- Email: [email protected]
Transfer of surface texture from silicon nitride rolls to stainless steel wire in cold-rolling
Journal of Materials Processing Technology 173 (2006) 394–400
Transfer of surface texture from silicon nitride rolls to stainless steel wire in cold-rolling Peter Andersson a,∗ , Michael Wild b , Jan Lev´en c , Bj¨orn Hemming d a
VTT Industrial Systems, P.O. Box 1702, FIN-02044 VTT, Espoo, Finland b Haldex Garphyttan Wire AB, S-71980 Garphyttan, Sweden c Mefos-Metallurgical Research Institute AB, Box 812, S-97125 Lule˚ a, Sweden d Centre for Metrology and Accreditation, Tekniikantie 1, FIN-02150 Espoo, Finland Received 6 January 2005; received in revised form 6 December 2005; accepted 7 December 2005
Abstract A set of cold-rolling experiments with ceramic rolls was carried out using facilities for the production of profiled steel wire. Based on the experiments, the strength of sintered silicon nitride was found sufficient for allowing cold-rolling of austenitic stainless steel wire. The wear rate of the silicon nitride rolls in the tests was low. Metal was transferred from the wire to the rolling tracks of the ceramic rolls. The surface texture of the rolls was reproduced on the rolled product. © 2006 Elsevier B.V. All rights reserved. Keywords: Silicon nitride; Cold-rolling; Wire rolling; Surface transfer; Wear; Surface roughness; Texture
1. Introduction Rolling is a three-dimensional plastic flow process used for the forming of metallic sheet material and profiles. In rolling, a work piece material is brought into a gap between two rolls. Due to the reduction of the material thickness in the work zone, the work piece material is accelerated first, before the neutral plane of the work zone, to the surface velocity of the rolls, then to a velocity higher than the surface velocity of the rolls, as seen in Fig. 1. When a wire material is rolled, this spreads laterally in the work zone during the thickness reduction, and the rolling process results in a cross-section that is lower and wider and has an area that is smaller than that of the work piece material [1]. The surface quality of cold-rolled sheet and wire materials is normally of great importance, as the rolled surface often forms a functional surface of an end product. The surface quality of the cold-rolled product is sensitive to the surface quality of the rolls, and for this reason particular emphasis is put on the roll surface finish. For certain purposes, textured rolls that give a desired surface structure to the cold-rolled material are used [2,3]. In addition to the roll surface modifications, alterations in
∗
Corresponding author. E-mail address: [email protected] (P. Andersson).
0924-0136/$ – see front matter © 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.jmatprotec.2005.12.009
the lubrication, velocity and degree of reduction, for instance, provide means to control the surface quality of the cold-rolled product [1–5]. For more than a decade, hybrid ball bearings composed of silicon nitride ceramic balls and steel races have been successfully used particularly in high-speed applications [6–8]. One of the benefits of the use of silicon nitride ceramic balls is the diminished need for lubricating oil, owing to the good surface quality that can be achieved on a silicon nitride surface by polishing. Similarly, silicon nitride rolls that operate against steel camshafts in engines have been successful [9]. Experience with hybrid ball bearings and ceramic cam followers has shown that silicon nitride is beneficial in rolling contacts, which has created interest in the use of this ceramic material in metal rolling applications. A question frequently raised when ceramic materials for machine elements and tools are discussed is the issue of the strength and reliability of these non-metallic materials. One of the central approaches of the present study was to evaluate the performance and reliability of silicon nitride rolls in actual steelrolling processes. The other approaches comprised tribological aspects of cold-rolling with ceramic rolls, including the transfer of the roll surface texture (surface profile waviness and surface roughness) onto the work piece material. In order to facilitate the evaluation, a set of cold-rolling experiments was carried
P. Andersson et al. / Journal of Materials Processing Technology 173 (2006) 394–400
395
Fig. 1. Schematic presentation of the cold-rolling of wire; h1 is the entry thickness and h2 the exit thickness of the wire. Due to cross-section area reduction, the wire exit velocity v2 is higher than the entry velocity v1 . The line NP represents the neutral plane, at which the contact pressure between roll and wire reaches a maximum and the roll and wire surfaces have the same velocity v. Wire tension at entry or exit zones can move the neutral plane, but was not applied in the present tests.
out using production facilities for cold-rolling of profiled steel wire. In the following, the preparations and procedures for the experiments and the post-test investigations are presented and discussed. 2. Experimental 2.1. Equipment The experiments were carried out using a cold-rolling mill at Haldex Garphyttan Wire AB. The rolling mill had four consecutive rolling passes, each with a two-high roll configuration, the last one of which accommodated the ceramic rolls. During the rolling procedure, rolling oil from a circuit with filtration was sprayed from nozzles onto the rolls and wire. An overview of the rolling mill is shown in Fig. 2 and a detailed view of the first ceramic roll pair before the tests is shown in Fig. 3.
Fig. 3. Exit zone of the final rolling pass with ceramic rolls 1 and 2 and two oil nozzles. Taylor–Hobson Form Talysurf diamond stylus equipment was used for measuring the surface roughness of the ceramic roll surfaces, the wire surfaces and the rolling tracks on the ceramic rolls. All the surface roughness measurements were done perpendicularly to the rolling direction of the wire and roll samples. Of the measurement results used in the present study, the surface roughness parameter Ra represents the arithmetic mean surface roughness [10], as an average value for five individual measurements of at least 5 × 0.25 mm length each. The Ra value, however, carries no information about the shape of the surface profile. The Rsk value expresses the surface profile skewness as a mean value for five individual measurements of at least 5 × 0.25 mm length each; for Rsk > 0 the surface profile is dominated by peaks while for Rsk < 0 the surface profile is dominated by valleys. The surface roughness parameter Rp indicates the highest peak and Rv the deepest valley of the surface profile within a measurement. The Rp and Rv values given in the tables are mean values for five individual measurements of at least 5 × 0.25 mm length each. All surface roughness values in the present work are based on filtered profiles from measurements using standard cut-off lengths (λc ) of 0.25 or 0.8 mm. The values for the wire samples are presented with reference to the upper and lower surfaces, of which the former were formed against upper rolls and the latter against lower rolls of the rolling equipment.
2.2. Materials
Fig. 2. The cold-rolling mill used for the present experiments. The final rolling pass is at the right end of the mill.
Four ceramic rolls, mounted on individual steel support shafts, were used in the experiments, in sets of two rolls each. The ceramic materials for the rolls, with a hollow cylinder shape, had been manufactured by cold isostatic pressing and gas pressure sintering (GPS) from Si3 N4 powder with a few percent of binder and sintering additives. The as-sintered hollow cylinders had been ground on their inner and outer surfaces, and mounted with a conical press fit on steel shafts. When tightly mounted on the shafts, the outer cylindrical surfaces of the
396
P. Andersson et al. / Journal of Materials Processing Technology 173 (2006) 394–400 Table 4 Surface roughness parameters for the rolling tracks of the silicon nitride ceramic rolls after the cold-rolling pre-test
Table 1 Silicon nitride ceramic rolls used in the cold-rolling experiments Roll identifications
Surface roughness parameters
Roll no.
Colour
Ra [m]
Rsk
Rp [m]
Rv [m]
1 2 3 4
Greyish black Light grey Dark grey Dark grey
0.20 0.13 0.03 0.04
−0.4 −3.3 −1.1 −0.1
0.62 0.47 0.11 0.16
0.99 2.4 0.29 0.21
Roll identification no. Surface roughness parameters for the rolling track Ra [m] Rsk Rp [m] Rv [m] 1 2
0.21−0.6 0.65 0.16−1.1 0.54
1.1 1.2
2.3. Experimental procedure Table 2 Proportions of principal alloying elements of the steel wire material Element
Proportion [%]
C
Si
Mn
Cr
Ni
0.1
0.5
6
17
5
ceramic rolls had been ground and polished concentrically with the shafts. The final dimensions of the roll surfaces were Ø 150 mm × 100 mm and the wall thickness was 20–20.5 mm. The individual rolls and selected surface roughness parameters are presented in Table 1. For comparison, it should be noted that the Ra value of metallic rolls during similar cold-rolling operations normally lies between 0.04 and 0.2 m. The surfaces of rolls 1 and 2 were generally flat with a high degree of open porosity, which was reflected by high Ra and Rv surface roughness values. Most of the pores were sufficiently small to be reflected by the surface roughness parameters. Rolls 3 and 4 were made from a dense material that gave a low surface roughness. The poor integrity of the materials in rolls 1 and 2 on the one hand, and the dense material used in rolls 3 and 4 on the other, makes it possible to compare the transfer of two different surface textures to a steel wire. The raw material for the rolling procedure was an austenitic stainless steel wire with a circular cross-section. The composition of the wire material is shown in Table 2. Before rolling, the steel wire had been annealed to obtain a yield strength of 300–450 N/mm2 , and acid pickled. Before the final flat rolling between the ceramic rolls, the wire had passed three consecutive steps of flat rolling between metallic rolls, with approximately 29% thickness reduction at each step. The thickness tolerance allowed for the flat rolled product was ±0.015 mm and the tensile strength requirement after rolling was below 1900 N/mm2 . With respect to the applications in which the rolled wire is used, a low Rsk value, i.e. a profile comprising more cavities and less asperities, is preferred. Prior to the tests, the surface roughness of flat-rolled wire of the same material and sizes within the present range, sampled from normal production, was measured; the results are shown in Table 3. The rolling oil was a mineral oil with additives containing zinc, phosphorus and sulphur, and water, at concentrations of a few hundred ppm each. The viscosity of the oil was 18.3 mm2 /s at 40 ◦ C. The circulating oil was continuously filtered, but contained micrometer-sized ferrous particles with alloying elements typical of carbon steel and stainless steel. At the time of the present tests, the proportion of the ferrous wear particles in the oil was roughly 1.5 wt%.
Table 3 Surface roughness parameters (see details in Section 2.1) for flat rolled austenitic steel wire from normal production Steel wire sample
The experiments consisted of the final flat rolling of steel wire between a pair of ceramic rolls. The steel wire was pre-rolled flat in the first three rolling steps, and in the final rolling stage the thickness reduction of the steel wire was approximately 26%. In the tests, the velocity of the steel wire after the final rolling stage was about 5 m/s. The rolling tests were carried out on spots located 10–15 mm from the drive-ends of the ceramic rolls. After the tests, the surface roughness of the roll, wire and rolling track surfaces was measured. Two categories of experiments were performed: (i) A pre-test with rolls 1 (upper) and 2 (lower), which were made of two different silicon nitride materials, both with a high degree of open porosity. The dimension of the steel wire rolled in the pre-test was 3.2 mm × 0.4 mm and the rolled quantity was about 43 km or 500 kg. (ii) The main test program with rolls 3 (upper) and 4 (lower), which were made of a dense silicon nitride material. The final dimensions of the rolled wire in the main test program were 3.2 mm × 0.4 mm, 3.8 mm × 0.5 mm and 3.6 mm × 0.7 mm. The total weight of the steel wire rolled in the main test program was 632 km or about 7000 kg.
3. Results 3.1. Pre-test Almost instantly in the pre-test, thin layers of bright metal were transferred from the wire to the rolling tracks of the ceramic rolls. The electrical conductivity of the layer, which was determined by measurements after the tests, revealed the metallic nature of the layer. The metallic transfer layers remained on the rolls throughout the test. At the end of the pre-test, when 21 km of steel wire had been rolled flat, the thickness of the metallic layer was roughly 0.1 m. The surface roughness of the rolling tracks is shown in Table 4. Quality control carried out on the wire produced in the pre-test showed that the surface of the wire was too rough, with high Ra values and a significant proportion of surface peaks (high Rp values), as seen in Table 5. The surfaces of ceramic rolls 1 and 2, which were used in the pre-test, comprised a significant amount of open porosity, which was reflected by a corresponding pattern of microscopic peaks, ridges and bulges on the surfaces of the rolled wire. An example of Table 5 Surface roughness parameters for a wire sample from the end of the pre-test
Surface roughness parameters
Dimensions [mm]
Surface
Ra [m]
Rsk
Rp [m]
Rv [m]
3.2 × 0.46 3.2 × 0.46 3.2 × 0.38 3.2 × 0.38
Upper Lower Upper Lower
0.05 0.05 0.08 0.06
0.3 0.1 0.2 0.1
0.25 0.22 0.43 0.29
0.18 0.20 0.32 0.26
Wire surface
Upper Lower
Surface roughness parameters Ra [m]
Rsk
Rp [m]
Rv [m]
0.16 0.15
−0.1 0.3
0.54 0.58
0.47 0.38
P. Andersson et al. / Journal of Materials Processing Technology 173 (2006) 394–400
397
Fig. 4. Bulges, marked “A”, “B” and “C” (left image, SEM), on the rolled wire surface owing to open porosity (right image, light optical microscopy) on the rolling track that had been covered with a thin layer of metal.
large-sized porosity on the metal-covered roll surface and the corresponding bulges on the wire surface is shown in Fig. 4. The tensile strength of the wire was about 1510 MPa. After 1.5 h of rolling in the pre-test, the surface temperature of the ceramic rolls was about 42 ◦ C and that of the steel wire coil 35 ◦ C. The temperature remained almost the same in subsequent tests. 3.2. Main test program In the main test program, which employed rolls with a higher surface quality, the steel wire produced was of uniform and acceptable quality. Surface roughness parameters for wire samples of different dimensions from the main test program are listed in Table 6. The thickness of the rolled wire showed slightly larger variations than for normal production wire, being however still within the tolerances specified. The strength of the wire samples was 1310–1630 MPa.
A layer of steel was transferred onto the rolling tracks of the ceramic rolls immediately at the beginning of the first test, and remained on the rolls during the main test program. At the end of the main test program, when 632 km or 7 tonnes of steel wire had been rolled on the same track, the thickness of the steel transfer layer was 0.3 m or less above the original surface of the roll, on both rolls 3 and 4. The thickest transfer layer sections occurred at the edges of the rolling track, as seen in the example in Fig. 5. In addition to the layer, narrow grooves and minor pits with a depth of 0.7 m or less had formed within the area covered by the metal layers, as shown in Fig. 6. Surface roughness parameters for the rolling tracks after all the main tests are shown in Table 7. The production rate obtained in the main test program was equal to the corresponding rate used in normal production with metallic rolls. The wear of the ceramic rolls was so small that it was unnecessary to change the rolling spot, i.e. to use more than a single rolling track. In cold-rolling of stainless steel with metallic rolls, the rolling spot normally has to be
Table 6 Surface roughness parameters for wire samples from the main test program Wire sample identifications
Surface roughness parameters
No.
Dimensions [mm]
Sampling
Wire surface
Ra [m]
Rsk
Rp [m]
Rv [m]
2
3.24 × 0.42
When 153 km of wire 3.24 mm × 0.42 mm had been rolled
Upper Lower
0.05 0.06
−1.6 −1.4
0.11 0.15
0.25 0.23
3
3.76 × 0.51
When 16 km of wire 3.76 mm × 0.51 mm had been rolled
Upper Lower
0.09 0.09
−1.7 −2.1
0.20 0.20
0.54 0.54
4
3.76 × 0.51
When 132 km of wire 3.76 mm × 0.51 mm had been rolled
Upper Lower
0.09 0.08
−2.7 −1.5
0.19 0.21
0.68 0.45
5
3.24 × 0.42
When 22 km of wire 3.24 mm × 0.42 mm had been rolled
Upper Lower
0.05 0.05
−1.6 −1.2
0.10 0.13
0.25 0.23
6
3.24 × 0.42
When 341 km of wire 3.24 mm × 0.42 mm had been rolled
Upper Lower
0.05 0.05
−1.8 −1.6
0.11 0.12
0.27 0.28
7
3.56 × 0.75
When 4 km of wire 3.56 mm × 0.75 mm had been rolled
Upper Lower
0.06 0.07
−5.4 −1.0
0.14 0.21
0.53 0.33
398
P. Andersson et al. / Journal of Materials Processing Technology 173 (2006) 394–400
4.1. Transfer layer formation on the roll surfaces
Fig. 5. Example showing variations across the track width in the thickness of the metallic layer, and the grooves formed, on roll 3 during the main test program. The rolling track is located approximately between the vertical dotted lines.
Fig. 6. Example showing a photomicrograph of the rolling track formed on roll 3 during the main test program. The edge regions of the track (dark areas) are covered by a metallic layer, while the silicon nitride roll material (light areas) is partly exposed in the central region of the track. The rolling direction is vertical (light optical microscope).
Table 7 Surface roughness parameters for the rolling track of the silicon nitride ceramic rolls after the main test program Roll identification no.
3 4
Surface roughness parameters for the rolling track Ra [m]
Rsk
Rp [m]
Rv [m]
0.04 0.06
−0.3 −0.3
0.16 0.21
0.22 0.29
changed quite frequently. Inspection of the position marks made prior to the tests showed that no tangential sliding motion had occurred between the rolls and the support shafts during the tests.
During the tests, the silicon nitride rolls revealed surface changes in the form of layers of transferred metal and as grooves or pits. The Ra and Rp values of the rolling tracks increased slightly. Most likely, the transferred metal layers that formed in the pre-test and main tests were mainly transferred to the ceramic surfaces during adhesive and micro-abrasive wear of the wire. A contribution from compaction of wear particles from the oil is possible. The thickness of the metal transfer layers on the ceramic rolls seems to be determined by layer growth and disruption caused by the sliding parameters including lubrication. In cold-rolling, transfer of metal from a wire to a roll surface is promoted by pressure and relative sliding between roll and wire. The sliding occurs in the direction of rolling, before and after the neutral plane (Fig. 1), and due to the lateral spread of wire material that occurs simultaneously with the reduction in the wire thickness. According to Carlsson [11], the highest contact pressure, at least in the first rolling pass, occurs over an area shaped like a horse shoe, symmetrically to the rolling direction. This is in agreement with the ferrous transfer layers in the present rolling tracks (Figs. 5 and 6), which were thicker at the track edges than in the track centre. At the same time the transfer of metal is limited by the oil film. Additives based on sulphur, zinc and phosphorus are commonly used for reducing the coefficient of sliding friction between wire and roll in normal cold-rolling processes. The additives act through the formation of soft and easily sheared reaction product layers on the wire and roll surfaces. In the present tests, however, protective boundary layers probably did not form at a sufficiently high rate with the present additive concentrations on the stainless steel and silicon nitride surfaces with low chemical activity. Owing to the lack of protective reaction films on the ceramic surfaces, the wire material was able to adhere to the silicon nitride material during sliding contact under the contact stress applied. Metallic transfer layers are known to form on metallic rolls, particularly in the cold-rolling of stainless steel and similar well-adhering metals [12]. With silicon nitride in oil-lubricated rolling and partially sliding contact with hardened carbon steel surfaces, such as camshafts or steel rings in hybrid ball bearings, ferrous transfer layers are not usually formed. In oil-lubricated sliding contacts with silicon nitride against stainless steel, where the frictional work is stronger than in rolling contacts, formation of ferrous transfer layers on the ceramic surfaces has been observed [13]. To a limited extent, ferrous transfer layers have been observed on silicon nitride surfaces after sliding tests against hardened carbon steel under lubrication with mineral oil [14–16].
4. Discussion 4.2. Wear phenomena at the ceramic rolls The results of these tests show that with the use of silicon nitride rolls, it is possible in practice to obtain high production rates and rolled products of good dimensional stability and surface quality. The rolls survived the tests without macroscopic damage and stayed firmly attached to the steel support shafts.
When sliding on another solid in the presence of mineral oil, ceramics normally wear through brittle fracturing, including grain pull-out on a micro-scale. Under similar conditions, metals wear mainly by abrasion and surface fatigue involving plastic deformation and dislocation movement.
P. Andersson et al. / Journal of Materials Processing Technology 173 (2006) 394–400
399
Wear scars with grooves and pits occurred on the rolling tracks after the tests. The deterioration of the ceramic surfaces probably occurred due to tangential forces caused by high friction and partial slip between roll and wire, which caused mechanical and thermal stresses in the silicon nitride surfaces. The partial slip resulted from velocity differences in the rolling direction before and after the neutral plane, and due to the lateral spread of the wire. The work hardening of the present steel wire material increased the wear of the rolls. Repeated stress concentrations at large open pores on the ceramic surfaces may have caused pore enlargement by edge cracking. It is most likely that some proportion of the damages on the ceramic surfaces during the tests was covered by subsequent layers of transferred steel and therefore was not observed by profilometry or microscopy. 4.3. Wire surface quality in the pre-test In the pre-test, the coarse and semi-deterministic [2] surface structures of the porous ceramic rolls were transferred to the surfaces of the steel wire by plastic deformation of the wire surface against the ceramic rolls in the rolling gap. The transfer of surface texture from roll to wire is obvious from the individual surface impressions shown in Fig. 4 and from the level of the surface roughness parameters; the arithmetic average surface roughness of the wire, Ra = 0.16/0.15 m, corresponds well to the roll surface roughness, Ra = 0.21/0.16 m, at the end of the pre-test, and it is significantly higher than that of the normal production wire. The peak height of the surface profile for the wire, Rp = 0.54/0.58 m, was about one-half of the valley depth of the roll surface profile, Rv = 1.1/1.2 m, at the end of the pre-test. Correspondingly, the valley depth of the wire, Rv = 0.47/0.38 m, was about two-thirds of the peak height of the roll surface profile Rp = 0.65/0.54 m, at the end of the pretest. The difference in the above Rp and Rv values, in the range 0.2–0.6 m, seems to be the lower limit for the replication of minor details of the coarse ceramic roll surfaces onto the steel wire, with the present materials, pressure, velocities, lubricant and partial slip between wire and roll involved. A comparison of mean values for the Ra, Rp and Rv surface roughness parameters for the wire and roll surfaces, respectively, is presented schematically in Fig. 7. The transfer of the roll surface asperity features to the wire surface profile was more complete than the transfer of the roll surface valley features to the wire surface profile in the pretest. A complete replication would have given opposed skewness (Rsk) values for the roll and wire surface profiles, respectively, which was not the case in the present tests. The oil film between roll and wire, and hydrostatically compressed oil in cavities in the ceramic surface, seems to have reduced the proportion of peaks in the surface profile of the steel wire and thus the Ra, Rp and Rsk values of the wire. The wear of the wire during rolling, obvious from the ferrous transfer layers on the rolls, as well as the known partial slip between wire and roll, oil in the roll–wire interface and the wire hardness, contributed to the limitation of the surface texture transfer from roll to wire. The observation of
Fig. 7. Mean values of the Ra, Rp and Rv surface roughness parameters, for roll and wire surfaces at the end of the pre-test and the main tests.
partial surface texture transfer is in agreement with observations in earlier studies by other authors [2–4]. 4.4. Wire surface quality in the main test program The Ra values for all the wire samples from the main test program (Ra = 0.05–0.09 m) were similar to the corresponding values for the normal production wire, and significantly lower than for the wire produced in the pre-tests. The variations in the Ra values for the steel wire samples taken in the main tests reflect changes in the rolling track profile. In general, rolled wire with smaller dimensions developed a better surface quality, probably due to the smaller forces involved and because only part of the metal-covered rolling track width was utilised in the rolling action. The range of steel wire Ra values from the main test program correlates well with the corresponding values for the ceramic rolls, which were Ra = 0.03–0.04 m before the main tests and Ra = 0.04–0.06 m after the tests. When both unused and used roll surfaces are considered, the transfer of the Ra value from the ceramic roll onto the rolled steel wire, which normally is the main concern in common quality assurance, took place with an accuracy of approximately 60 nm or better for all the wire samples from the main tests. This is in good agreement with the work by Plourabo´ue and Boehm, who report transfer of surface texture from steel roll to aluminium sheet on a scale down to 50 nm at a sheet thickness reduction of 30% [3]. The Rp values for the wire samples from the main test program were generally lower than those for the normal production wire samples. The Rp values for wire sample 7 taken at the end of all the main tests (Rp = 0.14/0.21 m) and the Rv values for the ceramic rolls (Rv = 0.22/0.29 m) at the end of the main tests show a fairly good correlation. The Rv values, or cavity deepness indicators, for the wire samples from the main test program were similar or higher than those for the normal production wire samples. Similarly, the Rsk values were slightly above zero for the normal production wire samples and clearly below zero, indicating cavities
400
P. Andersson et al. / Journal of Materials Processing Technology 173 (2006) 394–400
in the surface profile, for all the wire samples from the main test program. The Rv values for wire sample 7 taken at the end of the main tests, Rv = 0.53/0.33 m, and the Rp values for the ceramic rolls, Rp = 0.16/0.21 m, at the end of the main tests, however, have a weak correlation; the difference in the above Rp and Rv values is of the order of 0.1–0.3 m. On the other hand, the Rv values for wire sample 7 are close to those for the coarse wire from the pre-test, as shown in Fig. 7. The high Rv values for most of the wire surfaces that had been produced with ceramic rolls may be attributed to the metallic transfer layers on the rolls, which indicate high friction between roll and wire and a potential for scratching of wire surfaces during partial slip. In the main tests, individual scratches on generally smooth wire surface profiles caused strongly negative Rsk values. This is different from the effect in the pre-test, where the Rsk values remained close to zero in spite of high Rv values; the reason for this is that the Ra value was higher and the profile was symmetric because of high surface peaks or Rp values. However, before further conclusions are made, it should be taken into consideration that the Rp and Rv values are statistically less significant than the Ra value. As in the pre-test, the reason for the limitation in the transfer of surface texture from roll to wire is probably the oil film between them, and hydrostatically compressed oil in cavities in the ceramic surface that smoothens the roll surface during the rolling event. Other factors limiting the surface texture transfer are the partial slip, the difference in contact pressure at surface asperities and cavities, the wire material hardness, the time available for material flow and the wear of the wire during rolling. 5. Conclusions Based on the present experiments with silicon nitride rolls in a commercial cold-rolling mill, the following conclusions can be made: The surface texture of the smooth rolls was transferred onto the rolled product with an accuracy of 60 nm or better regarding the Ra value and a few hundred nanometers regarding the Rp and Rv values of the roll surface profile. The wear rate of the silicon nitride rolls used in the cold-rolling of steel was low. This reduced the need for adjustments of the rolling installation during rolling. Some wear due to microabrasion occurred on the ceramic surface. The profile changes caused by wear were partly compensated by ferrous material transfer from the wire. The strength of sintered silicon nitride was found sufficient for allowing cold-rolling of austenitic stainless steel wire with a small cross-section between a pair of rolls made from this ceramic. The rolls stayed firmly attached to the steel support shafts. Acknowledgements The present work was part of the Brite-Euram Project BRPRCT96-0343, Large Components of Silicon Nitride Ceramics for
Rolling Operations in the Steel-Working Industry. The authors are grateful for the financial support from the European Union that made the study possible. The authors express their thanks to Mr. Heinz-Withold Schmitz of H.C. Starck, Dr. Jean-Michel Drouin of Saint-Gobain Desmarquest, Dr. Rolf Wagner of FCT Ingenieurkeramik GmbH, Dr. Gerhard W¨otting of Ceramics For Industry CFI (presently part of H.C. Starck Ceramics) and Dr. Andreas Wagemann of BeaTec GmbH for the raw material powders and the production of the rolls. Mr. Ricard Ohlsson, Mr. ¨ Anders Lundvall, Mr. Ola Helmfridsson and Mr. Leif Osterdahl of Haldex Garphyttan Wire AB are acknowledged for their efforts during the cold-rolling actions, Mr. Jan Jansson of Haldex Garphyttan Wire AB for SEM microscopy and Mr. Ilkka Raeluoto of the Finnish Centre for Metrology and Accreditation for surface roughness measurements. The authors thank Mr. David Thureborn of Haldex Garphyttan Wire AB for his comments on the present work. References [1] G.D. Lahoti, S.L. Semiatin, Flat, bar and shape rolling, in: Metals Handbook, vol. 14, Forming and Forging, ninth ed., ASM International, Metals Park, OH, USA, 1988, ISBN 0-87170-020-4, pp. 343– 360. [2] R. B¨unten, K. Steinhoff, W. Rasp, R. Kopp, O. Pawelski, Development of a FEM model for the simulation of the transfer of surface structure in cold-rolling processes, J. Mater. Technol. 60 (1996) 369–376. [3] F. Plourabo´ue, M. Boehm, Multi-scale roughness transfer in cold metal rolling, Tribol. Int. 32 (1999) 45–57. [4] B. Ma, A.K. Tieu, C. Lu, Z. Jiang, An experimental investigation of steel surface characteristic transfer by cold rolling, J. Mater. Process. Technol. 125–126 (2002) 657–663. [5] H.C. Xie, D.R. Chen, X.M. Kong, An analysis of the three-dimensional surface topography of textured cold-rolled steel sheets, Tribol. Int. 32 (2) (1999) 83–87. [6] A.T. Galbato, R.T. Cundill, T.A. Harris, Fatigue life of silicon nitride balls, Lubr. Eng. 48 (11) (1992) 886–894. [7] Y.P. Chiu, P.K. Pearson, M. Dezzani, H. Daverio, Fatigue life and performance testing of hybrid ceramic ball bearings, Lubr. Eng. 52 (3) (1996) 198–204. [8] L. Wang, R.W. Snidle, L. Gu, Rolling contact silicon nitride bearing technology: a review of recent research, Wear 246 (2000) 159–173. [9] J.F. Braza, Ceramic roller evaluation in valve train tests, Tribol. Trans. 38 (1) (1995) 146–152. [10] ISO 4287:1997, International Standard. Geometrical Product Specifications (GPS)—Surface Texture: Profile Method—Terms, Definitions and Surface Texture Parameters, 25 pp. [11] B. Carlsson, The contact pressure distribution in flat rolling of wire, J. Mater. Process. Technol. 73 (1998) 1–6. [12] P. Montmitonnet, F. Delamare, B. Rizoulieres, Transfer layers and friction in cold metal strip rolling processes, Wear 245 (2000) 125–135. [13] X. Zhao, J. Liu, B. Zhu, H. Miao, Z. Luo, Tribological characteristics of Si3 N4 ceramic sliding on stainless steel, Wear 206 (1997) 76–82. [14] J. Denape, A. Marzinotto, J.A. Petit, Roughness effect of silicon nitride sliding on steel under boundary lubrication, Wear 159 (1992) 173– 184. [15] A.J. Winn, D. Dowson, J.C. Bell, The lubricated wear of ceramics, Part I: the wear and friction of silicon nitride, alumina and steel in the presence of a mineral oil based lubricant, Tribol. Int. 28 (6) (1995) 383–393. [16] T. Morimoto, K. Kamikawa, Friction and wear in silicon nitride – steel and cemented carbide – steel pairs in lubricated sliding, Tribol. Int. 29 (7) (1996) 537–546.
Transfer of surface texture from silicon nitride rolls to stainless steel wire in cold-rolling Peter Andersson a,∗ , Michael Wild b , Jan Lev´en c , Bj¨orn Hemming d a
VTT Industrial Systems, P.O. Box 1702, FIN-02044 VTT, Espoo, Finland b Haldex Garphyttan Wire AB, S-71980 Garphyttan, Sweden c Mefos-Metallurgical Research Institute AB, Box 812, S-97125 Lule˚ a, Sweden d Centre for Metrology and Accreditation, Tekniikantie 1, FIN-02150 Espoo, Finland Received 6 January 2005; received in revised form 6 December 2005; accepted 7 December 2005
Abstract A set of cold-rolling experiments with ceramic rolls was carried out using facilities for the production of profiled steel wire. Based on the experiments, the strength of sintered silicon nitride was found sufficient for allowing cold-rolling of austenitic stainless steel wire. The wear rate of the silicon nitride rolls in the tests was low. Metal was transferred from the wire to the rolling tracks of the ceramic rolls. The surface texture of the rolls was reproduced on the rolled product. © 2006 Elsevier B.V. All rights reserved. Keywords: Silicon nitride; Cold-rolling; Wire rolling; Surface transfer; Wear; Surface roughness; Texture
1. Introduction Rolling is a three-dimensional plastic flow process used for the forming of metallic sheet material and profiles. In rolling, a work piece material is brought into a gap between two rolls. Due to the reduction of the material thickness in the work zone, the work piece material is accelerated first, before the neutral plane of the work zone, to the surface velocity of the rolls, then to a velocity higher than the surface velocity of the rolls, as seen in Fig. 1. When a wire material is rolled, this spreads laterally in the work zone during the thickness reduction, and the rolling process results in a cross-section that is lower and wider and has an area that is smaller than that of the work piece material [1]. The surface quality of cold-rolled sheet and wire materials is normally of great importance, as the rolled surface often forms a functional surface of an end product. The surface quality of the cold-rolled product is sensitive to the surface quality of the rolls, and for this reason particular emphasis is put on the roll surface finish. For certain purposes, textured rolls that give a desired surface structure to the cold-rolled material are used [2,3]. In addition to the roll surface modifications, alterations in
∗
Corresponding author. E-mail address: [email protected] (P. Andersson).
0924-0136/$ – see front matter © 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.jmatprotec.2005.12.009
the lubrication, velocity and degree of reduction, for instance, provide means to control the surface quality of the cold-rolled product [1–5]. For more than a decade, hybrid ball bearings composed of silicon nitride ceramic balls and steel races have been successfully used particularly in high-speed applications [6–8]. One of the benefits of the use of silicon nitride ceramic balls is the diminished need for lubricating oil, owing to the good surface quality that can be achieved on a silicon nitride surface by polishing. Similarly, silicon nitride rolls that operate against steel camshafts in engines have been successful [9]. Experience with hybrid ball bearings and ceramic cam followers has shown that silicon nitride is beneficial in rolling contacts, which has created interest in the use of this ceramic material in metal rolling applications. A question frequently raised when ceramic materials for machine elements and tools are discussed is the issue of the strength and reliability of these non-metallic materials. One of the central approaches of the present study was to evaluate the performance and reliability of silicon nitride rolls in actual steelrolling processes. The other approaches comprised tribological aspects of cold-rolling with ceramic rolls, including the transfer of the roll surface texture (surface profile waviness and surface roughness) onto the work piece material. In order to facilitate the evaluation, a set of cold-rolling experiments was carried
P. Andersson et al. / Journal of Materials Processing Technology 173 (2006) 394–400
395
Fig. 1. Schematic presentation of the cold-rolling of wire; h1 is the entry thickness and h2 the exit thickness of the wire. Due to cross-section area reduction, the wire exit velocity v2 is higher than the entry velocity v1 . The line NP represents the neutral plane, at which the contact pressure between roll and wire reaches a maximum and the roll and wire surfaces have the same velocity v. Wire tension at entry or exit zones can move the neutral plane, but was not applied in the present tests.
out using production facilities for cold-rolling of profiled steel wire. In the following, the preparations and procedures for the experiments and the post-test investigations are presented and discussed. 2. Experimental 2.1. Equipment The experiments were carried out using a cold-rolling mill at Haldex Garphyttan Wire AB. The rolling mill had four consecutive rolling passes, each with a two-high roll configuration, the last one of which accommodated the ceramic rolls. During the rolling procedure, rolling oil from a circuit with filtration was sprayed from nozzles onto the rolls and wire. An overview of the rolling mill is shown in Fig. 2 and a detailed view of the first ceramic roll pair before the tests is shown in Fig. 3.
Fig. 3. Exit zone of the final rolling pass with ceramic rolls 1 and 2 and two oil nozzles. Taylor–Hobson Form Talysurf diamond stylus equipment was used for measuring the surface roughness of the ceramic roll surfaces, the wire surfaces and the rolling tracks on the ceramic rolls. All the surface roughness measurements were done perpendicularly to the rolling direction of the wire and roll samples. Of the measurement results used in the present study, the surface roughness parameter Ra represents the arithmetic mean surface roughness [10], as an average value for five individual measurements of at least 5 × 0.25 mm length each. The Ra value, however, carries no information about the shape of the surface profile. The Rsk value expresses the surface profile skewness as a mean value for five individual measurements of at least 5 × 0.25 mm length each; for Rsk > 0 the surface profile is dominated by peaks while for Rsk < 0 the surface profile is dominated by valleys. The surface roughness parameter Rp indicates the highest peak and Rv the deepest valley of the surface profile within a measurement. The Rp and Rv values given in the tables are mean values for five individual measurements of at least 5 × 0.25 mm length each. All surface roughness values in the present work are based on filtered profiles from measurements using standard cut-off lengths (λc ) of 0.25 or 0.8 mm. The values for the wire samples are presented with reference to the upper and lower surfaces, of which the former were formed against upper rolls and the latter against lower rolls of the rolling equipment.
2.2. Materials
Fig. 2. The cold-rolling mill used for the present experiments. The final rolling pass is at the right end of the mill.
Four ceramic rolls, mounted on individual steel support shafts, were used in the experiments, in sets of two rolls each. The ceramic materials for the rolls, with a hollow cylinder shape, had been manufactured by cold isostatic pressing and gas pressure sintering (GPS) from Si3 N4 powder with a few percent of binder and sintering additives. The as-sintered hollow cylinders had been ground on their inner and outer surfaces, and mounted with a conical press fit on steel shafts. When tightly mounted on the shafts, the outer cylindrical surfaces of the
396
P. Andersson et al. / Journal of Materials Processing Technology 173 (2006) 394–400 Table 4 Surface roughness parameters for the rolling tracks of the silicon nitride ceramic rolls after the cold-rolling pre-test
Table 1 Silicon nitride ceramic rolls used in the cold-rolling experiments Roll identifications
Surface roughness parameters
Roll no.
Colour
Ra [m]
Rsk
Rp [m]
Rv [m]
1 2 3 4
Greyish black Light grey Dark grey Dark grey
0.20 0.13 0.03 0.04
−0.4 −3.3 −1.1 −0.1
0.62 0.47 0.11 0.16
0.99 2.4 0.29 0.21
Roll identification no. Surface roughness parameters for the rolling track Ra [m] Rsk Rp [m] Rv [m] 1 2
0.21−0.6 0.65 0.16−1.1 0.54
1.1 1.2
2.3. Experimental procedure Table 2 Proportions of principal alloying elements of the steel wire material Element
Proportion [%]
C
Si
Mn
Cr
Ni
0.1
0.5
6
17
5
ceramic rolls had been ground and polished concentrically with the shafts. The final dimensions of the roll surfaces were Ø 150 mm × 100 mm and the wall thickness was 20–20.5 mm. The individual rolls and selected surface roughness parameters are presented in Table 1. For comparison, it should be noted that the Ra value of metallic rolls during similar cold-rolling operations normally lies between 0.04 and 0.2 m. The surfaces of rolls 1 and 2 were generally flat with a high degree of open porosity, which was reflected by high Ra and Rv surface roughness values. Most of the pores were sufficiently small to be reflected by the surface roughness parameters. Rolls 3 and 4 were made from a dense material that gave a low surface roughness. The poor integrity of the materials in rolls 1 and 2 on the one hand, and the dense material used in rolls 3 and 4 on the other, makes it possible to compare the transfer of two different surface textures to a steel wire. The raw material for the rolling procedure was an austenitic stainless steel wire with a circular cross-section. The composition of the wire material is shown in Table 2. Before rolling, the steel wire had been annealed to obtain a yield strength of 300–450 N/mm2 , and acid pickled. Before the final flat rolling between the ceramic rolls, the wire had passed three consecutive steps of flat rolling between metallic rolls, with approximately 29% thickness reduction at each step. The thickness tolerance allowed for the flat rolled product was ±0.015 mm and the tensile strength requirement after rolling was below 1900 N/mm2 . With respect to the applications in which the rolled wire is used, a low Rsk value, i.e. a profile comprising more cavities and less asperities, is preferred. Prior to the tests, the surface roughness of flat-rolled wire of the same material and sizes within the present range, sampled from normal production, was measured; the results are shown in Table 3. The rolling oil was a mineral oil with additives containing zinc, phosphorus and sulphur, and water, at concentrations of a few hundred ppm each. The viscosity of the oil was 18.3 mm2 /s at 40 ◦ C. The circulating oil was continuously filtered, but contained micrometer-sized ferrous particles with alloying elements typical of carbon steel and stainless steel. At the time of the present tests, the proportion of the ferrous wear particles in the oil was roughly 1.5 wt%.
Table 3 Surface roughness parameters (see details in Section 2.1) for flat rolled austenitic steel wire from normal production Steel wire sample
The experiments consisted of the final flat rolling of steel wire between a pair of ceramic rolls. The steel wire was pre-rolled flat in the first three rolling steps, and in the final rolling stage the thickness reduction of the steel wire was approximately 26%. In the tests, the velocity of the steel wire after the final rolling stage was about 5 m/s. The rolling tests were carried out on spots located 10–15 mm from the drive-ends of the ceramic rolls. After the tests, the surface roughness of the roll, wire and rolling track surfaces was measured. Two categories of experiments were performed: (i) A pre-test with rolls 1 (upper) and 2 (lower), which were made of two different silicon nitride materials, both with a high degree of open porosity. The dimension of the steel wire rolled in the pre-test was 3.2 mm × 0.4 mm and the rolled quantity was about 43 km or 500 kg. (ii) The main test program with rolls 3 (upper) and 4 (lower), which were made of a dense silicon nitride material. The final dimensions of the rolled wire in the main test program were 3.2 mm × 0.4 mm, 3.8 mm × 0.5 mm and 3.6 mm × 0.7 mm. The total weight of the steel wire rolled in the main test program was 632 km or about 7000 kg.
3. Results 3.1. Pre-test Almost instantly in the pre-test, thin layers of bright metal were transferred from the wire to the rolling tracks of the ceramic rolls. The electrical conductivity of the layer, which was determined by measurements after the tests, revealed the metallic nature of the layer. The metallic transfer layers remained on the rolls throughout the test. At the end of the pre-test, when 21 km of steel wire had been rolled flat, the thickness of the metallic layer was roughly 0.1 m. The surface roughness of the rolling tracks is shown in Table 4. Quality control carried out on the wire produced in the pre-test showed that the surface of the wire was too rough, with high Ra values and a significant proportion of surface peaks (high Rp values), as seen in Table 5. The surfaces of ceramic rolls 1 and 2, which were used in the pre-test, comprised a significant amount of open porosity, which was reflected by a corresponding pattern of microscopic peaks, ridges and bulges on the surfaces of the rolled wire. An example of Table 5 Surface roughness parameters for a wire sample from the end of the pre-test
Surface roughness parameters
Dimensions [mm]
Surface
Ra [m]
Rsk
Rp [m]
Rv [m]
3.2 × 0.46 3.2 × 0.46 3.2 × 0.38 3.2 × 0.38
Upper Lower Upper Lower
0.05 0.05 0.08 0.06
0.3 0.1 0.2 0.1
0.25 0.22 0.43 0.29
0.18 0.20 0.32 0.26
Wire surface
Upper Lower
Surface roughness parameters Ra [m]
Rsk
Rp [m]
Rv [m]
0.16 0.15
−0.1 0.3
0.54 0.58
0.47 0.38
P. Andersson et al. / Journal of Materials Processing Technology 173 (2006) 394–400
397
Fig. 4. Bulges, marked “A”, “B” and “C” (left image, SEM), on the rolled wire surface owing to open porosity (right image, light optical microscopy) on the rolling track that had been covered with a thin layer of metal.
large-sized porosity on the metal-covered roll surface and the corresponding bulges on the wire surface is shown in Fig. 4. The tensile strength of the wire was about 1510 MPa. After 1.5 h of rolling in the pre-test, the surface temperature of the ceramic rolls was about 42 ◦ C and that of the steel wire coil 35 ◦ C. The temperature remained almost the same in subsequent tests. 3.2. Main test program In the main test program, which employed rolls with a higher surface quality, the steel wire produced was of uniform and acceptable quality. Surface roughness parameters for wire samples of different dimensions from the main test program are listed in Table 6. The thickness of the rolled wire showed slightly larger variations than for normal production wire, being however still within the tolerances specified. The strength of the wire samples was 1310–1630 MPa.
A layer of steel was transferred onto the rolling tracks of the ceramic rolls immediately at the beginning of the first test, and remained on the rolls during the main test program. At the end of the main test program, when 632 km or 7 tonnes of steel wire had been rolled on the same track, the thickness of the steel transfer layer was 0.3 m or less above the original surface of the roll, on both rolls 3 and 4. The thickest transfer layer sections occurred at the edges of the rolling track, as seen in the example in Fig. 5. In addition to the layer, narrow grooves and minor pits with a depth of 0.7 m or less had formed within the area covered by the metal layers, as shown in Fig. 6. Surface roughness parameters for the rolling tracks after all the main tests are shown in Table 7. The production rate obtained in the main test program was equal to the corresponding rate used in normal production with metallic rolls. The wear of the ceramic rolls was so small that it was unnecessary to change the rolling spot, i.e. to use more than a single rolling track. In cold-rolling of stainless steel with metallic rolls, the rolling spot normally has to be
Table 6 Surface roughness parameters for wire samples from the main test program Wire sample identifications
Surface roughness parameters
No.
Dimensions [mm]
Sampling
Wire surface
Ra [m]
Rsk
Rp [m]
Rv [m]
2
3.24 × 0.42
When 153 km of wire 3.24 mm × 0.42 mm had been rolled
Upper Lower
0.05 0.06
−1.6 −1.4
0.11 0.15
0.25 0.23
3
3.76 × 0.51
When 16 km of wire 3.76 mm × 0.51 mm had been rolled
Upper Lower
0.09 0.09
−1.7 −2.1
0.20 0.20
0.54 0.54
4
3.76 × 0.51
When 132 km of wire 3.76 mm × 0.51 mm had been rolled
Upper Lower
0.09 0.08
−2.7 −1.5
0.19 0.21
0.68 0.45
5
3.24 × 0.42
When 22 km of wire 3.24 mm × 0.42 mm had been rolled
Upper Lower
0.05 0.05
−1.6 −1.2
0.10 0.13
0.25 0.23
6
3.24 × 0.42
When 341 km of wire 3.24 mm × 0.42 mm had been rolled
Upper Lower
0.05 0.05
−1.8 −1.6
0.11 0.12
0.27 0.28
7
3.56 × 0.75
When 4 km of wire 3.56 mm × 0.75 mm had been rolled
Upper Lower
0.06 0.07
−5.4 −1.0
0.14 0.21
0.53 0.33
398
P. Andersson et al. / Journal of Materials Processing Technology 173 (2006) 394–400
4.1. Transfer layer formation on the roll surfaces
Fig. 5. Example showing variations across the track width in the thickness of the metallic layer, and the grooves formed, on roll 3 during the main test program. The rolling track is located approximately between the vertical dotted lines.
Fig. 6. Example showing a photomicrograph of the rolling track formed on roll 3 during the main test program. The edge regions of the track (dark areas) are covered by a metallic layer, while the silicon nitride roll material (light areas) is partly exposed in the central region of the track. The rolling direction is vertical (light optical microscope).
Table 7 Surface roughness parameters for the rolling track of the silicon nitride ceramic rolls after the main test program Roll identification no.
3 4
Surface roughness parameters for the rolling track Ra [m]
Rsk
Rp [m]
Rv [m]
0.04 0.06
−0.3 −0.3
0.16 0.21
0.22 0.29
changed quite frequently. Inspection of the position marks made prior to the tests showed that no tangential sliding motion had occurred between the rolls and the support shafts during the tests.
During the tests, the silicon nitride rolls revealed surface changes in the form of layers of transferred metal and as grooves or pits. The Ra and Rp values of the rolling tracks increased slightly. Most likely, the transferred metal layers that formed in the pre-test and main tests were mainly transferred to the ceramic surfaces during adhesive and micro-abrasive wear of the wire. A contribution from compaction of wear particles from the oil is possible. The thickness of the metal transfer layers on the ceramic rolls seems to be determined by layer growth and disruption caused by the sliding parameters including lubrication. In cold-rolling, transfer of metal from a wire to a roll surface is promoted by pressure and relative sliding between roll and wire. The sliding occurs in the direction of rolling, before and after the neutral plane (Fig. 1), and due to the lateral spread of wire material that occurs simultaneously with the reduction in the wire thickness. According to Carlsson [11], the highest contact pressure, at least in the first rolling pass, occurs over an area shaped like a horse shoe, symmetrically to the rolling direction. This is in agreement with the ferrous transfer layers in the present rolling tracks (Figs. 5 and 6), which were thicker at the track edges than in the track centre. At the same time the transfer of metal is limited by the oil film. Additives based on sulphur, zinc and phosphorus are commonly used for reducing the coefficient of sliding friction between wire and roll in normal cold-rolling processes. The additives act through the formation of soft and easily sheared reaction product layers on the wire and roll surfaces. In the present tests, however, protective boundary layers probably did not form at a sufficiently high rate with the present additive concentrations on the stainless steel and silicon nitride surfaces with low chemical activity. Owing to the lack of protective reaction films on the ceramic surfaces, the wire material was able to adhere to the silicon nitride material during sliding contact under the contact stress applied. Metallic transfer layers are known to form on metallic rolls, particularly in the cold-rolling of stainless steel and similar well-adhering metals [12]. With silicon nitride in oil-lubricated rolling and partially sliding contact with hardened carbon steel surfaces, such as camshafts or steel rings in hybrid ball bearings, ferrous transfer layers are not usually formed. In oil-lubricated sliding contacts with silicon nitride against stainless steel, where the frictional work is stronger than in rolling contacts, formation of ferrous transfer layers on the ceramic surfaces has been observed [13]. To a limited extent, ferrous transfer layers have been observed on silicon nitride surfaces after sliding tests against hardened carbon steel under lubrication with mineral oil [14–16].
4. Discussion 4.2. Wear phenomena at the ceramic rolls The results of these tests show that with the use of silicon nitride rolls, it is possible in practice to obtain high production rates and rolled products of good dimensional stability and surface quality. The rolls survived the tests without macroscopic damage and stayed firmly attached to the steel support shafts.
When sliding on another solid in the presence of mineral oil, ceramics normally wear through brittle fracturing, including grain pull-out on a micro-scale. Under similar conditions, metals wear mainly by abrasion and surface fatigue involving plastic deformation and dislocation movement.
P. Andersson et al. / Journal of Materials Processing Technology 173 (2006) 394–400
399
Wear scars with grooves and pits occurred on the rolling tracks after the tests. The deterioration of the ceramic surfaces probably occurred due to tangential forces caused by high friction and partial slip between roll and wire, which caused mechanical and thermal stresses in the silicon nitride surfaces. The partial slip resulted from velocity differences in the rolling direction before and after the neutral plane, and due to the lateral spread of the wire. The work hardening of the present steel wire material increased the wear of the rolls. Repeated stress concentrations at large open pores on the ceramic surfaces may have caused pore enlargement by edge cracking. It is most likely that some proportion of the damages on the ceramic surfaces during the tests was covered by subsequent layers of transferred steel and therefore was not observed by profilometry or microscopy. 4.3. Wire surface quality in the pre-test In the pre-test, the coarse and semi-deterministic [2] surface structures of the porous ceramic rolls were transferred to the surfaces of the steel wire by plastic deformation of the wire surface against the ceramic rolls in the rolling gap. The transfer of surface texture from roll to wire is obvious from the individual surface impressions shown in Fig. 4 and from the level of the surface roughness parameters; the arithmetic average surface roughness of the wire, Ra = 0.16/0.15 m, corresponds well to the roll surface roughness, Ra = 0.21/0.16 m, at the end of the pre-test, and it is significantly higher than that of the normal production wire. The peak height of the surface profile for the wire, Rp = 0.54/0.58 m, was about one-half of the valley depth of the roll surface profile, Rv = 1.1/1.2 m, at the end of the pre-test. Correspondingly, the valley depth of the wire, Rv = 0.47/0.38 m, was about two-thirds of the peak height of the roll surface profile Rp = 0.65/0.54 m, at the end of the pretest. The difference in the above Rp and Rv values, in the range 0.2–0.6 m, seems to be the lower limit for the replication of minor details of the coarse ceramic roll surfaces onto the steel wire, with the present materials, pressure, velocities, lubricant and partial slip between wire and roll involved. A comparison of mean values for the Ra, Rp and Rv surface roughness parameters for the wire and roll surfaces, respectively, is presented schematically in Fig. 7. The transfer of the roll surface asperity features to the wire surface profile was more complete than the transfer of the roll surface valley features to the wire surface profile in the pretest. A complete replication would have given opposed skewness (Rsk) values for the roll and wire surface profiles, respectively, which was not the case in the present tests. The oil film between roll and wire, and hydrostatically compressed oil in cavities in the ceramic surface, seems to have reduced the proportion of peaks in the surface profile of the steel wire and thus the Ra, Rp and Rsk values of the wire. The wear of the wire during rolling, obvious from the ferrous transfer layers on the rolls, as well as the known partial slip between wire and roll, oil in the roll–wire interface and the wire hardness, contributed to the limitation of the surface texture transfer from roll to wire. The observation of
Fig. 7. Mean values of the Ra, Rp and Rv surface roughness parameters, for roll and wire surfaces at the end of the pre-test and the main tests.
partial surface texture transfer is in agreement with observations in earlier studies by other authors [2–4]. 4.4. Wire surface quality in the main test program The Ra values for all the wire samples from the main test program (Ra = 0.05–0.09 m) were similar to the corresponding values for the normal production wire, and significantly lower than for the wire produced in the pre-tests. The variations in the Ra values for the steel wire samples taken in the main tests reflect changes in the rolling track profile. In general, rolled wire with smaller dimensions developed a better surface quality, probably due to the smaller forces involved and because only part of the metal-covered rolling track width was utilised in the rolling action. The range of steel wire Ra values from the main test program correlates well with the corresponding values for the ceramic rolls, which were Ra = 0.03–0.04 m before the main tests and Ra = 0.04–0.06 m after the tests. When both unused and used roll surfaces are considered, the transfer of the Ra value from the ceramic roll onto the rolled steel wire, which normally is the main concern in common quality assurance, took place with an accuracy of approximately 60 nm or better for all the wire samples from the main tests. This is in good agreement with the work by Plourabo´ue and Boehm, who report transfer of surface texture from steel roll to aluminium sheet on a scale down to 50 nm at a sheet thickness reduction of 30% [3]. The Rp values for the wire samples from the main test program were generally lower than those for the normal production wire samples. The Rp values for wire sample 7 taken at the end of all the main tests (Rp = 0.14/0.21 m) and the Rv values for the ceramic rolls (Rv = 0.22/0.29 m) at the end of the main tests show a fairly good correlation. The Rv values, or cavity deepness indicators, for the wire samples from the main test program were similar or higher than those for the normal production wire samples. Similarly, the Rsk values were slightly above zero for the normal production wire samples and clearly below zero, indicating cavities
400
P. Andersson et al. / Journal of Materials Processing Technology 173 (2006) 394–400
in the surface profile, for all the wire samples from the main test program. The Rv values for wire sample 7 taken at the end of the main tests, Rv = 0.53/0.33 m, and the Rp values for the ceramic rolls, Rp = 0.16/0.21 m, at the end of the main tests, however, have a weak correlation; the difference in the above Rp and Rv values is of the order of 0.1–0.3 m. On the other hand, the Rv values for wire sample 7 are close to those for the coarse wire from the pre-test, as shown in Fig. 7. The high Rv values for most of the wire surfaces that had been produced with ceramic rolls may be attributed to the metallic transfer layers on the rolls, which indicate high friction between roll and wire and a potential for scratching of wire surfaces during partial slip. In the main tests, individual scratches on generally smooth wire surface profiles caused strongly negative Rsk values. This is different from the effect in the pre-test, where the Rsk values remained close to zero in spite of high Rv values; the reason for this is that the Ra value was higher and the profile was symmetric because of high surface peaks or Rp values. However, before further conclusions are made, it should be taken into consideration that the Rp and Rv values are statistically less significant than the Ra value. As in the pre-test, the reason for the limitation in the transfer of surface texture from roll to wire is probably the oil film between them, and hydrostatically compressed oil in cavities in the ceramic surface that smoothens the roll surface during the rolling event. Other factors limiting the surface texture transfer are the partial slip, the difference in contact pressure at surface asperities and cavities, the wire material hardness, the time available for material flow and the wear of the wire during rolling. 5. Conclusions Based on the present experiments with silicon nitride rolls in a commercial cold-rolling mill, the following conclusions can be made: The surface texture of the smooth rolls was transferred onto the rolled product with an accuracy of 60 nm or better regarding the Ra value and a few hundred nanometers regarding the Rp and Rv values of the roll surface profile. The wear rate of the silicon nitride rolls used in the cold-rolling of steel was low. This reduced the need for adjustments of the rolling installation during rolling. Some wear due to microabrasion occurred on the ceramic surface. The profile changes caused by wear were partly compensated by ferrous material transfer from the wire. The strength of sintered silicon nitride was found sufficient for allowing cold-rolling of austenitic stainless steel wire with a small cross-section between a pair of rolls made from this ceramic. The rolls stayed firmly attached to the steel support shafts. Acknowledgements The present work was part of the Brite-Euram Project BRPRCT96-0343, Large Components of Silicon Nitride Ceramics for
Rolling Operations in the Steel-Working Industry. The authors are grateful for the financial support from the European Union that made the study possible. The authors express their thanks to Mr. Heinz-Withold Schmitz of H.C. Starck, Dr. Jean-Michel Drouin of Saint-Gobain Desmarquest, Dr. Rolf Wagner of FCT Ingenieurkeramik GmbH, Dr. Gerhard W¨otting of Ceramics For Industry CFI (presently part of H.C. Starck Ceramics) and Dr. Andreas Wagemann of BeaTec GmbH for the raw material powders and the production of the rolls. Mr. Ricard Ohlsson, Mr. ¨ Anders Lundvall, Mr. Ola Helmfridsson and Mr. Leif Osterdahl of Haldex Garphyttan Wire AB are acknowledged for their efforts during the cold-rolling actions, Mr. Jan Jansson of Haldex Garphyttan Wire AB for SEM microscopy and Mr. Ilkka Raeluoto of the Finnish Centre for Metrology and Accreditation for surface roughness measurements. The authors thank Mr. David Thureborn of Haldex Garphyttan Wire AB for his comments on the present work. References [1] G.D. Lahoti, S.L. Semiatin, Flat, bar and shape rolling, in: Metals Handbook, vol. 14, Forming and Forging, ninth ed., ASM International, Metals Park, OH, USA, 1988, ISBN 0-87170-020-4, pp. 343– 360. [2] R. B¨unten, K. Steinhoff, W. Rasp, R. Kopp, O. Pawelski, Development of a FEM model for the simulation of the transfer of surface structure in cold-rolling processes, J. Mater. Technol. 60 (1996) 369–376. [3] F. Plourabo´ue, M. Boehm, Multi-scale roughness transfer in cold metal rolling, Tribol. Int. 32 (1999) 45–57. [4] B. Ma, A.K. Tieu, C. Lu, Z. Jiang, An experimental investigation of steel surface characteristic transfer by cold rolling, J. Mater. Process. Technol. 125–126 (2002) 657–663. [5] H.C. Xie, D.R. Chen, X.M. Kong, An analysis of the three-dimensional surface topography of textured cold-rolled steel sheets, Tribol. Int. 32 (2) (1999) 83–87. [6] A.T. Galbato, R.T. Cundill, T.A. Harris, Fatigue life of silicon nitride balls, Lubr. Eng. 48 (11) (1992) 886–894. [7] Y.P. Chiu, P.K. Pearson, M. Dezzani, H. Daverio, Fatigue life and performance testing of hybrid ceramic ball bearings, Lubr. Eng. 52 (3) (1996) 198–204. [8] L. Wang, R.W. Snidle, L. Gu, Rolling contact silicon nitride bearing technology: a review of recent research, Wear 246 (2000) 159–173. [9] J.F. Braza, Ceramic roller evaluation in valve train tests, Tribol. Trans. 38 (1) (1995) 146–152. [10] ISO 4287:1997, International Standard. Geometrical Product Specifications (GPS)—Surface Texture: Profile Method—Terms, Definitions and Surface Texture Parameters, 25 pp. [11] B. Carlsson, The contact pressure distribution in flat rolling of wire, J. Mater. Process. Technol. 73 (1998) 1–6. [12] P. Montmitonnet, F. Delamare, B. Rizoulieres, Transfer layers and friction in cold metal strip rolling processes, Wear 245 (2000) 125–135. [13] X. Zhao, J. Liu, B. Zhu, H. Miao, Z. Luo, Tribological characteristics of Si3 N4 ceramic sliding on stainless steel, Wear 206 (1997) 76–82. [14] J. Denape, A. Marzinotto, J.A. Petit, Roughness effect of silicon nitride sliding on steel under boundary lubrication, Wear 159 (1992) 173– 184. [15] A.J. Winn, D. Dowson, J.C. Bell, The lubricated wear of ceramics, Part I: the wear and friction of silicon nitride, alumina and steel in the presence of a mineral oil based lubricant, Tribol. Int. 28 (6) (1995) 383–393. [16] T. Morimoto, K. Kamikawa, Friction and wear in silicon nitride – steel and cemented carbide – steel pairs in lubricated sliding, Tribol. Int. 29 (7) (1996) 537–546.