CVD grown Graphene Microlms as a Promising Microscaled Solid Lubricant for the Lubrication of Silicon MEMS Devices

Friction, wear, stiction, adhesion, and absence of suitable lubrication methods are important challenges, severely restricting and limiting the expeditious development of Microelectromechanical system (MEMS) technology. This paper aims to explore the potential of chemical vapor deposition (CVD) grown graphene microlms for the lubrication of sliding silicon MEMS devices to reduce friction and wear problems. A novel silicon-based pin-on-disk friction-pair is designed to mimic sliding MEMS working conditions. Pure graphene-based microlms are fabricated on the Cu substrate via a CVD method and transferred to the silicon substrate via the PMMA transfer method. To investigate microlms' surface quality and morphology, microlms are characterized via Raman spectroscopy, AFM, and SEM. For the tribological performance evaluation, different tribological tests were conducted using the microtribometer. Results show that microlms remarkably reduced the friction coecient and wear in the MEMS devices; however, microlms' tribological performance depends on the roughness and the number of lms on the specimen. This remarkable tribological performance suggests that graphene microlms have the potential to increase the reliability and wear lifespan of MEMS devices. It is foreseeable that this lubrication method can be a step towards the expeditious industrial development of silicon MEMS devices.

for lubrication of MEMS thrust pad bearing and achieved low coe cients of friction at high speeds [7].
Several types of coatings such as diamond-like carbon coatings [8], amorphous carbon thin-lms [9], uorinated carbon lms [10], Langmuir-Blodgett lms [11], self-assembled monolayers [12] are in practice to solve the micro-scaled tribological problems where liquid lubricants cannot be employed due to precision requirement of the small parts of the MEMS devices. Thus, coatings of two-dimensional (2D) materials such as graphene, molybdenum disul de (MoS2), and h-BN are receiving widespread attention as a potential solid lubricant because of their remarkable mechanical and tribological properties [13].
Among these 2D materials, graphene is preferred in tribological applications due to its thermal stability, low friction coe cient, high wear resistance, weak van der Waals forces, and controllable thickness [14,15]. The layered graphene structure makes it an excellent solid lubricant, which can be used in nano or microsystems with oscillating, rotating, and sliding contacts to reduce static friction, friction, and wear [16,17]. Liang, Bu [18] deposited graphene oxide lm on a silicon specimen via electrophoretic deposition (EPD) and showed that lms have high stiffness, good antiwear, and antifriction qualities. Kim, Lee [19] had grown graphene lms on the Cu and Ni and transferred them onto the Si substrate, and exhibited graphene lms effectively reduced the adhesion and friction forces. Liu, Li [20] tested the coatings of uorinated graphene on steel contacts, achieved low friction and wear in different environments, and shown low wear and friction results.
This research aims to develop the graphene micro lms on the silicon MEMS devices via the CVD method and investigate their tribological performances. Surface quality and morphology, and structure of micro lms were characterized via Raman spectroscopy and AFM. Different tribological tests were carried out on a microtribometer to understand the graphene lms' possible lubrication mechanism. After tribological tests, the surface morphology of worn specimens was carried out by SEM. The focus is on the effect of the different numbers of graphene lms on MEMS devices' friction and wear. The excellent tribological performance indicates that CVD-grown graphene micro lms can be a potential solid lubricant for MEMS devices.
2 Experimental Design

Materials and Reagents
A single crystalline polished silicon wafer of 2 inches (thickness 500 µm) was used to fabricate MEMS devices' sliding parts. Ultrapure water and deionized water were used to wash the specimens in the Ultrasonic bath. The list of reagents and their usage is shown in Table 1. For the development of a sliding MEMS device, a pin-on-disk friction-pair was fabricated by loading a stationary upper specimen (a cylindrical silicon pin) of 2 mm against a reciprocating lower specimen (a rectangular silicon disk) of 15⊆10 mm. Figure 1. It shows the SEM photo of the upper specimen. A safety margin of 0.5 mm wide is kept to prevent the surface pattern's structure from being damaged during laser cutting. As a result, the actual diameter of the upper specimen is 3 mm. The pin's surface pattern is 100 µm wide, the spacing of each pattern is 100 µm, and the depth is 50 µm, and it was fabricated via DRIE.

Preparation of graphene-based micro lms
For the chemical vapor deposition (CVD) growth of graphene micro lms, copper (Cu) and nickel substrates are mostly preferred catalysts. Since this research aims to study the tribological properties of graphene micro lms grown by CVD on silicon, a Cu substrate has been chosen to support the lm mechanically. A Cu foil was cut into the same dimensions as the silicon wafer and was cleaned in acetone and deionized water to remove the impurities. Before starting the CVD process, the substrate was annealed at 900°C, and then graphene lms were developed by the CVD process details that can be seen anywhere else [21][22][23][24].
In the next step, the prepared lms were transferred onto the silicon substrates by the poly (methyl methacrylate) (PMMA) method, described in the literature [25] using the following steps.
(i) For transfer, a sacri cial lm of PMMA solution was sprayed on the graphene-coated Cu substrate. (ii) PMMA lm was dried at 120 ℃ for 5 minutes in an oven so that PMMA lm can adequately adhere to the Cu substrate. (iii) Two Petri dishes were lled with FeCl 3 solution; corners of the sample were held in the solution with tweezers to etch away the graphene on the edges of the Cu foil, the sample was rinsed to clean the impurities of etchants. (iv) Then, to remove the Cu foil, the sample was placed in the FeCl 3 solution for 2 h, and the copper substrate was removed. (v) As a result, a strip of PMMA was left behind with graphene grown by CVD on one side. (vi) The sample was taken out from the FeCl 3 solution, and the sample was placed in a petri dish lled with deionized water to remove the residual impurities of the FeCl 3 solution. (vii) Afterward, samples were removed from the deionized water and were dried at about 110 ºC for about 1 hour. (viii) For the removal of the PMMA layer, the sample was placed in the acetone solution. This process is repeated three times every time the new acetone solution was poured, and the sample was placed for 30 minutes. (ix) After that, the sample was taken out from the acetone solution, and then acetone was blown out in normal air. Then the sample was placed in an oven for about 30 minutes at 90 ℃. Figure 3 shows the real-time photos of the nal products of graphene micro lm on silicon substrates.

Microtribological Characterization
After preparing micro lms on silicon samples, the surface and structural evaluation of micro lms Raman spectra were acquired for the original graphene micro lms by Raman spectroscopy (JY-HR800) at an argon − iron laser wavelength of 514 nm as the excitation source. AFM was used to study the surface morphology and quality of samples at a scanning range of 2×2 µ m, and the scanning rate was 1.01 Hz. After tribological tests, the surface morphology of worn specimens was carried out by SEM to study the lubrication mechanism and antiwear performance of the graphene micro lms.

Tribological testing
A pin-on-disk microtribometer investigated the graphene micro lms' microscale tribological behavior under different tribological parameters at room temperature and in the normal ambient environment. The prepared silicon-based pin was used as a xed upper counterpart. A graphene-coated lower specimen disk was mounted on the tribometer's reciprocating drive, which slides reciprocally at a distance of 6 mm.
Before the tribological test, all the prepared samples were cleaned in an acetone solution and dried before use. Three sets of the tribological test were carried out to investigate the effect of the different number of micro lms, loads, and reciprocating speeds. After the tribological testing, specimens were taken down from the tribometer, placed in the ultrasonic cleaner for 10 mins to remove the specimens' wear impurities. Later, deionized water and 99.9% absolute were used to clean the specimen and dried it with hot air. Then, to investigate lubrication and antiwear performance of graphene micro lms, the worn samples' wear scar morphologies were characterized through SEM.

Characterization of Micro lms
For the microstructural evaluation of original graphene micro lms, Raman spectra were performed before the friction tests. For the original graphene micro lms, the peak appears at 1270-1450 cm − 1 , caused by the lattice vibration leaving the center of the Brillouin region and characterizing the defects or edges of graphene samples. The D peak appears at 1270 cm-1, the G peak, caused by the in-plane vibration of the carbon atom, appears near 1580 cm-1. The 2D peak appears near to 2700 cm-1, which is the secondorder Raman peak of two phonon resonance. The intensity ratio of D and G peaks (ID/IG) is typically used to estimate graphene defect sites' concentration. The ID/IG ratio of the original graphene coating is 0.8. Figure 4 shows the obtained Raman spectra of the sample with one graphene lm. It can be seen from the Raman characterization results of the sample that it is consistent with the typical graphene Raman spectra, which proves that the graphene lm has been successfully prepared on the surface of monocrystalline silicon, and the graphene lm has no defects.
AFM was used to characterize the surface morphology and quality of graphene micro lms and bare silicon surfaces. Figure 5 shows the AFM images of the samples with and without graphene micro lms. Figure 5 (a) shows the surface morphology of the sample without graphene lms. It can be seen that the surface of the sample is uneven and has roughness. However, every roughness peak's overall height distribution is relatively uniform without signi cant variations, showing the silicon specimen's processing quality is according to the experimental requirements. Figure 5 (b) shows the AFM images of the sample with one graphene micro lm. It is worth noting that the sample's surface quality has been signi cantly improved. The overall roughness is reduced, and the surface looks much smoother than the sample's surface without the graphene lm. Figure 5 (c) shows the AFM images of the sample with 2 micro lms of graphene. It can be seen that the surface quality is better than both samples. Topography photographs of graphene-coated samples reveal that the graphene was reasonably smooth. However, some ne PMMA residue and pollutants were observed on the surface of graphene lms. Corresponding to the silicon surface, which has higher roughness peaks, the overall surface quality signi cantly improved with graphene lms' deposition.

Tribological Performance
For the tribological performance evaluation of graphene micro lms, three different tribological experiments were carried out under different tribological conditions such as different loads, sliding speeds, and different numbers of graphene lms. (
The experiment's reciprocating speed is 5 mm/s, the applied load is 10 N, and the experimental time is 10 min. The measured friction coe cients of all four samples with different micro lms are compared in Fig. 6. It can be seen that the sample with 0 lms exhibits the highest friction coe cient, and the curve has high uctuations. The high friction coe cient can be assumed: the silicon surface has high roughness, contributing to a high friction coe cient. At the same time, uctuations in the curves are because of debris produced during the running-in process.
On the other hand, samples with the graphene lms had shown a remarkable reduction in friction coe cient. They achieved friction coe cients of 0.39, 0.31, and 0.25 for each sample, respectively, which suggests that graphene is most likely the reason for the friction reduction. It is perceived that the friction coe cient was decreasing with an increase in the number of lms. Because the friction coe cient of sample 4 was the lowest than all other samples, and the friction coe cient of sample 1 was the highest. This reduction is because the number of lms had increased the load-bearing capacity and reduced the samples' roughness.
Our results show a trend of friction reduction with the increase in the number of lms, and the friction coe cient increase with an increase in the surface's roughness. This shows that graphene lms can be used to improve the friction performance of the friction-pair. (

2) Effect of Different Speeds
For this tribological testing, a sample with two graphene lms was selected, installed on the tribometer, and tribological tests were carried out at different reciprocating speeds of 5, 10, 15, and 20 mm/s. Figure 7 presents the coe cients of friction at different speed conditions. The friction coe cient for a 5 mm/s friction curve starts with high values due to surface roughness. After small-time, contaminants once all roughness on the friction track were grounded, it became stable and nally reached a value of about 0.35. However, this friction coe cient is the highest among all other curves. This is concluded that this reciprocating speed was much lower than other speeds and was not enough to properly ground all the roughness in the surfaces, which leads to the higher friction coe cient.
Further, for 10 mm/s, the friction curve starts at 0.2, increases, and becomes stable at 0.3, and remained constant with no uctuations in the curve. The friction coe cient for 15 mm/s friction curve starts at 0.25, becomes stable at the same value, and has the lowest uctuations. Conversely, the friction coe cient for 20 mm/s friction curve starts at 0.01 and becomes stable at 0.24.
This friction reduction can be attributed to the high speed, which had a smooth surface roughness of the graphene lms quickly and helped gain the lowest friction coe cient. However, at low speeds, it had taken time to smooth the graphene lm's roughness and was unable to achieve sound reduction as it has gained at high speeds. Our results showed a higher friction coe cient at lower speeds and a lower friction coe cient at higher speeds. (

3) Effect of Different Applied Loads
To study the effect of different applied loads on a graphene lm, testing was carried out at 10, 20, and 30 N, while other conditions were kept the same. Figure 8 shows the friction coe cient curves under different loads. It can be seen that the smaller loads of graphene lms have a good reduction in friction coe cient, while at higher loads, it has a higher friction coe cient. When the load applied was 10 N, the friction coe cient remains very low and steady at about 0.3 during the reciprocating time. Higher loads of 20 N and 30 N friction coe cient were much higher, and it was about 0.35 and 0.41, respectively.
These results indicate that the graphene lms had given better protection to the specimen and increased the specimen's load-bearing capacity. However, this protection seems dependent on the applied load because when a load was small, it had the lowest friction coe cient, and when the load was high, the graphene lm was worn out or quickly removed out of the wear track. Its bene cial effect is lost. Films were damaged and led to a high friction coe cient.
Therefore, it is concluded that the graphene lms are vulnerable to higher loads because at higher loads after initial run-in time, the protective graphene layer is perhaps removed out of the wear track. The sliding surfaces experienced more and more metal-to-metal contact and higher adhesion levels and hence achieved higher friction. The most extended durability or the graphene lm's lifetime was achieved under the lowest load 10 N. However. It is stated clearly that this experiment was performed on the three-layered sample. It is maybe possible that when the sample has more lms, it could show better friction of coe cient. Our results have shown a lower friction coe cient at lower applied loads and a higher friction coe cient at higher applied loads.

Wear Performance
After the tribological testing, specimens were taken down from the tribometer, placed in the ultrasonic cleaner for 10 mins to remove the samples' wear impurities. Later, deionized water and 99.9% absolute were used to clean the specimen and dried it with hot air. The surface morphology of the specimens was analyzed with SEM, and images at different magni cations were taken. Figure 9 shows the SEM images of rubbed specimens. Figure 9 (a) shows that the specimen's SEM images without graphene lms can be seen.
The specimen had deep furrows and wear marks. Wear track is highlighted with a box on the image. This can be attributed to the presence of micro convex bodies and the surface roughness of the specimen.
These bodies caused the peeling of material from the abrasive marks and resulted in higher and deep wear marks. Figure 9 (b). shows the SEM images of rubbed silicon devices with one graphene lm. It can be seen that with the employment of graphene lms, the wear conditions are improved. Wear marks are still there, but their depth and quantity are much less than the silicon sample without graphene lm. This standing of wear is possibly due to the upper specimen's reciprocating during testing, carries the wear particles on the wear track, and results in the specimen's abrasion marks. This is assumed because the graphene lm's employment had increased the load-bearing capacity by increasing the contact between the specimen and helped friction-pair to bear more load. As a result, abrasion marks wear conditions were reduced than the specimen's wear conditions without graphene lms. Figure 9 (c). shows the SEM images of rubbed silicon devices with two graphene lms. It can be observed that wear condition is improved with the fabrication of two lms of graphene on the silicon because the thickness of graphene lms helped friction pair to bear more load and increased its load-bearing capacity. As a result, abrasion marks and wear conditions were reduced than the specimen's wear conditions without graphene lms. Figure 9 (d). shows the SEM images of rubbed silicon devices with three graphene lms. It can be observed that wear condition is improved with the fabrication of three lms of graphene on the silicon devices because the thickness of graphene lms helped friction pair to bear more load and increased its load-bearing capacity. Very slight and light wear marks can be seen in the highlighted box. It is concluded that graphene lms had a good load-bearing capacity and can improve the silicon devices' wear.
It is concluded that graphene has good antiwear properties; however, this wear performance is dependent on the number of lms, and the roughness in the lms, roughness in the lms is also a cause to increase the wear. Because of that, a sample with 1 and 2 lms had also experienced little wear because it is shown in the samples' AFM images; they have surface roughness. On the whole, it has been concluded that graphene lms have excellent wear reduction qualities, and they can be an excellent solid lubricant for the small-scaled silicon-based devices with moving contacts.

Conclusions
In summary, pure graphene-based micro lms were fabricated on the Cu substrate via a CVD technique, and developed lms were transferred on the silicon substrate by the PMMA transfer method. It is intended that fabricated graphene micro lms can be a potential lubricant for the lubrication of sliding silicon MEMS devices. After developing lms on silicon specimens, tribological testing was carried out by a commercial microtribometer, and the surface morphology of the worn samples was characterized with SEM. Results show that the graphene-based micro lms had remarkably reduced the friction coe cient and wear. However, micro lms' tribological performance depends on the roughness and the number of micro lms on the specimen. Therefore, it can be concluded that graphene micro lms developed via the CVD method have the potential to reduce friction and prolong the wear life of silicon MEMS and can be an effective lubricant for lubricating silicon MEMS devices.