3D Printing versus LIGA for CMOS Integrated MEMS

Most micromachining methods on silicon wafers are subtractive, meaning material is patterned and etched away. LIGA and 3D printing are the exception, in which the micromachined structures are “grown” using various printing or deposition methods. LIGA, a German acronym for Lithographie, Galvanoformung, Abformung, is a MEMS technology developed heavily in the 1990s (1,2) and is still being used for creating new devices and processes (3). Both 3D printing and LIGA are additive manufacturing (AM) methods. Both AM techniques have been integrated into or onto electronic substrates like printed circuit boards (PCBs) using 2D ink jet (4,5) and 3D printing (6,7), as well as LIGA structures on CMOS wafers (2). This article will compare these additive manufacturing methods with regard to MEMS/CMOS wafer integration.

3D Printed Circuits

3D printers for electronics applications are leveraging years of work in the 1980s on the 2D metal and dielectric film ink jet printing onto hybrid ceramic circuits (4) and porcelain coated steel substrates (5). Thermal decomposition of metallo-organic and inorganic compounds was used to product metal films. Conductive and insulating adhesives and nanopartilces have also been printed to form PCBs. More recently, Nanodimension ( https://www.nano-di.com/ ) has fabricated 3D printed circuit boards with integrated passive components. The push for smaller 3D printed structures has led to a variety of laser (8) and selective electroplating (9-14) print methods of fine metal features with micro to nanometer dimensional cross-sections, as shown in Figure 1. Northeastern University developed methods for printing adhesive conductive and dielectric layers onto substrates (15,16). This method of micro & nanostructure manufacturing uses double oxide deposition and etching (DODE) process. The DODE approach is also called nanoscale offset printing and uses a reusable template enables the assembly of nanomaterials into a sheet of structures followed by their transfer onto a flexible substrate or even wafer.  

Exaddon ( https://www.exaddon.com/ ) developed a fine metal 3D printer based on an atomic force microscope (AFM) using hollow cantilevers for local liquid dispensing of an electroplatable solution (10-12). A nanofluidic channel in the cantilever allows soluble molecules to be dispensed through a submicrometer aperture in the AFM tip. The sensitive AFM force feedback allows for the controlled approach of the tip to a sample for extremely local metal electroprinting on surfaces in liquid environments. Only electroplatable metals like gold and copper can be printed with this method, and it certainly could be used to plate onto the metal bond pads of a PCB or wafer.

Fraunhofer has developed a method of printing very fine metal linewidths using laser melt methods (8), see the top right of Figure 1. Entire metal MEMS (MicroElectroMechanical System) substrates and discrete sensors have been 3D printed using direct metal laser powder sintering (DMLS) as shown in Figure 2 (17,18). Titanium, plastic and stainless steel micromachined wafers have been fabricated with this method  https://3dprintwafer.com/.  Silicon and glass are prone to cracking so due to thermal shock concerns,  laser melt printing may not be a good match for silicon wafer integration. Wafer to wafer bonding may be an approach that could work. However, when the material properties of wafers do not match, the wafer stack can shear apart at the bond interface, during thermal cycling (19).  While 3D printed metal wafers could be bonded to a silicon wafers, 3D printing via electroplating or adhesive conductive nanoparticles has a faster integration path for on-wafer MEMS-CMOS devices. 

AM using LIGA

LIGA micromachining is based on electroplating metal in a high aspect ratios photoresist mold. The LIGA gear on the bottom left of the introductory figure was made using an X-ray synchrotron exposure source (1) for the exposure of the photoresist and so can product vertical structures that are 100’s of micron thick. The LIGA on CMOS structures from prior studies (2,20-22) used a conventional wafer fab photo lithography tool with resist thicknesses of around 20 microns. This technology was the first CMOS integrated AM process. Figure 3 shows the process flow for a plated metal LIGA micromachine on top of a CMOS wafer. LIGA MEMS tunneling accelerometers and gyroscopes (23,24), fuel nozzles and capacitive gyroscopes (2,20,25), were developed with this electroplated AM technology in the 1990’s.

LIGA vs 3D Printing

What are the differences between LIGA and 3D printing? 3D printing and LIGA are subsets within AM. There are many different methods of 3D printing, laser, e-beam, plating. LIGA uses a photoresist 2D pattern to grown a metal micromachine vertically. A sacrificial layer, see Figure 3, is used to fabricate an overhang on a silicon wafer. Only a single overhang can be printed with LIGA in this manner. 3D printing can produce multiple overhangs of almost any angle. Fabrication complexity is another factor when comparing LIGA and 3D printing. To fabricate a plated microstructure on a CMOS wafer, like that shown in Figure 3 and 4, with the LIGA process, takes 40 to 60 individual wafer steps. These processing step include wafer clean, resist coating, exposure, develop, plating etc. The major process steps are illustrated on the right of Figure 3, with an SEM of a released nickel ring resonator shown on the left side of Figure 3 and 4.  3D printing of a single free-standing metal structures is much simpler and, in most cases, faster. LIGA has been used to simultaneously manufacture hundreds of devices on a single 125mm CMOS wafer, as shown in Figure 4, back in the 1990’s and 2000’s. If LIGA were to be used today on 300mm CMOS wafers, it could produce tens of thousands of devices on a single wafer and a quarter of a million chips in a 25 wafer lot. The minimum LIGA linewidth in Fig. 3 and 4 back in the 1990’s, was about 8 microns, although 4-5 microns test structures were made. In 2020 minimum CMOS linewidths are down below 10 nanometers on wafers, so a metal linewidth for a LIGA structure could be in the 100’s of nanometers with plating solution issues being the limiting factor. In contrast, most 3D printing methods are still producing a single structure at a time, although some printing techniques have incorporated multiple lasers to speed up the print rate per part, and Ti MEMS wafer have been printed with multiple chips per wafer (18). Multiple PCBs can also be printed in sheet form (15,16).  

LIGA Lessons Learned

The new 3D printed micro and nano printers and processes based on electroprinting, can learn from the more than decade push to commercialize metal electroplated LIGA micromachines and sensors. These earlier devices were introduced to the automotive and aerospace market place and as such had to undergo extensive reliability testing and iterative improvements (21,22). Electroplated structures made of gold and pure nickel (23,24) were found to be “soft” compared to silicon and prone to creep, positional change over time for cantilevers as well as during shock and drop testing. Both comb and ring resonator made of pure nickel were limited in the amount of drive amplitude that could be employed (23-25). High drive or amplitude of motion over long time periods resulted in changes with time due to work hardening or cyclic fatigue and this affect was accelerated at higher operational temperatures for pure nickel. In some cases, the work hardening change in the nickel spring constant could be recovered after annealing at temperatures of 55 to 120?C (24). As has been noted in other studies of nickel plating (26), alloying could increase the hardness and Q of the plated resonator structures which reduced the work hardening and drop sensitivity of the LIGA integrated CMOS gyros (24).

To protect the LIGA on CMOS microstructures from physical damage and particle contamination, wafer to wafer bonding was used to provide a chip scale package for these MEMS devices (20). Some of the wafer bonding methods that had worked with silicon devices, like glass reflow and anodic bonding, were found to be incompatible with electroplated structures. The 400 to 450?C wafer bonding processes resulted in grain growth and warpage of the plated nickel cantilevers. A lower temperature, < 300?C, solder wafer bond process (20) had to be developed for the LIGA on CMOS products. The wafer bonded stack and a few MEMS chips are shown on the right of Figure 4. 

Another problem with nickel and nickel alloy sensors was that the ferromagnetic nature of nickel resulted in sensor output shifts in the presence of external magnet fields. This could be found near speakers and electric motors in a vehicle.  Figure 5 shows that a special Mu metal shield had to designed into the LIGA sensor package (21) to prevent low frequency magnetic fields from degrading sensor performance. Mu-metal is a nickel–iron soft ferromagnetic alloy with very high permeability often used for shielding sensitive electronic from magnetic fields. 

Conclusions

Both LIGA and 3D printing offer approaches to micromachine additive manufacturing on top of CMOS wafers. 3D printing is certainly faster than more traditional wafer processing, but has more difficulty in higher volume MEMS manufacturing when printing on a silicon wafer. Just because products can be made with a certain technology doesn’t mean it will be commercially successful. This applies to Additive Manufacturing just like other new technologies. For inertial sensors metal LIGA technology lost out to silicon and polysilicon in the early 2000’s. Cost, manufacturing infrastructure, yield and design/process integration will sort out which competing technology win’s out in the market. We will see this with AM across many products and applications. New applications and AM technologies offers opportunities for smart MEMS/CMOS integration in the future.

To learn more about 3D printing of sensors and MEMS devices attend Sensors Expo and the preconference symposiums on June 22, 2020 in San Jose.  I will be talking about 3D printing and sensors at Symposium #4.

https://sensorsexpoconference2020.sched.com/event/Zluv

References

1. Guckel, “High-aspect-ratio micromachining via deep X-ray lithography,” Proc. IEEE, Vol.86, (8), 1586–1593, (1998).

2. Chang, Sparks et al., “An electroformed CMOS integrated angular rate sensor,” Sensors and Actuators, A66, pp.138-143, (1998).

3. Malekabadi and Paoloni, “UV-LIGA microfabrication process for sub-terahertz waveguides utilizing multiple layered SU-8 photoresist,” Journal of Micromechanics and Microengineering, Vol 26, No 9 (2016)

4. Vest, Tweedle, Buchanan, “Ink jet printing of hybrid circuits,” International Journal of Microcircuits and Electronic Packaging 6(1), pp. 261-267 · January (1983).

5. Sparks and Vest, "Copper films from aqueous solutions of copper nitrate trihydrate," Thin Solid Films, vol.200, p.77, (1991).

6. Macdonald, et al. "3D printing for the rapid prototyping of structural electronics." IEEE, 13 Mar. 2014, pp. 234-242., doi:10.1109/ACCESS.2014.2311810, (2014).

7. Kiesel, et al. "Practical 3D printing of antennas and RF electronics." dtic.mil/docs/citations/AD1041830 Mar. (2017).

8. Vialva, “Fraunhofer IMM scientists sharpen lasers for nanoscale metal 3D printed structures,” https://3dprintingindustry.com/news/fraunhofer-imm-scientists-sharpen-lasers-for-nanoscale-metal-3d-printed-structures-153718/ April (2019).

9. Grüter et al. “FluidFM as a lithography tool in liquid: Spatially controlled deposition of fluorescent nanoparticles.” Nanoscale 5, pp. 1097-1104, doi: 10.1039/c2nr33214k (2012).

10. Hirt et al. “Local surface modification via confined electrochemical deposition with FluidFM.” RSC Adv. 5, pp. 84517-84522, doi: 10.1039/C5RA07239E (2015).

11. Hirt et al. “Template-free 3D microprinting of metals using a force-controlled nanopipette for layer-by-layer electrodeposition,” Advanced Materials, volume 28, issue 12, pp. 2311–5, (2016).

12. Ercolano et al., “Additive manufacturing of sub-micron to sub-mm metal structures with hollow AFM cantilevers,” Micromachines, 11(1),p. 6,  https://doi.org/10.3390/mi11010006 (2020). https://www.exaddon.com/

13. Perkins, New process allows 3-D printing of nanoscale metal structures,” Caltech, https://www.caltech.edu/news/new-process-allows-3-d-printing-nanoscale-metal-structures-81373 Feb (2018).

14. Vyatskikh, et al., “Additive manufacturing of 3D nano-architected metals,” Nature Communications, 9, No. 593. ISSN pp.2041-1723. (2018).

15. Cho, et al., “High–rate nanoscale offset printing process using directed assembly and transfer of nanomaterials,”. Advanced Materials, 27, pp. 1759–1766, (2015).

16. Yilmaz,et al., “High-rate assembly of nanomaterials on insulating surfaces using electro-fluidic directed assembly,” ACS Nano, 11 (8), pp 7679–7689, (2017).

17. Sparks, “Using 3D metal printing for flow and pressure sensors,” Flow Control, Vol. 24, No.2, pp. 23-25, Feb. (2019).

18. Sparks, “The advantages of using additive micromanufacturing in the fabrication of MEMS wafers and sensors,” Commercial Micro Manufacturing, Vol. 12, No.6, pp. 20-26, Dec. (2019). https://3dprintwafer.com/

19. Sparks, “Metal-based wafer Level and 3D-printed packaging,” Chip-Scale Review, Vol. 22, No.4, pp.14-17, Jul-Aug, (2018).

20. Sparks, et al., “Chip-scale packaging of a gyroscope using wafer bonding,” Sensors and Materials, Vol. 11, p.197-207, (1999).

21. Sparks, Chia, Zarabadi, “The reliability of resonant micromachined sensors and actuators,” Proceedings of the Spring SAE Conf., No. 2001-01-0618, SAE, (2001).

22. Sparks, Chia, Jiang, “Cyclic Fatigue and Creep of Electroformed Micromachines,” Sensors and Actuators A, 95 (1), pp.61-68, (2001).

23. Kubena, Atkinson, Robinson, Stratton, “A new high‐performance surface‐micromachined tunneling accelerometer fabricated using nanolithography,” J. Vac. Sci. Technol. B 14, p.4029 (1996).

24. Kubena et al, “New miniaturized tunneling-based gyro for inertial measurement applications,” J. Vac. Sci. Technol. B, 17, p.2948 (1999).

25. Bernstein, et al., “A micromachined comb-drive tuning fork rate gyroscope,” MEMS, IEEE, p.143, (1993).

26. Atanassovy and Schils, “Deposition of Ni-based alloys with addition of manganese and sulfur,” Plating & Surface Finishing, Vol. 49, July (1996).