The Effects of Friction and Temperature in the Chemical–Mechanical Planarization Process

Chemical–mechanical planarization (CMP) represents the preferred technology in which both chemical and mechanical interactions are combined to achieve global planarization/polishing of wafer surfaces (wafer patterns from metal with a selective layer, in this paper). CMP is a complex process of material removal process by friction, which interferes with numerous mechanical and chemical parameters. Compared with chemical parameters, mechanical parameters have a greater influence on the material removal rate (MRR). The mechanical parameters manifest by friction force (Ff) and heat generated by friction in the CMP process. The Ff can be estimated by its monitoring in the CMP process, and process temperature is obtained with help of an infrared rays (IR) sensor. Both the Ff and the MRR increase by introducing colloidal silica (SiO2) as an abrasive into the selective layer CMP slurry. The calculated wafer non-uniformity (WNU) was correlated with the friction coefficient (COF). The control of Ff and of the slurry stability is important to maintain a good quality of planarization with optimal results, because Ff participates in mechanical abrasion, and large Ff may generate defects on the wafer surface. Additionally, the temperature generated by the Ff increases as the SiO2 concentration increases. The MRR of the selective layer into the CMP slurry showed a non-linear (Prestonian) behavior, useful not only to improve the planarization level but to improve its non-uniformity due to the various pressure distributions. The evaluation of the Ff allowed the calculation of the friction energy (Ef) to highlight the chemical contribution in selective-layer CMP, from which it derived an empirical model for the material removal amount (MRA) and validated by the CMP results. With the addition of abrasive nanoparticles into the CMP slurry, their concentration increased and the MRA of the selective layer improved; Ff and MRR can be increased due to the number of chemisorbed active abrasive nanoparticles by the selective layer. Therefore, a single abrasive was considered to better understand the effect of SiO2 concentration as an abrasive and of the MRR features depending on abrasive nanoparticle concentration. This paper highlights the correlation between friction and temperature of the SiO2 slurry with CMP results, useful to examine the temperature distribution. All the MRRs depending on Ef after planarization with various SiO2 concentrations had a non-linear characteristic. The obtained results can help in developing a CMP process more effectively.


Introduction
In a friction pair, a selective transfer can certainly be made if, in the contact area, there exists an adequate lubricant (such as glycerin), a copper-based material (in this case, bronze), relative movement, and favorable energy [1]. By investigating the CMP process of the selective layer, in which copper was the predominant element (≥85%), it was found that under normal conditions, its surface is oxidized. The rates of oxide removal differ depending on the applied load, the removal depth, and the slurry used. There is and sizes of some abrasive particles, while Kim in ref. [14] analyzed the tribological state during the CMP process, how it influences the pad surface, and its characterization.
To elucidate the copper CMP mechanism, the mechanical parameters of the copper CMP process were analyzed and approached. Lee et al. in ref. [15] analyzed the mechanical aspects of the MRR phenomenon, and Levert et al. in ref. [16] researched the oxide's influence on the friction force in the copper CMP process. Furthermore, they proved the dependence of the friction force on the nature of the oxide film generated by H 2 O 2 during copper CMP. Instead, Luan et al. [12] and Stojadinovic et al. [17] proved that the copper MRR grows as the abrasive's concentration increases. Additionally, the MRR depends on the covered surface by the dispersed abrasives particles, the sizes, and the abrasive particles' shape [18]. Both at copper CMP and at selective-layer CMP, the physical defects (such as scratching and etching), are related to the friction force during CMP. As a result, they allow an important understanding of the friction characteristics of the selective-layer CMP process. Thus, this paper investigates the friction, thermal effect, and the MRRs of the selective-layer CMP process depending on SiO 2 concentrations in acid slurry (a mixture of an oxidizer, corrosion inhibitor, a complexing agent, and a surfactant) by monitoring the friction force and the stability of the slurry used.
The results of this investigation can help in developing a CMP process more effectively, with an optimal MRR to achieve a quality planarization.

Materials and Methods
The friction and thermal investigation in the CMP process on a selective-layer surface were carried out on a wafer pattern with a selective layer. The wafer pattern from metal was made of OLC45 steel (equivalent AISI 1045), previously coated on one face with a selective layer (thickness of about 500 nm) by friction with the CuSn12T bronze (equivalent CC483K).
The planarization experiments were performed by using a CP-4 CMP planarization (polisher) installation with a polyurethane pad and an acid slurry mixed with an oxidizer, corrosion inhibitor, complexing agent, and surfactant. The CMP planarizer/polisher used was of rotational type, and the pressing normal down-force between the wafer and the planarization pad was achieved with the help of the air pressure from a supply source and was uniformly applied to the wafer carrier. The applied pressure ranged from 0.5 psi to 3.0 psi, and the rotational speed normal at planarization ranged between 0 and 100 rpm. The CP-4 CMP planarizer was properly equipped with in situ monitoring sensors to signal the (detected) friction force and temperature during the CMP process. After being converted and amplified, the signals emitted by sensors were transferred to the CMP analysis software DAQ 21.8 (a data acquisition system). The infrared (IR) sensor used for monitoring the friction force was placed at the polishing head edge. The signals of temperature and friction force were displayed in real-time on a monitor. The dynamic friction force signal was registered with a piezoelectric sensor attached to the planarization head in the form of voltage, and then was amplified and converted. All these signals were analyzed with the help of the CMP analysis software.

Theoretical Aspects
The measurement points of the process temperature between the wafer and pad were at the contact area edge. Regarding the measured temperature in the CMP process, although not considered the real temperature, the technique used can indirectly indicate the characteristics of the process temperature. Using the America Society of Testing Materials (ASTM F1530-94), standard deviation uniformity can be determined using the wafer nonuniformity (WNU), with the relationship: where, σ is standard deviation and MRR avg is the average removal rate of the material.
Implicitly, the WNU calculation also demonstrates the non-uniformity of the CMP process temperature, so by measuring the temperature at the edge of the wafer-pad contact area, we cannot say with certainty that it represents the real temperature. However, by measuring the temperature at several points along the edge of the wafer-pad contact area with the help sensors, and analyzing the signals emitted and transferred to the data analysis software DAQ 21.8, we can consider with sufficient accuracy that these measurements represent the temperature characteristics of the CMP process.
The MRR expresses itself with the Preston equation [19] in the form: Equation (2) shows the MRR dependence by the Prestonian constant, k, the pressure, p, and relative speed, v r . The constant, k, includes the slurry characteristics of planarization equipment, pad, and other environmental factors.
Friction energy E f [20] can be determined using Equation (3) in the CMP process, namely: where: F f is friction force and t is planarization (polishing) time.
The expression for E f in Equation (3) represents the mechanical contribution in the MRR during planarization/polishing by CMP. Although the mechanical contribution has a greater influence on the MRR than the chemical one, it should not be ignored when studying the thermal effects in the CMP process. This is also due to the fact that the chemical reaction plays a great role in the CMP process.
Thus, the multiplication between the activation energy (E a ) and the molecular frequency defines the chemical reaction rate. Additionally, with the growth of molecular kinetic energy, which surpasses E a , the process temperature increases, and the chemical reaction probability is elevated. For this reason, the reactive constant (K) from the Arrhenius-type equation [21] becomes larger and is given by Equation (4): where A is the Arrhenius constant, E a is the activation energy, R is the gas universal constant, respectively, and T is temperature.
In the selective-layer CMP process, understanding the MRR features depending on nanoparticle concentration could be useful for the MRR calculation per particle. According to Equation (2) the MRR can be established with Preston's equation and it depends on p and v r . However, the MRR during the CMP process may also be expressed depending on the E f . Thus, Stojadinovic et al. in ref. [17] presented the MRR calculation on a single abrasive (proposed by Tamboli et al. in ref. [22]), considering the MRR produced by the planarization pad. Hence, if it is admitted that the abrasive nanoparticles are nearly spherical, their distribution and size can be neglected, and all participate in planarization, then per an abrasive nanoparticle single, the MRR calculation equation can be rewritten, thus: where MRA is the material removal amount; MRR pad only is the MRR at an abrasive concentration of 0.0 wt.%; ρ is the abrasive density; d is the abrasive size; Φ m is the slurry mass flow during the CMP; and W is the fractional weight of the abrasive in the slurry. In fact, the E f is a physical representation of the environmental conditions, pressure, relative motion, material characteristics, surface conditions of the chemically etched wafers, and other planarization/polishing conditions. The use of E f can make it easy to represent where (v r t 0 F f dt) represents the mechanical contribution in the MRR during planarizing/polishing, and C c (Å/min kJ) is referred to as a chemical contribution [23] in the MRA of the CMP. Therefore, the multiplication between E f and abrasive concentration condition may be defined as the mechanical energy rate accompanying the CMP process. Thus, the MRA combined with the interaction between chemical and mechanical energy is expressed with Equation (6) and the ratio between the MRA and mechanical energy can be an index of chemical contribution (C c ) in the CMP process, i.e., the C c would be affected by the mechanical condition of the CMP process.

Friction Characteristics in the CMP Process
The aim of the CMP process is to obtain the global planarization/polishing of the surfaces of the wafers (here covered with a selective layer), in which mechanical and chemical interactions are combined to achieve an efficient MRR. Mechanical interactions are manifested by F f and frictional heat in the CMP process. By introducing SiO 2 as an abrasive in the selective-layer CMP slurry, both F f and MRR increase, and the calculated WNU is correlated with the COF. Additionally, the F f allows the calculation of the E f to facilitate the representation of mechanical action/abrasion during the CMP process.
Thus, Figure 1 shows the variation of MRR and WNU depending on E f . As shown in Figure 1, E f is related in particular to the MRR, not to the WNU. It is observed as MRR increases with the increasing E f . In the CMP process of the oxides, the input energy caused by the relative speed, the chemicals, and of contact is composed of energy, E f , thermal, E t , and vibration, E v . Among these components of energy, E f and E t participate at the MRR, and E v is dispersed in the surrounding environment [15,16]. According to Figure 1, the MRR is dependent on the E f in the CMP process; however, the MRR is related to F f through the quantity removed per unit length.
where ·( ) represents the mechanical contribution in the MRR during planarizing/polishing, and Cc (Å/min kJ) is referred to as a chemical contribution [23] in the MRA of the CMP.
Therefore, the multiplication between Ef and abrasive concentration condition may be defined as the mechanical energy rate accompanying the CMP process. Thus, the MRA combined with the interaction between chemical and mechanical energy is expressed with Equation (6) and the ratio between the MRA and mechanical energy can be an index of chemical contribution (Cc) in the CMP process, i.e., the Cc would be affected by the mechanical condition of the CMP process.

Friction Characteristics in the CMP Process
The aim of the CMP process is to obtain the global planarization/polishing of the surfaces of the wafers (here covered with a selective layer), in which mechanical and chemical interactions are combined to achieve an efficient MRR. Mechanical interactions are manifested by Ff and frictional heat in the CMP process. By introducing SiO2 as an abrasive in the selective-layer CMP slurry, both Ff and MRR increase, and the calculated WNU is correlated with the COF. Additionally, the Ff allows the calculation of the Ef to facilitate the representation of mechanical action/abrasion during the CMP process.
Thus, Figure 1 shows the variation of MRR and WNU depending on Ef. As shown in Figure 1, Ef is related in particular to the MRR, not to the WNU. It is observed as MRR increases with the increasing Ef. In the CMP process of the oxides, the input energy caused by the relative speed, the chemicals, and of contact is composed of energy, Ef, thermal, Et, and vibration, Ev. Among these components of energy, Ef and Et participate at the MRR, and Ev is dispersed in the surrounding environment [15,16]. According to Figure 1, the MRR is dependent on the Ef in the CMP process; however, the MRR is related to Ff through the quantity removed per unit length. The factors that influence the WNU in the CMP process are largely divided into processing and equipment conditions, chemicals, and consumables. Therefore, the WNU depends on the CMP planarization mechanical characteristics in the work process if there  The factors that influence the WNU in the CMP process are largely divided into processing and equipment conditions, chemicals, and consumables. Therefore, the WNU depends on the CMP planarization mechanical characteristics in the work process if there are no changes in the consumables and the slurry composition. Experimentally, it turned out that there was a connection between the friction coefficient (COF) and the WNU. Thus, Figure 2 shows the evolution of the MRR and WNU depending on COF. The COF was introduced to characterize the friction effect and was determined by measuring F f and establishing the resultant normal force during the CMP process. According to Figure 2, as the COF increases, the WNU also grows, due to the resultant normal force's position [13,20]. The resultant normal force is determined from the moment's equilibrium equation relative to the planarization (polishing) head center, as it moves away from the planarization/polishing platen. are no changes in the consumables and the slurry composition. Experimentally, it turned out that there was a connection between the friction coefficient (COF) and the WNU. Thus, Figure 2 shows the evolution of the MRR and WNU depending on COF. The COF was introduced to characterize the friction effect and was determined by measuring Ff and establishing the resultant normal force during the CMP process. According to Figure  2, as the COF increases, the WNU also grows, due to the resultant normal force's position [13,20]. The resultant normal force is determined from the moment's equilibrium equation relative to the planarization (polishing) head center, as it moves away from the planarization/polishing platen.    Figure 3a shows the variation Ef and average Ff with various rotating speeds at constant pressure, p. While n(vr) increases, the Ef also increases; however, the average Ff decreases because the removed amount per unit time is prolonged with an increase in n(vr). Moreover, the decrease in Ff with growing n(vr) also takes place due to the support provided by the boundary layer between the wafer and pad formed as a result of the suspension flow via the pressure dynamic, p [16]. Figure 3b shows the variation of Ff and Ef with various p of planarization/polishing at an n(vr) constant. With the increase in p of planarization/polishing, both the Ff and Ef grow due to the real contact area influence, caused by the p of planarization.

Thermal Effect in the CMP Process
The temperature generated by the friction to the pad-wafer contact in the CMP process is a critical parameter for the MRR. At the same time, this temperature is inevi-  are no changes in the consumables and the slurry composition. Experimentally, it turned out that there was a connection between the friction coefficient (COF) and the WNU. Thus, Figure 2 shows the evolution of the MRR and WNU depending on COF. The COF was introduced to characterize the friction effect and was determined by measuring Ff and establishing the resultant normal force during the CMP process. According to Figure  2, as the COF increases, the WNU also grows, due to the resultant normal force's position [13,20]. The resultant normal force is determined from the moment's equilibrium equation relative to the planarization (polishing) head center, as it moves away from the planarization/polishing platen.    Figure 3a shows the variation Ef and average Ff with various rotating speeds at constant pressure, p. While n(vr) increases, the Ef also increases; however, the average Ff decreases because the removed amount per unit time is prolonged with an increase in n(vr). Moreover, the decrease in Ff with growing n(vr) also takes place due to the support provided by the boundary layer between the wafer and pad formed as a result of the suspension flow via the pressure dynamic, p [16]. Figure 3b shows the variation of Ff and Ef with various p of planarization/polishing at an n(vr) constant. With the increase in p of planarization/polishing, both the Ff and Ef grow due to the real contact area influence, caused by the p of planarization.

Thermal Effect in the CMP Process
The temperature generated by the friction to the pad-wafer contact in the CMP process is a critical parameter for the MRR. At the same time, this temperature is inevi-  Figure 3a shows the variation E f and average F f with various rotating speeds at constant pressure, p. While n(v r ) increases, the E f also increases; however, the average F f decreases because the removed amount per unit time is prolonged with an increase in n(v r ). Moreover, the decrease in F f with growing n(v r ) also takes place due to the support provided by the boundary layer between the wafer and pad formed as a result of the suspension flow via the pressure dynamic, p [16]. Figure 3b shows the variation of F f and E f with various p of planarization/polishing at an n(v r ) constant. With the increase in p of planarization/polishing, both the F f and E f grow due to the real contact area influence, caused by the p of planarization.

Thermal Effect in the CMP Process
The temperature generated by the friction to the pad-wafer contact in the CMP process is a critical parameter for the MRR. At the same time, this temperature is inevitable during the CMP process because it is provoked by F f due to the abrasive nanoparticles and the CMP slurry chemistry. Therefore, Figure 4 illustrates how the CMP process friction characteristics are dependent on the material mechanical and chemical removal. Thus, the friction is caused the most by the abrasive nanoparticles in the CMP slurry, as seen in Figure 4a. Additionally, it is observed that the CMP process temperature, depending on the pressure, p (see Figure 4b), generated through de-ionized water, is bigger than that that generated through the slurry. The de-ionized water was used in the experiments to significantly highlight the effect (influence) of the abrasive nanoparticles in the slurry, as seen in Figure 4. Therefore, the process temperature also contributes to the MRR, which is caused by F f in the CMP process. table during the CMP process because it is provoked by Ff due to the abrasive nanoparticles and the CMP slurry chemistry. Therefore, Figure 4 illustrates how the CMP process friction characteristics are dependent on the material mechanical and chemical removal. Thus, the friction is caused the most by the abrasive nanoparticles in the CMP slurry, as seen in Figure 4a. Additionally, it is observed that the CMP process temperature, depending on the pressure, p (see Figure 4b), generated through de-ionized water, is bigger than that that generated through the slurry. The de-ionized water was used in the experiments to significantly highlight the effect (influence) of the abrasive nanoparticles in the slurry, as seen in Figure 4. Therefore, the process temperature also contributes to the MRR, which is caused by Ff in the CMP process. Therefore, during the friction process in the CMP, the change in the CMP process temperature becomes essential. The effects of the slurry nanoparticle concentration and size were investigated by many researchers [16,[24][25][26], often with contradictory results, because it influences the MRR during the CMP process.
This discrepancy explains itself based on the size concentration and the properties of the nanoparticles in the slurry and the planarization technique. At the same time, it indicates that the different planarization processes can become dominant for a certain planarization system as a mode of CMP application.
To solve the differences between the applied planarization rates, two material removal mechanisms were explained based on the silica (SiO2) CMP [2,8,15,27]. Figure 5 highlights the MRR distribution and Ff obtained with 50 nm nanoparticle size SiO2 slurry depending on the slurry abrasive concentration (from 0.0 to 30 wt.%). At a low abrasive nanoparticle concentration in the slurries (0.0-7.5 wt.%), the MRR increases with growth in the concentration of abrasive nanoparticles, indicating the mechanical removal mechanism. The surface micrographs of the wafer obtained with an AFM from Figure 6 present the planarized surface in the base-slurry ( Figure 6a) and scratches on the planarized surface in the slurry with SiO2 abrasive concentration of 15 wt.%, after the CMP process ( Figure 6b).
The surface roughness of the wafer planarized in the base-slurry (see Figure 6a) was Ra of 1.512 nm, starting from Ra (before planarization) of 4409 nm, measured with a Portable Optical Profilometer (JR100, Nanovea SRL, European Office, Rivalta di Torino, Italy). This roughness of the wafer surface was measured after one hour of chemicalmechanical planarization.
The AFM micrograph (see Figure 6b) shows that the nanoparticles started sliding and rolling on the wafer as a result of the friction and the MRR. Additionally, the depth of the scratches caused by slurry SiO2 abrasive concentration was less than 7 nm. Therefore, during the friction process in the CMP, the change in the CMP process temperature becomes essential. The effects of the slurry nanoparticle concentration and size were investigated by many researchers [16,[24][25][26], often with contradictory results, because it influences the MRR during the CMP process.
This discrepancy explains itself based on the size concentration and the properties of the nanoparticles in the slurry and the planarization technique. At the same time, it indicates that the different planarization processes can become dominant for a certain planarization system as a mode of CMP application.
To solve the differences between the applied planarization rates, two material removal mechanisms were explained based on the silica (SiO 2 ) CMP [2,8,15,27]. Figure 5 highlights the MRR distribution and F f obtained with 50 nm nanoparticle size SiO 2 slurry depending on the slurry abrasive concentration (from 0.0 to 30 wt.%). At a low abrasive nanoparticle concentration in the slurries (0.0-7.5 wt.%), the MRR increases with growth in the concentration of abrasive nanoparticles, indicating the mechanical removal mechanism. The surface micrographs of the wafer obtained with an AFM from Figure 6 present the planarized surface in the base-slurry (Figure 6a) and scratches on the planarized surface in the slurry with SiO 2 abrasive concentration of 15 wt.%, after the CMP process (Figure 6b).
The surface roughness of the wafer planarized in the base-slurry (see Figure 6a) was Ra of 1.512 nm, starting from Ra (before planarization) of 4409 nm, measured with a Portable Optical Profilometer (JR100, Nanovea SRL, European Office, Rivalta di Torino, Italy). This roughness of the wafer surface was measured after one hour of chemicalmechanical planarization. It is notable that MRR reaches its maximum value at an abrasive nanoparticle concentration of 7.5 wt.%, after which a significant decrease occurs. This change is caused by the motion modification of the nanoparticles, and an increase in their number leads to the load per nanoparticle decreasing when in contact with the wafer surface. Then, for the abrasive concentrations between 7.5 and 15 wt.%, the nanoparticles start to roll faster rather than slide on the wafer surface [16]. The surface AFM micrograph at 15 wt.% shows pitting deformations rather than scratches on the wafer surface, demonstrating the rolling motion of the abrasive nanoparticles ( Figure 6b). Therefore, at higher abrasive nanoparticle concentrations, MRR is primarily due to chemical interactions which significantly reduce the mechanical removal action. Figure 7 describes the MRR depending on the concentration of abrasive nanoparticles (in the range 0.0-7.5 wt.%) with the 50 nm abrasive size. For planarization, a relative speed of 0.75 m/s, a pressure of 0.05 MPa, and a slurry flow of 150 mL/min were used. It can be observed that as the abrasive nanoparticle concentration increased, the MRR of the selective layer also increased. Thus, the MRR was about 85 nm/min when the abrasive nanoparticle concentration was 0.0 wt.%. This value is the mechanical removal/abrasion result between the pad asperities and the selective layer, respectively, by the selective-layer chemical dissolution. When 0.5 wt.% SiO2 was added in the selective-layer CMP slurry, the MRR reached approximately 132 nm/min. The MRR increased rapidly until the abrasive nanoparticle concentration of 2.5 wt.%, and then increased slowly as the abrasive nanoparticle concentration grew from 2.5 wt.% to 7.5 wt.%. It is notable that MRR reaches its maximum value at an abrasive nanoparticle concentration of 7.5 wt.%, after which a significant decrease occurs. This change is caused by the motion modification of the nanoparticles, and an increase in their number leads to the load per nanoparticle decreasing when in contact with the wafer surface. Then, for the abrasive concentrations between 7.5 and 15 wt.%, the nanoparticles start to roll faster rather than slide on the wafer surface [16]. The surface AFM micrograph at 15 wt.% shows pitting deformations rather than scratches on the wafer surface, demonstrating the rolling motion of the abrasive nanoparticles ( Figure 6b). Therefore, at higher abrasive nanoparticle concentrations, MRR is primarily due to chemical interactions which significantly reduce the mechanical removal action. Figure 7 describes the MRR depending on the concentration of abrasive nanoparticles (in the range 0.0-7.5 wt.%) with the 50 nm abrasive size. For planarization, a relative speed of 0.75 m/s, a pressure of 0.05 MPa, and a slurry flow of 150 mL/min were used. It can be observed that as the abrasive nanoparticle concentration increased, the MRR of the selective layer also increased. Thus, the MRR was about 85 nm/min when the abrasive nanoparticle concentration was 0.0 wt.%. This value is the mechanical removal/abrasion result between the pad asperities and the selective layer, respectively, by the selective-layer chemical dissolution. When 0.5 wt.% SiO2 was added in the selective-layer CMP slurry, the MRR reached approximately 132 nm/min. The MRR increased rapidly until the abrasive nanoparticle concentration of 2.5 wt.%, and then increased slowly as the abrasive nanoparticle concentration grew from 2.5 wt.% to 7.5 wt.%. The AFM micrograph (see Figure 6b) shows that the nanoparticles started sliding and rolling on the wafer as a result of the friction and the MRR. Additionally, the depth of the scratches caused by slurry SiO 2 abrasive concentration was less than 7 nm.
It is notable that MRR reaches its maximum value at an abrasive nanoparticle concentration of 7.5 wt.%, after which a significant decrease occurs. This change is caused by the motion modification of the nanoparticles, and an increase in their number leads to the load per nanoparticle decreasing when in contact with the wafer surface. Then, for the abrasive concentrations between 7.5 and 15 wt.%, the nanoparticles start to roll faster rather than slide on the wafer surface [16]. The surface AFM micrograph at 15 wt.% shows pitting deformations rather than scratches on the wafer surface, demonstrating the rolling motion of the abrasive nanoparticles (Figure 6b). Therefore, at higher abrasive nanoparticle concentrations, MRR is primarily due to chemical interactions which significantly reduce the mechanical removal action. Figure 7 describes the MRR depending on the concentration of abrasive nanoparticles (in the range 0.0-7.5 wt.%) with the 50 nm abrasive size. For planarization, a relative speed of 0.75 m/s, a pressure of 0.05 MPa, and a slurry flow of 150 mL/min were used. It can be observed that as the abrasive nanoparticle concentration increased, the MRR of the selective layer also increased. Thus, the MRR was about 85 nm/min when the abrasive nanoparticle concentration was 0.0 wt.%. This value is the mechanical removal/abrasion result between the pad asperities and the selective layer, respectively, by the selectivelayer chemical dissolution. When 0.5 wt.% SiO 2 was added in the selective-layer CMP slurry, the MRR reached approximately 132 nm/min. The MRR increased rapidly until the abrasive nanoparticle concentration of 2.5 wt.%, and then increased slowly as the abrasive nanoparticle concentration grew from 2.5 wt.% to 7.5 wt.%. In support of these observations, in situ Ff measurements showed an evolutionary trend of the total Ff, similar to that of the MRR depending on the abrasive nanoparticle concentration in the slurry with the size of 50 nm (see Figure 5).
Additionally, it was proved that in situ measurements during the CMP process are correlated with the MRR response presented in ref. [7], i.e., the load per particle decreases with the increasing abrasive nanoparticle concentration, at which point these start rolling, leading to the reduction of Ff. Figure 8 shows the variation Ff and planarization/polishing temperature with the planarization time in the CMP process, under the same conditions as those shown in Figure  7. The friction force signals are presented in Figure 8a, and those of the process temperature in Figure 8b, depending on the planarization time. The friction appears in the early stage of planarization, because the Ff is the effect of air pressure, respectively, of the relative movement between the wafer-pad when the static Ff is exceeded. Additionally, the average Ff increases with the growth of the abrasive nanoparticle concentration, changing from 18.50 N to 57.00 N. During the CMP, colloidal SiO2 is chemisorbed by the passivation selective layer, because the adhesion forces of the abrasives are usually about twice as large as the van der Waals forces due to the covalent bonds present [9]. Thus, the friction behavior and MRR in the CMP process could be influenced by the chemical absorption of SiO2 into the selective layer. The Ff and MRR could also increase as the abrasive nanoparticle concentration increases as a result of the chemisorbed active abrasives number by the selective layer. As can be seen from Figure 8b, during the CMP process the In support of these observations, in situ F f measurements showed an evolutionary trend of the total F f , similar to that of the MRR depending on the abrasive nanoparticle concentration in the slurry with the size of 50 nm (see Figure 5).
Additionally, it was proved that in situ measurements during the CMP process are correlated with the MRR response presented in ref. [7], i.e., the load per particle decreases with the increasing abrasive nanoparticle concentration, at which point these start rolling, leading to the reduction of F f . Figure 8 shows the variation F f and planarization/polishing temperature with the planarization time in the CMP process, under the same conditions as those shown in Figure 7. In support of these observations, in situ Ff measurements showed an evolutionary trend of the total Ff, similar to that of the MRR depending on the abrasive nanoparticle concentration in the slurry with the size of 50 nm (see Figure 5).
Additionally, it was proved that in situ measurements during the CMP process are correlated with the MRR response presented in ref. [7], i.e., the load per particle decreases with the increasing abrasive nanoparticle concentration, at which point these start rolling, leading to the reduction of Ff. Figure 8 shows the variation Ff and planarization/polishing temperature with the planarization time in the CMP process, under the same conditions as those shown in Figure  7. The friction force signals are presented in Figure 8a, and those of the process temperature in Figure 8b, depending on the planarization time. The friction appears in the early stage of planarization, because the Ff is the effect of air pressure, respectively, of the relative movement between the wafer-pad when the static Ff is exceeded. Additionally, the average Ff increases with the growth of the abrasive nanoparticle concentration, changing from 18.50 N to 57.00 N. During the CMP, colloidal SiO2 is chemisorbed by the passivation selective layer, because the adhesion forces of the abrasives are usually about twice as large as the van der Waals forces due to the covalent bonds present [9]. Thus, the friction behavior and MRR in the CMP process could be influenced by the chemical absorption of SiO2 into the selective layer. The Ff and MRR could also increase as the abrasive nanoparticle concentration increases as a result of the chemisorbed active abrasives number by the selective layer. As can be seen from Figure 8b, during the CMP process the The friction force signals are presented in Figure 8a, and those of the process temperature in Figure 8b, depending on the planarization time. The friction appears in the early stage of planarization, because the F f is the effect of air pressure, respectively, of the relative movement between the wafer-pad when the static F f is exceeded. Additionally, the average F f increases with the growth of the abrasive nanoparticle concentration, changing from 18.50 N to 57.00 N. During the CMP, colloidal SiO 2 is chemisorbed by the passivation selective layer, because the adhesion forces of the abrasives are usually about twice as large as the van der Waals forces due to the covalent bonds present [9]. Thus, the friction behavior and MRR in the CMP process could be influenced by the chemical absorption of SiO 2 into the selective layer. The F f and MRR could also increase as the abrasive nanoparticle concentration increases as a result of the chemisorbed active abrasives number by the selective layer. As can be seen from Figure 8b, during the CMP process the planarization temperature increased while growing the concentration of abrasive nanoparticles. This could be related to the E f , because it would hasten the realization of a protective layer (soluble passivation layer) due to the chemical dissolution. Lee et al. in ref. [28] and Kawaguchi in ref. [29] confirmed that the inhibitor leads to the soluble passivation layer (protective layer) formation, indicating the selective-layer surface characteristic zone immersed in the slurry. Additionally, they demonstrated that the soluble layer of passivation is removed through the abrasive nanoparticle mechanical action because the protective layer covering the amount by inhibitor decreased on the selective-layer surface. As a result, the number of active abrasive nanoparticles has an effect on the growth of the F f ; hence, the protective layer removal amount could also be increased.
The experimental results obtained in this paper during the CMP process were useful to examine the temperature distribution in the planarization pad. Additionally, a kinematic analysis of temperature distribution was performed to understand its growth, considering the friction characteristics. Therefore, the results of this examination can help in developing a more effective CMP process by estimating the temperature distribution in a planarization pad. Figure 9 shows MRR and E f variation for a single abrasive depending on the abrasive concentration. As shown in Figure 9a, with the increased abrasive nanoparticle concentration, the MRR per single abrasive decreases. The probability of the active abrasive nanoparticle being in contact with the wafer, even by their number growth at a high concentration or with the pad asperities, is possible during the CMP process. It is proven that the MRA by using an abrasive nanoparticle is bigger than the MRR only by the pad asperities because the planarization pad has viscoelastic a property from being made of polyurethane. During the CMP, the pressure applied on the contact surface between a single abrasive and the wafer could be reduced, but due to the active abrasives number, the contact area increases, and a higher MRA could be obtained.
particles. This could be related to the Ef, because it would hasten the realization of a protective layer (soluble passivation layer) due to the chemical dissolution. Lee et al. in ref. [28] and Kawaguchi in ref. [29] confirmed that the inhibitor leads to the soluble passivation layer (protective layer) formation, indicating the selective-layer surface characteristic zone immersed in the slurry. Additionally, they demonstrated that the soluble layer of passivation is removed through the abrasive nanoparticle mechanical action because the protective layer covering the amount by inhibitor decreased on the selective-layer surface. As a result, the number of active abrasive nanoparticles has an effect on the growth of the Ff; hence, the protective layer removal amount could also be increased.
The experimental results obtained in this paper during the CMP process were useful to examine the temperature distribution in the planarization pad. Additionally, a kinematic analysis of temperature distribution was performed to understand its growth, considering the friction characteristics. Therefore, the results of this examination can help in developing a more effective CMP process by estimating the temperature distribution in a planarization pad. Figure 9 shows MRR and Ef variation for a single abrasive depending on the abrasive concentration. As shown in Figure 9a, with the increased abrasive nanoparticle concentration, the MRR per single abrasive decreases. The probability of the active abrasive nanoparticle being in contact with the wafer, even by their number growth at a high concentration or with the pad asperities, is possible during the CMP process. It is proven that the MRA by using an abrasive nanoparticle is bigger than the MRR only by the pad asperities because the planarization pad has viscoelastic a property from being made of polyurethane. During the CMP, the pressure applied on the contact surface between a single abrasive and the wafer could be reduced, but due to the active abrasives number, the contact area increases, and a higher MRA could be obtained.
The average Ef which is used for the characterization of friction behavior is time-dependent and cannot influence the shape of the friction curve. Hence, this characterization in the CMP process would be more reasonable with a singular Ef because it depends on the planarization time and the friction curve shape. The Ef, depending on abrasive concentration, for a single abrasive is plotted in Figure 9b, where it is observed that the Ef decreases with the increasing abrasive nanoparticle concentration. Instead, the Ef increases, because on a single abrasive the applied pressure increases if the active abrasive nanoparticle number decreases.  The average E f which is used for the characterization of friction behavior is timedependent and cannot influence the shape of the friction curve. Hence, this characterization in the CMP process would be more reasonable with a singular E f because it depends on the planarization time and the friction curve shape. The E f , depending on abrasive concentration, for a single abrasive is plotted in Figure 9b, where it is observed that the E f decreases with the increasing abrasive nanoparticle concentration. Instead, the E f increases, because on a single abrasive the applied pressure increases if the active abrasive nanoparticle number decreases. Figure 10 shows the MRA versus E f in different conditions of pressures and relative speeds that generate F f after the selective layer CMP process. It can be observed that by decreasing the abrasive nanoparticle concentration and increasing the E f , it is obtains higher MRRs.  Figure 10 shows the MRA versus Ef in different conditions of pressures and relative speeds that generate Ff after the selective layer CMP process. It can be observed that by decreasing the abrasive nanoparticle concentration and increasing the Ef, it is obtains higher MRRs. When the MRA is plotted represented depending on Ef, the curves show similar trends at different concentrations (see Figure 10), except that at the low values of Ef (close to zero), MRA increased rapidly after 0.5 kJ and continued to grow relatively slower (with a slower rate). It seems that the MRA, depending on Ef, is adequate to effectively characterize each concentration value of the abrasive nanoparticles, while the MRA, depending on Ff , is useful (by the Ff monitoring) to estimate MRR during CMP. Between the calculated MRA with Equation (5) and that experimentally determined, there were no significant differences, indicating that the theoretical and experimental results are in correlation, which proves (confirms) the planarization slurry stability in the CMP process. When the roughness and deformation of the surface are minimal, the surface quality is considered to be optimal. Therefore, the optimal unfolding of the CMP process presents importance for both nanoparticle-nanoparticle and nanoparticle-substrate interactions [30]. Thus, the surfactant presence can influence the lubrication between the abrasives and the planarization surface [28,29]; as a result, the Ff can decrease, leading to the reduction of the MRR.
To this, the applied force calculation during planarization on a single abrasive was necessary [7]. This was possible by determining the contact area resulting from the applied pressure between the pad and the nanoparticle by the pad's head [16], which is consistent with the modeling from previous studies [12,18,25,28]. Hence, it is possible for the MRR to be controlled by the lubrication effect introduced by the surfactant through the friction forces between the wafer and the abrasive nanoparticles. Additionally, studies with surfactants have shown that lateral forces exhibit nonlinear behavior with respect to normal forces ( Figure 11). When the MRA is plotted represented depending on E f , the curves show similar trends at different concentrations (see Figure 10), except that at the low values of E f (close to zero), MRA increased rapidly after 0.5 kJ and continued to grow relatively slower (with a slower rate). It seems that the MRA, depending on E f , is adequate to effectively characterize each concentration value of the abrasive nanoparticles, while the MRA, depending on F f , is useful (by the F f monitoring) to estimate MRR during CMP. Between the calculated MRA with Equation (5) and that experimentally determined, there were no significant differences, indicating that the theoretical and experimental results are in correlation, which proves (confirms) the planarization slurry stability in the CMP process. When the roughness and deformation of the surface are minimal, the surface quality is considered to be optimal. Therefore, the optimal unfolding of the CMP process presents importance for both nanoparticle-nanoparticle and nanoparticle-substrate interactions [30]. Thus, the surfactant presence can influence the lubrication between the abrasives and the planarization surface [28,29]; as a result, the F f can decrease, leading to the reduction of the MRR.
To this, the applied force calculation during planarization on a single abrasive was necessary [7]. This was possible by determining the contact area resulting from the applied pressure between the pad and the nanoparticle by the pad's head [16], which is consistent with the modeling from previous studies [12,18,25,28]. Hence, it is possible for the MRR to be controlled by the lubrication effect introduced by the surfactant through the friction forces between the wafer and the abrasive nanoparticles. Additionally, studies with surfactants have shown that lateral forces exhibit nonlinear behavior with respect to normal forces ( Figure 11).
Another approach for quantifying the friction forces that appear during the CMP process resulted from the observation that the MRR values compared with the baseline slurry were lower (see Figure 11). In-depth studies by simulation with AFM on the friction in situ between single-particle and the substrate in suspension with a surfactant highlighted the response of friction force. These simulations demonstrated that the slurry chemistry, including the composition, pH, and adhesion tension of the surfactant, could be adjusted to ensure the necessary friction force for optimal planarization performance [25,30]. A certain surface-active surfactant has an optimum concentration range, which ensures that the dispersion ability results in a better surface quality. Therefore, surfactants can be used to modify the nanoparticle-substrate and the nanoparticle-nanoparticle interactions in the CMP process, s, with the goal of optimizing it based on the selected/desired performance (surfactant and the optimum concentration range). Another approach for quantifying the friction forces that appear during the CMP process resulted from the observation that the MRR values compared with the baseline slurry were lower (see Figure 11). In-depth studies by simulation with AFM on the friction in situ between single-particle and the substrate in suspension with a surfactant highlighted the response of friction force. These simulations demonstrated that the slurry chemistry, including the composition, pH, and adhesion tension of the surfactant, could be adjusted to ensure the necessary friction force for optimal planarization performance [25,30]. A certain surface-active surfactant has an optimum concentration range, which ensures that the dispersion ability results in a better surface quality. Therefore, surfactants can be used to modify the nanoparticle-substrate and the nanoparticle-nanoparticle interactions in the CMP process, s, with the goal of optimizing it based on the selected/desired performance (surfactant and the optimum concentration range).

Conclusions
The frictional and thermal features of the selective-layer CMP process were analyzed in correlation with the MRR by the monitoring of Ff depending on SiO2 concentration in an acid slurry mixed with an oxidizer, corrosion inhibitor, complexing agent, and surfactant.
The Ff can decrease by increasing the planarization pad rotating speed at a constant pressure, and the slurry boundary layer support formed between the wafer and pad. Additionally, the Ff can increase due to modification of the real contact area with increasing pressure, and WNU calculated from the balance moments equation in ratio with the planarization head center was in correlation with the COF.
The slurry SiO2 abrasive nanoparticles generate Ff (based on which the Ef is calculated/determined) and, implicitly, the CMP process temperature. The Ef generated during CMP influences the modification of the CMP process temperature, presents importance in unfolding the chemical reactions and accelerates the MRR by the surface material activating, which is to be removed.
The colloidal SiO2 abrasive during the CMP helps remove the protective layer (passivation layer) by mechanical abrasion. By the addition of abrasive nanoparticles into the CMP slurry, the MRA of the selective layer has been improving, respectively; as the active abrasive nanoparticle concentration grows, it is possible that the Ff and MRR increase due to chemisorbed nanoparticle number by the selective layer.
The MRR of the selective layer on the wafer surface into the CMP slurry, depending on Ef after planarization with SiO2 concentrations, varied and had a non-linear characteristic, i.e., a non-linear (Prestonian) behavior. This behavior is useful to improve planarization level and wafer surface non-uniformity due to the various pressure distributions.
Therefore, in the CMP process, the Ff participates in mechanical abrasion of the selective layer; instead, a large Ff can generate defects, such as scratches and etching. It is Figure 11. Friction force measurements on a selective-layer wafer in slurries with the pH8, a particle size of 50 nm, and surfactants at different concentrations, with the help of an AFM.

Conclusions
The frictional and thermal features of the selective-layer CMP process were analyzed in correlation with the MRR by the monitoring of F f depending on SiO 2 concentration in an acid slurry mixed with an oxidizer, corrosion inhibitor, complexing agent, and surfactant.
The F f can decrease by increasing the planarization pad rotating speed at a constant pressure, and the slurry boundary layer support formed between the wafer and pad. Additionally, the F f can increase due to modification of the real contact area with increasing pressure, and WNU calculated from the balance moments equation in ratio with the planarization head center was in correlation with the COF.
The slurry SiO 2 abrasive nanoparticles generate F f (based on which the E f is calculated/determined) and, implicitly, the CMP process temperature. The E f generated during CMP influences the modification of the CMP process temperature, presents importance in unfolding the chemical reactions and accelerates the MRR by the surface material activating, which is to be removed.
The colloidal SiO 2 abrasive during the CMP helps remove the protective layer (passivation layer) by mechanical abrasion. By the addition of abrasive nanoparticles into the CMP slurry, the MRA of the selective layer has been improving, respectively; as the active abrasive nanoparticle concentration grows, it is possible that the F f and MRR increase due to chemisorbed nanoparticle number by the selective layer.
The MRR of the selective layer on the wafer surface into the CMP slurry, depending on E f after planarization with SiO 2 concentrations, varied and had a non-linear characteristic, i.e., a non-linear (Prestonian) behavior. This behavior is useful to improve planarization level and wafer surface non-uniformity due to the various pressure distributions.
Therefore, in the CMP process, the F f participates in mechanical abrasion of the selective layer; instead, a large F f can generate defects, such as scratches and etching. It is very important to obtain a planarized surface of good quality and with optimal results, which is possible by controlling F f .
The oxide present on the metallic selective layer surface during CMP affects the selective-layer MRR (is responsible for MRR), respectively, and the layer oxidation rate and the oxide film solubility seem to condition the MRR of the selective layer for the slurry used.
For optimal results and the planarization/polishing performance, the slurry composition should be closely monitored during the CMP process, because the effect is significant.
The slurry's chemical composition must be completely free of any other soft or hard nanoparticles (regardless of concentration) because it would lead to inconsistent MRRs, even with the deterioration of the surface quality. Thus, slurry stabilization in extreme conditions presents importance at the solid-liquid interface, and it is conducted by using surfactants. In this way, the CMP process is improved and has beneficial (positive) effects on the surface quality (minimal deformations and roughness).
At the same time, in the CMP process, the selective layer surface chemical state has an important role in the MRR of the layer.
To verify the chemical contribution in the selective layer CMP, an empirical model for the MRA was derived from the E f equation and validated by the CMP results.
Additionally, by reducing at a minimum the surface defects, with an optimal MRR, it can attain a quality planarization. This requirement can be met in the CMP process through abrasive nanoparticle stability and properties or by slurry chemistry engineering, respectively, by the control of the mechanical and chemical interactions.