Creating three-dimensional magnetic functional microdevices via molding-integrated direct laser writing

Magnetically driven wireless miniature devices have become promising recently in healthcare, information technology, and many other fields. However, they lack advanced fabrication methods to go down to micrometer length scales with heterogeneous functional materials, complex three-dimensional (3D) geometries, and 3D programmable magnetization profiles. To fill this gap, we propose a molding-integrated direct laser writing-based microfabrication approach in this study and showcase its advanced enabling capabilities with various proof-of-concept functional microdevice prototypes. Unique motions and functionalities, such as metachronal coordinated motion, fluid mixing, function reprogramming, geometrical reconfiguring, multiple degrees-of-freedom rotation, and wireless stiffness tuning are exemplary demonstrations of the versatility of this fabrication method. Such facile fabrication strategy can be applied toward building next-generation smart microsystems in healthcare, robotics, metamaterials, microfluidics, and programmable matter.


Supplementary Notes Solution to contamination introduced by multi-step molding
The molded structures via multi-step molding may suffer from contamination, when the geometry has a large surface area exposed at the photoresist surface, which can be covered by following molded materials. But for deep structures with small exposed surface area, such as the microcilia array, it is not a major problem because the covering material is much thinner than the underneath structures thus the effect is negligible. If the designed molding structure is a thin layer with large surface area, we suggest using a slightly thicker photoresist layer and print the structure underneath the photoresist surface. To make the structure accessible to the developer, extra tunnels need to be printed to connect the structure and the photoresist surface.
Another solution is to spin-coat a thin layer of photoresist (few microns) after the curing process of each molding step to seal the molded structure and protect it from future contamination.

Potential of reprogramming individual µM-bits
There is another reason why we choose to use CrO2 besides its low coercivity. Previous work [1][2][3] has shown that CrO2 has a low Curie temperature at 398 K and a large wide-spectrum photoabsorbance. These two physical properties of CrO2 make it easy to be demagnetized and remagnetized upon light illumination. Using this ferromagnetic material equips the micromachines with potentials to be selectively reprogrammed via localized light illumination.
Our preliminary results validate the feasibility of localized reprogram of the m directions at a micrometer scale. We fabricated microcilia with a 3 µm-thickness and 100 µm-length using the molding method and filled with CrO2-PDMS. We magnetized all cilia along one direction.
And we scanned some of the microcilia with the same laser we use for 2PP 3D lithography while applying a small (20 mT) realigning B field in the opposite direction. We observed opposite motion directions from these microcilia under magnetic fields. However, more technical issues need to be resolved, such as laser transmission for thicker structures, integrating an automatically-changing realigning B field with the laser scanning process, among others.
Step 3: Fill the mold with magnetized 1:1 NdFeB: Ecoflex 00-30 mixture, then align the particles along the desired direction using a uniform external B field. Cure the elastomeric mixture on a 50 ℃-hotplate for 15 minutes.
Step 4: Repeat step 2 and step 3 three more times to mold all four phases magnetic microcilia.
Step 5: Pour a layer of pristine Ecoflex 00-30 on the surface of the photoresist and cure it for 15 minutes at 50 ℃. This elastomeric layer will function as a soft substrate for the microcilia

Fabrication steps of double-layer µM-bits (Supplementary Figure 7)
Step 1: Print the supporting bases for the bottom layer rotor rings (diameter 80 µm, height 50 µm) using IP-S.
Step 3: Dissolve the photoresist layer with acetone after the elastomeric composite is cured. Step 4: Prepare a layer of 300 µm thick AZ-IPS 6090 photoresist. Repeat the casting photoresist, vacuuming (1 minute), spin-coating (600 rpm for 10 seconds), soft baking (80 ℃ for 3 minutes then ramp up to 110 ℃ for 6 minutes) process for two times to achieve a layer thickness of 150 µm. Expose the top rotor ring structures (power: 20%, speed: 13 mm s -1 ) and obtain the molding cavities after PEB (100 ℃ for 100 seconds) and developing (4 minutes). Fill the mold with unmagnetized NdFeB-Ecoflex and cure it for 15 minutes under 50 ℃.
Step 5: Magnetize all the rotor rings in the wafer plane with a 1.8 T-magnetic field.
Step 6: Dissolve the photoresist in acetone. Print the top layer mechanical structures using IP-S.
The phase between the bottom and top layers can be realized by changing the angle between the stopper and magnetization. After developing in IPA for 10 minutes, put the sample in an ultrasonic bath for 30 seconds to release both supporting bases and thus enable the stoppers to rotate freely.
For the fabrication of three-layer µM-bits, repeat the photoresist preparation process in step 4 for two times to prepare a 600 µm thick AZ-IPS 6090 photoresist layer, followed by the same procedure to stack the third layer on the second layer. Supplementary Fig. 7 Fabrication steps of the double-layer µM-bits. Blue arrows indicate the applied magnetizing B field direction. Supplementary Fig. 8 SEM images of three-layer µM-bits and a two-layer lattice structure. a, Three-layer µM-bits. Each layer is composed of nine µM-bits, and each µM-bit has a magnetic ring. b, Two-layer magnetic lattice structure. Each layer has 144 unit cells (12×12).