Evaluation o racture strength o TiN thin ilms deposited on WC-Co substrate T. Takamatsu 1, Y. Miyoshi 1, H. Tanabe 1, T. Itoh 2 1 The University o Shiga Preecture, Japan 2 NSK Ltd., Japan Abstract To evaluate the racture strength o TiN thin ilms deposited on the hard metal substrate WC-Co, and to investigate the inluence o the deposition conditions (bias voltage V B ) on the racture strength o TiN thin ilms, the sphere indentation test was carried out to determine the ring crack initiation strength σ,m in TiN thin ilms deposited on two kinds o WC-Co substrates diering in hardness using sphere indenters o varying diameter. TiN thin ilms 2µm thick were deposited by dc magnetron sputtering under various V B. Based on the probabilistic theory assuming a two-parameter Weibull distribution, the averages o the racture strength σ% o TiN thin ilms without residual stress under conditions o uniorm tensile stress and the residual stress σ% R o thin ilms were predicted rom the distribution characteristics o σ,m. The main results were as ollows: the average σ% is almost independent o sphere indenter and substrate hardness, and decreases with increasing V B ; the variation in σ% is mainly due to the grain size o thin ilms; the residual stress σ% R increases with increasing V B, and this tendency is qualitatively consistent with the measurements obtained by the X-ray diraction method. 1. Introduction Various coating technologies using ceramic thin ilms have been applied to allow use o cutting tools and machine components under more severe conditions. There have been many reports regarding the inluence o coating conditions on the mechanical properties o thin ilms deposited on metal substrates, such as hardness, wear properties, and adhesion strength. The deposition conditions are determined based on evaluation o the properties o thin ilms. However, a method or absolute evaluation o the racture strength o thin ilms has not been established, and there have been ew studies regarding the inluence o deposition conditions on racture strength o thin ilms. 1 Department o Mechanical Systems Engineering, The Univ. o Shiga Preecture, 2500 Hassaka-cho, Hikone, Shiga 522-8533, Japan 2 1-6-3 Ohsaki, Shinagawa-Ku, Tokyo, Japan Recently, we investigated the inluence o deposition conditions on the mechanical properties o titanium nitride TiN thin ilms deposited on steel substrates by dc magnetron spattering [1]. The hardness o thin ilms and the adhesion strength clearly changed according to the deposition conditions, such as bias voltage V B and gas pressure. The changes in mechanical properties o thin ilms can be explained based on the residual stress o thin ilms. The present study was perormed to investigate the validity o racture strength evaluation o ceramic thin ilms deposited on hard metal substrates by a sphere indentation test method, and to clariy the inluence o deposition conditions on the racture strength o thin ilms. Furthermore, the validity o the prediction o residual stress on thin ilms rom the results by the sphere indentation test was also examined.
2. Sphere Indentation Test Figure 1 shows a schematic o the sphere indentation test. When a sphere indenter o a given diameter 2R is pressed on the thin ilm, a ring crack initiates on the thin ilm at a certain load. Ring cracks occur rom micro-cracks in the thin ilm due to tensile stress in the radial direction slightly outside the region o contact. Ichikawa et al. demonstrated that sphere indentation tests are eective or evaluation o the racture strength o the surace layer in bulk ceramics [2]. Recently, we demonstrated the validity o sphere indentation tests to evaluate racture toughness o micro-cracks in the surace layer o silicon nitride Si 3 N 4 bulk material [3]. Furthermore, Ichikawa et al. reported that sphere indentation tests are possible to evaluate the racture strength o thin ilms deposited on hard material substrates [4] [5]. In the previous study [6], we carried out sphere indentation tests using the tungsten-carbide WC-Co substrate with deposited alumina Al 2 O 3 thin ilms and investigated the inluence o the diameter o a sphere indenter, the hardness o substrate, and the thickness o thin ilms on the ring crack initiation strength o Al 2 O 3 thin ilms. The ring crack initiation strength σ,m o thin ilms was obtained by equation (1). E 1 + νs (1 2 νs) P = (1) E 1+ ν 2π r E is Young s modulus, ν is Poisson s ratio, and and s denote the thin ilm and the substrate, respectively. P is the ring crack initiation load, and r is the ring crack radius. This equation was introduced in approximation rom Hertz s elastic contact theory based on the assumptions that both thin ilm and substrate show only elastic deormation, Young s modulus o a thin ilm is smaller than that o the substrate, and ilm thickness is suiciently small compared with the contact circle radius [4]. As the average value o σ,m is dependent on the diameter 2R o a spherical indenter, σ,m cannot be regarded as an absolute property o thin ilms. Thereore, based on the probabilistic theory assuming a two-parameter Weibull distribution, the distribution unction o σ,m was introduced in consideration o the residual stress o thin ilms [5]. σ,m s ( ) 2 ( ) 3. Experimental Methods Two types o WC-Co hard material were used as the substrate, the F-type and the EM-type. The Vickers hardness HV is 2050 or F-type and 1648 or EM-type. Three kinds o HIP-Si 3 N 4 balls with diameters o 3.96, 5.95, and 7.93mm were used as sphere indenters. Young s modulus and Poisson s ratio o the substrate, sphere indenter, and TiN ilm are shown in Table 1. Figure 1: Schematic o sphere indentation test method Figure 2 shows the shape and dimensions o the specimen. The surace o 40 20mm 2 was inished using a diamond grinding stone (surace roughness Ra=0.05µm). TiN with ilm thickness
Tale 1: Mechanical properties o the material Ltd., S3000N). tested Figure 2: Shape and dimensions o the specimen. Table 2: Deposition conditions o TiN thin ilm o about 2.5µm was deposited on the grinding surace using a dc magnetron sputtering equipment (Hirano Tecseed Co., Ltd.) as in the previous study [1]. Deposition conditions are shown in Table 2. Bias voltage V B was 0V, 30V, or 60V. Ater deposition processing, the specimens were heated in a vacuum urnace or 30minutes at 400 C to improve the adhesion strength o thin ilms. TiN thin ilm hardness H u was measured using an ultra-micro hardness testing machine (Mitutoyo Co., Ltd., MZT-3) and peel load was determined using a scratch testing machine (Rhesca Co., Ltd., CSR-01). The residual stress o the TiN thin ilm σ R was determined by X-ray stress measurement (Mac Science, Ltd., M03XHF 22 ). The microstructure o the thin ilm was observed using a scanning electron microscope (Hitachi, Sphere indentation tests were carried out at a cross-head speed o 0.03~0.09mm/min using an Instron type testing machine (Tokyo Testing Machine Inc., SC-5M12). The ring crack initiation load P was determined using an AE analysis equipment (Showa Electric Laboratory Co., Ltd., SAS-5000). Applied load was interrupted at the load that detected AE by ring crack initiation, and load was unloaded at high speed. The contact area was observed using a digital microscope (Keyence Co., VH-8000); and data were considered valid when only a single ring crack was observed, and the diameter o the ring crack 2r measured. The distribution characteristics o σ,m on each specimen were obtained rom many sphere indentation tests, changing the indentation position ater each test. Furthermore, the apparent racture toughness K % c o thin ilms was evaluated using the same IF method as used or ceramic bulk material. 4. Experimental Results and Discussion Figure 3 shows an example o a ring crack in the TiN thin ilm using 2R=7.93mm. In the experiments in the present study, clear plastic. deormation was not observed on the substrate or the thin ilm at ring crack initiation. Figure 3: Example o a ring crack (F-type, 2R=7.93mm, V B =-60V)
Figure 4: Relationship between r / a and P Figure 5: Comparison between IF-method and Sphere Indentation method Figure 4 shows the relationship between the ratio r /a and P or various 2R, where a is the contact circle radius and was obtained using equation (2): (2) 2 2 a 3 1 1 3 4 R νb νs = + E P b Es R is the radius o the sphere indenter, and subscript b denotes the sphere indenter. Although equation (2) was obtained in Hertz s theory with a coating-less substrate, the equation is eective in approximation under conditions in which a is suiciently large compared with ilm thickness. The values o r /a were almost constant and independent o P and 2R, although P varied widely with the diameter 2R. The average values o r /a were 1.08~1.20. Figure 5 shows the ring crack initiation strength σ,m (2R=3.96 mm, V B = 60 V) and the apparent racture toughness K % c obtained rom the IF method. The inluence o substrate hardness on K % c is clear, but there was little inluence on σ,m. However, when plastic deormation o the substrate is large at ring crack initiation, σ,m will change with substrate hardness. Figure 6 shows examples o the microstructure o the thin ilm o F-type substrate with V B =0V (a) V B =0V (b) V B =-60V Figure 6: Microstructure o thin ilms and 60V. The crystal grain size increases with increasing bias voltage V B, and the crystal direction also diers rom V B. Figure 7 shows the relationship between the residual stress σ R o thin ilms and the bias voltage V B or both types o substrate. σ R was determined by X-ray stress measurement.
Figure 7: Relationship between σ R and V B (X-ray stress measurement) Fig.9 Variation o σ,m (Weibull probability Paper) Figure 8: Relationship between Η u and V B Figure 10: Relationship between σ,m and V B Figure 8 shows the relationship between the hardness H u o thin ilms and V B or both types o substrate. The values o σ R are almost 0 in the case o V B =0V, and σ R becomes compressive and increases with increasing V B. The values o H u increase similarly according to the increase in V B. The above tendency o σ R and H u to V B is the same as that reported previously. As described in the previous study [1], H u is proportional to σ R, i.e., H u is governed by σ R. Although the scatter o adhesion strength was large, the adhesion strength did not clearly depend on V B and the substrate hardness compared with that in H u. Figure 9 shows the Weibull probability o variations in ring crack initiation strength σ,m o F-type substrate with dierent 2R in the case o V B = 60V. It can be seen that σ,m nearly conorms to a two-parameter Weibull distribution, and that σ,m depends on 2R. The same result was obtained in the case o the EM-type substrate. Figure 10 shows the relationship between the average σ,m and V B in both substrates. The values o σ,m or both substrates increase with increasing V B. The variation in σ,m is considered to be due to the residual stress σ R and the microstructure o thin ilms. The increasing tendency o σ,m cannot be explained rom the changes in crystal grain size shown in Fig.6. That is, the tendency o σ,m mentioned in Fig.10 is thought to be mainly due to the change in σ R, as shown in Fig.7.
Ichikawa et al. introduced equation (3) o the distribution-unction F(σ,m ) obtained based on the probabilistic theory (concept o the eective area) assuming a two-parameter Weibull distribution [5]. 2 o 3 K σ,m γ σ ( ) 1 exp 2 R F σ - 2 r,m = - ç α r ê- çè ø π dr (3) Here, it was assumed that the residual stress σ R o thin ilms is uniorm. m o and η o are the shape parameter and scale parameter o σ,m or a standard area A o (=1.0mm 2 ) under conditions o uniorm tensile stress and σ R =0. r is the distance rom the center o the contact circle. α, γ, and K are expressed according to the ollowing equation. s é ê ë ò E (1 + ν s ) α = E (1 + ν ) r ' 2 2 3π 1 νb 1 νs R K = + 2 Eb (4) Es 1 2ν r æ ö ç çè η ø o γ = s r a ö m A o ù ú û determined using equation (5) rom the distribution properties o σ,m with 2R=3.96mm (m, η). Figure 11 shows the relationship between the average σ,m and 2R. The symbols indicate the average and standard deviation o experimental values o F-type substrate, and the dotted line shows the predicted relationship. As shown in the igure, the predicted relationship and the experiments are almost in agreement. That is, the prediction o the σ,m 2R relationship in the case o V B =0 using equation (3) is eective. Then, the racture strength σ% o thin ilms under a uniorm tensile stress ield without residual stress o thin ilms and the residual stress σ% R o thin ilms were predicted rom the distribution properties o σ,m based on the probabilistic theory, assuming a two-parameter Weibull distribution. The averages o η o or each V B. σ% were calculated using m o and The integration range in equation (3) is rom r= r to r=r' where the tensile stress in the radial direction on the surace o the ilm is almost zero. In the case o V B =0, the relationship between the average values o σ,m and the diameter 2R o a sphere indenter was predicted on the assumption o σ R =0. That is, in the case o σ R =0, integration o equation (3) can be calculated directly, and m o and η o are obtained rom the ollowing equations. m o =m-2 η 6 2 πγ K mo + 2 = (5) η ( mo 1) α Ao o 2 1 mo Thereore, in the present study, m o and η o were Figure 11: Relationship between σ,m and 2R Figure 12 shows the relationship between average σ% and V B. As shown in the igure, σ% tends to decrease with increases in V B. This tendency o σ% in relation to V B is the opposite o that o σ,m shown in Fig.10. Although the inluence o
crystal direction in the microstructure described in Fig.6 is unknown, the decreasing tendency o σ% is considered mainly due to changes in the crystal grain size. In comparison with σ,m shown in Fig.10, the dependence o σ% on 2R is small. That is, σ% is considered an absolute evaluation value. (1) is an approximation, and the residual stress in thin ilms is not uniorm but changes. That is, the results o recent studies [7][8] indicated that the residual stress near the surace is smaller than that on the inside. Further studies are required to evaluate the residual stress o thin ilms based on the results o sphere indentation tests. Figure 13 shows the relationship between the prediction o the residual stress σ% R o thin ilms and V B. It can be seen that the value o σ% R 5. Conclusions To estimate the racture strength o TiN thin ilms deposited on WC-Co substrates, to investigate the inluence o the deposition conditions on the racture strength o thin ilms, and to predict the residual stress o thin ilms rom the distribution characteristics o the ring crack initiation strength o thin ilms, sphere indentation tests were carried out using two types o WC-Co substrate diering in hardness using sphere indenters o varying diameter. TiN thin ilm was deposited by dc magnetron sputtering under various V B. The ollowing conclusions were obtained. Figure 12: Relationship betweenσ% and V B There is little inluence o substrate hardness on the ring crack initiation strength σ,m o thin ilms, but apparent racture toughness K % c determined by the IF-method is clearly dependent on the substrate hardness. The residual stress σ R o thin ilms measured by the X-ray diraction method and the hardness H u o thin ilms increase with increasing bias voltage V B regardless o the substrate hardness. The crystal grain size also shows a tendency to increase with increasing V B. Figure 13: Relationship betweenσ% increases with increasing V B, and this tendency R and V B is in agreement with that determined by X-ray stress measurement as shown in Fig.7. However, the values o σ% R are considerably smaller than those obtained by X-ray stress measurements. Equation The scatter o σ,m nearly conorms to a two-parameter Weibull distribution and the average σ,m is dependent on the diameter o a sphere indenter 2R regardless o V B and substrate hardness.
Mechanical Engineers, Series A, Vol.60, The value o σ,m increases with increasing V B. No.575, pp.1538-1544, 1994(in Japanese). This variation in σ,m is due to the changes in σ R o thin ilms. The relationship between the average σ,m and 2R can be predicted rom the distribution properties o σ,m based on the probabilistic theory assuming a two-parameter Weibull distribution. The residual stress σ% R predicted rom the distribution properties o σ,m is considerably smaller than that o X-ray stress measurement, but the prediction o the σ R V B relationship is quantitatively in agreement with that o X-ray stress measurement. The average racture strength o thin ilmσ% predicted rom the distribution properties o σ,m under a uniorm stress ield and without the residual stress o thin ilm is almost independent o 2R. Thereore, the sphere indentation test method is eective to evaluate the absolute racture strength o thin ilms. The racture strength o thin ilmσ% decreases according to increases in V B. The tendency o σ% increase is mainly due to the crystal grain size o thin ilms. [3] Takamatsu, T., Miyoshi, Y., Tanabe, H., Segawa, M., Transactions o the Japan Society o Mechanical Engineers, Series A, Vol.71, No.708, pp.1147-1152, 2005(in Japanese). [4] Kagawa, H., Ichikawa, H., Takamatsu, T., Kuwano, H., Transactions o the Japan Society o Mechanical Engineers, Series A,Vol.60, No.570, pp.416-420, 1994(in Japanes). [5] Kagawa, H., Ichikawa, H., Takamatsu, T., Kuwano, H., Transactions o the Japan Society o Mechanical Engineers, Series A, Vol.60, No.570, pp.396-401, 1994(in Japanese). [6] Takamatsu, T., Ichikawa, H., Matsumura, T., Kawasaki,T., Journal o the Society o Materials Science, Japan, Vol.47, No.8, pp.819-824, 1998(in Japanese). [7] Tanaka, K, Akiba, Y., Kawai, M., Itho, T., Journal o the Society o Materials Science, Japan, Vol.54, No.7, pp.704-709, 2005(in Japanese). [8] Takago, S., Yasui, H., Kuritsu, K., Sasaki, T., Hirose, S., Sakurai, K., BUNSEKI KAGAKU, Vol.55, No.6, pp.405-410, 2006(in Japanese). 6. Reerences [1] Tanabe, H., Miyoshi, Y., Takamatsu, T., Sugiura, H., Journal o the Society o Materials Science, Japan, Vol.51, No.6, pp.694-700, 2002(in Japanese). [2] Ichikawa, H., Ohogushi, K., Takamatsu, T., Transactions o the Japan Society o