- Email: [email protected]
Nanoscale friction and wear properties of silicon wafer under different lubrication conditions
Applied Surface Science 282 (2013) 25–31
Contents lists available at SciVerse ScienceDirect
Applied Surface Science journal homepage: www.elsevier.com/locate/apsusc
Nanoscale friction and wear properties of silicon wafer under different lubrication conditions Xiaochun Chen, Yongwu Zhao ∗ , Yongguang Wang, Hailan Zhou, Zhifeng Ni, Wei An School of Mechanical Engineering, Jiangnan University, Wuxi, Jiangsu 214122, PR China
a r t i c l e
i n f o
Article history: Received 8 September 2012 Received in revised form 24 April 2013 Accepted 26 April 2013 Available online 6 June 2013 Keywords: Friction and wear property Silicon wafer Water lubrication Hydrogen peroxide lubrication Wear mechanism
a b s t r a c t The nanoscale friction and wear properties of single crystal silicon wafer under different lubrication conditions are studied in this paper. The experiments were performed with Si3 N4 ball sliding on the surface of silicon wafer under four different lubrication conditions: dry friction, water lubrication, hydrogen peroxide lubrication and the static hydrogen peroxide dry friction. The results from the experiments have been analyzed showing the different friction and wear properties of the silicon wafer in different lubrication conditions. It is concluded that the wear rates under the water lubrication and under the hydrogen peroxide lubrication are both small, the chemical reactions are facilitated by the mechanical processes when the load and the sliding speed reach certain levels. This is mainly resulted by the enhanced lubricant performance with the formed silicon hydroxide Si(OH)4 film. Under the water lubrication, the wear is found in a way of material removed in molecule scale. Under the hydrogen peroxide lubrication, the wear is mainly caused by the spalling of micro-cracks. Under the dry friction condition, the wear is found being adhesive wear. And under the static peroxide dry friction, the wear is prevailing adhesive wear. These results are essential and valuable to the development of the efficient and environmental-friendly slurry for the chemical mechanical polishing (CMP) process. © 2013 Elsevier B.V. All rights reserved.
1. Introduction Single crystal silicon is an important material for semiconductor industry. Silicon wafer is widely used in the integrated circuit (IC) devices, high-density information storage devices, and microelectric-mechanical systems (MEMS) [1,2]. The fabrication and application of these delicate devices require high surface quality of IC chips. This has become one of the major concerns [3]. Up to now, CMP has been proven the best global planarization technique which can satisfy the requirements on surface quality of silicon wafer and chips in the semiconductor industry [4]. With the scientific progress in nanotechnology in recent years, CMP has been developed rapidly. However, the material removal mechanism of the CMP process is not yet clearly understood [5]. Many scientists are extensively working from different aspects to reveal the insights of the CMP technology in order to gain the effective method to remove the material with minimized damage to the polished surface quality, and minimized damage to the environment [3,6].
∗ Corresponding author at: School of Mechanical Engineering, Jiangnan University, No. 1800, Lihu Ave, Wuxi, Jiangsu 214122, China. Tel.: +86 510 8591 0562; fax: +86 510 8532 8262. E-mail addresses: [email protected] (X. Chen), [email protected] (Y. Zhao), [email protected] (Y. Wang), [email protected] (H. Zhou), [email protected] (Z. Ni), [email protected] (W. An). 0169-4332/$ – see front matter © 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.apsusc.2013.04.148
In a typical CMP process in the semiconductor industry, the silicon wafer is carried by the polishing head and pushed onto the polishing pad. The polishing head and the polishing pad rotate, oppositely to each other at a relative speed. Meantime the polishing slurry is fed onto the pad and flows into the polishing interface between the wafer and the pad. The material removal in the CMP process is a synthetic function of chemical and mechanical actions [7]. The friction and wear property of the polished material in a given lubrication condition is an essential factor as well to the quality of the polishing surface and the material removal rate. Therefore, to study on the friction and wear properties of the silicon wafer is an important part of silicon wafer CMP research. By using a servo-hydraulic reciprocating sliding apparatus, Wang et al. [8] studied the effects of water and oxygen on the tribochemical wear of single crystal silicon (1 0 0) against SiO2 ball. The wear tests were performed under four different atmosphere conditions, those were, pure nitrogen, dry air, humid air and pure oxygen. The results from the experiments indicate that the increase in hydrophilicity of silicon (1 0 0) surface will induce an obvious increase of the friction coefficient and the wear depth of silicon surface. Moreover, the oxygen plays a very important role in the wear behavior. The wear depth on silicon surface was highest due to the strong oxidation reaction under the condition of pure oxygen atmosphere. Varenberg et al. [9] studied nano fretting wear on the silicon surface by using scanning probe microscopy. Partial and gross slip fretting with displacement amplitudes from 5 to 500 nm
26
X. Chen et al. / Applied Surface Science 282 (2013) 25–31
Si 3 N4 Ball
Load
Friction Coefficient vs. Load
Reciprocating Direction 1.2
dry friction water lubrication H2O2 lubrication
1.1
static H 2O2 dry friction
Silicon Wafer Fig. 1. Schematic of ball-on-flat sliding.
Friction Coefficient
1 0.9 0.8 0.7 0.6 0.5 0.4
were used for the study, and the results from their study show a substantial increase of the friction at the transition from partial to gross slip and a significant difference between damage surfaces in the two fretting regimes. Although the study was concentrating on the nano fretting wear of the silicon surface which was not chemical modified, the results from the study are good reference for the study of this paper to understand the friction and wear behavior of the modified silicon surface. Singh and Yoon [10] studied the microand nano-frictional properties of chemically and topographically modified Si (1 0 0) surfaces. The results indicate that at the microand nano-scales, the modified Si (1 0 0) surfaces show enhanced friction behavior when compared to the bare Si (1 0 0) surfaces. The modified Si (1 0 0) surfaces exhibit lower friction values. Singh and Yoon found that at the nano-scale, the lower friction values of modified surfaces were due to the lower intrinsic adhesion and contact areas. And with the chemically modified surfaces, the reduction of the contact area was because of the lower interfacial energies. Silicon wafers are thin and brittle slices, which are formed of highly pure and nearly defect-free single crystal silicon. For the application in the IC and MEMS, silicon wafers must have the super surface quality, and be free of defects after fabrication [11–13]. CMP is now the only ideal process to achieve global planarization of silicon wafers [4]. However, it is still lack of fundamental understanding of this process although many researchers have studied it intensively in the past decades [14–16]. Meanwhile traditional CMP slurry contains a large number of chemical substances, such as oxidants, corrosion inhibitors and chelating agents. All these chemicals generate great deal of wastewater and cause heavy environmental pollution. With the strengthening of environmental protection, safety and health consciousness, green polishing slurry is desired in order to replace traditional slurry. It has been becoming research’s hotspot in the world in recent years [6]. In this paper, nanoscale friction and wear properties of single crystal silicon wafer under different lubrication conditions are studied. At the same time, the results from this study are supporting to use deionized water and H2 O2 as possible less harmful green slurry for the CMP process. 2. Experimental The p- Si (1 0 0) single crystal silicon wafers were selected as the test material. The samples were cut from the wafers with a rectangle size of 20 mm × 20 mm. The nanoscale friction and wear tests were performed on a reciprocating UMT-2 micro-tribotester at standard room temperature of 20 ± 1 ◦ C with a relative humidity of 45–55%. The type of frictional contact was ball-on-plate contact, as illustrated in the Fig. 1, a Si3 N4 ceramic ball with the diameter of 4 mm slides on the 20 mm × 20 mm sized sample of p- Si (1 0 0) rectangle silicon wafer. During the experiment, a new Si3 N4 ball is used only one time for each single test in order to minimize the effect of the wear of the ball on the wear of the wafer
0.3 20
30
40
50
60
70
80
90
100
110
120
Load (mN) Fig. 2. Results from experiments about relation between friction coefficient and load.
silicon surface. Before the sliding experiment, both the Si3 N4 ball and the silicon wafer were dipped into acetone and cleansed with an ultrasonic cleaner for 10 min. The test load was 30–110 mN, the sliding speed was 5.33–10.66 mm/s, the time of friction and wear test was 2 min, the amplitude of the reciprocation sliding is 2 mm. A group of friction and wear experiments were conducted under different lubricating conditions namely dry friction, water lubrication, H2 O2 lubrication and static H2 O2 dry friction. Deionized water was used as lubrication medium in water lubrication experiments. 30% H2 O2 was used as lubrication medium in H2 O2 lubrication experiments. In the static H2 O2 dry friction experiments, the wafer samples were immersed in the 30% H2 O2 liquid for 30 min before the experiments, then taken out and got it dried afterward. During the sliding tests in so-called static H2 O2 dry friction experiments no lubrication medium was used. The UMT-2 micro-tribotester collected and recorded the friction coefficients from the sliding experiments automatically. The surface morphology was characterized with Phase Shift MicroXAM-3D, by measuring the length, the width and the depth of the wear marks. The wear rate and the wear depth of single scratch are therefore able to be computed for further analyses. Furthermore, it was pointed out a mathematical modeling for the contact pressure between the ball and surface would be benefit for the understanding of the sliding wear mechanism, which would be addressed in future. 3. Results and discussion 3.1. The effects of load on friction coefficient and wear rate A series of experiments have been conducted to investigate the relationship between the friction coefficient and the applied load. The friction coefficients between the Si3 N4 balls and the single crystal silicon wafers were collected and recorded by the Micro-tribotester from the sliding experiments under four different lubrication conditions. The sliding speed was 8.00 mm/s, the applied loads were 30 mN, 50 mN, 70 mN, 90 mN, 110 mN. The coefficients obtained from the experiments are shown in the Fig. 2. From the Fig. 2, it can be seen that the friction coefficient is becoming smaller with the increasing of the load from 30 mN to 70 mN under all four different lubrication conditions, and there is no big difference with friction coefficient when the load increases above 70 mN.
X. Chen et al. / Applied Surface Science 282 (2013) 25–31
27
Friction Coefficient vs. Speed
Wear Rate vs. Load 0.65
1.2
dry friction water lubrication H2O2 lubrication
1
dry friction water lubrication H2O2 lubrication
0.6 0.55
static H 2O2 dry friction
Friction Coefficient
Wear Rate (10-6mm3/s)
static H 2O2 dry friction 0.8
0.6
0.4
0.5 0.45 0.4 0.35 0.3 0.25
0.2 0.2 0.15
20
30
40
50
60
70
80
90
100
110
120
Load (mN) Fig. 3. Results from experiments about relation between wear rate and load.
Izhak has revisited the Cattaneo-Mindlin Concept of Interfacial Slip in Tangentially Loaded Compliant Bodies [17], the general tendency of decreasing static friction coefficient with increasing normal preload was observed experimentally. In the Izhak’s study, the static friction coefficient decreases sharply with increasing dimensionless normal preload in the early elastic-plastic loading range, and at higher dimensionless normal preloads, the reduction rate of the friction coefficient diminishes, and further the static friction coefficient approaches a constant value. The results in the Fig. 2 regarding the dependency of the sliding friction coefficient on the normal load are in accordance with the results regarding the dependency of the static friction coefficient on the normal preload which are obtained by Izhak, both do obey the classical laws of friction. In the dry friction and the water lubrication experiments, the friction coefficients are nearly equal when the load is 30 mN. And the friction coefficients under the H2 O2 lubrication and the static H2 O2 dry friction experiments are very close to each other as well. When the load increases to 50 mN, the friction coefficients under water lubrication and static H2 O2 dry friction are almost same. The maximum of friction coefficient is under dry friction condition, while the minimum of the friction coefficient is under H2 O2 lubrication condition. The Fig. 2 shows that the friction coefficient deceases rapidly with the increasing of the load when the load is less than 70 mN, and the friction coefficient reaches highest value under the dry friction and the minimum of the friction coefficient is under the water lubrication when the load increases to 70 mN. The friction coefficients under the static H2 O2 dry friction are greater than that under H2 O2 lubrication in all respective loads in the experiments. The wear rate is computed with the length, the width and the depth of the wear mark, which are characterized and measured by Phase Shift MicroXAM-3D. The Fig. 3 shows the relation between the wear rate and the load. From the Fig. 3, it can be seen that the wear rate under dry friction is highest one in all four lubrication conditions when the load is 30 mN. With the load increasing in all experiments of four different lubrication conditions, the wear rates are all increasing, the wear rates under dry friction and static H2 O2 dry friction increase rapidly. When the load is 50 mN, the wear rates under dry friction and static H2 O2 dry friction are much greater than the wear rates under water lubrication and H2 O2 lubrication. The wear rates under water lubrication and H2 O2 lubrication increases only slightly with the load increasing from 30 mN to
5
6
7
8
9
10
11
12
Speed (mm/s) Fig. 4. Results from experiments about relation between friction coefficient and speed.
110 mN. And the wear rates under H2 O2 lubrication are slightly greater than that under water lubrication with all applied loads. The situation with the wear rates under dry friction and static H2 O2 dry friction is different. At the load of 30 mN the wear rate under dry friction is much greater than that under static H2 O2 dry friction, which is very close to the wear rates under water lubrication and H2 O2 lubrication with the same applied load. When the load increases to 50 mN, the wear rate under dry friction is slightly lower than that under H2 O2 dry friction, both are much greater than those under water lubrication and H2 O2 lubrication. With the load increasing, the wear rates under dry friction and H2 O2 dry friction are both increasing to same level, which are extremely greater than the wear rates out of water lubrication and H2 O2 lubrication experiments. The wear rates under water lubrication and H2 O2 lubrication are both slightly increasing with the load increasing. The curves for these two wear rates in the Fig. 3 are quite flat. However, at each load the wear rates under H2 O2 are greater than those under water lubrication. 3.2. The effects of sliding speed on friction coefficient and wear rate Another series of experiments have been conducted to investigate the relationship between the friction coefficient and the sliding speed. The friction coefficients under four different lubrication conditions between the Si3 N4 balls and the single crystal silicon wafers were collected and recorded by the Micro-tribotester. The applied normal load was 70 mN, the sliding speeds were 5.33 mm/s, 6.66 mm/s, 8.00 mm/s, 9.33 mm/s, 10.66 mm/s. The results are shown in the Fig. 4. The Fig. 4 shows the relationship curves of friction coefficients versus the sliding speeds under four different lubrication conditions with the applied normal load at 70 mN. In dry friction experiment, with the sliding speed increasing from 5.33 mm/s to 6.66 mm/s, 8.00 mm/s and 9.33 mm/s the friction coefficient becomes smaller, especially when the speed increases from 6.66 mm/s to 8.00 mm/s the friction coefficient decreased dramatically from 0.5451 to 0.3856. But the friction coefficient raises from 0.3744 to 0.3977 when the sliding speed increases from 9.33 mm/s to 10.66 mm/s. In the H2 O2 lubrication experiment, the friction coefficient increases with the sliding speed increasing. In the water lubrication and the static H2 O2 dry friction experiments the friction
28
X. Chen et al. / Applied Surface Science 282 (2013) 25–31
Wear Rate vs. Speed dry friction water lubrication H2O2 lubrication
1.2
static H 2O2 dry friction
Wear Rate (10-6mm3/s)
1
0.8
0.6 Fig. 6. Topography of wear surface from dry friction experiment.
0.4
0.2
4
5
6
7
8
9
10
11
Speed (mm/s) Fig. 5. Results from experiments about relation between wear rate and speed.
coefficients are dropping down with the increasing of the sliding speed. The relationship between the wear rate and the sliding speed from this series of experiments is shown in the Fig. 5. The wear rates under the dry friction experiments and the static H2 O2 dry friction experiments increase rapidly with the increasing of the sliding speed from 5.33 mm/s to 9.33 mm/s and reaches the highest value, the wear rates at the sliding speed of 10.66 mm/s drops down under both the dry friction and the static H2 O2 dry friction experiments. In the water lubrication and the H2 O2 lubrication experiments, the wear rates increases slightly with the increasing of the sliding speed, and the wear rates under the H2 O2 lubrication are a little bit greater than the wear rates under the water lubrication with all test speeds. 3.3. Discussions on friction coefficient and wear rate The series of experiments aimed at understanding the friction and wear properties of single crystal silicon under different lubrication conditions. Further analyses have been carried through on all above results of the experiments. In the dry friction experiments, when the sliding speed is kept unchanged, as illustrated in the Fig. 2 and in the Fig. 3, the friction heat between the frictional interfaces increases significantly with the increasing of the load. Thus the contact area under the friction heat is more conducive for creating oxide product on the friction surfaces, which induces the decreasing of friction coefficient. Meanwhile, the friction heat causes a temperature rise in the friction surfaces, which makes the single crystal silicon tending to be easier deformable [18], thus the contact area between the Si3 N4 ball and the single crystal silicon increases under the load pressure, which leads to the increasing of wear rate. However, based on the authors’ investigation the temperature increased or the energy dissipated during the sliding tests is still not exactly measurable in current experimental condition. When the load reaches 70 mN, the friction coefficient is gradually reaching a plateau as the load increases further. The reason for this is mainly related to the oxide layer in the friction zone which is formed during the friction reaction. When the applied load is greater than 70 mN, the oxide layer contacts with the Si3 N4 ball directly during the sliding test, the friction and wear phenomenon reflects the Si3 N4 ball against the oxide layer. So the friction and wear tends to a steady stage, the friction coefficient and the wear rate reaches a plateau. In the dry friction experiments, when the
applied normal load is kept unchanged, as illustrated in the Fig. 4 and in the Fig. 5, the friction and wear properties of single crystal silicon are found differently under different sliding speeds. In the range of the sliding speed less than 8.00 mm/s, the friction heat increases significantly with the sliding speed increasing, which is beneficial to the formation of oxide layer and makes single crystal silicon tending deformable, leading to the friction coefficient decreasing sharply and the wear rate increasing rapidly. When the sliding speed reaches 9.33 mm/s and after, severe friction reactions will happen in the frictional interfaces with the speed increasing, and the temperature of the frictional interfaces becomes extremely high. The micro debris from the scratches in the high temperature zone forms very hard layer, which makes the friction coefficient increase. Meanwhile, the real contact area is enlarged in the abrasion interface, thus further results in a decline of the wear rate [19]. The Fig. 6 shows the worn surface of the single crystal silicon from the dry friction experiment with 70 mN load at 8.00 mm/s sliding speed. It can be seen that the worn surface is rough, and there are serious particle adhesion and obvious groove, the wear mechanism is mainly adhesive wear. In the water lubrication experiments, the friction heat accelerates the chemical reactions between silicon and water, and generates synthetic film of silicon peroxide on the surface of silicon wafer [8,20]. When water flows into the frictional interfaces, silicon peroxide reacts with water, while the Si3 N4 reacts with water to have strengthened the tribochemical reactions in the frictional interfaces [21]. SiO2 + 2H2 O = Si(OH)4
(1)
Si3 N4 + 16H2 O = 3Si(OH)4 + 4NH4 OH
(2)
The new generated Si(OH)4 film functions as lubricant and lowers the shear strength in the frictional interfaces. The hardness of the film is lower than the hardness of the silicon wafer surface. On the other hand, water flow plays as coolant and forms water film in between of the frictional interfaces, which reduces the adhesive wear significantly and reduces the friction coefficient and wear rate effectively. When the load is 30 mN in the Fig. 2, the friction coefficient is greater than that from the dry friction test, the reason for this is primarily because that the load is not heavy enough to accelerate the tribochemical reactions in the frictional interfaces compared to those with the load greater than 30 mN. When the load reaches 70 mN and after, the mechanical action is greater than the chemical reaction with the load increasing, the balance is broken. The chemical reaction is accelerated by the mechanical actions during the process [5]. This directly results in the decrement of the friction coefficients and the wear rates. It can be seen that both tend to be stabilized at a certain level from the Fig. 3. The topography of the worn surface from water lubrication experiment can be seen in the Fig. 7. The worn mark is a clean ploughed groove.
X. Chen et al. / Applied Surface Science 282 (2013) 25–31
29
Fig. 7. Topography of wear surface from water lubrication experiment.
In the hydrogen peroxide lubrication experiments, hydrogen peroxide is a specific oxidant, which possesses both oxidation and reduction. The 30% hydrogen peroxide was used in the experiments. The quantity of the hydrogen ions is becoming less because of the reaction in the 30% hydrogen peroxide solution. It becomes slightly acidic and further weakens the oxidation layer on the surface of single crystal silicon. On the other side, hydrogen peroxide can act as a proton donor as well as a proton acceptor parallel [22]. As proton acceptor, molecules of hydrogen peroxide are known to form stronger bonds with silicon dioxide other than water molecules as proton donors to form bonds with silicon dioxide. So the tribochemical reaction under hydrogen peroxide lubrication is weaker than that under water lubrication, which leads that the friction coefficient and the wear rate are greater than that under water lubrication when the loads are over 70 mN. Also, because of the reduction of hydrogen ions, the adhesive oxidation layer on the surface of the single crystal silicon is minimized. Therefore, the adhesive wear is significantly reduced. The Fig. 8 shows that the worn surface is relatively smooth, with a few peeling pits of micro-cracks and micro-flakes. The wear mechanism is mainly spalling, which is caused by micro-cracks. The wear rates under hydrogen peroxide lubrication are greater than those under water lubrication, as illustrated in the Fig. 2 and the Fig. 3. In the experiments with unchanged load at 70 mN, the friction coefficients increase rapidly with the sliding speed increasing, as shown in the Fig. 4. In this period, the top layer of the single crystal silicon is modified by the oxidation and reduction of the hydrogen peroxide. The Si OH− bonds are formed on the
surface, the top layer is assumed softer than the silicon. When the sliding speed is raised, the tribochemical reactions becomes more intensive, the corrosive wear on the surface of single crystal silicon is aggravated. This leads to an increscent friction coefficient. The Fig. 5 shows that the wear rates under hydrogen peroxide lubrication do not change much with the sliding speed increasing. In the experiments under the static hydrogen peroxide dry friction, the wafers of single crystal silicon were immersed in the 30% hydrogen peroxide solution for 30 min before the tests started. During this immersing period, the oxide silicon peroxide film on the surface of single crystal silicon fully contacts the water molecules in the hydrogen peroxide solution, the molecules of silicon peroxide reacts with hydroxyl, the bond of Si O converts to the bond of Si OH− , which eventually forms Si(OH)4 soft layer. When the load is 30 mN, the lubrication of Si(OH)4 effectively reduces the friction coefficient and wear rate. With the load increasing, the Si(OH)4 soft layer is removed gradually, therefore the friction and wear properties of single crystal silicon become similar to that under dry friction, as illustrated in the Fig. 2 and in the Fig. 3. In the experiments with unchanged load, the results are similar to that from the dry friction experiments due to the same reason of the soft layer removed on the surface. The Fig. 9 shows the worn surface from the static hydrogen peroxide dry friction experiments. Obviously, the characteristics seen here are very close to those from the dry friction test which is shown in the Fig. 6.
Fig. 8. Topography of wear surface from hydrogen peroxide lubrication experiment.
Fig. 9. Topography of wear surface from static hydrogen peroxide dry friction experiment.
30
X. Chen et al. / Applied Surface Science 282 (2013) 25–31
Table 1 The wear depth of single scratch under four different lubrication conditions with speed at 8.00 mm/s. Load (mN)
Wear depth from dry friction (nm)
Wear depth from water lubrication (nm)
Wear depth from H2 O2 lubrication (nm)
Wear depth from static H2 O2 dry friction (nm)
30 50 70 90 110
1.029 1.066 1.252 1.275 1.458
0.063 0.066 0.068 0.087 0.089
0.109 0.122 0.153 0.181 0.201
0.350 1.156 1.274 1.279 1.452
3.4. Discussions on the wear depth of single scratch
4. Conclusions
The wear depths of single scratches on the surface of single crystal silicon in these series of experiments are collected and given in the Table 1. The minimum wear depth under dry friction is 1.029 nm. Under water lubrication, the wear depth of single scratch is the smallest with the minimum value of 0.063 nm and the maximum value of 0.089 nm. The wear depths of single scratches under hydrogen peroxide lubrication are slightly greater than that from water lubrication experiments, the minimum value is 0.109 nm and the maximum value is 0.201 nm. Under the static hydrogen peroxide corrosion dry tests, the minimum wear depth is 0.35 nm at 30 mN. When the load is raised to 50 mN, the wear depth is close to that from dry friction test. The wear depth of single scratch is at 10−11 m magnitude under the water lubrication tests and at 10−10 m magnitude under the hydrogen peroxide lubrication tests. Therefore, it can be assumed that the material removal is happened at the molecular scale under these two conditions [7,23] and the wear mechanism under the water lubrication is considered to be of molecule-scale removal process. In this regard, the study is exploring the nanoscale friction and wear properties of the test material single crystal silicon. In the meantime, the wear rates are both small under the water lubrication and the hydrogen peroxide lubrication. The Fig. 2 shows that the maximum wear rate is 0.0303 × 10−6 mm3 /s and the minimum is 0.0109 × 10−6 mm3 /s under water lubrication, the maximum wear rate is 0.0467 × 10−6 mm3 /s and the minimum is 0.0262 × 10−6 mm3 /s under the hydrogen peroxide lubrication. It is reasonable to assume that the silicon surface is without damage after the processing under the certain conditions. The Fig. 10 shows the wear depths of single scratches under different loads with the sliding speed at 8.00 mm/s.
1) The wear mechanism under dry frication is regarded to be of adhesive wear. The wear mechanism under water lubrication is considered to be of molecule-scale removal process. The wear mechanism under hydrogen peroxide lubrication is mainly spalling caused by micro-cracks. Under the static hydrogen peroxide dry friction, when the soft layer is removed, the wear mechanism is same as that under dry frication, which is mainly adhesive wear. 2) The minimum of the wear depth of single scratch is 0.063 nm and the minimum of wear rate is 0.0109 × 10−6 mm3 /s under the water lubrication, while the minimum of the wear depth of single scratch is 0.109 nm and the minimum of wear rate is 0.0262 × 10−6 mm3 /s under the hydrogen peroxide lubrication. From the study in this paper, the hydrogen peroxide is promising material used as the valuable component for the CMP slurry, it is possible to realize green chemical mechanical polishing without damage under certain mechanical action. 3) Under water lubrication and hydrogen peroxide lubrication, when the load is between 30 mN and 70 mN with the sliding speed being 8.00 mm/s, the tribochemical reaction is enhanced to form Si(OH)4 film which possesses lubrication performance in the frictional interface. Thus, the mechanical action accelerates the chemical reaction under the above conditions.
Wear Depth vs. Load 1.8
dry friction water lubrication H2O2 lubrication
1.6
Wear Depth (nm)
1.2 1 0.8 0.6 0.4 0.2
20
30
40
50
This research work was financially supported in part by the Fundamental Research Funds for the Central Universities (JUSRP10909), the State Key Laboratory of Tribology of Tsinghua University (SKLTKF10B04), National Natural Science Foundation of China (51005102, 51175228/E050903) and the Scientific Research Foundation for the Returned Overseas Chinese Scholars, State Education Ministry (20111139). References
static H 2O2 dry friction
1.4
Acknowledgements
60
70
80
90
100
110
120
Load (mN) Fig. 10. Schematic of the wear depths of single scratches under different loads with speed at 8.00 mm/s.
[1] X.C. Li, J.J. Lu, S.R. Yang, Sliding wear behavior and phase transformation of single crystal silicon, Tribology 24 (4) (2004) 326–331. [2] Y.X. Zhang, W. Gao, R.K. Kang, D.M. Guo, TEM observation on the ground damage of monocrystalline silicon wafers, Journal of Semiconductors 29 (8) (2008) 1552–1556. [3] C.Y. Liu, W.E. Fu, T.Y. Lin, C.S. Chang, J.S. Chen, Nanoscale surface roughness characterization by full field polarized light-scattering, Optics and Lasers in Engineering 49 (2011) 145–151. [4] X.K. Hu, Z.T. Song, H.B. Wang, W.L. Liu, Z.F. Zhang, Investigation on the controllable growth of monodisperse silica colloid abrasives for the chemical mechanical polishing application, Microelectronic Engineering 87 (2010) 1751–1755. [5] Y. Wang, Y.W. Zhao, X. Chen, Chemical mechanical planarization from macroscale to molecular-scale, Materials and Manufacturing Processes 27 (2012) 641–649. [6] Y.G. Wang, L.C. Zhang, A. Biddut, Chemical effect on the material removal rate in the CMP of silicon wafers, Wear 270 (2011) 312–316. [7] Y.W. Zhao, L. Chang, S.H. Kim, A mathematical model for chemical-mechanical polishing based on formation and removal of weakly bonded molecular species, Wear 254 (2003) 332–339. [8] X.D. Wang, J.X. Yu, L. Chen, L.M. Qian, Z.R. Zhou, Effects of water and oxygen on the tribochemical wear of monocrystalline Si (1 0 0) against SiO2 sphere by simulating the contact conditions in MEMS, Wear 271 (2011) 1681–1688.
X. Chen et al. / Applied Surface Science 282 (2013) 25–31 [9] M. Varenberg, I. Etsion, G. Halperin, Nanoscale fretting wear study by scanning probe microscopy, Tribology Letters 18 (4) (2005) 493–498. [10] R.A. Singh, E.S. Yoon, Friction of chemical and topographically modified Si (1 0 0) surfaces, Wear 263 (2007) 912–919. [11] Y.R. Jeng, P.Y. Huang, A material removal rate model considering interfacial micro-contact wear behavior for chemical mechanical polishing, Journal of Tribology 127 (2005) 190–197. [12] P.B. Zantye, A. Kumar, A.K. Sikder, Chemical mechanical planarization for micro-electronics applications, Material Science Engineering, R 45 (2004) 89–220. [13] C.C. Wei, J.H. Horng, A.C. Lee, J.F. Lin, Analyses and experimental confirmation of removal performance of silicon oxide film in the chemical mechanical polishing (CMP) process with pattern geometry of concentric groove pads, Wear 270 (2011) 172–180. [14] S.W. Youn, C.G. Kang, Effect of nanoscratch conditions on both deformation behavior and wet-etching characteristics of silicon (1 0 0) surface, Wear 261 (2006) 328–337. [15] J.F. Luo, D.A. Dornfeld, Effects of abrasive size distribution in chemical mechanical planarization: modeling and verification, IEEE Transactions on Semiconductor Manufacturing 16 (3) (2003) 469–476.
31
[16] X.S. Han, Y.Z. Hu, S.Y. Yu, Investigation of material removal mechanism of silicon wafer in the chemical mechanical polishing process using molecular dynamics simulation method, Applied Physics A 95 (2009) 899–905. [17] I. Etsion, Revisiting the Cattaneo-Mindlin concept of interfacial slip in tangentially loaded compliant bodies, Journal of Tribology 132 (2010) 020801. [18] W.C. O’Mara, R.B. Herring, L.P. Hunt, Handbook of Semiconductor Silicon Technology, Noyes Publications, Park Ridge, 1990. [19] Y.W. Zhao, L. Chang, A model of asperity interactions in elastic-plastic contact of rough surfaces, Journal of Tribology 123 (2001) 857–864. [20] X.G. Zhang, Electrochemistry of Silicon and Its Oxide, Kluwer Academic Publishers, New York, 2001. [21] H.W. Liu, B. Bhushan, Nanotribological characterization of molecularly thick lubricant films for application to MEMS/NEMS by AFM, Ultramicroscopy 97 (2003) 321–340. [22] R. Manivannan, S. Ramanathan, The effect of hydrogen peroxide on polishing removal rate in CMP with various abrasives, Applied Surface Science 255 (2009) 3764–3768. [23] L. Si, D. Guo, J.B. Luo, X.C. Lu, Monoatomic layer removal mechanism in chemical mechanical polishing process: a molecular dynamics study, Journal of Applied Physics 107 (6) (2010) 064310.
Contents lists available at SciVerse ScienceDirect
Applied Surface Science journal homepage: www.elsevier.com/locate/apsusc
Nanoscale friction and wear properties of silicon wafer under different lubrication conditions Xiaochun Chen, Yongwu Zhao ∗ , Yongguang Wang, Hailan Zhou, Zhifeng Ni, Wei An School of Mechanical Engineering, Jiangnan University, Wuxi, Jiangsu 214122, PR China
a r t i c l e
i n f o
Article history: Received 8 September 2012 Received in revised form 24 April 2013 Accepted 26 April 2013 Available online 6 June 2013 Keywords: Friction and wear property Silicon wafer Water lubrication Hydrogen peroxide lubrication Wear mechanism
a b s t r a c t The nanoscale friction and wear properties of single crystal silicon wafer under different lubrication conditions are studied in this paper. The experiments were performed with Si3 N4 ball sliding on the surface of silicon wafer under four different lubrication conditions: dry friction, water lubrication, hydrogen peroxide lubrication and the static hydrogen peroxide dry friction. The results from the experiments have been analyzed showing the different friction and wear properties of the silicon wafer in different lubrication conditions. It is concluded that the wear rates under the water lubrication and under the hydrogen peroxide lubrication are both small, the chemical reactions are facilitated by the mechanical processes when the load and the sliding speed reach certain levels. This is mainly resulted by the enhanced lubricant performance with the formed silicon hydroxide Si(OH)4 film. Under the water lubrication, the wear is found in a way of material removed in molecule scale. Under the hydrogen peroxide lubrication, the wear is mainly caused by the spalling of micro-cracks. Under the dry friction condition, the wear is found being adhesive wear. And under the static peroxide dry friction, the wear is prevailing adhesive wear. These results are essential and valuable to the development of the efficient and environmental-friendly slurry for the chemical mechanical polishing (CMP) process. © 2013 Elsevier B.V. All rights reserved.
1. Introduction Single crystal silicon is an important material for semiconductor industry. Silicon wafer is widely used in the integrated circuit (IC) devices, high-density information storage devices, and microelectric-mechanical systems (MEMS) [1,2]. The fabrication and application of these delicate devices require high surface quality of IC chips. This has become one of the major concerns [3]. Up to now, CMP has been proven the best global planarization technique which can satisfy the requirements on surface quality of silicon wafer and chips in the semiconductor industry [4]. With the scientific progress in nanotechnology in recent years, CMP has been developed rapidly. However, the material removal mechanism of the CMP process is not yet clearly understood [5]. Many scientists are extensively working from different aspects to reveal the insights of the CMP technology in order to gain the effective method to remove the material with minimized damage to the polished surface quality, and minimized damage to the environment [3,6].
∗ Corresponding author at: School of Mechanical Engineering, Jiangnan University, No. 1800, Lihu Ave, Wuxi, Jiangsu 214122, China. Tel.: +86 510 8591 0562; fax: +86 510 8532 8262. E-mail addresses: [email protected] (X. Chen), [email protected] (Y. Zhao), [email protected] (Y. Wang), [email protected] (H. Zhou), [email protected] (Z. Ni), [email protected] (W. An). 0169-4332/$ – see front matter © 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.apsusc.2013.04.148
In a typical CMP process in the semiconductor industry, the silicon wafer is carried by the polishing head and pushed onto the polishing pad. The polishing head and the polishing pad rotate, oppositely to each other at a relative speed. Meantime the polishing slurry is fed onto the pad and flows into the polishing interface between the wafer and the pad. The material removal in the CMP process is a synthetic function of chemical and mechanical actions [7]. The friction and wear property of the polished material in a given lubrication condition is an essential factor as well to the quality of the polishing surface and the material removal rate. Therefore, to study on the friction and wear properties of the silicon wafer is an important part of silicon wafer CMP research. By using a servo-hydraulic reciprocating sliding apparatus, Wang et al. [8] studied the effects of water and oxygen on the tribochemical wear of single crystal silicon (1 0 0) against SiO2 ball. The wear tests were performed under four different atmosphere conditions, those were, pure nitrogen, dry air, humid air and pure oxygen. The results from the experiments indicate that the increase in hydrophilicity of silicon (1 0 0) surface will induce an obvious increase of the friction coefficient and the wear depth of silicon surface. Moreover, the oxygen plays a very important role in the wear behavior. The wear depth on silicon surface was highest due to the strong oxidation reaction under the condition of pure oxygen atmosphere. Varenberg et al. [9] studied nano fretting wear on the silicon surface by using scanning probe microscopy. Partial and gross slip fretting with displacement amplitudes from 5 to 500 nm
26
X. Chen et al. / Applied Surface Science 282 (2013) 25–31
Si 3 N4 Ball
Load
Friction Coefficient vs. Load
Reciprocating Direction 1.2
dry friction water lubrication H2O2 lubrication
1.1
static H 2O2 dry friction
Silicon Wafer Fig. 1. Schematic of ball-on-flat sliding.
Friction Coefficient
1 0.9 0.8 0.7 0.6 0.5 0.4
were used for the study, and the results from their study show a substantial increase of the friction at the transition from partial to gross slip and a significant difference between damage surfaces in the two fretting regimes. Although the study was concentrating on the nano fretting wear of the silicon surface which was not chemical modified, the results from the study are good reference for the study of this paper to understand the friction and wear behavior of the modified silicon surface. Singh and Yoon [10] studied the microand nano-frictional properties of chemically and topographically modified Si (1 0 0) surfaces. The results indicate that at the microand nano-scales, the modified Si (1 0 0) surfaces show enhanced friction behavior when compared to the bare Si (1 0 0) surfaces. The modified Si (1 0 0) surfaces exhibit lower friction values. Singh and Yoon found that at the nano-scale, the lower friction values of modified surfaces were due to the lower intrinsic adhesion and contact areas. And with the chemically modified surfaces, the reduction of the contact area was because of the lower interfacial energies. Silicon wafers are thin and brittle slices, which are formed of highly pure and nearly defect-free single crystal silicon. For the application in the IC and MEMS, silicon wafers must have the super surface quality, and be free of defects after fabrication [11–13]. CMP is now the only ideal process to achieve global planarization of silicon wafers [4]. However, it is still lack of fundamental understanding of this process although many researchers have studied it intensively in the past decades [14–16]. Meanwhile traditional CMP slurry contains a large number of chemical substances, such as oxidants, corrosion inhibitors and chelating agents. All these chemicals generate great deal of wastewater and cause heavy environmental pollution. With the strengthening of environmental protection, safety and health consciousness, green polishing slurry is desired in order to replace traditional slurry. It has been becoming research’s hotspot in the world in recent years [6]. In this paper, nanoscale friction and wear properties of single crystal silicon wafer under different lubrication conditions are studied. At the same time, the results from this study are supporting to use deionized water and H2 O2 as possible less harmful green slurry for the CMP process. 2. Experimental The p- Si (1 0 0) single crystal silicon wafers were selected as the test material. The samples were cut from the wafers with a rectangle size of 20 mm × 20 mm. The nanoscale friction and wear tests were performed on a reciprocating UMT-2 micro-tribotester at standard room temperature of 20 ± 1 ◦ C with a relative humidity of 45–55%. The type of frictional contact was ball-on-plate contact, as illustrated in the Fig. 1, a Si3 N4 ceramic ball with the diameter of 4 mm slides on the 20 mm × 20 mm sized sample of p- Si (1 0 0) rectangle silicon wafer. During the experiment, a new Si3 N4 ball is used only one time for each single test in order to minimize the effect of the wear of the ball on the wear of the wafer
0.3 20
30
40
50
60
70
80
90
100
110
120
Load (mN) Fig. 2. Results from experiments about relation between friction coefficient and load.
silicon surface. Before the sliding experiment, both the Si3 N4 ball and the silicon wafer were dipped into acetone and cleansed with an ultrasonic cleaner for 10 min. The test load was 30–110 mN, the sliding speed was 5.33–10.66 mm/s, the time of friction and wear test was 2 min, the amplitude of the reciprocation sliding is 2 mm. A group of friction and wear experiments were conducted under different lubricating conditions namely dry friction, water lubrication, H2 O2 lubrication and static H2 O2 dry friction. Deionized water was used as lubrication medium in water lubrication experiments. 30% H2 O2 was used as lubrication medium in H2 O2 lubrication experiments. In the static H2 O2 dry friction experiments, the wafer samples were immersed in the 30% H2 O2 liquid for 30 min before the experiments, then taken out and got it dried afterward. During the sliding tests in so-called static H2 O2 dry friction experiments no lubrication medium was used. The UMT-2 micro-tribotester collected and recorded the friction coefficients from the sliding experiments automatically. The surface morphology was characterized with Phase Shift MicroXAM-3D, by measuring the length, the width and the depth of the wear marks. The wear rate and the wear depth of single scratch are therefore able to be computed for further analyses. Furthermore, it was pointed out a mathematical modeling for the contact pressure between the ball and surface would be benefit for the understanding of the sliding wear mechanism, which would be addressed in future. 3. Results and discussion 3.1. The effects of load on friction coefficient and wear rate A series of experiments have been conducted to investigate the relationship between the friction coefficient and the applied load. The friction coefficients between the Si3 N4 balls and the single crystal silicon wafers were collected and recorded by the Micro-tribotester from the sliding experiments under four different lubrication conditions. The sliding speed was 8.00 mm/s, the applied loads were 30 mN, 50 mN, 70 mN, 90 mN, 110 mN. The coefficients obtained from the experiments are shown in the Fig. 2. From the Fig. 2, it can be seen that the friction coefficient is becoming smaller with the increasing of the load from 30 mN to 70 mN under all four different lubrication conditions, and there is no big difference with friction coefficient when the load increases above 70 mN.
X. Chen et al. / Applied Surface Science 282 (2013) 25–31
27
Friction Coefficient vs. Speed
Wear Rate vs. Load 0.65
1.2
dry friction water lubrication H2O2 lubrication
1
dry friction water lubrication H2O2 lubrication
0.6 0.55
static H 2O2 dry friction
Friction Coefficient
Wear Rate (10-6mm3/s)
static H 2O2 dry friction 0.8
0.6
0.4
0.5 0.45 0.4 0.35 0.3 0.25
0.2 0.2 0.15
20
30
40
50
60
70
80
90
100
110
120
Load (mN) Fig. 3. Results from experiments about relation between wear rate and load.
Izhak has revisited the Cattaneo-Mindlin Concept of Interfacial Slip in Tangentially Loaded Compliant Bodies [17], the general tendency of decreasing static friction coefficient with increasing normal preload was observed experimentally. In the Izhak’s study, the static friction coefficient decreases sharply with increasing dimensionless normal preload in the early elastic-plastic loading range, and at higher dimensionless normal preloads, the reduction rate of the friction coefficient diminishes, and further the static friction coefficient approaches a constant value. The results in the Fig. 2 regarding the dependency of the sliding friction coefficient on the normal load are in accordance with the results regarding the dependency of the static friction coefficient on the normal preload which are obtained by Izhak, both do obey the classical laws of friction. In the dry friction and the water lubrication experiments, the friction coefficients are nearly equal when the load is 30 mN. And the friction coefficients under the H2 O2 lubrication and the static H2 O2 dry friction experiments are very close to each other as well. When the load increases to 50 mN, the friction coefficients under water lubrication and static H2 O2 dry friction are almost same. The maximum of friction coefficient is under dry friction condition, while the minimum of the friction coefficient is under H2 O2 lubrication condition. The Fig. 2 shows that the friction coefficient deceases rapidly with the increasing of the load when the load is less than 70 mN, and the friction coefficient reaches highest value under the dry friction and the minimum of the friction coefficient is under the water lubrication when the load increases to 70 mN. The friction coefficients under the static H2 O2 dry friction are greater than that under H2 O2 lubrication in all respective loads in the experiments. The wear rate is computed with the length, the width and the depth of the wear mark, which are characterized and measured by Phase Shift MicroXAM-3D. The Fig. 3 shows the relation between the wear rate and the load. From the Fig. 3, it can be seen that the wear rate under dry friction is highest one in all four lubrication conditions when the load is 30 mN. With the load increasing in all experiments of four different lubrication conditions, the wear rates are all increasing, the wear rates under dry friction and static H2 O2 dry friction increase rapidly. When the load is 50 mN, the wear rates under dry friction and static H2 O2 dry friction are much greater than the wear rates under water lubrication and H2 O2 lubrication. The wear rates under water lubrication and H2 O2 lubrication increases only slightly with the load increasing from 30 mN to
5
6
7
8
9
10
11
12
Speed (mm/s) Fig. 4. Results from experiments about relation between friction coefficient and speed.
110 mN. And the wear rates under H2 O2 lubrication are slightly greater than that under water lubrication with all applied loads. The situation with the wear rates under dry friction and static H2 O2 dry friction is different. At the load of 30 mN the wear rate under dry friction is much greater than that under static H2 O2 dry friction, which is very close to the wear rates under water lubrication and H2 O2 lubrication with the same applied load. When the load increases to 50 mN, the wear rate under dry friction is slightly lower than that under H2 O2 dry friction, both are much greater than those under water lubrication and H2 O2 lubrication. With the load increasing, the wear rates under dry friction and H2 O2 dry friction are both increasing to same level, which are extremely greater than the wear rates out of water lubrication and H2 O2 lubrication experiments. The wear rates under water lubrication and H2 O2 lubrication are both slightly increasing with the load increasing. The curves for these two wear rates in the Fig. 3 are quite flat. However, at each load the wear rates under H2 O2 are greater than those under water lubrication. 3.2. The effects of sliding speed on friction coefficient and wear rate Another series of experiments have been conducted to investigate the relationship between the friction coefficient and the sliding speed. The friction coefficients under four different lubrication conditions between the Si3 N4 balls and the single crystal silicon wafers were collected and recorded by the Micro-tribotester. The applied normal load was 70 mN, the sliding speeds were 5.33 mm/s, 6.66 mm/s, 8.00 mm/s, 9.33 mm/s, 10.66 mm/s. The results are shown in the Fig. 4. The Fig. 4 shows the relationship curves of friction coefficients versus the sliding speeds under four different lubrication conditions with the applied normal load at 70 mN. In dry friction experiment, with the sliding speed increasing from 5.33 mm/s to 6.66 mm/s, 8.00 mm/s and 9.33 mm/s the friction coefficient becomes smaller, especially when the speed increases from 6.66 mm/s to 8.00 mm/s the friction coefficient decreased dramatically from 0.5451 to 0.3856. But the friction coefficient raises from 0.3744 to 0.3977 when the sliding speed increases from 9.33 mm/s to 10.66 mm/s. In the H2 O2 lubrication experiment, the friction coefficient increases with the sliding speed increasing. In the water lubrication and the static H2 O2 dry friction experiments the friction
28
X. Chen et al. / Applied Surface Science 282 (2013) 25–31
Wear Rate vs. Speed dry friction water lubrication H2O2 lubrication
1.2
static H 2O2 dry friction
Wear Rate (10-6mm3/s)
1
0.8
0.6 Fig. 6. Topography of wear surface from dry friction experiment.
0.4
0.2
4
5
6
7
8
9
10
11
Speed (mm/s) Fig. 5. Results from experiments about relation between wear rate and speed.
coefficients are dropping down with the increasing of the sliding speed. The relationship between the wear rate and the sliding speed from this series of experiments is shown in the Fig. 5. The wear rates under the dry friction experiments and the static H2 O2 dry friction experiments increase rapidly with the increasing of the sliding speed from 5.33 mm/s to 9.33 mm/s and reaches the highest value, the wear rates at the sliding speed of 10.66 mm/s drops down under both the dry friction and the static H2 O2 dry friction experiments. In the water lubrication and the H2 O2 lubrication experiments, the wear rates increases slightly with the increasing of the sliding speed, and the wear rates under the H2 O2 lubrication are a little bit greater than the wear rates under the water lubrication with all test speeds. 3.3. Discussions on friction coefficient and wear rate The series of experiments aimed at understanding the friction and wear properties of single crystal silicon under different lubrication conditions. Further analyses have been carried through on all above results of the experiments. In the dry friction experiments, when the sliding speed is kept unchanged, as illustrated in the Fig. 2 and in the Fig. 3, the friction heat between the frictional interfaces increases significantly with the increasing of the load. Thus the contact area under the friction heat is more conducive for creating oxide product on the friction surfaces, which induces the decreasing of friction coefficient. Meanwhile, the friction heat causes a temperature rise in the friction surfaces, which makes the single crystal silicon tending to be easier deformable [18], thus the contact area between the Si3 N4 ball and the single crystal silicon increases under the load pressure, which leads to the increasing of wear rate. However, based on the authors’ investigation the temperature increased or the energy dissipated during the sliding tests is still not exactly measurable in current experimental condition. When the load reaches 70 mN, the friction coefficient is gradually reaching a plateau as the load increases further. The reason for this is mainly related to the oxide layer in the friction zone which is formed during the friction reaction. When the applied load is greater than 70 mN, the oxide layer contacts with the Si3 N4 ball directly during the sliding test, the friction and wear phenomenon reflects the Si3 N4 ball against the oxide layer. So the friction and wear tends to a steady stage, the friction coefficient and the wear rate reaches a plateau. In the dry friction experiments, when the
applied normal load is kept unchanged, as illustrated in the Fig. 4 and in the Fig. 5, the friction and wear properties of single crystal silicon are found differently under different sliding speeds. In the range of the sliding speed less than 8.00 mm/s, the friction heat increases significantly with the sliding speed increasing, which is beneficial to the formation of oxide layer and makes single crystal silicon tending deformable, leading to the friction coefficient decreasing sharply and the wear rate increasing rapidly. When the sliding speed reaches 9.33 mm/s and after, severe friction reactions will happen in the frictional interfaces with the speed increasing, and the temperature of the frictional interfaces becomes extremely high. The micro debris from the scratches in the high temperature zone forms very hard layer, which makes the friction coefficient increase. Meanwhile, the real contact area is enlarged in the abrasion interface, thus further results in a decline of the wear rate [19]. The Fig. 6 shows the worn surface of the single crystal silicon from the dry friction experiment with 70 mN load at 8.00 mm/s sliding speed. It can be seen that the worn surface is rough, and there are serious particle adhesion and obvious groove, the wear mechanism is mainly adhesive wear. In the water lubrication experiments, the friction heat accelerates the chemical reactions between silicon and water, and generates synthetic film of silicon peroxide on the surface of silicon wafer [8,20]. When water flows into the frictional interfaces, silicon peroxide reacts with water, while the Si3 N4 reacts with water to have strengthened the tribochemical reactions in the frictional interfaces [21]. SiO2 + 2H2 O = Si(OH)4
(1)
Si3 N4 + 16H2 O = 3Si(OH)4 + 4NH4 OH
(2)
The new generated Si(OH)4 film functions as lubricant and lowers the shear strength in the frictional interfaces. The hardness of the film is lower than the hardness of the silicon wafer surface. On the other hand, water flow plays as coolant and forms water film in between of the frictional interfaces, which reduces the adhesive wear significantly and reduces the friction coefficient and wear rate effectively. When the load is 30 mN in the Fig. 2, the friction coefficient is greater than that from the dry friction test, the reason for this is primarily because that the load is not heavy enough to accelerate the tribochemical reactions in the frictional interfaces compared to those with the load greater than 30 mN. When the load reaches 70 mN and after, the mechanical action is greater than the chemical reaction with the load increasing, the balance is broken. The chemical reaction is accelerated by the mechanical actions during the process [5]. This directly results in the decrement of the friction coefficients and the wear rates. It can be seen that both tend to be stabilized at a certain level from the Fig. 3. The topography of the worn surface from water lubrication experiment can be seen in the Fig. 7. The worn mark is a clean ploughed groove.
X. Chen et al. / Applied Surface Science 282 (2013) 25–31
29
Fig. 7. Topography of wear surface from water lubrication experiment.
In the hydrogen peroxide lubrication experiments, hydrogen peroxide is a specific oxidant, which possesses both oxidation and reduction. The 30% hydrogen peroxide was used in the experiments. The quantity of the hydrogen ions is becoming less because of the reaction in the 30% hydrogen peroxide solution. It becomes slightly acidic and further weakens the oxidation layer on the surface of single crystal silicon. On the other side, hydrogen peroxide can act as a proton donor as well as a proton acceptor parallel [22]. As proton acceptor, molecules of hydrogen peroxide are known to form stronger bonds with silicon dioxide other than water molecules as proton donors to form bonds with silicon dioxide. So the tribochemical reaction under hydrogen peroxide lubrication is weaker than that under water lubrication, which leads that the friction coefficient and the wear rate are greater than that under water lubrication when the loads are over 70 mN. Also, because of the reduction of hydrogen ions, the adhesive oxidation layer on the surface of the single crystal silicon is minimized. Therefore, the adhesive wear is significantly reduced. The Fig. 8 shows that the worn surface is relatively smooth, with a few peeling pits of micro-cracks and micro-flakes. The wear mechanism is mainly spalling, which is caused by micro-cracks. The wear rates under hydrogen peroxide lubrication are greater than those under water lubrication, as illustrated in the Fig. 2 and the Fig. 3. In the experiments with unchanged load at 70 mN, the friction coefficients increase rapidly with the sliding speed increasing, as shown in the Fig. 4. In this period, the top layer of the single crystal silicon is modified by the oxidation and reduction of the hydrogen peroxide. The Si OH− bonds are formed on the
surface, the top layer is assumed softer than the silicon. When the sliding speed is raised, the tribochemical reactions becomes more intensive, the corrosive wear on the surface of single crystal silicon is aggravated. This leads to an increscent friction coefficient. The Fig. 5 shows that the wear rates under hydrogen peroxide lubrication do not change much with the sliding speed increasing. In the experiments under the static hydrogen peroxide dry friction, the wafers of single crystal silicon were immersed in the 30% hydrogen peroxide solution for 30 min before the tests started. During this immersing period, the oxide silicon peroxide film on the surface of single crystal silicon fully contacts the water molecules in the hydrogen peroxide solution, the molecules of silicon peroxide reacts with hydroxyl, the bond of Si O converts to the bond of Si OH− , which eventually forms Si(OH)4 soft layer. When the load is 30 mN, the lubrication of Si(OH)4 effectively reduces the friction coefficient and wear rate. With the load increasing, the Si(OH)4 soft layer is removed gradually, therefore the friction and wear properties of single crystal silicon become similar to that under dry friction, as illustrated in the Fig. 2 and in the Fig. 3. In the experiments with unchanged load, the results are similar to that from the dry friction experiments due to the same reason of the soft layer removed on the surface. The Fig. 9 shows the worn surface from the static hydrogen peroxide dry friction experiments. Obviously, the characteristics seen here are very close to those from the dry friction test which is shown in the Fig. 6.
Fig. 8. Topography of wear surface from hydrogen peroxide lubrication experiment.
Fig. 9. Topography of wear surface from static hydrogen peroxide dry friction experiment.
30
X. Chen et al. / Applied Surface Science 282 (2013) 25–31
Table 1 The wear depth of single scratch under four different lubrication conditions with speed at 8.00 mm/s. Load (mN)
Wear depth from dry friction (nm)
Wear depth from water lubrication (nm)
Wear depth from H2 O2 lubrication (nm)
Wear depth from static H2 O2 dry friction (nm)
30 50 70 90 110
1.029 1.066 1.252 1.275 1.458
0.063 0.066 0.068 0.087 0.089
0.109 0.122 0.153 0.181 0.201
0.350 1.156 1.274 1.279 1.452
3.4. Discussions on the wear depth of single scratch
4. Conclusions
The wear depths of single scratches on the surface of single crystal silicon in these series of experiments are collected and given in the Table 1. The minimum wear depth under dry friction is 1.029 nm. Under water lubrication, the wear depth of single scratch is the smallest with the minimum value of 0.063 nm and the maximum value of 0.089 nm. The wear depths of single scratches under hydrogen peroxide lubrication are slightly greater than that from water lubrication experiments, the minimum value is 0.109 nm and the maximum value is 0.201 nm. Under the static hydrogen peroxide corrosion dry tests, the minimum wear depth is 0.35 nm at 30 mN. When the load is raised to 50 mN, the wear depth is close to that from dry friction test. The wear depth of single scratch is at 10−11 m magnitude under the water lubrication tests and at 10−10 m magnitude under the hydrogen peroxide lubrication tests. Therefore, it can be assumed that the material removal is happened at the molecular scale under these two conditions [7,23] and the wear mechanism under the water lubrication is considered to be of molecule-scale removal process. In this regard, the study is exploring the nanoscale friction and wear properties of the test material single crystal silicon. In the meantime, the wear rates are both small under the water lubrication and the hydrogen peroxide lubrication. The Fig. 2 shows that the maximum wear rate is 0.0303 × 10−6 mm3 /s and the minimum is 0.0109 × 10−6 mm3 /s under water lubrication, the maximum wear rate is 0.0467 × 10−6 mm3 /s and the minimum is 0.0262 × 10−6 mm3 /s under the hydrogen peroxide lubrication. It is reasonable to assume that the silicon surface is without damage after the processing under the certain conditions. The Fig. 10 shows the wear depths of single scratches under different loads with the sliding speed at 8.00 mm/s.
1) The wear mechanism under dry frication is regarded to be of adhesive wear. The wear mechanism under water lubrication is considered to be of molecule-scale removal process. The wear mechanism under hydrogen peroxide lubrication is mainly spalling caused by micro-cracks. Under the static hydrogen peroxide dry friction, when the soft layer is removed, the wear mechanism is same as that under dry frication, which is mainly adhesive wear. 2) The minimum of the wear depth of single scratch is 0.063 nm and the minimum of wear rate is 0.0109 × 10−6 mm3 /s under the water lubrication, while the minimum of the wear depth of single scratch is 0.109 nm and the minimum of wear rate is 0.0262 × 10−6 mm3 /s under the hydrogen peroxide lubrication. From the study in this paper, the hydrogen peroxide is promising material used as the valuable component for the CMP slurry, it is possible to realize green chemical mechanical polishing without damage under certain mechanical action. 3) Under water lubrication and hydrogen peroxide lubrication, when the load is between 30 mN and 70 mN with the sliding speed being 8.00 mm/s, the tribochemical reaction is enhanced to form Si(OH)4 film which possesses lubrication performance in the frictional interface. Thus, the mechanical action accelerates the chemical reaction under the above conditions.
Wear Depth vs. Load 1.8
dry friction water lubrication H2O2 lubrication
1.6
Wear Depth (nm)
1.2 1 0.8 0.6 0.4 0.2
20
30
40
50
This research work was financially supported in part by the Fundamental Research Funds for the Central Universities (JUSRP10909), the State Key Laboratory of Tribology of Tsinghua University (SKLTKF10B04), National Natural Science Foundation of China (51005102, 51175228/E050903) and the Scientific Research Foundation for the Returned Overseas Chinese Scholars, State Education Ministry (20111139). References
static H 2O2 dry friction
1.4
Acknowledgements
60
70
80
90
100
110
120
Load (mN) Fig. 10. Schematic of the wear depths of single scratches under different loads with speed at 8.00 mm/s.
[1] X.C. Li, J.J. Lu, S.R. Yang, Sliding wear behavior and phase transformation of single crystal silicon, Tribology 24 (4) (2004) 326–331. [2] Y.X. Zhang, W. Gao, R.K. Kang, D.M. Guo, TEM observation on the ground damage of monocrystalline silicon wafers, Journal of Semiconductors 29 (8) (2008) 1552–1556. [3] C.Y. Liu, W.E. Fu, T.Y. Lin, C.S. Chang, J.S. Chen, Nanoscale surface roughness characterization by full field polarized light-scattering, Optics and Lasers in Engineering 49 (2011) 145–151. [4] X.K. Hu, Z.T. Song, H.B. Wang, W.L. Liu, Z.F. Zhang, Investigation on the controllable growth of monodisperse silica colloid abrasives for the chemical mechanical polishing application, Microelectronic Engineering 87 (2010) 1751–1755. [5] Y. Wang, Y.W. Zhao, X. Chen, Chemical mechanical planarization from macroscale to molecular-scale, Materials and Manufacturing Processes 27 (2012) 641–649. [6] Y.G. Wang, L.C. Zhang, A. Biddut, Chemical effect on the material removal rate in the CMP of silicon wafers, Wear 270 (2011) 312–316. [7] Y.W. Zhao, L. Chang, S.H. Kim, A mathematical model for chemical-mechanical polishing based on formation and removal of weakly bonded molecular species, Wear 254 (2003) 332–339. [8] X.D. Wang, J.X. Yu, L. Chen, L.M. Qian, Z.R. Zhou, Effects of water and oxygen on the tribochemical wear of monocrystalline Si (1 0 0) against SiO2 sphere by simulating the contact conditions in MEMS, Wear 271 (2011) 1681–1688.
X. Chen et al. / Applied Surface Science 282 (2013) 25–31 [9] M. Varenberg, I. Etsion, G. Halperin, Nanoscale fretting wear study by scanning probe microscopy, Tribology Letters 18 (4) (2005) 493–498. [10] R.A. Singh, E.S. Yoon, Friction of chemical and topographically modified Si (1 0 0) surfaces, Wear 263 (2007) 912–919. [11] Y.R. Jeng, P.Y. Huang, A material removal rate model considering interfacial micro-contact wear behavior for chemical mechanical polishing, Journal of Tribology 127 (2005) 190–197. [12] P.B. Zantye, A. Kumar, A.K. Sikder, Chemical mechanical planarization for micro-electronics applications, Material Science Engineering, R 45 (2004) 89–220. [13] C.C. Wei, J.H. Horng, A.C. Lee, J.F. Lin, Analyses and experimental confirmation of removal performance of silicon oxide film in the chemical mechanical polishing (CMP) process with pattern geometry of concentric groove pads, Wear 270 (2011) 172–180. [14] S.W. Youn, C.G. Kang, Effect of nanoscratch conditions on both deformation behavior and wet-etching characteristics of silicon (1 0 0) surface, Wear 261 (2006) 328–337. [15] J.F. Luo, D.A. Dornfeld, Effects of abrasive size distribution in chemical mechanical planarization: modeling and verification, IEEE Transactions on Semiconductor Manufacturing 16 (3) (2003) 469–476.
31
[16] X.S. Han, Y.Z. Hu, S.Y. Yu, Investigation of material removal mechanism of silicon wafer in the chemical mechanical polishing process using molecular dynamics simulation method, Applied Physics A 95 (2009) 899–905. [17] I. Etsion, Revisiting the Cattaneo-Mindlin concept of interfacial slip in tangentially loaded compliant bodies, Journal of Tribology 132 (2010) 020801. [18] W.C. O’Mara, R.B. Herring, L.P. Hunt, Handbook of Semiconductor Silicon Technology, Noyes Publications, Park Ridge, 1990. [19] Y.W. Zhao, L. Chang, A model of asperity interactions in elastic-plastic contact of rough surfaces, Journal of Tribology 123 (2001) 857–864. [20] X.G. Zhang, Electrochemistry of Silicon and Its Oxide, Kluwer Academic Publishers, New York, 2001. [21] H.W. Liu, B. Bhushan, Nanotribological characterization of molecularly thick lubricant films for application to MEMS/NEMS by AFM, Ultramicroscopy 97 (2003) 321–340. [22] R. Manivannan, S. Ramanathan, The effect of hydrogen peroxide on polishing removal rate in CMP with various abrasives, Applied Surface Science 255 (2009) 3764–3768. [23] L. Si, D. Guo, J.B. Luo, X.C. Lu, Monoatomic layer removal mechanism in chemical mechanical polishing process: a molecular dynamics study, Journal of Applied Physics 107 (6) (2010) 064310.