grows as blocky grain boundary allotriomorphs and pushes the carbon ahead of the growth front until pearlite forms between neighboring IRs. Apparently, rolling or forging deformation is quite effective in aligning the IRs of the solidified ingots into planar arrays, because the ferrite appears as planar bands parallel to the deformation plane separated by bands of pearlite. The ferrite/pearlite bands of sword 8 were probably produced by this type of banding caused, most likely, by the microsegregation of phosphorus.

A strong body of evidence has been obtained^16-18 that supports the theory that the layered structures in the normal hypereutectoid Damascus steels are produced by a mechanism similar to the mechanism causing ferrite/pearlite banding in hypoeutectoid steels with one important difference in ferrite/pearlite banding, the bands form on a single thermal cycle. For example, the ferrite/pearlite bands can be destroyed by complete austenitization at low temperatures (just above the A₃ temperature) followed by rapid cooling and are then reformed in a single heat up to austenite, followed by an adequately slow cool.²⁶ (Low-temperature austenitization is required to avoid homogenization of the microsegregated X element.) The carbide bands of the wootz Damascus steel are destroyed by a complete austenitization at low temperatures (just above the A_cm temperature) followed by cooling at all rates, slow or fast. However, if the steel is then repeatedly cycled to maximum temperatures of around 50-100°C below A_cm, the carbide bands will begin to develop after a few cycles and become clear after 6-8 cycles.

The formation mechanism of the carbides clustered selectively along the IRs during the cyclic heating of the forging process is not resolved. It seems likely, however, that it involves a selective coarsening process, whereby cementite particles lying on the IRs slowly become larger than their neighbors lying on dendrite regions and crowd them out. A model for such a selective coarsening process has been presented.¹⁷ During the heat-up stage of each thermal cycle, the smaller cementite particles will dissolve, and only the larger particles will remain at the forging temperature, which lies just below the A_cm temperature. The model requires the segregated impurity atoms lying in the IRs to selectively reduce the mobility of the cementite/austenite interfaces in those regions. Larger particles would then occur in the IRs at the forging temperature. They probably maintain their dominance on cool down because one would not expect the small particles that had dissolved to renucleate on cool down in the presence of the nearby cementite particles. These near-by particles would provide sites for cementite growth prior to adequate local supercooling sufficient to nucleate new particles.

EXISTING DAMASCUS BLADES QUALITY

Based on this experience, it seems likely that the fraction of Indian crucible steel that was successfully forged into the damascened blades was probably quite small; the majority of surviving wootz Damascus blades probably display low-quality surface patterns. Craddock²⁹ has come to this same conclusion based on an analysis of the literature on damascene-patterned steels. The results on the four Moser blades studied by Zschokke support this same conclusion. These blades were supposedly representative of good-quality damascened blades from the east, and yet of the four, only sword 9 displays the high-quality Fe₃C bands characteristic of the best museum-quality wootz Damascus blades.

WHY THE ART WAS LOST

One of the big mysteries of wootz Damascus steel has been why the art of making these blades was lost. The vanadium levels provide the basis for a theory. Based on our studies, it is clear that to produce the damascene patterns of a museum-quality wootz Damascus blade the smith would have to fulfill at least three requirements. First, the wootz ingot would have to have come from an ore deposit that provided significant levels of certain trace elements, notably, Cr, Mo, Nb, Mn, or V. This idea is consistent with the theory of some authors³⁰ who believe the blades with good patterns were only produced from wootz ingots made in southern India, apparently around Hyderabad. Second, the data of Table IV confirm previous knowledge that wootz Damascus blades with good patterns are characterized by a high phosphorus level. This means that the ingots of these blades would be severely hot short, which explains why Breant's⁹ 19th century smiths in Paris could not forge wootz ingots. Therefore, as previously shown,¹⁵ successful forging would require the development of heat-treating techniques that decarburized the surface in order to produce a ductile surface rim adequate to contain the hot-short interior regions. Third, a smith who developed a heat-treatment technique that allowed the hot-short ingots to be forged might still not have learned how to produce the surface patterns, because they do not appear until the surface decarb region is ground off the blades; this grinding process is not a simple matter.

The smiths that produced the high-quality blades would most likely have kept the process for making these blades a closely guarded secret to be passed on only to their apprentices. The smiths would be able to teach the apprentices the second and third points listed, but point one is something they would not have known. There is no difference in physical appearance between an ingot with the proper minor elements present and one without. Suppose that during several generations all of the ingots from India were coming from an ore body with the proper amount of minor elements present, and blades with good patterns were being produced. Then, after a few centuries, the ore source may have been exhausted or become inaccessible to the smithing community; therefore, the technique no longer worked. With time, the smiths who knew about the technique died out without passing it on to their apprentices (since it no longer worked), so even if a similar source was later found, the knowledge was no longer around to exploit it. The possible validity of this theory could be examined if data were available on the level of carbide-forming elements in the various ore deposits in India used to produce wootz steel.

ACKNOWLEDGEMENTS

References

ABOUT THE AUTHORS

J.D. Verhoeven is currently a professor in the Materials Science and Engineering Department at Iowa State University. A.H. Pendray is currently president of the Knifemakers Guild. W.E. Dauksch is retired as vice president and general manager of Nucor Steel Corporation.

For more information, contact J.D. Verhoeven, Iowa State University, Materials Science and Engineering Department, 104 Wilhelm Hall, Ames, Iowa 50011; (515) 294-9471; fax (515) 294-4291; jver@iastate.edu.