Inductively coupled plasma etching of bulk tungsten for MEMS applications
ABSTRACT
As a promising material for bulk micromachining for MEMS applications, tungsten can be used as microneedles
for medicine injection and tools for micro-inject molding, ultrasonic machining, and electrical discharge
machining. However, it is difficult to fabricate microstructures with high aspect ratios and small feature sizes due
to the excellent material of tungsten properties, such as chemical stability, high hardness, and high melting point.
Here, an inductively coupled plasma (ICP) deep etching process is developed for bulk tungsten, and its process
parameters are studied. After optimizing the process parameters, three recipes were developed for different
MEMS applications. A patterned tungsten structure with a depth of more than 300 µm was etched by tungsten
(W) ICP deep etching (WIDE) at an etch rate of 2.73 µm/min and etch selectivity of 35. The WIDE process
produced a tungsten microstructure with a high aspect ratio (> 20), a small feature size (< 3 µm), and a surface
roughness of Ra 30 nm. To demonstrate the potential application prospect of deep etching bulk of tungsten,
tungsten microneedles with a positive sidewall angle and several non-silicon substrates for MEMS applications
were achieved based on the WIDE process.
1. Introduction
Tungsten (W) has the highest melting point of all metals and an ultrahigh hardness [1,2]. It also has a high chemical inertness, good heat stability, excellent tensile strength, good conductivity, and anti-radiation qualities, which give tungsten-based MEMS excellent application prospects. Tungsten microneedles with a high aspect ratio (HAR) and diameter of less than 150 µm can efficiently and painlessly puncture the skin to extract a blood sample or inject medicine due to tungsten’s high hardness and good longitudinal elastic modulus [3,4]. Because of its high melting point and good heat stability, tungsten has been used in applications that require high operating temperatures, such as microelectrodes, launcher tips, and rocket engine nozzles [5,6]. Because of tungsten’s anti-radiation qualities, tungsten is also suited for micro-metal gratings in X-ray phase-contrast imaging systems applied in combination with CT technology to realize the early detection of cancer [7]. Its CMOS process compatibility and CTE similar to silicon allow tungsten to be used for ULSI and advanced packaging applications [8,9].
Bulk non-silicon substrates machined by tungsten mold overcome the limitation of silicon-based MEMS [10]. Micro-injection molding (µ-IM) molds the polymer substrate for bio-MEMS microfluidic devices due to its excellent heat conductivity, hardness, and tensile strength [11–13]. Micro-electrical discharge machining (µ-EDM) machines bulk metal microstructures due to its high melting points and good conductivity [14]. Micro-ultrasonic machining (µ-USM) machines hard and brittle materials with high transparency due to its high hardness [15, 16]. Micro hot-embossing (µ-HE) molds glass [17] and [18] copper substrates due to its good heat stability and hardness.
However, the micromachining of tungsten is difficult because of its hardness, chemical stability, and chemical stability. Tungsten-based MEMS applications require structures with small feature sizes, HAR, and excellent surface profiles. Traditional CNC machining methods show very high tool wear for tungsten machining and cannot be used to machine the small feature sizes required for MEMS applications [19,20]. The electrode molds of µ-EDM and micro electrochemical machining are tough to machine and experience high wear and cost [4,21–23]. The µ-wire EDM and electrochemical machining can fabricate only simple parallel structures or through-holes [19,20]. Moreover, the machining side gaps limit the minimum feature size of the tungsten microstructure. Therefore, it is challenging to machine small feature sizes and HAR bulk tungsten structures with straight sidewalls and positive angles required for MEMS applications.
Bulk metal inductively coupled plasma (ICP) etching is a MEMS process for removing substrate materials by plasma etching, which has a high resolution, small feature size, HAR, batch mode pattern machining, and excellent machining surface quality. McDonald et al. first reported the bulk metal plasma deep etching of MARIO titanium deep plasma etching, which obtained functionalities not possible with a traditional micromechanical material system [24]. Then, the influences of different ICP parameters, gas combinations, and masks on the bulk titanium ICP etching were researched [10,25,26]. Bulk molybdenum ICP etching in our previous research has been reported with a SU8 mask and chlorine-based gas etching [27–30]. However, due to its excellent chemical stability, tungsten is difficult to etch, especially at a high etch rate and large depth. Tungsten film dry etching has been researched as a ULSI interconnect material for the past 40 years [31,32]. Still, the thin film etching mechanism is different from that of deep bulk etching because of thin-film properties, surface effects, and substrate electrical properties. Moreover, the thin film etch rate and etch depth are much lower than the requirements of MEMS applications.
Here, a bulk tungsten (W) ICP deep etch (WIDE) process is proposed and developed. The machining and characterized parameter were systematically studied by variation process parameters, including the chamber pressure, ICP power, RF power, substrate temperature, etching gas flow, and different gas combinations. Process parameters were optimized based on study results to achieve a HAR bulk tungsten microstructure with straight sidewalls and small feature sizes etched by the WIDE process at a high etch rate (ER) and etch selectivity (ES). Several tungsten-based MEMS applications were also achieved to prove the application prospects of the WIDE process.
2. Experimental
2.1. Raw materials and experimental devices
A size of 1 × 1 cm, 500 µm thick bulk tungsten sample with patterned aluminum (Al) hard mask was customized from Hicomp Co., Ltd. An 8 µm-thick Al film was deposited on the polished surface (< Ra 10 nm) of the tungsten sample by a magnetron sputtering system and then patterned by dry etching. Al achieved a high ES for tungsten etching as a hard mask material because of aluminum fluoride’s high boiling point (1291 ◦C) [33]. This allowed it to achieve an extremely low ER in fluorine-based gases. The sample was attached to a 4-inch Al wafer substrate with perfluoropolyether oil and then loaded into the plasma etcher for the WIDE process. An SI 500 ICP plasma etcher (Sentech, Berlin, Germany) was used to etch the bulk tungsten samples. The plasma is excited through an antenna evenly distributed on the top surface of the reactor chamber by a 13.56 MHz frequency that can reach 2500 W. A negative bias is applied to the substrate by a 13.56 MHz frequency generator connecting the substrate holder, which up to 600 W and accelerates the plasma vertically to the substrate surface. Process gases (SF6, C4F8, O2, and CF4) are injected through a showerhead at the sidewall top of the chamber, and flow valves control their flow rates. A throttle valve connecting the exhaust pump controls the chamber pressure from 0.1 Pa to 8 Pa. The substrate temperature is regulated with a temperature range from − 30 to 250 ◦C by a thermocouple, helium backside flow, and coolant system.
2.2. Methods and characterization
A single-factor experimental method was used to investigate the influence of process parameters on the WIDE process. Only a single process parameter was varied at a time, and all other parameters were held constant at the following values: chamber pressure of 0.6 Pa, ICP power of 1000 W, RF power of 150 W, substrate temperature of 150 ◦C, and SF6 flow rate of 80 sccm. The etching time was 10 min. To characterize the WIDE process, the machining parameters showing the machining ability and characterization parameters showing machining quality were evaluated. The machining parameters included ER and ES, and the characterization parameters included depthundercut ratio (DUR), sidewall straightness, sidewall angle, and surface roughness. The parameters are shown in Fig. 1a, which is a schematic of typical ICP deep-etching structural parameters. ER is the speed of the etching process, which is necessary to achieve deep etching to ensure the deepest possible etching within limited time and costs. ER is
given by:
ER = h/t (1)
where h is the etching depth of bulk tungsten, measured by a DektakXT surface step profiler (Bruker, MA, USA) after removing the Al film hard mask by hydrochloric acid (HCl), and t is the etching time. ES is the ratio of ER between the bulk tungsten and mask. The thickness of mask material determines the maximum etching depth and influences the etching resolution. ES is given by
ES = h/thk − (H − h) (2)
where thk is the original thickness of the Al film hard mask, and H is the total depth of the tungsten etched structure with the remaining Al hard mask. Compared with the Bosch Process, which is a cyclic process that includes a deposition step to protect the sidewalls, ICP deep etching without a deposition step produces isotropic etching because of the chemical etching by radicals. The isotropic etching lead to undercut, sidewall un-straightness and sidewall angle [34,35]. Less undercut at a larger depth can improve the feature size resolution and aspect ratio of deep etching. The DUR is the ratio of depth and single-side undercut of
WIDE and is given by:
DUR = h/1/2 ∗ (W − w) (3)
where W is the original width of the patterned Al film mask, and w is the top side surface width of the tungsten structure measured by an optical microscope. Sidewall straightness and angle determine the sidewall gap and minimal feature size. A straight sidewall and positive angle (< 90◦) reduce the demolding strength of μ-IM and µ-HE. Moreover, the sharp tip by the positive sidewall angle helps microneedles pierce the skin. Sidewall straightness with a positive or right angle (≤ 90◦) is an ideal sidewall profile for most MEMS applications. The low surface roughness without a grass-like structure can improve the etching surface quality. The sidewall straightness and sidewall angle of the WIDE process were characterized by a JSM-6390 SEM (JEOL, Tokyo, Japan). Surface roughness is measured by a WYKO NT100 optical profilometer (Veeco, New York, USA).
3. Results and discussion
3.1. WIDE process mechanism
The chemical reaction between radicals and the material to be etched is fundamental to ICP deep etching. The chemical reaction removes the material, which generates volatile and non-volatile etch product at a substrate temperature. The volatile etch product can be removed and promote the etch process, while non-volatile etch product deposit and passivate the surface, thus inhibiting etching. Reducing the substrate temperature can enhance the passivation, helping to control the etching profile. In addition, ICP etching includes physical etching through ion bombardment and a mixture of physical and chemical etching at the substrate surface [36]. The physical bombardment of ions induced by RF power assists the bottom passivation removing to improve the etching profile, surface roughness, and chemical etching. However, physical etching is non-selective and increases the etching rate of the mask,which decreases the ES.
[photo]
Fig. 1. (a) Schematic of typical ICP deep etching structure parameters. (b) Schematic of bulk tungsten etching by WIDE process.
[photo]
Fig. 2. The effects of process parameters on ER, ES, DUR, and surface roughness: (a) chamber pressure, (d) ICP power, (g) RF power, and (k) substrate temperature. The profile of tungsten etching structure under different process parameters: chamber pressure of (b) 0.6 Pa and (c) 1.5 Pa, ICP power of (e) 500 W and (c) 1200 W, RF power of (h) 100 W, (i) 225 W, and (j) 450 W, substrate temperature of (l) 25 ◦C, (m) 125 ◦C, and (n) 200 ◦C.
Tungsten-based-MEMS applications require a HAR tungsten structure with a straight sidewall profile and good surface quality, which is etched at a high ER and ES. A schematic of bulk tungsten ICP etching is shown in Fig. 1b. High ES and ES and good surface quality are determined by the highly efficient chemical reactions that require more radicals and more efficient etch product removal. More radicals can be generated by using a higher molecule density and ionization, which requires a higher pressure and ICP power. However, excessively high pressure decreases the distance of molecules to make it impossible for electrons to gain enough energy to ionize the gas, which inhibits glow discharge and molecular ionization. When the ICP power is higher than a specific power, the increase of molecular ionization is limited, which also makes the etching process unstable. The gas flow rate only slightly affects the number of radicals in a single-gas etching system because the pressure determines the number of molecules. SF6 was chosen as the tungsten etching gas because a higher concentration of fluorine atoms can be produced in SF6 plasma, and tungsten fluoride is easily volatilized and removed due to its lower boiling point (17.5 ◦C) [33]. A higher substrate temperature promotes passivation removal, which exposes the surface to be etched.
A HAR structure with a straight sidewall profile was achieved by anisotropic etching with a low undercut, which required excellent sidewall passivation protection and etching directionality. A lower temperature can lead to more etch product deposition and passivate the sidewall. The etch product deposition improves the anisotropic etching but inhibits chemical reactions and reduces the ER and surface quality. Physical bombardment induced by a higher RF power can assist in the vertical removal of the passivation layer on the surface to promote anisotropic etching. However, if the RF is too high, ES will decrease significantly.
The WIDE process is a complex and multi-parameter process. Studying the effects of different process parameters on the WIDE process is of great significance for developing a process to etch bulk tungsten structures for MEMS applications.
3.2. Influence of process parameters
Fig. 2 is the effects of process parameters (chamber pressure, ICP power, RF power, and substrate temperature) on the WIDE process. The chamber pressure and ICP power strongly influenced the WIDE process, and as the pressure increased, ER, ES, and DUR increased significantly, but roughness decreased (Fig. 2a, d). More ions and radicals were generated as the number of gas molecules increased induced by the pressure increased and as more atomic fluorine ionized or dissociated by ICP power increased, which all led to more chemical etching. In this regime, chemical etching increased more than physical etching because more fluorine radicals participated in the chemical reaction, but bombardment did not increase significantly due to the constant RF power [36]. Tungsten ER increased significantly, but the Al mask ER by physical bombardment was almost unchanged, which increased ES. Even though the chemical reaction increased undercut, more chemical reactions in the vertical direction decreased its influence on DUR. However, even if the chemical etching showed directionality under assistance by physical bombardment, there was more chemical isotropic etching of the sidewalls, which made the sidewall profile non-straight (Fig. 2b, c, e, f). An excessively fast chemical etching prevented the etch product from being immediately removed and deposited on the etching surface, leading to a low-quality surface. When the pressure and ICP power were above 1.0 Pa and 1500 W, respectively, the radicals’ generation was limited by each other constant value, which makes the ER, ES, and DUR not continuously increase. The chamber pressure had a more pronounced effect than the ICP power because a higher molecular concentration increased the random collisions between particles, which generated more plasma.
ER increased slightly upon increasing the RF power (Fig. 2g). Even though tungsten etching was more dependent on chemical etching, physical bombardment enhanced the physical etching and assisted in removing the etch product deposited on the surface, which improved the chemical etching. The increased physical bombardment also improved the etching surface roughness by removing the deposited etch product, as shown in Fig. 2. ES significantly decreases with increasing the RF power because physical etching controlled by RF power increases the ER of Al mask. A higher RF power improved the sidewall verticality by improving the etching anisotropy. DUR also increased with the RF power because a higher RF power improved the etching anisotropy. However, the high-energy physical bombardment when using too high of an RF power (450 W, Fig. 2j) removed the edges of the Al mask, which led to etching the edge of tungsten under the removal mask. The edge removal also decreased the DUR when the RF power was above 150 W. A higher temperature continuously improved the ER because it increased the molecule energy of chemical reactions between tungsten and fluorine radicals, which enhanced chemical etching. A higher temperature also encouraged the complete volatilization of the etch product to prevent their surface deposition and from influencing the surface quality and subsequent chemical reactions. This improved the ER and surface quality of the etched surface (Fig. 2k)and sidewalls (Fig. 2l–n). However, a higher temperature also led to the isotropic etching having a greater influence on the sidewall straightness.
In the single-gas etching system, the most critical influence process parameters of the chemical etching plasma density are the chamber pressure and ICP power. The gas flow rate less influences the WIDE process and is to supply enough gas in the chamber and assist the etch product exhaust.
3.3. Gas mixture etching mode
The bulk deep ICP etching process was combined with chemical etching and passivation deposition. Effective passivation of the sidewall helps reduce the minimum feature size, control the sidewall angle, and improve the structure’s aspect ratio and etch resolution. The deposits not removed quickly enough can inhibit or even stop the chemical etching reaction, which causes passivation deposition. In the single SF6 gas etching system, the main deposit is fluoride tungsten. However, the WIDE process aims to remove tungsten by the continuous generation of the fluoride tungsten, so it is essential to remove fluoride tungsten effectively. Therefore, the passivation performance of the sidewall is poor in the SF6 single etching gas system. In this section, the auxiliary gases of O2 and C4F8 were used to study the gas mixture mode of bulk tungsten deep etching to improve the sidewall passivation protection and anisotropic etching.
3.3.1. O2 ratio
Identical to SiOF4 in Si ICP etching process, WOF4 with a higher boiling point (187.5 ◦C) [33] can be generated by adding O2 gas to the WIDE process, which passivates the sidewalls and reduces isotropic etching. Deposits on the bottom surface can be removed by the vertical physical bombardment introduced by the RF power to ensure continuous etching. The schematic of this process is shown in Fig. 3a. The addition of auxiliary gases increases the passivation and also dilutes the concentration of original SF6 gas under a constant chamber pressure. Fig. 3b shows the effect of different percentages of O2 gas on ER, ES, DUR, and roughness. The total flow rate of SF6 and O2 was 80 sccm, and the other process parameters were not changed. As the O2 percentage increased, DUR increased significantly, but ES decreased. The chemical etching decreased due to a decrease in the concentration of SF6 gas as the percentage of O2 increased. Increasing the oxygen plasma concentration improved the etching anisotropy and passivation of sidewalls (Figs. 2b & 3c) and enhanced the vertical physical bombardment. This decreased ES due to the etching of the Al film mask. ER reached its peak at 20 % O2 and then decreased significantly. The peak ER with 20 % O2 is a comprehensive effect of complex physical and chemical processes on increasing ER. First, the addition of O2 enhanced the physical bombardment to help remove the passivation layer on the surface, thus improving the chemical etching and surface quality. Then, the physical etching due to vertical bombardment also enhanced the ER increase. Third, oxygen ion bombardment of the surface increased the molecular activity of substances to promote chemical reactions. Last, the combination of oxygen plasma with SFx+ prevented the recombination of SFx+ with fluorine radicals, thus increasing the density of fluorine radicals and enhancing the chemical etching. However, upon decreasing the SF6 concentration, there were not enough fluorine radicals to support chemical etching, which decreased ER. Excessive deposits also reduced the surface quality (Fig. 3b, d). Even though adding O2 increased the sidewall passivation and improved anisotropic etching and undercut (Fig. 3e), it was still apparent and limited the formation of HAR structures.
[photo]
Fig. 3. (a) Schematic of anisotropic etching and sidewall passivation by SF6 and O2 gas by WIDE process. (b) The effects of different O2 gas percentages from 0 % to 60 % on ER, ES, DUR, and surface roughness. (c) The profile of the etched tungsten structure under 20 % O2 gas. The sidewall profile (d) and undercut (e) of the etched tungsten structure under 50 % O2 gas.
3.3.2. Mixture of SF6, C4F8, and O2
As an important passivation gas, C4F8 can generate macromolecular polymer deposits to perform excellent sidewall protection during deep etching process, which has been used in the deposition step of the silicon Bosch etching process [37] and W film RIE etching [38,39]. C4F8 molecule will be dissociated and ionized by ICP power to form CFx molecules that can polymerize and form nCF2 long-chain macromole
cules, which can deposit on the etching surface and passivate the sidewall. However, the excellent passivation layer is hard to remove, which significantly inhibits vertical chemical etching, leading to a very low ER. The vertical physical bombardment of heavy molecules also increases the ER of the Al mask and reduces ES.
O2 is added to improve the ER and ES of SF6 and C4F8 mixture etching. O2 can promote the chemical etching of SF6, and its physical bombardment helps remove C4F8 deposits, thereby increasing ER. The addition of O2 also improves ES by decreasing the percentage of C4F8 to reduce physical bombardment on the Al mask. Fig. 4a shows a schematic of sidewall passivation after adding C4F8 and O2.
The different combinations of flow rates of SF6, O2, and C4F8 are shown in Table 1, and the other process parameters remained unchanged.
Fig. 4b shows the effects of C4F8 and O2 gas addition with different ratios on the ER, ES, and surface roughness, in which the data of DUR was not evaluated because of the extremely low undercut due to C4F8 passivation (DUR > 35). The percentage of C4F8 significantly affected ER and ES (Group 1# & 2#), but a low ratio of C4F8 did not protect the sidewall effectively, as shown in Fig. 4c–f.
Adding O2 under a constant ratio of C4F8 and SF6 (Group 3#) greatly reduced ER and ES. Reducing the SF6 concentration decreased the fluorine radicals available for chemical etching. Physical bombardment by O2 enhanced the Al mask ER, which decreased ES. Moreover, the oxide passivation deposition and original deposition of C4F8 led to over-passivation, resulting in grass-like structure formation on the etching bottom and surface roughness increasing (Fig. 4g, h).
O2 was added by reducing the percentage of C4F8, while the SF6 concentration remained unchanged (Group 4#). The SF6 and O2 ratio (4:1) was the same as the ratio of the highest ER of SF6 and O2 mixture etching mode, which promoted chemical etching. Reducing the concentration of C4F8 decreased the macromolecule deposition on the surface and physical bombardment of the Al mask, which improved the ER and ES. The sidewall passivation protection by oxide deposition compensated for the effect of a reduction in C4F8 deposition. The etching profile was excellent without grass-like structure generation, as shown in Fig. 4i, j.
[photo]
Fig. 4. (a) Schematic of anisotropic etching and passivation of the sidewall by using SF6, O2 and C4F8 gases during the WIDE process. (b) The effects of C4F8 and O2 gas addition with different ratios on the ER, ES, and surface roughness. The profile and undercut of the etched tungsten structure under different ratios of SF6 / O2 / C4F8: (c) & (d) 40/ 0/ 40, (e) & (f) 40/ 0/ 20, (g) & (h) 35/ 10/ 35, and (i) & (j) 40/ 10/30.
[excel]
[excel]
3.4. Optimized recipes
Based on the above results, three WIDE recipes were developed and optimized to satisfy different MEMS applications, which are shown in Table 2. Recipe 1 improved the concentration of radicals by increasing the chamber pressure and ICP power, which significantly improved the chemical etching to achieve a large etching depth and a high ER. Meanwhile, reducing the RF power improved ES. A patterned tungsten structure more than 300 µm deep was etched at an ER of 2.73 µm/min, which is shown in Fig. 5a, b.
To obtain a HAR tungsten structure, recipe 2 was optimized from the SF6, C4F8, and O2 mixture mode results because the mode achieved full passivation, which inhibited isotropic etching. ER and ES were improved by increasing the chamber pressure and ICP to improve chemical etching and enhance the sidewall passivation at a lower substrate temperature. An aspect ratio of 20 and feature size of less than 3 µm tungsten structure with vertical sidewall angle was achieved by recipe 2 at an ER of 0.72 µm/min and ES of 13. (Fig. 5c, d) Increasing the physical bombardment by increasing the RF power decreased the etching surface roughness to below Ra 30 nm.
Even though the excellent sidewall angle assists in the demolding process in μ-IM and µ-HE, the vertical sidewall also enhances the demolding pressure, which influences the molding microstructures and mold lifetime [11,17,18]. A positive sidewall angle was obtained by etching with recipe 3, which was optimized by reducing the ratio of O2 and C4F8 to increase the chemical etching and decrease the sidewall passivation. The RF power decreased to reduce the Al mask ER. The positive sidewall angle promotes machining tungsten microneedles with a sharply angled tip to extract blood samples and inject medicine (Fig. 5e). The tungsten mold with a positive sidewall angle was etched at an ER of 1.08 µm/min and an ES of more than 20, as shown in Fig. 5f.
[photo]
Fig. 5. Tungsten microstructures etched at an ER of 2.73 µm/min: (a) 300 µm depth patterned tungsten structure, (b) Bulk tungsten gear. HAR tungsten microstructures with etching surface roughness of < Ra 30 nm: (c) A reverse word ’by’ tungsten microstructure with HAR of > 20 and feature size of < 3 µm, (d) A high aspect ratio tungsten gear with a slight tundercut. Tungsten microstructure with positive sidewall angle: (e) Tungsten microneedle, (f) Bulk tungsten mold for μ-IM and µ-HE.
3.5. Applications of non-silicon micromachined by bulk tungsten mold
Based on the optimized recipes of the WIDE process, bulk tungsten molds were etched to machine the non-silicon substrate for MEMS applications, as shown in Fig. 6. For μ-EDM and μ-USM subtractive processes, the etching depth of the mold is the most important to ensure the removal of chips and does not influence the polished surface. During the batch mode μ-EDM machining, electric discharges are fired between the patterned electrode and the workpiece, which can wear away both of them with the higher wear ratio of the workpiece [40]. A die steel micro-structure with a depth of 41.2 µm and surface roughness of Ra 190 nm was machined by a patterned tungsten electrode with a large etching height, as shown in Fig. 6c. Due to the high melting point and low conductivity of tungsten electrodes, the electrode wear rate and machining rate are 37 % and 0.48 µm/min. μ-USM process machines the workpiece by the vibration of the tool imparts momentum to the abrasive particles suspended in the slurry which bombard the workpiece to remove its surface material [41]. Tungsten is an excellent tool material with high hardness, ensuring low tool wear as the machining occurs predominantly at the brittle workpiece surface. Fig. 6d shows borosilicate glass microstructures with a depth of 56 µm, surface roughness of Ra 220 nm, and minimal feature size of 11 µm machined by μ-USM using a patterned tungsten cutting tool at a machining rate of 11.2 µm/min and mold wear rate of 2.34 %. In the μ-IM process, the melting polymer is injected into a microstructured mold insert, and then it is cooled and the part demolded to form the polymer microstructures [42]. For the molding machining process a mold with excellent sidewall and bottom surface quality is essential, which can influence the workpiece machining surface quality. Due to the hysteresis effect of high aspect ratio microstructure in injection molding, lowering the temperature of the mold can cause a condensation layer to form on the surface of the mold, which cannot be effectively filled, [43] and there is an obvious curve angle [44]. By optimization of the mold temperature in the μ-IM process, polymer spherical structures with a diameter of 31 µm are molded using a microholes tungsten mold with a positive angle at a lower mold temperature, as shown in Fig. 6e. A higher mold temperature is applied on a tungsten mold with a high aspect ratio slit to mold a honeycomb-like superhydrophobic polymer structure with a minimum size of 12 µm and an aspect ratio of 5:1, as shown in Fig. 6f.
[photo]
Fig. 6. (a) Bulk tungsten mold etched by the WIDE process. (b) Schematic of non-silicon substrate machined by bulk tungsten mold. (c) A die steel micro-structure with a depth of 41.2 µm and surface roughness of Ra 190 nm machined by μ-EDM using a patterned tungsten electrode at an electrode loss rate of 37 % and machining rate of 0.48 µm/min. (d) Borosilicate glass microstructures with a depth of 56 µm, surface roughness of Ra 220 nm and minimal feature size of 11 µm machined by μ-USM using a patterned tungsten cutting tool at a machining rate of 11.2 µm/min and mold loss rate of 2.34 %. (e) Polymer spherical structures with a diameter of 31 µm molded by μ-IM using a microholes tungsten mold with a positive angle sidewall. (f) A honeycomb-like superhydrophobic polymer structure with a minimum size of 12 µm molded by μ-IM using a high aspect ratio slit tungsten mold.
4. Conclusion
Here, a deep ICP etching process for bulk tungsten is presented and characterized systematically. The process parameters (chamber pressure, ICP power, RF power, substrate temperature, and SF6 flow rate)
were researched to determine their influence on the WIDE process. A higher chamber pressure, ICP power, and substrate temperature promoted chemical reactions during the WIDE process, which enhanced ER, ES, and surface quality but led to isotropic sidewall etching. Increasing the RF power inhibited isotropic etching but decreased the ES due to the physical etching of the hard mask. A higher chamber pressure, ICP power, and moderate substrate temperature and RF power promoted the further development of the WIDE process. O2 and C4F8 were added to enhance the passivation of the sidewall to achieve a HAR structure with a straight sidewalls. A 20 % ratio of O2 addition effectively improved the ER, ES, and sidewall protection of the WIDE process. Adding C4F8 to SF6 and O2 at a specific ratio achieved excellent sidewall passivation, which inhibited undercut and sidewall un-straightness.
Based on the research results, the optimized process recipes were achieved to improve the WIDE application prospect. A HAR (> 20) tungsten microstructure with an excellent sidewall (vertical, straight,
and smooth) and bottom surface (Ra 30 nm) was etched by the WIDE process. The tungsten with HAR microstructure was used to machine a micro-groove (HR >5) on borosilicate glass by μ-USM. Microneedle arrays with positive sidewall angles were also achieved. The tungsten mold with a positive sidewall angle was applied to the μ-IM process for polymer micro molding. A deep (> 300 µm) patterned tungsten was etched at a high ER (2.73 µm/min) and ES (35) and then used to micromachine die steel by μ-EDM. The WIDE process provides further potential for realizing novel tungsten-based MEMS devices and nonsilicon-based MEMS applications.
CRediT authorship contribution statement
Yanming Xia: Conceptualization, Methodology, Validation, Formal analysis, Investigation, Resources, Data curation, Writing, Visualization.
Zetian Wang: Investigation, Resources, Data curation. Lu Song: Conceptualization, Methodology. Wei Wang: Conceptualization,
Methodology. Jing Chen: Conceptualization, Writing, Supervision, Project administration, Funding acquisition. Shenglin Ma: Methodology, Writing, Supervision, Project administration, Funding acquisition.
Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Data Availability
Data will be made available on request.
Acknowledgment
This work was supported by the National Key Research and Development Program of China under Grant No. 2019YFB2204900.