【正文】 C). An M 35 tool steel (position C wt.%, Cr wt.%, Mo wt.%, W wt.%, V wt.%, Co wt.%) coated with a layer of hard metal (WC–18%Co, hardness ～88 HRA) deposited by the air plasma spray (APS) technique was selected as a reference, highly wear resistant material. Tribological tests Dry sliding tests were carried out using a puter controlled slideroncylinder tribometer (Fig. 1). The stationary sliders were constituted by the material under investigation, in the form of prismatic bars (5mm5mm50 mm). The counterfacing material was a ceramic coating consisting of Al2O3 (87 wt.%) and TiO2 (Rockwell hardness HRD = 60, surface roughness Ra = ) deposited onto the rotating cylinder. The tests were carried out under applied loads of 5 and 25N and sliding speed of ?1, for sliding distances up to 5 km, at room temperature (20–25 ?C), in laboratory air (relative humidity in the range 50–60%). Both friction resistance and system wear (. cumulative wear of both slider and cylinder) were continuously measured by means of a bending load cell and a displacement transducer, respectively, and were recorded as a function of the sliding distance. At the end of each test, wear scar depths were measured on both slider and cylinder by means of a stylus profilometer (pickup curvature radius, 5 _m), recording line profiles perpendicularly to the wear scar. The resistance of the boride coatings to abrasive wear was evaluated using a microscale abrasion tester (MSAT), which is based on a ballcratering geometry [8,9]. The rotation of a sphere against a flat specimen in the presence of small abrasive particles generates a wear crater with an imposed spherical geometry within the material. Basically the rig consists, as shown in Fig. 2, of a hard martensitic steel sphere (radius R = mm, hardness HV 1000) rotating against the specimen under investigation in presence of an abrasive slurry (an aqueous suspension of SiC particles 4–5 _m in size, with an initial concentration of g cm?3), maintained and replenished at the contact region by a slow constant drip feed (～ cm3 min?1). A contact load of was used and the sliding speed was ?1. The diameter b of the spherical cap produced on the specimen by abrasion was measured with a calibrated optical microscope, and the value of b was used to calculate both the peration depth h and the wear volume V: h ≈ b2/8R (1) V ≈ πb4/64R (2) where b ＜＜ R. If the depth of wear craters is lower than the thickness of the coating, a simple model for abrasive wear of bulk materials (equivalent to the Archard equation for sliding wear) can be used and leads to the following equation for the wear volume V: V = kSN (3) where S is the total sliding distance of the sphere relative to the specimen surface, N the normal load, k the wear coefficient or specific wear rate. Eq. (3) allows the wear rate of the material to be calculated for each set of experiments, . for each region of the boride coating exposed by the layerbylayer method. 3 Results and discussion Fig. 3 shows a typical microstructure of polyphase boride coatings grown on Armco iron. The columnar morphology of both FeB and Fe2B layers and the crack propagated along the interface between the FeB–Fe2B interface should be noted. The coatings grown on medium carbon steel under the same thermochemical conditions displayed lower values of maximum thickness and less pronounced columnarity, . lower differences between maximum and minimum values of thickness. The coating thickness is influenced by alloying elements in the metal substrate (and especially by chromium) which can modify the active boron diffusivity by entering the iron boride lattice. Also the metal–boron reactivity can be changed by the modifications occurring in the position of the surface region of the metal substrate (depletion or concentration of elements), as a consequence of redistribution phenomena involving alloying element[6]. Columnarity at the interfaces has been explained as a consequence of locally intense stress fields and lattice distortions near the tips of the first, acicular nuclei of reaction products [7]. The brittleness at the FeB–Fe2B interface, in turn, has been ascribed to the stress states induced in that region by the difference between the thermal expansion coefficients of the two iron borides, that of FeB being about three times higher [1]. Also the difference in elasticity between iron borides is considerable, the value of FeB being about half of that displayed by Fe2B. Fig. 4 shows two microhardness profiles measured on the same crosssection as in Fig. 3. The significantly lower hardness that is displayed in the outer part of the coating by one of the profiles (the other one starts from an appreciable distance from the external surface) is to be considered realistic, in the light of the already mentioned evidence that the outermost, few micrometers thick region of the boride coatings is crystallographically disordered and friable [7]. The decrease in hardness as the distance from the external surface increases is to be ascribed to the presence of unborided iron zones between boride columns, that bee larger on going towards the bulk of the specimen. The high hardness displayed by inner regions of the coating (profile no. 2) should be noted. Fig. 5 shows XRD patterns recorded at different depths of the same boride coating, after a progressive removal of material. The FeB (0 0 2) and Fe2B (0 0 2) reflections are much more intense than those observed for boride powders. In particular, a very intense Fe2B (0 0 2) peak can be seen even in the XRD patte