Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

# Rapidly separable microneedle patch for the sustained release of a contraceptive

## Abstract

Women often have limited access to contraception, and barrier methods have low acceptance and a high failure rate, mostly due to incorrect use, which can result in unplanned pregnancies. Sustained-release formulations of contraceptive hormones are available, yet typically require their administration by trained personnel. Here, we report the design of a microneedle patch with rapidly separable biodegradable polylactic acid and polylactic-co-glycolic acid needles, and its application for the continuous release of levonorgestrel—a contraceptive hormone. Bubble structures between each microneedle and the patch backing allow the microneedles to efficiently penetrate skin under compression, and to snap off under shear within five seconds after patch administration. In rats, the microneedle patch was well tolerated, leaving little visible evidence of use, and maintained plasma concentrations of the hormone above the human therapeutic level for one month. Further development of the rapidly separable microneedle patch for self-administered, long-acting contraception could enable women to better control their fertility.

## Main

Despite great advances in contraceptive methods, 85 million pregnancies, representing 40% of all pregnancies worldwide, were unintended in 20121. Many pregnancies are terminated, resulting in 56 million abortions per year2. The high incidence of unintended pregnancy brings heavy economic and emotional burden to women and society at large. A primary reason for unintended pregnancy is the lack of contraceptive methods that meet the needs of diverse populations of women at various stages of their reproductive life cycle.

Non-hormonal contraceptive methods, such as condoms and diaphragms, provide physical barriers for pregnancy protection, but these barrier methods, even when accompanied by spermicide, usually have high failure rates due in part to poor patient acceptance and compliance with correct use3,4. Hormonal contraceptives, such as oral pills, vaginal rings, intrauterine devices, subdermal injections and implants5,6, generally provide a better level of effectiveness, but either require frequent dosing, which has significant compliance problems7, or delivery by healthcare professionals, which can be especially problematic in low-income countries8. Hence, there has been tremendous interest in a contraceptive that is safe and effective, enables long-term contraception, facilitates good patient access and compliance through self-administration, and has low cost suitable for use globally.

A number of different contraceptives are safe, effective and low cost. Some are long acting through use of sustained-release formulations9, but options for self-administration are limited. A well-established method of sustained release involves encapsulating drug in biodegradable polymers, such as polylactic acid (PLA) and polylactic-co-glycolic acid (PLGA), which slowly release the drug by diffusion and/or polymer degradation. This approach is utilized in many pharmaceutical products (for example, Lupron Depot)10,11 and investigated as injectable formulations or depot formulations for contraception12,13,14. However, these formulations generally require administration by trained professionals, thereby limiting patient access, and safety is hampered by possible needle re-use and needle-based injuries15. These issues are particularly important in developing countries.

To address the limitations of current contraceptive methods, this study introduces a rapidly separable microneedle patch for sustained release of contraceptive hormone. Instead of injecting sustained-release formulations by needle and syringe, a microneedle patch is briefly and painlessly applied to the skin to break off and embed biodegradable microneedles for the slow release of contraceptive hormone.

Microneedles are micrometre-scale structures that pierce just below the skin’s surface to administer drugs in a minimally invasive manner16. Most previous experience with microneedles has involved bolus delivery of drugs17,18,19,20,21 and vaccines22,23,24,25,26 using either coated or water-soluble microneedles, with limited work on microneedle patches for sustained release27,28,29. One study reported the use of dissolvable microneedles for the delivery of levonorgestrel (LNG) for emergency contraception. These patches were worn for up to 2 h and did not provide sustained drug release21. Previous studies have demonstrated that microneedle patches are painless30,31, can be self-administered with minimal training32,33,34 and are otherwise well tolerated in clinical trials33,34,35,36. Building off previous literature, the development of microneedle patches for long-acting contraception requires the formulation of microneedles for slow release of contraceptive hormones and engineering of microneedles that can be embedded in the skin after simple and rapid separation from the patch backing.

Given these objectives and challenges, we designed a microneedle patch for the sustained release of contraceptive hormone by encapsulating LNG in microneedles made from biodegradable polymers (PLGA and PLA). The microneedles were designed with an air bubble between the microneedle and patch backing, thereby enabling simple and rapid separation of the microneedles from the backing on insertion into the skin, after which the patch is removed and discarded as non-sharps waste; this should facilitate good patient compliance through self-administration. Detached microneedles subsequently undergo slow biodegradation in the skin for the sustained release and systemic delivery of encapsulated LNG for more than one month. We believe that this microneedle patch can be manufactured at low cost (that is, at a competitive cost compared with current long-acting injectable contraceptives) and can thereby increase women’s access to contraception and protection against unwanted pregnancy. Here, we present this microneedle patch and study its performance in vitro, ex vivo and in vivo.

## Results

### Design and fabrication of rapidly separable microneedle patches

When designing microneedle patches for long-acting contraception, we selected LNG as the contraceptive hormone because it is widely used in long-acting contraceptive products (for example, Norplant37) and requires a low daily dose (for example, 25–30 µg d−1 (ref. 38)). PLGA and PLA were selected as microneedle materials because these biodegradable polymers are biocompatible, mechanically strong and can be formulated for slow drug release for weeks to months10,11. The patch backing was formulated using polyvinyl alcohol (PVA) and sucrose, which are safe, water-soluble materials. To facilitate separation of the microneedles from the patch in the skin, we created an air bubble at the interface of the microneedle and patch backing (Fig. 1), designed to be strong under compression during insertion into the skin, but weak under shear applied to the patch after skin insertion. The patch backing contains a tapered pedestal at each microneedle base to facilitate deep microneedle insertion into the skin.

We moulded microneedles by casting an organic solvent (dioxane/tetrahydrofuran, 70/25% v/v) to solubilize PLA, PLGA and LNG, and 5% v/v water to slow evaporation during fabrication (Fig. 2a). Polymer and LNG were filled into mould cavities by centrifugation to form the microneedles and enhance microneedle strength by minimizing void formation. Next, an aqueous PVA and sucrose backing solution was applied to the mould, which entrapped an air bubble due to poor wetting of the dried polymer microneedles by the aqueous backing solution (Fig. 2b). The resulting microneedle patch (that is, ‘bubble-microneedle’ patch) comprised a 10 × 10 array of microneedles in ~0.5 cm2 mounted on a slightly larger, rigid tape. This patch was designed small enough to simplify transportation and storage, but large enough for convenient patient handling (Fig. 1b).

It was important to control the air bubble size at the base of each microneedle because the bubble structure determined the mechanical strength of the microneedle–backing interface. The bubble size was controlled by adjusting the backing solution volume applied during the second cast, as the increased weight of larger volumes of backing solution forced more air from the microneedle–backing interface (Fig. 2c). Varying the backing solution volume between 30 and 90 μl created bubble structures measuring 310–105 μm in depth (one-way analysis of variance (ANOVA), P = 0.0064) (Fig. 2d). The bubble structures extended into the patch backing pedestals, thereby not altering the size and shape of the microneedles (Fig. 2c).

### Mechanical performance of rapidly separable microneedle patches

To investigate whether bubble-microneedle patches have sufficient mechanical strength to penetrate the skin under compression but still detach in the skin under mild shear, we determined that the microneedle strength during compression decreased with increasing bubble size when measured using 100-microneedle arrays (one-way ANOVA, P = 0.042) (Fig. 3a) and individual microneedles (Supplementary Fig. 1). Although bubble-microneedle patches were weaker than solid-microneedle patches (without bubbles), microneedle patches with the largest bubbles (that is, 30 µl backing solution and 310 µm bubbles) tolerated compressive forces of ≥0.15 N needle−1, which is expected to enable skin puncture without breaking39.

In contrast, bubble-microneedle patches were easily broken under shear forces of 0.05–0.08 N needle−1 (Fig. 3b), which can easily be applied by hand40. Solid-microneedle patches required a significantly larger shear of 0.157 ± 0.001 N needle−1 to deform (two-tailed Student’s t-test, P < 0.001), and these solid-microneedle patches bent the microneedles without fracture, indicating that shear force would not break off the microneedles in the skin without bubbles.

### Application of rapidly separable microneedle patches to skin ex vivo

To determine whether microneedle patches could rapidly separate when applied to skin, we pressed bubble-microneedle patches into porcine skin (Fig. 4). Microneedles were loaded with Nile red dye for visualization (Fig. 4a). Microneedles penetrated the skin and, after applying gentle shear (~0.07 N needle−1) by thumb 5 s after patch application, microneedles detached from the patch backing and remained embedded in the skin (Fig. 4c). After microneedle separation, there was little residual red dye in the patch (Fig. 4d), further evidencing the efficient delivery of microneedles into the skin. Histological sections showed that the microneedles separated fully within the skin below its surface (Fig. 5). Gently and repeatedly scraping sites of microneedle patch treatment with a swab did not remove microneedles from the skin (Supplementary Fig. 2).

Although ~100% of microneedles penetrated the skin (Fig. 4e and Supplementary Fig. 3), >95% of bubble microneedles detached from the patch backing and >90% of encapsulated dye (simulating encapsulated hormone) was delivered into the skin. In contrast, only 15% of solid microneedles detached and <10% of dye was delivered into the skin—significantly less than for the bubble microneedles (two-tailed Student’s t-test, P = 0.00024 for detachment efficiency; P = 0.00030 for dye delivery efficiency). Taken together, these results show the rapid and efficient separation and high delivery efficiency of bubble-microneedle patches.

### Release of LNG from rapidly separable microneedle patches in vitro

LNG release from bubble-microneedle patches was performed in vitro using release media of saline containing 0–25% ethanol, which was added to better simulate in vivo release kinetics41. This is often faster than release in vitro42,43. LNG release showed no initial burst release on day 1 (Fig. 6a), and the LNG release kinetics was fairly constant over time (ranging from ~0.3 to ~2.2% LNG released per day, depending on the ethanol concentration) (one-way ANOVA, P < 0.001). Using 25% ethanol, all LNG was released within 45 d. These data indicate that sustained release of LNG from bubble-microneedle patches is possible and may achieve the target delivery timeframe of at least one month. Additional formulation optimization may be needed to achieve more specific target release times.

In addition, microneedle patches were made by encapsulating LNG in microneedles made of highly water-soluble PVA and sucrose. These microneedles exhibited burst releases of 60–90% of LNG, and all LNG was released within 6–12 d (Supplementary Fig. 4). All LNG was not released immediately, probably due to slow dissolution of sparingly water-soluble LNG, as opposed to resistance from the highly water-soluble PVA and sucrose microneedle matrix.

### LNG pharmacokinetics from rapidly separable microneedle patches in vivo

When bubble-microneedle patches (encapsulating hydrophobic Nile red dye) were manually applied to rat skin in vivo and gently sheared after 5 s, microneedles penetrated the skin, broke off from the patch backing and were fully embedded under the skin surface, as shown by histology (Fig. 7a). Fluorescence imaging of the skin surface shows the dye release kinetics during microneedle biodegradation in the skin (Fig. 7b). An array of fluorescent spots corresponding to microneedles embedded in the skin was seen initially, followed by gradual dimming over time. Variable fluorescence intensity at the site of each microneedle may be due to different depths of microneedle insertion into the skin, resulting in different amounts of skin between each embedded microneedle and the skin surface that absorbs and scatters light. Quantitative analysis similarly shows steady decay in fluorescence, corresponding to slow and continuous release kinetics, with most fluorescence gone after 45 d (Fig. 7c). These release kinetics mirror those of LNG release, as shown in Fig. 6c (see Supplementary Fig. 5).

In addition, the application of a water-soluble microneedle patch made of PVA and sucrose and loaded with red dye also generated an array of bright fluorescent spots in the skin, but they disappeared within 18 h (Supplementary Fig. 6). This rapid disappearance shows that dye could be cleared from the skin within 1 d, but encapsulation in separable PLGA and PLA microneedles extended release by >1 month (Fig. 7).

To assess LNG pharmacokinetics from bubble-microneedle patches, rats were administered: (1) an LNG-loaded bubble-microneedle patch; (2) a blank bubble-microneedle patch containing no LNG; or (3) no treatment (Fig. 6b). Rats administered LNG-loaded microneedle patches exhibited LNG plasma concentrations that increased to a peak concentration (Cmax) of 1.05 ± 0.14 ng ml−1 (mean ± s.d.) at a time (Tmax) of 6.0 ± 1.9 d post-application (Supplementary Table 1). After that, LNG levels slowly decreased (one-way ANOVA, P = 0.015), remaining above 200 pg ml−1 (which is the therapeutic level in humans41,44) for 30 d, then hovered near the therapeutic level until 45 d, after which LNG concentrations dropped to insignificant levels by 60 d.

Pharmacokinetic analysis shows relatively faster LNG absorption for the first 30 d, followed by slower absorption, with >95% absorption after 45 d (one-way ANOVA, P < 0.0001) (Fig. 6c). This slow, continuous LNG absorption profile in vivo is similar to LNG release kinetics in vitro (using 25% ethanol release media; Fig. 6a) and dye release in vivo (Fig. 7c). The in vivo LNG release profile is also similar to the target kinetics for a once-per-month contraceptive patch. The area under the curve (AUC) for LNG delivery from bubble microneedles of 598 ± 141 ng h ml−1 (Fig. 6b and Supplementary Table 1) indicates ~70% bioavailability compared with intravenous LNG injection (Supplementary Fig. 8 and Supplementary Table 1). Rats receiving blank microneedle patches or no treatment did not achieve LNG concentrations above background noise (Supplementary Fig. 9).

Bubble-microneedle patch administration of LNG was well tolerated by rats, without erythema, oedema or other signs of irritation during the 60-d study. Histological analysis after study completion showed no evidence of changes to the skin architecture, inflammatory cells or other signs of tissue damage (Supplementary Fig. 10).

## Discussion

Due to the high rate of unintended pregnancy, there is an urgent need to develop contraceptive methods that better meet women’s needs. While many contraceptives are safe and effective, only some enable long-term contraception, only a subset of these have low costs suitable for use globally, and still fewer have these capabilities in combination with self-administration45. The ability to self-administer a longer-acting contraceptive can provide women with greater access and autonomy, and has been shown to improve method continuation45.

To address these needs, we designed a patch containing LNG-loaded microneedles made from biodegradable polymer that break off in the skin and deliver contraceptive hormones to maintain plasma levels in rats above the human therapeutic level for more than one month. We believe that this approach is safe and effective because LNG has a long history of clinical contraceptive use46, the biodegradable polymers PLGA and PLA have been used safely in many medical products47, and microneedle patches (albeit not with separable microneedles) have been used successfully in clinical trials for bolus drug and vaccine delivery33,34,35.

### Rapid separation of microneedles

Long-term contraception can be enabled by encapsulating LNG in biodegradable microneedles for slow release, and designing microneedles that can simply and rapidly separate from the patch backing so that the user only wears the patch for a few seconds, after which there is little or no evidence of patch use. The discreet nature of this approach is important not only for cosmetic reasons, but also to address cultural, religious and other pressures that can require women to use family planning inconspicuously48. Previous studies with microneedle patches of similar dimensions also showed them to be painless30. In previous studies, bubbles have been incorporated into microneedle patches to help localize drugs within microneedles, but not to enable rapid microneedle separation49,50. Previous studies have also developed patches for rapid microneedle separation, using methods requiring complex, multi-step fabrication19,24,51,52,53.

### Controlled release of LNG from microneedles

The bubble-microneedle patch achieved sustained release of LNG, maintaining an LNG concentration above the human therapeutic level (200 pg ml−1) for 1 month in rats. Although the average LNG plasma concentration was up to ~1 ng ml−1, the therapeutic window for LNG is relatively large54, and marketed LNG-releasing products generate LNG plasma levels up to 1.5 ng ml−1 (ref. 55), indicating that an elevated LNG plasma concentration is acceptable.

While additional formulation optimization is needed, a once-per-month patch could be a convenient contraceptive option for many women. The return to fertility (all LNG cleared from the body) within one month after stopping patch use should also be attractive. The microneedle patch could alternatively be reformulated for release over shorter (weekly) or longer (biannually) times, to address the needs of different users. The dose per patch could be increased (for longer delivery times or to load a dose suitable for human use) by increasing drug loading, the microneedle size, the number of microneedles or other parameters.

There was no burst release of LNG from bubble-microneedle patches in vitro or in vivo, although burst release is commonly seen in other biodegradable-polymer controlled-release systems56,57. We hypothesize that burst release does not happen in the bubble-microneedle patches because a film of largely drug-free polymer forms on the microneedle surfaces, possibly due to solvent migration into the mould that concentrates or precipitates PLGA and PLA at the microneedle–mould interface. In addition, we believe that there is faster LNG redistribution within the mould due to its smaller molecular size, and/or possible phase separation into a polymer-rich phase and a polymer-poor phase. These hypotheses need further examination in future studies.

The microneedle patches were made from PLGA (which facilitated controlled LNG release) and a small amount of PLA (which provided additional mechanical strength to the microneedles). While essentially all LNG was released in vivo within two months, some biodegradable polymer may have remained in the skin. The literature indicates that the PLGA and PLA used in this study should biodegrade on timescales of 1–2 months and ~24 months10,58, respectively, to oligomers and monomers of lactic and glycolic acid, which can be safely cleared from the body10. It is worth noting that the total amounts of PLGA and PLA in a patch are ~1.08 and ~0.12 mg, respectively. For comparison, Lupron Depot, which has been safely administered to patients since Food and Drug Administration approval in 1989, contains 66 mg of PLGA in the 1 month formulation and 265 mg of PLA in the 4 month formulation59. We therefore expect that PLGA and PLA administered by microneedle patch should be safely cleared from the body.

### Cost of a rapidly separable microneedle patch

We expect that bubble-microneedle patches can be mass produced for <US$1 each60. Operations to make microneedle patches are familiar to pharmaceutical manufacturing; for example, mixing solutions, casting liquid formulations as films, drying and packaging. These operations are also similar to those used to prepare a lyophilized vial for injection61, suggesting a similar cost. The cost of long-acting injectable contraceptives ranges from >US$100 year−1 in the United States to <US$10 year−1 when purchased by the United Nations for developing countries (https://www.unfpaprocurement.org/products). Therefore, a once-per-month bubble-microneedle patch that costs ~US$10 year−1 is able to compete cost-wise with some existing products. In addition, the ability to self-administer a microneedle patch would reduce the overall service delivery costs and provide a further cost advantage compared with many longer-acting methods. Additional work is required to validate these predictions of bubble-microneedle patch costs.

### Study limitations and future work

A limitation of this study is that it was conducted in a relatively small number of animals. To move this research closer to future use in public health, technical advances are needed to increase the LNG dose from a rat dose to a human dose. This will require increasing the LNG per patch to ~1 mg per once-per-month patch38, which is 3–4 times larger than the LNG dose used here (0.3 mg patch−1). To demonstrate the feasibility of scaling up to a human dose, we fabricated patches containing 20 × 20 arrays of microneedles (Supplementary Fig. 11a), which could be inserted and detached in the skin (Supplementary Fig. 11b) and contained 1.52 ± 0.08 mg LNG per patch. While 10 × 10 arrays have been shown to insert into skin by thumb33, the manual insertion of larger patches needs to be studied and optimized. The LNG release profile(s) will need to be optimized as well, to match user preferences (for example, for duration). Incorporating a shearing mechanism into the patch is also necessary so that microneedle detachment does not rely on users correctly applying shear force. Developing validated in vitro–in vivo correlations for microneedle release performance will facilitate decision making through the design optimization process. Acceptability of the bubble-microneedle patch should be studied as part of the process, to optimize patch design for diverse cultures and settings. Translation of the work to clinical trials assessing both safety and efficacy in large populations will be needed to validate the study findings. Additional issues associated with the future translation of bubble-microneedle patches for long-acting contraception are discussed in the Supplementary Discussion.

### Outlook

High rates of unplanned pregnancy around the world bring economic and emotional burden to women, families and society at large. To provide greater access to contraception, we developed a delivery system for contraception based on a microneedle patch designed to enable self-administration of a long-acting contraceptive that is safe, effective and low cost. Simple administration is a critical feature of the microneedle patch. This was achieved by enabling manual application of the patch, requiring it to stay on the skin for just five seconds, and generating no biohazardous sharps waste. Once broken off, microneedles embedded under the skin’s surface were well tolerated, with faint evidence of patch application, thereby providing a discreet method of contraception—an important factor for many women. Contraceptive levels in plasma were maintained above the human therapeutic level for at least one month, then tapered off to background levels by two months, which should enable return to fertility. These findings suggest that a rapidly separable microneedle patch for self-administered, long-acting contraception could enable women around the world to better control their fertility.

## Methods

### Study design

This study was designed with the objective of developing a microneedle patch with rapidly separable microneedles that slowly release LNG and maintain an LNG plasma concentration above the human therapeutic level for one month. The approach was first to formulate a microneedle patch that met the following criteria: (1) sharp tips and mechanical strength suitable for penetration into skin; (2) incorporation of a bubble at the microneedle–patch backing interface that enables rapid microneedle separation in skin after the application of mild shear; (3) encapsulation of LNG in microneedles formulated to release it at a steady rate that maintains an LNG plasma concentration above the human therapeutic level for one month; (4) use of well-established biocompatible materials; (5) generation of no sharps waste; and (6) expectation of simple and painless self-administration by patients, based on the literature. The resulting microneedle patch was studied in vitro and in vivo in the rat to assess the ability of the patch to meet these criteria.

### Fabrication of rapidly separable microneedle patches

Polydimethylsiloxane (Dow Corning) moulds were used to fabricate the microneedle patches. The microneedles were arranged in a 10 × 10 array with a centre-to-centre interval of 600 μm in an area of 7 mm × 7 mm, and each microneedle was conical with a base radius of 150 μm, a height of 600 μm and a tip radius of ~10 μm. The patch backing contained an array of pedestals (base diameter 600 μm, top diameter 150 μm and height 350 μm) that were positioned at the base of each microneedle to elevate the microneedles above the base of the backing.

Microneedle patch fabrication involved sequentially casting two solutions onto the mould. The first casting solution contained 5% (w/v) solids dissolved in a mixture of dioxane, tetrahydrofuran and water (70, 25 and 5% v/v). The solids were composed of PLGA, PLA and LNG (72, 8 and 20% w/w). The formulation containing PLGA and PLA in a ratio of 90:10 was selected so that PLGA would provide primary control over the drug release rate, and PLA was added to increase the mechanical strength. The casting solution was made by dissolving 0.45 g PLGA (a 50:50 molar ratio of lactide:glycolide; inherent viscosity: 0.59 dl g1; Durect) and 0.05 g PLA (inherent viscosity: 1.02 dL g−1; Durect) in 2 ml dioxane (Sigma–Aldrich), then adding a solution of 0.125 g LNG (Industriale Chimica Srl) in 3.375 ml tetrahydrofuran (Thermo Fisher Scientific) and finally mixing them together with additional dioxane and deionized water to obtain the final casting solution.

To fabricate the blank microneedle patches, no LNG was added in the polymer solution, which contained 5% (w/v) solids composed of PLGA and PLA (90 and 10% w/w) in dioxane and deionized water (95 and 5% v/v). To fabricate the microneedle patches containing Nile red (Sigma–Aldrich), 20 mg Nile red powder was added into the blank casting solution without LNG. Next, 20 µl of the casting solution was applied to the top of the microneedle mould and centrifuged at 3,200g for 2 min to fill the mould. Then, 20 μl dioxane was applied to the top of the mould and centrifuged at 3,200g for 2 min to wash the residual casting solution on the top of the mould into the mould cavities. The loading and washing process was repeated three more times to fully fill the mould, then the mould was placed in a 60 °C oven with a vacuum for 12 h to dry.

After that, the second casting solution, consisting of 18% (w/v) PVA (molecular weight: 6,000 Da; Sigma–Aldrich) and 18% (w/v) sucrose (Sigma–Aldrich) in deionized water, was gently applied to the dried polydimethylsiloxane mould surface to form the patch backing. During this casting, an air bubble could be trapped between the microneedle and the pedestal of the patch backing, such that the bubble size could be controlled by adjusting the volume (30, 50, 70 or 90 µl) of the second casting solution. After drying in the chemical hood for 2 h, the mould was placed in a desiccator for 2 d at room temperature (20–25 °C) for complete drying, after which the patch was carefully peeled from the mould and stored in a desiccator until use.

### Microneedle patch mechanical properties

The mechanical properties of the solid-microneedle patches and rapidly separable microneedle patches containing bubble structures of different sizes were measured using a displacement-force test station (Force Gauge; Mark-10) (Fig. 3). Briefly, to test microneedle patches under compression, a single patch was attached to a rigid stainless-steel platform positioned vertically (microneedles facing up), and the test station sensor probe approached the microneedles in the vertical direction at a speed of 0.1 mm s−1. The initial distance between the sensor and microneedle tips was 1 cm. Displacement and force measurements began when the sensor first touched the microneedle tips and continued until the sensor travelled 0.4 mm from the microneedle tips towards the patch backing.

To test microneedle patches under shear, a single microneedle patch was attached to a rigid platform positioned horizontally (microneedles facing to the side). The starting position was 1 cm away from the top row of microneedles, and the sensor approached the microneedles in the vertical direction at a speed of 0.1 mm s−1. Displacement and force began when the sensor first touched the microneedles and continued until the sensor travelled 2.1 mm parallel to the patch backing.

### Skin insertion of microneedle patches ex vivo

To evaluate the penetration, separation, retention and delivery efficiency of the microneedle patches, patches loaded with fluorescent dye (Nile red) were inserted into stretched porcine skin ex vivo by pressing with a thumb for 5 s, then gently sliding to one side along the skin surface to apply a shear force to separate the microneedles from the patch backing. After separation, the skin containing separated microneedles was examined by optical microscopy (Olympus) to identify the detached microneedles embedded in the skin. In some cases, a swab was gently and repeatedly scraped across the site of the microneedle patch treatment for 10 s to remove any detached microneedles that were partially protruding above the skin surface. To assess only the penetration of the microneedle patches, patches were applied to the skin using a vertical force only, then immediately removed. The skin was covered with gentian violet solution (Humco) for 10 min to stain the sites of microneedle penetration, then cleaned with alcohol swabs to remove residual dye from the skin surface. The penetration, separation and retention efficiency were calculated by dividing the number of coloured spots (that is, due to gentian violet staining or the presence of fluorescent microneedles in the skin) by the number of microneedles in the patch (that is, 100).

Microneedle patches were applied to the skin manually to better simulate actual use. To estimate the forces applied during insertion and detachment, the investigator pressed his thumb against the force gauge with a force similar to that applied to the microneedle patches. The compressive force during microneedle patch insertion and shear force during microneedle detachment were estimated to be ~0.25 and ~0.07 N needle−1, respectively.

To evaluate the delivery efficiency of the microneedle patch, the fluorescence intensity from the dye in the microneedle patch before and after skin insertion, as well as the fluorescence from the dye on the skin surface, were measured by quantitative image analysis (Microplate Reader; Bio-Rad). The dye delivered in the skin was quantified by subtracting the amount of dye in the residual backing and on the skin surface from that in the microneedle patch before insertion. The delivery efficiency was calculated by dividing the delivered dye in the skin by the amount of dye in the microneedle patch before insertion. Finally, the skin was frozen, then cut into 10 µm sections for histological analysis.

### LNG release from microneedle patches in vitro

To evaluate the in vitro release of LNG from microneedle patches and predict the release of LNG in vivo, we used phosphate-buffered saline with Tween-80 (PBST), with different concentrations of ethanol as the release medium41. Specifically, one microneedle patch was placed into 1 l PBST (with varying concentrations of ethanol) in a glass vessel. The PBST solution contained 137 mM NaCl, 2.68 mM KCl, 10.14 mM Na2HPO4, 1.76 mM KH2PO4 and 0.02% (w/v) Tween-80; ethanol was added to the PBST to a final concentration of 0, 2, 10 or 25% (v/v) ethanol. The glass vessel was incubated in a shaker water bath at 37 °C and shaken at 80 r.p.m. At pre-determined time points (0, 1, 3, 6, 12, 18, 24, 30, 36, 43, 49, 54 and 60 d), 1 ml release medium was collected and replaced with the same amount of fresh medium. Collected samples were analysed by ultra-performance liquid chromatography–mass spectrometry (Waters) to quantify the LNG concentration. LNG was separated on an Acquity UPLC BEH C18 column (100 mm × 2.1 mm i.d.; 1.7 µm particle size) at 50 °C. The mobile phase was a mixture of acetonitrile containing 0.1% formic acid, and water containing 0.1% formic acid (8:2 ratio, v/v). The flow rate was 0.3 ml min−1, with an injection volume of 10 µl. The detection of LNG was performed by electrospray ionization mass spectrometry in the positive ion mode. The target analyte of LNG (M + H+; m/z = 313.4) was used for quantification62.

### LNG pharmacokinetics after release from microneedle patches in vivo

The LNG pharmacokinetics was evaluated in adult female Sprague Dawley rats (200 ± 12 g) by applying an LNG-loaded microneedle patch to each rat while under isoflurane anaesthesia. The rats’ dorsal skin was shaved before the application of microneedle patches, taking care not to damage the skin during shaving. All animal studies were performed with the approval of the Georgia Institute of Technology Institutional Animal Care and Use Committee. The experiments and handling of rats were conducted in accordance with federal, state and local regulations.

To investigate polymer biodegradation and the release of dye from the PLGA and PLA microneedles in rats, microneedle patches containing Nile red were administered to the rats using the methods described above for ex vivo microneedle patch application to porcine skin, after which the rats were imaged by fluorescence microscopy (Olympus) using a consistent imaging setup for all rats (for example, fluorescence excitation light intensity and image capture exposure time) on different days after microneedle application (days 0, 1, 7, 15, 30, 45 and 60). The fluorescence intensity of the microneedles embedded in rat skin was quantified by analysing the fluorescence images using ImageJ (National Institutes of Health). As a control group, a water-soluble microneedle patch containing Nile red was applied to rat skin in vivo and kept on the skin for 15 min to allow the microneedles to fully dissolve. The rat skin was then imaged at 0, 4, 8, 12 and 18 h post-administration, and the fluorescence intensity was quantified using the same method. As an additional control, we exposed a solution containing 10 mg ml−1 Nile red in dioxane to ambient light for 18 h. There was no significant difference in the fluorescence intensity of the solution between the exposed sample and (1) a freshly prepared sample or (2) a similar sample that was left in the dark for 18 h (Supplementary Fig. 7).

To study the pharmacokinetics of LNG release from separable microneedles, rats were randomly divided into three groups. The first received LNG-loaded microneedle patches, the second received blank microneedle patches (without LNG) and the third did not receive any microneedle patches. A power analysis indicated that a sample size of 8 rats per group would be sufficient to distinguish pharmacokinetic profiles in animals receiving LNG from those administered a blank microneedle patch (containing no LNG), with 95% confidence. The primary endpoint of the animal study was an LNG plasma concentration above the human therapeutic level for one month. The secondary endpoint was irritation at the site of microneedle patch administration. All data collected in this study were retained; no outliers were excluded.

Blood samples (~500 µl) were drawn from the tail vein at different times after microneedle patch application: 0 h, 12 h, 24 h, 3 d, 7 d, 10 d, 14 d, 17 d, 21 d, 24 d, 28 d, 31 d, 35 d, 38 d, 42 d, 45 d, 49 d, 52 d, 55 d and 60 d. The plasma was then separated by centrifuging the blood samples at 2,000g for 15 min at 4 °C, and underwent subsequent analysis by enzyme-linked immunosorbent assay (Thermo Fisher Scientific) following the manufacturer’s instructions, to determine the LNG concentration. To evaluate the biocompatibility of LNG delivery from separable microneedle patches, rats were euthanized by CO2 asphyxiation at the end of the study (that is, 60 d after microneedle patch application) and the tissue surrounding the patch application site was excised. This tissue was fixed in 10% neutral buffered formalin for 2 d at 4 °C, then embedded in paraffin after complete dehydration, cut into sections of 5 µm thickness and stained using haematoxylin and eosin for histological analysis.

### Pharmacokinetic analysis

Pharmacokinetic parameters were calculated using non-compartmental pharmacokinetic analysis63. These parameters included: the observed maximum plasma concentration (Cmax); the time when Cmax was achieved (Tmax); the elimination rate constant of LNG (Ke), which was estimated by fitting the data from the terminal phase of the plasma concentration versus time profile following intravenous LNG injection in the control group by log-linear regression to estimate the slope; the area under the plasma concentration–time curve from time zero to the time of the last detection using the linear trapezoidal rule (AUC0–t); and the AUC from time zero to infinity (AUC0–inf). The bioavailability of LNG delivered from the microneedles was calculated from the ratio of the dose-normalized AUC values after microneedle patch administration and intravenous LNG injection. The Wagner–Nelson method was used to estimate the percentage of LNG absorbed in vivo, and numerical deconvolution was applied to the LNG plasma concentration versus time profiles64.

### Statistical analysis

All of the results presented in this study are means ± s.d. Statistical analysis was perform using a two-sided Student’s t-test or ANOVA, with the software Origin. P < 0.05 was considered significant.

### Reporting Summary

Further information on research design is available in the Nature Research Reporting Summary linked to this article.

## Data availability

The authors declare that all data supporting the results in this study are available within the paper and its Supplementary Information. Source data for the figures in this study are available from figshare with the identifier https://doi.org/10.6084/m9.figshare.6025748.

## References

1. 1.

Sedgh, G., Singh, S. & Hussain, R. Intended and unintended pregnancies worldwide in 2012 and recent trends. Stud. Family Plan. 45, 301–314 (2014).

2. 2.

Sedgh, G. et al. Abortion incidence between 1990 and 2014: global, regional, and subregional levels and trends. Lancet 388, 258–267 (2016).

3. 3.

Rose, E. et al. The validity of teens’ and young adults’ self-reported condom use. Arch. Pediatr. Adolesc. Med. 163, 61–64 (2009).

4. 4.

Macaluso, M. et al. Mechanical failure of the latex condom in a cohort of women at high STD risk. Sex. Transm. Dis. 26, 450–458 (1999).

5. 5.

Galzote, R. M., Rafie, S., Teal, R. & Mody, S. K. Transdermal delivery of combined hormonal contraception: a review of the current literature. Int. J. Womens Health 9, 315–321 (2017).

6. 6.

Mansour, D., Inki, P. & Gemzell-Danielsson, K. Efficacy of contraceptive methods: a review of the literature. Eur. J. Contracept. Reprod. Health Care 15, S19–S31 (2010).

7. 7.

Petitti, D. B. et al. Stroke in users of low-dose oral contraceptives. New Engl. J. Med. 335, 8–15 (1996).

8. 8.

Halpern, V. et al. Towards the development of a longer-acting injectable contraceptive: past research and current trends. Contraception 92, 3–9 (2015).

9. 9.

Prescott, G. M. & Matthews, C. M. Long-acting reversible contraception: a review in special populations. Pharmacotherapy 34, 46–59 (2014).

10. 10.

Makadia, H. K. & Siegel, S. J. Poly lactic-co-glycolic acid (PLGA) as biodegradable controlled drug delivery carrier. Polymers (Basel) 3, 1377–1397 (2011).

11. 11.

Tyler, B., Gullotti, D., Mangraviti, A., Utsuki, T. & Brem, H. Polylactic acid (PLA) controlled delivery carriers for biomedical applications. Adv. Drug Deliv. Rev. 107, 163–175 (2016).

12. 12.

Sun, Y. et al. Synchronic release of two hormonal contraceptives for about one month from the PLGA microspheres: in vitro and in vivo studies. J. Control. Release 129, 192–199 (2008).

13. 13.

Lee, B. K., Yun, Y. & Park, K. PLA micro- and nano-particles. Adv. Drug Deliv. Rev. 107, 176–191 (2016).

14. 14.

Wu, L. F., Janagam, D. R., Mandrell, T. D., Johnson, J. R. & Lowe, T. L. Long-acting injectable hormonal dosage forms for contraception. Pharm. Res. 32, 2180–2191 (2015).

15. 15.

Dicko, M. et al. Safety of immunization injections in Africa: not simply a problem of logistics. Bull. World Health Organ. 78, 163–169 (2000).

16. 16.

Kim, Y. C., Park, J. H. & Prausnitz, M. R. Microneedles for drug and vaccine delivery. Adv. Drug Deliv. Rev. 64, 1547–1568 (2012).

17. 17.

Li, G. H., Badkar, A., Nema, S., Kolli, C. S. & Banga, A. K. In vitro transdermal delivery of therapeutic antibodies using maltose microneedles. Int. J. Pharm. 368, 109–115 (2009).

18. 18.

Yu, J. C. et al. Microneedle-array patches loaded with hypoxia-sensitive vesicles provide fast glucose-responsive insulin delivery. Proc. Natl Acad. Sci. USA 112, 8260–8265 (2015).

19. 19.

Chen, M. C., Lin, Z. W. & Ling, M. H. Near-infrared light-activatable microneedle system for treating superficial tumors by combination of chemotherapy and photothermal therapy. ACS Nano 10, 93–101 (2016).

20. 20.

Ye, Y. Q. et al. Microneedles integrated with pancreatic cells and synthetic glucose-signal amplifiers for smart insulin delivery. Adv. Mater. 28, 3115–3121 (2016).

21. 21.

Yao, G. T. et al. Novel dissolving microneedles for enhanced transdermal delivery of levonorgestrel: in vitro and in vivo characterization. Int. J. Pharm. 534, 378–386 (2017).

22. 22.

Sullivan, S. P. et al. Dissolving polymer microneedle patches for influenza vaccination. Nat. Med. 16, 915–920 (2010).

23. 23.

DeMuth, P. C. et al. Vaccine delivery with microneedle skin patches in nonhuman primates. Nat. Biotech. 31, 1082–1085 (2013).

24. 24.

Chen, M. C., Huang, S. F., Lai, K. Y. & Ling, M. H. Fully embeddable chitosan microneedles as a sustained release depot for intradermal vaccination. Biomaterials 34, 3077–3086 (2013).

25. 25.

Chen, X. F. et al. Dry-coated microprojection array patches for targeted delivery of immunotherapeutics to the skin. J. Control. Release 139, 212–220 (2009).

26. 26.

DeMuth, P. C. et al. Polymer multilayer tattooing for enhanced DNA vaccination. Nat. Mater. 12, 367–376 (2013).

27. 27.

Park, J. H., Allen, M. G. & Prausnitz, M. R. Polymer microneedles for controlled-release drug delivery. Pharm. Res. 23, 1008–1019 (2006).

28. 28.

Zhang, Y. Q. et al. Locally induced adipose tissue browning by microneedle patch for obesity treatment. ACS Nano 11, 9223–9230 (2017).

29. 29.

DeMuth, P. C., Garcia-Beltran, W. F., Ai-Ling, M. L., Hammond, P. T. & Irvine, D. J. Composite dissolving microneedles for coordinated control of antigen and adjuvant delivery kinetics in transcutaneous vaccination. Adv. Funct. Mater. 23, 161–172 (2013).

30. 30.

Gill, H. S., Denson, D. D., Burris, B. A. & Prausnitz, M. R. Effect of microneedle design on pain in human volunteers. Clin. J. Pain 24, 585–594 (2008).

31. 31.

Haq, M. I. et al. Clinical administration of microneedles: skin puncture, pain and sensation. Biomed. Microdevices 11, 35–47 (2009).

32. 32.

Norman, J. J. et al. Microneedle patches: usability and acceptability for self-vaccination against influenza. Vaccine 32, 1856–1862 (2014).

33. 33.

Rouphael, N. G. et al. The safety, immunogenicity, and acceptability of inactivated influenza vaccine delivered by microneedle patch (TIV-MNP 2015): a randomised, partly blinded, placebo-controlled, phase 1 trial. Lancet 390, 649–658 (2017).

34. 34.

Daddona, P. E., Matriano, J. A., Mandema, J. & Maa, Y. F. Parathyroid hormone (1-34)-coated microneedle patch system: clinical pharmacokinetics and pharmacodynamics for treatment of osteoporosis. Pharm. Res. 28, 159–165 (2011).

35. 35.

Hirobe, S. et al. Clinical study and stability assessment of a novel transcutaneous influenza vaccination using a dissolving microneedle patch. Biomaterials 57, 50–58 (2015).

36. 36.

Uppuluri, C. T. et al. Microneedle-assisted transdermal delivery of Zolmitriptan: effect of microneedle geometry, in vitro permeation experiments, scaling analyses and numerical simulations. Drug Dev. Ind. Pharm. 43, 1292–1303 (2017).

37. 37.

Polaneczky, M., Slap, G., Forke, C., Rappaport, A. & Sondheimer, S. The use of levonorgestrel implants (Norplant) for contraception in adolescent mothers. New Engl. J. Med. 331, 1201–1206 (1994).

38. 38.

Sivin, I. Risks and benefits, advantages and disadvantages of levonorgestrel-releasing contraceptive implants. Drug. Saf. 26, 303–335 (2003).

39. 39.

Prausnitz, M. R. Microneedles for transdermal drug delivery. Adv. Drug Deliv. Rev. 56, 581–587 (2004).

40. 40.

Bao, S. & Silverstein, B. Estimation of hand force in ergonomic job evaluations. Ergonomics 48, 288–301 (2005).

41. 41.

Wang, S. H. et al. Controlled release of levonorgestrel from biodegradable poly(d,l-lactide-co-glycolide) microspheres: in vitro and in vivo studies. Int. J. Pharm. 301, 217–225 (2005).

42. 42.

Zolnik, B. S. & Burgess, D. J. Evaluation of in vivo–in vitro release of dexamethasone from PLGA microspheres. J. Control. Release 127, 137–145 (2008).

43. 43.

Doty, A. C. et al. Mechanisms of in vivo release of triamcinolone acetonide from PLGA microspheres. J. Control. Release 256, 19–25 (2017).

44. 44.

Fotherby, K. Levonorgestrel—clinical pharmacokinetics. Clin. Pharmacokinet. 28, 203–215 (1995).

45. 45.

Kohn, J. E. DMPA self-administration can improve contraceptive access, continuation, and autonomy. Lancet Glob. Health 6, E481–E482 (2018).

46. 46.

Novikova, N., Weisberg, E., Stanczyk, F. Z., Croxatto, H. B. & Fraser, I. S. Effectiveness of levonorgestrel emergency contraception given before or after ovulation—a pilot study. Contraception 75, 112–118 (2007).

47. 47.

Anderson, J. M. & Shive, M. S. Biodegradation and biocompatibility of PLA and PLGA microspheres. Adv. Drug Deliv. Rev. 64, 72–82 (2012).

48. 48.

Higginbottom, G. M. A. et al. “I have to do what I believe”: Sudanese women’s beliefs and resistance to hegemonic practices at home and during experiences of maternity care in Canada. BMC Pregnancy Childb. 13, 51 (2013).

49. 49.

Wang, Q. L., Zhu, D. D., Liu, X. B., Chen, B. Z. & Guo, X. D. Microneedles with controlled bubble sizes and drug distributions for efficient transdermal drug delivery. Sci. Rep. 6, 38755 (2016).

50. 50.

Chu, L. Y., Choi, S. O. & Prausnitz, M. R. Fabrication of dissolving polymer microneedles for controlled drug encapsulation and delivery: bubble and pedestal microneedle designs. J. Pharm. Sci. 99, 4228–4238 (2010).

51. 51.

Chu, L. Y. & Prausnitz, M. R. Separable arrowhead microneedles. J. Control. Release 149, 242–249 (2011).

52. 52.

Zhu, D. D., Wang, Q. L., Liu, X. B. & Guo, X. D. Rapidly separating microneedles for transdermal drug delivery. Acta Biomater. 41, 312–319 (2016).

53. 53.

Zhu, D. D., Chen, B. Z., He, M. C. & Guo, X. D. Structural optimization of rapidly separating microneedles for efficient drug delivery. J. Ind. Eng. Chem. 51, 178–184 (2017).

54. 54.

Abrams, L. S., Skee, D. A., Natarajan, J., Wong, F. A. & Lasseter, K. C. Multiple-dose pharmacokinetics of a contraceptive patch in healthy women participants. Contraception 64, 287–294 (2001).

55. 55.

Sivin, I. et al. First week drug concentrations in women with levonorgestrel rod or Norplant (R) capsule implants. Contraception 56, 317–321 (1997).

56. 56.

Huang, X. & Brazel, C. S. On the importance and mechanisms of burst release in matrix-controlled drug delivery systems. J. Control. Release 73, 121–136 (2001).

57. 57.

Wang, J., Wang, B. A. & Schwendeman, S. P. Characterization of the initial burst release of a model peptide from poly(d,l-lactide-co-glycolide) microspheres. J. Control. Release 82, 289–307 (2002).

58. 58.

Avgoustakis, K. in Encyclopedia of Biomaterials and Biomedical Engineering 2259–2269 (Informa, New York, 2008).

59. 59.

Lupron Depot (Leuprolide Acetate for Depot Suspension) (AbbVie, 2016).

60. 60.

Larraneta, E., Lutton, R. E. M., Woolfson, A. D. & Donnelly, R. F. Microneedle arrays as transdermal and intradermal drug delivery systems: materials science, manufacture and commercial development. Mat. Sci. Eng. R 104, 1–32 (2016).

61. 61.

Overcashier, D. E., Patapoff, T. W. & Hsu, C. C. Lyophilization of protein formulations in vials: investigation of the relationship between resistance to vapor flow during primary drying and small-scale product collapse. J. Pharm. Sci. 88, 688–695 (1999).

62. 62.

Penning, T. M., Lee, S. H., Jin, Y., Gutierrez, A. & Blair, I. A. Liquid chromatography-mass spectrometry (LC-MS) of steroid hormone metabolites and its applications. J. Steroid Biochem. Mol. Biol. 121, 546–555 (2010).

63. 63.

Gabrielsson, J. & Weiner, D. Non-compartmental analysis. Methods Mol. Biol. 929, 377–389 (2012).

64. 64.

Gibaldi, M. & Perrier, D. Pharmacokinetics 2nd edn (M. Dekker, New York, 1982).

## Acknowledgements

We thank D. Owen, G. S. Kopf and J. Ayres of FHI 360 for valuable technical discussions and review of the manuscript, and D. Bondy and A. Troxler for administrative support. This publication is made possible by the generous support of the American people through the U.S. Agency for International Development (USAID) and was prepared under a subcontract funded by Family Health International under Cooperative Agreement No. AID-OAA-15-00045, funded by USAID. The content of this publication does not necessarily reflect the views, analysis or policies of FHI 360, USAID or the United States Government, nor does any mention of trade names, commercial products, or organizations imply endorsement by FHI 360, USAID or the United States Government.

## Author information

Authors

### Contributions

W.L., J.T., S.P.S. and M.R.P. designed the project. W.L. and M.R.P. wrote the manuscript, with contributions from R.N.T., J.T., M.R.F. and S.P.S. W.L., R.N.T. and J.T. performed the experiments. All authors analysed and interpreted the data.

### Corresponding author

Correspondence to Mark R. Prausnitz.

## Ethics declarations

### Competing interests

M.R.P. is an inventor of patents licensed to companies developing microneedle-based products, a paid advisor to companies developing microneedle-based products, and a founder/shareholder of companies developing microneedle-based products (Micron Biomedical). This potential conflict of interest has been disclosed and is managed by Georgia Tech and Emory University.

Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

## Supplementary information

### Supplementary Information

Supplementary figures, table, discussion and references.

## Rights and permissions

Reprints and Permissions

Li, W., Terry, R.N., Tang, J. et al. Rapidly separable microneedle patch for the sustained release of a contraceptive. Nat Biomed Eng 3, 220–229 (2019). https://doi.org/10.1038/s41551-018-0337-4

• Accepted:

• Published:

• Issue Date:

• ### Solvent-Free Polycaprolactone Dissolving Microneedles Generated via the Thermal Melting Method for the Sustained Release of Capsaicin

• Jaehong Eum
• , Youseong Kim
• , Daniel Um
• , Jiwoo Shin
• , Huisuk Yang
•  & Hyungil Jung

Micromachines (2021)

• ### Microneedle Technology: An Insight into Recent Advancements and Future Trends in Drug and Vaccine Delivery

• Jitendra Gupta
• , Reena Gupta
•  & Vanshita

ASSAY and Drug Development Technologies (2021)

• ### Microneedle array systems for long-acting drug delivery

• Lalit K. Vora
• , Kurtis Moffatt
• , Ismaiel A. Tekko
• , Alejandro J. Paredes
• , Fabiana Volpe-Zanutto
• , Deepakkumar Mishra
• , Ke Peng
• , Raghu Raj Singh Thakur
•  & Ryan F. Donnelly

European Journal of Pharmaceutics and Biopharmaceutics (2021)

• ### Microneedles for Extended Transdermal Therapeutics: A Route to Advanced Healthcare

• Suman Pahal
• , Vivek Ghate
• , Utkarsh Bhutani
• , Mangalore Manjunatha Nayak
• , Dinesh Narasimhaiah Subramanyam
•  & Praveen Kumar Vemula

European Journal of Pharmaceutics and Biopharmaceutics (2021)

• ### Promotion of collagen deposition during skin healing through Smad3/mTOR pathway by parathyroid hormone-loaded microneedle

• Zhixiao Yao
• , Tong Xue
• , Hao Xiong
• , Chuandong Cai
• , Xudong Liu
• , Fei Wu
• , Shen Liu
•  & Cunyi Fan

Materials Science and Engineering: C (2021)