Assessment of geometrical dimensions and puncture feasibility of microneedles manufactured by micromilling

Microneedles are an emerging technology designed to deliver drugs into human tissue. In this work, we assess the microneedle’s manufacturability by employing micromilling with a minimum quantity lubrication (MQL) system. A set of AISI 316L square pyramidal microneedles was fabricated and characterized using dimensional and surface metrology. Needle height (Hn), base length (Lb), tip radius (Rt), and the tip’s angle (θ) were studied. Additionally, surface roughness was quantified to correlate surface topography damage with tool wear (Dr). Experimental data shows tip truncation after manufacturing 30 needles (i.e., a tip radius between ~32 μm and 49 μm for manufacturing 10 and 30 needles, respectively). Additionally, to evaluate the effect of the tip’s morphology on the proficiency of the microneedles for a puncture, a numerical analysis to study the impact of tip truncation length (Tt) on puncture with an in silico assessment using COMSOL Multiphysics was performed. Data and insights from this work suggest that micromilling microneedle arrays is viable, considering the number of needles machined according to the cutting parameters selected to ensure functionality.


Introduction
Microneedles (MNs) are structures within the micro-scale used for transdermal drug delivery. These devices are an alternative to traditional hypodermic delivery resulting in a self-administered non-invasive method [1]. Microneedles have been classified traditionally as solid, hollow, dissolving, and coated. Among them, solid MNs excel in their capacity to penetrate the external layer of the skin and the primary barrier to drug transport (stratum corneum). Several drugs (i.e., small molecules, macromolecules, and vaccines across the skin [1][2][3][4][5][6][7][8]) have been used to penetrate the skin with outstanding performance.
Geometrical features such as height, base geometry, base diameter or base length, and tip radius are critical in the fabrication and functionality of microneedles. For example, tip radius impacts skin penetration and can represent a greater risk to the body in case of fracture [1]. Gill et al. studied the effect of microneedle geometry on pain; their results indicated that microneedle length and the number of microneedles have the most substantial impact on pain [9]. Singh et al. explained that typical microneedles vary from 25 to 2500 μm in height to overpass the stratum corneum (SC) [10]. Additionally, Nalluri et al. showed that a 1.5 mm height demonstrated the best cell activation in the dermis, increasing drug absorption [11]. Van Mulder et al. demonstrated that needle length should not exceed 700 μm to guarantee intradermal puncture in children [12].
Furthermore, Loizidou et al. fabricated microneedles with different base geometry (i.e., triangular, hexagonal, and square), resulting in the square base geometry with greater skin penetration depth and good mechanical strength [13]. Moreover, two relevant features, tip radius and insertion force, remain to be considered more thoroughly. Tip radius is a feature that strongly depends on the manufacturing process. It has been reported that typical microneedles vary this characteristic from 1 to 25 μm [10], and other experimentations have reported tip radii above 50 μm [14]. It has been found that the force required for insertion scaled with tip sharpness thus relates to the force of insertion with pain [15,16]. Although several in vivo and in vitro studies have been performed to obtain the applied force needed for inserting different microneedle geometries, there is no trend between geometry features and needle insertion capability. A research summary is presented in Table 1 for the required insertion forces per solid microneedles of different geometries, dimensions, and materials. According to Table 1, insertion forces  vary between 30 and 70 mN per microneedle for different  needle geometries. Each material (i.e., metals, polymers, and ceramics) for manufacturing microneedles has disadvantages related to its manufacture and the physical or mechanical properties of the needle functionality. For example, polymers such as silicone are easy to manufacture but brittle for application, while metals have demonstrated superior biocompatibility, although they are highly corrosive [22]. Kundu et al. performed stainless steel microneedles in a 5 × 5 array with a height of 500 μm using micromilling for applying agrochemicals into vascular bundles of plant tissue [23]. Additionally, hollow microneedles were manufactured in 304 stainless steel using micromilling and laser drilling with a length greater than 1500 μm to reach blood capillaries for point-of-care applications [24]. Due to the drug delivery method (i.e., a solid needle achieves skin penetration, and then a transdermal patch is placed or drug delivery through coated needles), solid metallic microneedles have gained interest in dermatology [25].
Several manufacturing processes have been studied for the fabrication of solid metallic microneedles. These are classified as additive and subtractive methods. Subtractive  [26]. Additionally, silicon material is brittle, and its design freedom is limited [27]. For example, Pt-based is of great interest for transdermal drug delivery. However, fabrication methods are complex for this material and involve multiple steps [28]. Several efforts have recently been made to improve cutting parameters in micromilling complex microgeometries [14,[29][30][31]. Implementing a minimum quantity lubrication (MQL) system can improve surface quality and acquire greater precision in machined surfaces [13]. Needle shapes are challenging due to the cutting strategies implemented in CAM (computer aided manufacturing) software which can influence needle application. For example, researchers have manufactured microneedles arrays in polymethylmethacrylate (PMMA) using micromilling to evaluate tool path strategies [32] and insertion effectiveness [33].
Considering that geometric features are critical for microneedles' design, fabrication, and functionality. This paper studied the needle's geometrical dimensions, surface roughness, and tool wear. Besides, an in silico assessment using a finite element method (FEM) software was performed to explore geometrical deviations' effect on skin penetration.

Material
The material used for experimental trials was AISI 316LVM stainless steel bars (M. Vincent & Associates Specialty Metals, USA) with a 12 mm height, 38.5 mm width, and 65 mm length.

Microneedles design
A set of 5 linear arrays were modeled using the solidworks software (Dassault Systems, France). Figure 1 presents the isometric view of the needle array and a detailed view of the tip. Each array consisted of 10 square pyramids, 0.5 mm base length (L b ) and 0.5 mm height (H n ), separated by 2.5 mm between centers. A nominal tip angle of 60° and a designed tip radius (R t ) of 0 μm were selected. Compared to the typical matrix array, a simplified linear array was created to measure needle geometrical features.

Experimental setup
The experiments were performed in a 3-axis vertical machining center (Makino F3, USA) with a maximum spindle speed of 30,000 RPM. Flat and ball micro end milling tools were mechanically adjusted to the collet and positioned in the   Fig. 2. Cutting parameters used for the micromilling assessment are listed in Table 2. Two experimental replications were made for each feed rate of the finishing operation. A total of 5 arrays per workpiece and 200 microneedles were manufactured.

CAM strategies
Three operations were programmed on the NX software (Siemens, Munich, Germany). A roughing operation using a planar mill setup with a follow-part path was set. A semi-finishing operation for the shaping of microneedles using a contour mill setup with a spiral path in the outward direction was programmed, and a 0.01 mm part side stock of material was left for the finishing operation. A finishing operation using a contour mill set up with the same path setting as the previous operation was set. For the roughing operation, a tungsten carbide 0.8 mm diameter flat end mill (Seco Tools, 920ML008MEGAT) was used; for the semi-finishing operation, a 0.8 mm diameter flat end mill was used (Mitsubishi, MS2SSD0080); and for the finishing operation, a 0.2 mm diameter ball end mill (Mitsubishi, MS2SBR0010S04) was employed. Table 3 provides additional details on the micromilling tools.

MQL system
The applied lubricant was a micro-drop of vegetable oil TRI-Cool MD-1 (TRICO, WI, USA). The compressed air rate was fixed to 12 ml/h with a rotameter FR2A16BVBN (key instrument, Hatfield, PA, USA). The MQL system used for the experiment was a MINI COOL MC1730 (NOGA, Shlomi, Israel) of a single spray unit with a control valve using a magnet and a Loc-Line® flexible hose. A magnetic base was installed on the right side of the spindle, and the tip of the nozzle was placed 30 mm off the tip of each tool with an inclination angle of 15°. Lubricant droplets were sprayed on a glass surface for 1 s, and 30 drops were measured.
Experimental data indicated a mean drop diameter value of ~37 ± 13 μm.

Geometrical and surface characterization
Geometrical features and surface topography were measured using a focus variation microscope infinite focus XL200 (Alicona, Raaba, Austria). A 10× objective lens, a vertical resolution of 0.1 μm, a lateral resolution of 4.0 μm, and an exposure time of 20 ms were selected. The profile form measurement module was used to quantify needle height (H n ), the base of the needle length (L b ), tip radius (R t ), and tip angle (θ). For surface roughness (Ra) characterization, microneedles were placed at a 90° angle. For this measurement, a cutoff length (L c ) of 800 μm and a path length of 4 mm were selected. Geometrical and dimensional characterization was conducted every ten microneedles up to a consecutive quantity of 50 microneedles.

Tool wear characterization
For semi-finishing and roughing operations, the tool wear was not quantified. Ball end mill tools were measured using an infinite focus microscope XL200 (Alicona, Raaba, Austria). Tools were measured each time a microneedle array was machined (i.e., ten needles). Each ball nose micro tool was placed at a 30° angle and analyzed with a 20× objective lens. 3D scans were performed and overlapped to analyze visual differences between datasets. Tool diameter was measured from the tip to an intersection plane at 40 μm on the 3D figure. Tool diameter reduction (D r ) was quantified as the difference between the initial diameter (D 0 ) and the diameter measured every ten needles (D f ) and calculated as:

In silico assesment
To assess the tip wear's effect on the microneedles' proficiency and reach the required tissue depth, a numerical modeling was developed using COMSOL Multiphysics version 5.6 (COMSOL INC, Burlington, MA). Simplified versions of a microneedle and skin tissue were used to represent the insertion process during a puncturing process. The structural mechanic module and the solid mechanic physics were employed with linear-elastic modeling to perform stationary and linear buckling analyses [34,35]. Table 4 shows the physical property details of AISI 316 LVM from the literature [36][37][38] for the FEM simulation. A skin model was comprised using three cylinders representing the stratum corneum, epidermis, and dermis (see Fig. 3). The materials were nearly incompressible with the modules of elasticity and Poisson's ratio described in Table 5.
The boundary conditions were defined on the three-dimensional model, as shown in Fig. 4a. The bottom and lateral surfaces of the geometry were restrained to movement. For the insertion study, a displacement of 0.39 (y displacement ) mm downward was applied for a set of experiments considering a previous study by Chang et al. [18]. Prior to the in-silico assessment, a tetrahedral mesh dependency analysis was performed employing an axial force of 30 mN (F applied ) considering ~ 330,000 elements and von Mises stresses. Tip truncation length (T t ) was evaluated considering experimental results. Experimentally, it represents the difference between the needle height designed (i.e., 500 μm) and the needle height measured. This truncation was observed experimentally due to two causes; firstly, when 3D geometry was not complete by the tool wear, and secondly when the zero point was not adjusted correctly in the Z axis. In silico assessment was performed to evaluate the effect of tip truncation length (i.e., increased each 10 μm) from 10 to 100 μm. Figure 4b shows the assessed geometry compared to an ideal microneedle tip.

Geometrical and surface characterization
Needle height and base length dimensions were measured (see Fig. 5). Data shows that height tends to be lower than expected (500 μm) (Fig. 5a). In contrast, the base lengths were above the nominal dimension (500 μm) (Fig. 5b). Additionally, a feed per flute of 4 μm/flute showed lesser variability than the 3 μm/flute that presented an abrupt change in height after 30 needles manufactured.
A designed tip angle of 60° and a designed tip radius of 0 μm from the side view were selected. Figure 6 summarizes the dimensional characterization of these features. Tip radius (Fig. 6a) ranged from approximately 41 to 82 μm and 32 to 76 μm for 3 and 4 μm/flute, respectively. While a mean tip angle of 52.4° and 47.4° was calculated for feed per flute of 3 and 4 μm/flute, respectively (Fig. 6b).
Surface metrology based on focus variation microscopy has enabled the surface characterization, wear, and damage of freeform surfaces [6]. Our measurements were performed in this apparatus because it combines the depth of focus of a vertical system with vertical scanning for a complete   For surface characterization, average and ten-point mean roughness were measured (Fig. 8a, b, respectively). Data suggests that feed rate did not have a significant impact on surface roughness, especially within the first 30 needles machined. This behavior can be attributed to the application of the MQL system. Due to the nature of micromilling, the flood coolant method was counterproductive because of the large quantity of cutting fluid used for a micro-scale product.

Tool wear measurements
Tool diameters were reported for every ten machined microneedles, and their reduction is presented in Fig. 9. Regarding tool wear, the maximum decrease in tool  Figure 10 illustrates the tool morphology, (Fig. 10a, b) corresponds to the initial tool diameter before machining. Figure 10c, d presents the tool diameter after 30 microneedles were manufactured, where it is observed material adhered to the tip. Figure 10e shows a broken cut edge, and Fig. 10f presents a flattening effect on the same zone. According to Câmara et al., a low feed rate tends to increase surface roughness, which is probably caused by a plowing phenomenon [39].
Our results indicate the same trend; the lowest values of feed rate (3 μm/flute) indicate the lower values of surface roughness, although material adhered to the tool tip presented in Fig. 10c, d, which causes the highest values of surface roughness reported in Fig. 8 (a,b) when 30 needles were manufactured.

Insertion simulation
An in-silico assessment was performed to evaluate the effect on the performance of different microneedle geometries. The study comprises a displacement and stress analysis for a set of microneedle tips considering the properties of a simplified skin model. The model was adopted from a previous work from Serrano et al. [8], where an epidermis thickness of 0.12 mm, from which 20 μm was defined by the stratum corneum. The module of elasticity of the stratum corneum of young adult skin was selected at ~26 MPa, although this value varies accordingly to gender, race, age, and anatomical region [8,40]. We evaluated the effect on the stress considering dullness and truncation of a microneedle tip (T t ) (see Fig. 11). For truncation deviations, the departure results in the Von Mises stresses show wide variation for relatively  Inspection of the images obtained using surface metrology (Fig. 7) provided a framework reference for some plausible operation conditions, as depicted in Fig. 12 (a-d).
The asterisks (*) in Fig. 12 show details regarding a representation of the numerical data presented in Fig. 11. The data show that for microneedles with a truncated geometry, the Von Mises stress maximum is found around the tip's periphery (see Fig. 12(a-d)). While it is correlated with the truncated area (A t ), the lowest the truncated area, the distribution of Von Misses stress is more uniform. Hence, truncated geometries with the highest truncated tip length due to breakage or caused by the manufacturing process can hinder the device's capability.
On the assessment, the von Mises stresses were evaluated for the skin layers and are summarized in Table 6. Data shows that regardless of the changes in the geometry of the microneedle, the stress on the inner layers remains constant. In contrast, the maximum vertical displacement is reduced when the truncated tip length increases, and hence, the crosssectional area of the MN decreases. Figure 13 describes the maximum Von Mises stress for the stratum corneum. Data suggests that truncation of a microneedle implies a reduction in the capacity of the microneedle for puncture.

Discussion
Needle height H n and base length L b dimensions complied with the medical requirements, which are 25 to 700 μm for penetration through the stratum corneum [41,42]. An increase in microneedle height can lead to intensified pain [9]. Van Mulder et al. reported that the optimal microneedle height should not exceed 700 μm for application in children ages: 8 weeks to 18 years old [12]. While Gill et al. concluded that the microneedle width base does not significantly impact pain, the shape of the base compromises the insertion into the skin [9]. The microneedle shape is defined by considering the limitations of the micro-machining process and their biomedical needs [32].
According to Gill H.S. et al., tip angle is not significant for insertion or pain for users compared with tip sharpness [9], while tip radius resulted in the most difficult variable to control, causing amorphous tips [42,43]. Further studies can be performed to evaluate microneedle insertion and its influence on skin puncture. It is important to note that the tip radius is highly dependent on the manufacturing process. It has been reported that typical microneedles vary this characteristic from 1 to 25 μm [10], and other experiments have reported tip radii above 50 μm [14]. In our results, the tip radius is increased for both feed rates above the microneedle 30. Other researchers have proposed an "S" type tool path strategy for micromilling PMMA pyramidal 2 mm height and square base of 0.7775 mm with sharper microneedle tips. They have also reported broken tips during the last few passes of the operation [32]. Therefore, further experiments must be performed to study the effect of tool path and strategy on tip geometry.
Cutting fluids involve cooling the tool's temperature, providing lubrication at the tool chip interface, improving the surface finish, and reducing adhesion between the tool-workpiece interface [44,45]. The proper use of these improves tool life and produces minimum burrs compared to dry-cutting conditions. Previous studies from our research group showed that tip diameter was the most affected feature by the cutting regime, showing a broad advantage of applying minimum quantity lubrication in terms of tool wear, surface roughness, and geometrical features in the fabrication of AISI 316L microneedles [14]. The data from this work is consistent with the previous work. Comparing the tool diameter reduction and the tool morphology presented in Figs. 9 and 10, the lowest feed rate generates less tool diameter reduction; however, the material is observed in the tool tip after manufacturing 30 needles for both feed rate values. According to Bissacco et al. [46], using ball nose tools in micromilling relates to this attached material and, consequently, to an increment of surface roughness.
Several methods to test insertion force are used in the literature, and no unique testing standard exists. Ah Kim et al. tried magnesium microneedles distributed in a 15 × The geometry was based on pyramidal needles with a base width and height of 300 and 600 μm, respectively. These arrays were tested on the center of artificial skin by applying 5 kg f and their results indicated that after elastic deformation, the force increased from 0.041 to 0.05 kg f (0.4 to 0.49 N), while the failure load for one needle was calculated around 0.00047 to 0.00051 kgf (4 to 5 mN) [47]. Sakamoto et al. designed an applicator for applying the spear-shaped microneedle to the skin surface. A polydimethylsiloxane (PDMS) sheet was used to simulate human skin, and the needle tested was a sharp four-sided pyramid with a height of 0.3 mm. Their results indicated that the needle tip was successfully entirely embedded into the PDMS with a force of 64.9 N and a speed of 150 N/s [48]. Mahony et al. determined that using the manual application, the maximum force applied in a 1 cm 2 microneedle array in a human subject is 30 N. He developed a microneedle insertion test in 36 males using an aluminum road to mount the needle array. This array consisted of 10 to 100 needles, assuring 95% penetration for forces more significant than 15 mN per needle (~1.2 N per array) [49]. In our simulation study, similar results were obtained for stainless steel microneedles (i.e., a tip truncation of 30 μm (experimentally 30 microneedles manufactured) resulted in a simulated insertion force of 6 mN). Further studies can be performed to study experimental insertion force in artificial skin and its compressive strength.
For in-silico assessment, Chang et al. concluded that the force distribution of microneedles increased with a decrease in the base diameter; in contrast, pyramid-shaped microneedles have the lowest distribution forces compared with beveled tip, cone, and tapered cone microneedles [18]. However, needle tip dimensional data was not reported. In our work, square pyramidal needles resulted in the highest values of maximum Von Misses stress in stratum corneum with lower values of truncated length (i.e., 10 μm). According to Kong et al., there is a linear relationship between the insertion force and the tip area; therefore, the smaller the tip area, the lower the insertion force [50], and similar results were obtained in this work. The progression of the change in the geometrical features from a sharp profile in Fig. 12a to a truncated structure in Fig. 12d is hinted because of the tool wear during a micromilling process. Experimentally, a tip truncation length lower than ~20 μm was obtained after manufacturing 30 needles. This truncation corresponds to a tip area below 214 μm 2 and a simulated insertion force of 3 mN. In contrast, when 40 needles are manufactured, the tip truncation length is between 30 and 40 μm and corresponds to an insertion force between 7 and 11 mN. Abidin et al. reported microneedle deformation in skin puncture with hollow conic shape microneedles; their stress results indicated that the maximum stress at the tip of the microneedle is 628 MPa and a resistive force of around 0.03 mN is required to pierce the human skin [34]. In our experiments, the maximum stresses at the tip were found with the lowest value of truncated tip length (Tt = 10 μm), resulting in 192 MPa. Additionally, no deformation is reported because the stress acting on the needle tip is less than the stainless steel yield strength (205 MPa). According to Kong et al. [50], the insertion needle force required for stratum corneum thickness of 20 μm is ~0.3 N with silicon needles. In our study, the stratum corneum thickness was fixed at 20 μm, and the variations on the tip area (A t ) caused by a truncated tip resulted in insertion forces between 0.002 and 0.04 N. These results differ due to the needle material used in experiments, specifically the material stiffness. In terms of the needle area, according to Kong et al. [50], a tip needle area of 5000 μm 2 , is required an insertion force less than 0.37 N; however, this force tends to have a linear trend and to increase (insertion force between 0.4 and 1.1 N) when higher values of tip area are used. These results are in accordance with our simulation results. Further experiments could be performed to determine the minimum force required to puncture the skin and to ensure drug absorption by stratum corneum.

Conclusions
Conclusions can be summarized as follow: • The feed per flute, comparing 3 μm/flute vs. 4 μm/flute, does not seem to significantly influence the microneedle surface roughness (Ra), remaining consistent up to 30 and 40 needles; it was shown that it is possible to obtain Ra values below 1 μm with the presented cutting parameters.
• Regarding tool wear (D r ), a maximum diameter reduction of ~8.2% and ~9.2% were found for 3 and 4 μm/flute of feed per flute, respectively. For a diameter reduction below 7%, approximately 30 microneedles can be machined for both feed rates.
• Simulations of insertion were carried out to check the functionality of the microneedles, with a truncated tip length (T t ) of 10 μm, representing a tip area of 54 μm 2 ; maximum needle von misses were obtained without apparent failure of the material.
• The study concludes that micromilling is viable for producing metal microneedle arrays. According to the specifications and simulation of insertion, the optimal number of needles to be machined is 30. This ensures functionality and a reduction of variability in geometrical characteristics.