technologies (e.g. chemical muco-adhesives and mechanical anchoring devices) require complicated systems and/or have a large form factor to provide effective preloading and control, limiting their ability to be integrated into ingestible devices [4]- [6]. Recently, the development of bio-inspired technology has provided insight for passive tissue anchoring using barbed MNs [7]- [10]. The barbed MN structures provide the advantage of a small tissue penetration force with a higher pull-out force, resulting in reduced preloading and robust tissue anchoring performance. However, their performances are limited by the resolution of conventional microfabrication methods and materials with the appropriate stiffness to achieve a high pull-out force to penetration force ratio (PPR).

Previously, we designed, fabricated, and demonstrated a parasite-inspired spiny MN (SMN) [11], closely mimicking the tissue anchoring proboscis of spiny headed worms [12]. Benefiting from high-fidelity direct laser writing (DLW) fabrication, the MN precisely replicates the biomimetic model with high-resolution features (ex. fine surface barb with tip diameter around 1 µm). The results demonstrate robust tissue anchoring performances on porcine small intestine tissues with a 25-mN pull-out force and more than an order of magnitude lower penetration force (1.6mN) [11], providing a simple and robust tissue-anchoring component with a PPR of over 10.

This work presents a passive, hybrid tissue-anchoring mechanism via incorporating micro-springs with barbed MNs within a compact structure, leveraging the state-of-the-art 3-D DLW technology to fabricate the hybrid structure on flexible substrates (Fig. 1). Specifically, each MN is built on top of a micro-spring that is compressed and constrained by a layer of dissolvable polymer. The spring-microneedle unit (SMU) can be actuated via the dissolution of polymer in aqueous media to release the MN for tissue anchoring.

Here, we successfully demonstrate the DLW fabrication of the SMU on a flexible polyimide surface and the compression of the SMU within dissolvable polyethylene glycol/PEG. We also demonstrate the efficacy of the polymer dissolution with DI water at body temperature (37 • C) showing 140 µm of MN displacement within 103 mins. The SMU demonstrates good mechanical performance with a stable compression profile and an adequate release force capable of effective tissue anchoring. Additionally, we further investigate SMNs via characterizing the effect of Young's modulus and the surface microhook arrangement. The results provide an array of options for optimizing the tissue-anchoring performance to assist the future design of SMNs (e.g. tuning the preloading actuation force and tissue-anchoring strength). Above all, Fig. 1.

Images of hybrid tissue-anchoring apparatus with (A) a 1 × 3 spring-microneedle unit (SMU) array wrapping around a pill-sized dummy capsule (d: 9.5 mm, w: 5 mm, scale bars: 2 mm), the zoomed-in image shows that each SMU consists of a barbed MN on top of a conical microspring. (B) Illustration of the tissue-anchoring mechanism with a compressed SMU (left) and released SMU (right) penetrating within the GI tissue, activated via PEG dissolution (4,000Mw, Sigma-Aldrich). The released SMU height (975 µm) is taller than the capsule trench (750 µm) so that the MN will be able to interact with the tissue when released. the hybrid, compact, and passive mechanism provides great opportunities to address the challenges of deploying/anchoring ingestible resident devices in the GI tract.

Fig. 1A shows a conceptual ingestible resident capsule integrated with a 1 × 3 SMU array compactly assembled on the surface trench (750 µm depth) of the capsule package using a flexible polyimide substrate (Kapton tape). Each MN is 260 µm tall with 48 barbs, allowing the robust tissueanchoring performance demonstrated in the previous work. The backward facing barbs near the tip of the MN play an important role in enabling the low tissue penetration force and more than 10-fold higher pull-out force [11].

The conical micro-spring design allows reduced solid height for a compact design as each active coil is partially recessed within the next larger coil. The overall volume of the SMU is only about 5% of the overall capsule package volume. The conical design also provides lateral actuation stability as the base coils have larger diameters (250 µm to 100 µm) with less tendency to buckle than conventional compression springs. The design contains 4 spring coils with an 80-µm wire diameter, allowing stable directional actuation with an estimated spring constant of 340 N/m and 100 mN peak compression force at 300 µm displacement. These calculations are based on the conical spring model [12], as well as the measured Young's modulus and shear strength of the IP-S photoresist [14], [15]. The design parameters and calculations of the conical spring are listed in Table I.

The passive tissue-anchoring mechanism is depicted in Fig. 1B. Before actuation, the SMU array is compressed in solid PEG within the trench of the capsule (the overall SMU height is about 975 µm). After PEG dissolution in a 37 • C aqueous environment (similar to the GI environment), the SMU array is actuated to anchor the MN tip into the GI tissue.

The SMU displays high-fidelity features with up to 1-µm resolution (Fig. 2). The conical spring demonstrates uniform spring wire diameter measured of 80 µm and an evenly distributed pitch of adjacent coils (Fig. 2A and Fig. 2B). The barbs on the MN surface are backward facing with sharp tips (Fig. 2C), which is essential for tissue anchoring [11]. The overall height of an uncompressed SMU is about 975 µm.

The SMU is fabricated utilizing a high-precision DLW technology (Photonic Professional GT, Nanoscribe GmbH, Germany), which is based on highly localized interactions between a femtosecond laser and photosensitive material. First, the 3D model of the SMU is created via SolidWorks (Dassault System, France) computer-aided design (CAD) software. The CAD file, in the format of stereolithography (STL), is then imported and processed in the DeScribe software for fabrication coding. The design is then printed using the Dip-in Laser Lithography (DiLL) objective (25X magnification, NA = 0.8) with a negative-tone IP-S photoresist (Nanoscribe GmbH, Authorized licensed use limited to: University of Maryland College Park. Downloaded on June 14,2020 at 13:06:24 UTC from IEEE Xplore. Restrictions apply. Germany, suitable for mesoscale prototyping compared to other photoresist and used by [10]) on Kapton tape on ITO-coated glass. The coded file is loaded into the DLW software, Nanowrite, for fabrication under the DLW settings of 50 mW laser power and 100 mm/s scan speed (galvo mode) to achieve 1-µm lateral slicing layer height, 1-µm hatching distance, and 26-minute fabrication time. After the laser writing, the sample is immersed in propylene glycol monomethyl ether acetate (PGMEA) to develop for 5 minutes, followed by 2-min of isopropyl alcohol (IPA) cleaning.

Fig. 3 shows the bottom-up DLW fabrication process of a single SMU on a Kapton tape covered ITO coated glass substrate. The base and conical spring are printed first, followed by the barbed MN structure. The transparency of Kapton tape allows the system to automatically locate the refractive interface on the ITO substrate surface. Therefore, the laser interaction is confined within the Kapton tape during the first 25 µm of printing and does not cure the photoresist. As a result, the as-fabricated SMU has a base thickness of about 25 µm instead of the designed value of 50 µm (Fig. 2D). This phenomenon of laser interaction can be observed in Fig. 3A (no photoresist is cured) and later Fig. 3B (the SMU base fabrication). Due to the limitation of the 300-µm by 300-µm lateral laser working area, the SMU is split into 5 parts to complete, with 2-µm crosslinking stitching overlapping. At t = 16 min, Fig. 3C shows that the microspring (250-µm base diameter, 80-µm wire diameter) sits within the central building block so that it is free from the risk of low connection strength at the overlapping region. At t = 25 min (Fig. 3D), the barb tips are first created without any support due to the bottom-up fabrication required to accommodate the microspring fabrication, which is notably different from the DLW process demonstrated previously for barbed MNs [11]. PDMS well and hold for 2 min so that the PEG will solidify and hold the compressed SMU in place (Fig. 4B). The SMU compression (225 µm) equals the height difference of the original SMU (975 mm) and the PDMS well (750 µm).

Fig. 5 displays the gradual dissolution of PEG upon rinsing with 37 • C water (body temperature) to verify the passive SMU actuation mechanism. Initially, the SMU is embedded in PEG after assembly (Fig. 5A). As body temperature (38 • C) water droplets (∼0.06 µL) are gradually added onto the assembly, the SMU tip slowly appears with a measured height from the base (including the substrate and the SMU) of 1.47 mm (h 1 ) and 1.48 mm (h 2 ) at t = 13 min (Fig. 5B) and t = 16 min (Fig. 5C), respectively. At t = 25 min, the entire MN appears with a measured height of 1.56 mm (h 3 ) (Fig. 5D). At t = 72 min, the majority of the SMU appears with some PEG and trapped within the space of the microspring with a measured height of 1.61 mm (h 4 ) (Fig. 5E). At t = 103 min, the majority of the SMU appears and the final height remains at 1.61 mm (h 5 ) (Fig. 5F). Because the initial displacement at t = 0 cannot be directly measured (the microneedle is not visible when fully embedded in the PEG), the estimated minimum displacement recovery is 140 µm, determined via subtracting the displacement between h 5 and h 1 . This measured actuation displacement is large enough for the MN to interact with GI tissue, confirming the autonomous actuation capability. Note that the dissolution time for the maximum actuation can be tuned by altering the molecular weight of the selected PEG (the higher molecular weight, the slower dissolution) [15]. Future work includes demonstrating a faster dissolving mechanism to allow MNs to penetrate the tissue before the capsule to be pushed away. Additionally, the design of the polymer coating allows the future design of targeted actuation of the MNs based on the media pH profile capable of targeting specific GI locations (ex. the stomach has low pH while the small intestines have neutral pH) [16].

The fabricated conical micro-spring requires mechanical characterization to verify stability and compression failure. Fig. 6 presents the typical compression and release profile of the SMU showing that the spring can withstand a 300-µm compression displacement and fully recover after release. There was no spring coil damage observed during the test. Also, the SMU remains adhered to the polyimide substrate during the test. The results indicate that the design characteristics are appropriate for the SMU actuation.

The SMU is also quantitively tested using a universal testing machine (Model 5565, Instron, MA, USA) with tensile and compressive modes containing a +/-50 N load cell and 3D-printed fixture assembled to the upper sensing column to compress the SMU samples. During the test, an SMU is placed under the movable sensing column of the tester. After balancing the load cell, the upper movable column moves towards and pushes the SMU with its fixture at the speed of 0.05 mm/s until the compression displacement reaches 300 µm. After the compression, the upper column moves upwards to release the SMU at the same speed. Throughout the test, the upper sensing column records time, displacement, and force measurement. This process is video recorded using a stereomicroscope (Leica M125, Germany) and a digital camera (Sony α6000, Japan).

Fig. 7A presents a typical compression and release force measurement profile of an SMU, showing that the compressional force increases as the displacement increases. Before releasing, the compression force decreases from 59.3 mN to 32.7 mN, indicating stress relaxation. The difference between the compression and release (hysteresis) indicates that the SMU spring has a considerably large energy dissipation, possibly due to the internal friction of the building block material (IP-S, Nanoscribe Gmbh, Germany). The hold time of the compression of the SMUs are about 30s. Future work needs to address the total relaxation of the SMU to demonstrate long-term actuation performance.

Fig. 7B shows that the average compression force increases up to 59 mN over a 300-µm displacement, showing an average spring constant of 197 mN/mm. The lower calibration values compared to the analytical estimations are attributed to the variation in the mechanical property of the cured photoresist (IP-S, Nanoscribe Gmbh, Germany). The releasing force, measured after stress relaxation, demonstrates 8 mN at 100-µm SMU actuation relative to the maximally compressed state (equivalent to a 200-µm displacement from its position before loading), ultimately indicating sufficient mechanical rigidity to withstand the 1.6mN MN penetration force [11] and the 8µN peristaltic force [18]". Additionally, cyclical mechanical tests of both single SMU and 1 × 3 SMU array are presented in Table II actuation demonstrated a large enough actuation force for robust tissue anchoring (>2 mN per MN). Notably, the 1 × 3 SMU array presents a much higher actuation force, possibly due to secondary parasitic motions of individual SMUs (e.g. lateral collapse) that provide interactive supports to each other, demonstrating the advantage of an SMU array for future capsule integration.

Our previous study of the bio-inspired, tissue-anchoring MNs demonstrated exceptionally low tissue penetration forces while maintaining high pull-out forces [11], with an over 10-fold enhancement in pull-out/penetration performance compared to the state-of-the-art [7]- [9]. Here, we further investigated the photoresist material properties and the spiny microhook patterns to guide the future design of the SMU to adjust the actuation force and tissue-anchoring strength.

Fig. 8 presents the effect of presence of the microhooks, needle material, hook base diameter, and microhook pattern density on tissue anchoring performance by comparing the pristine SMN design to barbless microneedles, microneedles printed with a softer photoresist material, microneedles equipped with larger microhooks, and microneedles with fewer microhooks.

The microneedle with no surface microhooks (Type I) relies solely on the frictional force between the needle surface and the surrounding tissue in this anchoring process, and as a result, there is no significant difference between the penetration and pull-out forces with both measuring low force levels. SMNs fabricated with softer photoresist (Type II) provide less effective anchoring (>4-fold lower pull-out force) compared to the pristine SMN (Type III), indicating the importance of material stiffness (Young's modulus: 1 GPa vs. 4.6 GPa, [14]) in achieving robust mechanical interlocking between the microhooks and the intestinal tissue. These findings are consistent with the evolutionary success of the spiny headed worm's effective parasitic tissue anchoring structure (sclerotized surface hooks). Additionally, SMNs with a doubled hook base diameter (Type IV), showed only a slight increase in pull-out force while the penetration force showed a 2-fold increase compared to the pristine SMN. Note that the hook tip diameter is the same as the pristine SMN (1 µm) and the needle trunk dimensions remains the same. This suggests that increasing the hook size brings marginal improvement to the strength of tissue anchoring while the increased contact area of the larger microhooks leads to greater penetration resistance, ultimately resulting in a lower PPR. On the other hand, SMNs with a reduced number of microhooks (Type V, larger microhook axial spacing) demonstrated slightly decreased penetration force (0.3 mN), similar pullout force (30 mN), and even higher PPR (∼100). Overall, these investigations determined that a smaller microhook base diameter and lower pattern density are beneficial for low penetration force, while microneedle stiffness increases the required pull-out force for more robust anchoring.

The results suggest that appropriate mechanical stiffness plays an essential role in better tissue anchoring (pull-out), while surface microhook dimensions and pattern density primarily influence tissue penetration.

Overall, this work successfully demonstrated a hybrid, passive tissue-anchoring prototype for compact ingestible system integration for prolonged GI interventions. The above characterizations confirm the reliability of the SMU for effective spring compression with a dissolvable polymer and passive actuation. Future works include demonstrating pH-sensitive polymer dissolution for targeted SMU actuation, PEG dissolution and tissue anchoring with the SMU array in a benchtop setting, as well as a test on the dummy capsule that utilizes the SMU package in an in vitro model of a GI tract.