Fabrication of thermoplastic microfluidic probes for nanospray desorption electrospray ionization mass spectrometry

Abstract

Nanospray desorption electrospray ionization mass spectrometry (nano-DESI MS) uses specially designed probes to achieve a stable liquid bridge on the sample surface, ensuring consistent analyte extraction. Here, we present a mold-based fabrication protocol for cyclic olefin copolymer (COC) microfluidic probes (MFPs). We describe mold design, probe fabrication, and probe setup for nano-DESI MS. Four mold fabrication methods, including computer numerical control (CNC) machining, metal 3D printing, photolithography, and selective laser-assisted etching (SLE), were evaluated, with SLE providing the highest structural precision and enabling reproducible channel fabrication. Performance evaluation using lysophosphatidylethanolamine (LPE) 17:1 standard demonstrated stable liquid bridge formation, clear spatial separation, reproducible signal response, and low background interference. This mold-based fabrication strategy produces reproducible, low-cost probes and offers an accessible approach for nano-DESI MS.

1. Introduction

Nanospray desorption electrospray ionization (nano-DESI) is a liquid extraction-based ambient mass spectrometry imaging (MSI) method, in which analytes are extracted into a dynamic liquid bridge between a specially designed probe and sample surface. The extracted analytes are then delivered to a mass spectrometer inlet where they are ionized by electrospray-like ionization [1]. Nano-DESI MSI has been used for spatial mapping and quantification of a wide range of analytes with high spatial resolution and sensitivity [2–12]. Several probe designs have been used in nano-DESI MSI. Conventional nano-DESI probes are assembled from two fused-silica capillaries positioned at ~90°. Despite the simplicity of capillary-based probes, the relative position of the capillaries can shift during an experiment, altering the size of the liquid bridge. More robust sampling has been achieved with microfluidic probes (MFPs) that integrate two microfluidic channels into a single device [13]. MFPs have demonstrated sensitivity and spatial resolution comparable to those of conventional capillary-based nano-DESI probes [14]. Moreover, owing to the stable liquid bridge, MFPs enhance the throughput of nano-DESI MSI by enabling high scan speeds of up to 400 μm/s [15].

We have developed several methods for fabricating MFPs [13] using either glass [16] or thermoplastic materials [17]. Glass MFPs offer superior solvent compatibility and longer lifespan [18], whereas thermoplastic MFPs provide a cost-effective alternative [17]. Cyclic olefin copolymer (COC) was chosen as the probe material because of its low cost, mechanical strength, and excellent resistance to polar solvents compatible with electrospray ionization [19]. The fabrication of thermoplastic MFPs involves three main steps: pattern imprinting, hot-press bonding, and trimming the outer shape of MFPs. Previously, we reported a fabrication process in which the channel pattern was imprinted using a metal wire [17], a method not suitable for mass production. Here, we present a detailed protocol for fabricating thermoplastic MFPs using custom-designed molds. These molds contain fixed microfluidic channel patterns that can be reused multiple times. We systematically evaluated multiple mold designs and MFP fabrication techniques to compare their advantages and limitations. Thermoplastic MFPs are well-suited for rapid imaging of analytes on surfaces with moderate spatial resolution and for high-throughput localized surface analysis.

2. Materials

The equipment and supplies listed here pertain specifically to the fabrication of the COC MFPs. Details of the equipment and supplies required for nano-DESI MSI experiments are provided in our previous work [20].

Reagents.

• HPLC-MS-grade MeOH (Fisher Chemical, cat. no. A452–4)

• HPLC-grade water (Fisher Chemical, cat. no. W5–4)

• Lysophosphatidylcholine (LPC) 17:1 (Avanti Polar Lipids, cat. no. 855677)

• Lysophosphatidylethanolamine (LPE) 17:1 (Avanti Polar Lipids, cat. no. 856707)

• MinuteWeld epoxy glue (J-B Weld Company, cat. no. 50101)

Equipment.

• Hot press machine (MTI Corporation, model YLJ-HP60).

• SSZ Binocular Stereo zoom microscope (Scienscope, model SZ-BD-B2).

• Ultrasonic cleaner (VWR, model B3500A-MT).

• Syringe pump (Legato 180, KD Scientific, cat. no. 788180)

• Inline micropositioners with micrometers (Quater Research & Development, model no. XYZ 500 MIM, inline).

• Magnetic base for XYZ 500 (Quater Research & Development, cat. no. 20232).

• Articulating arm with an alligator clip (Siskiyou, model no. MXB-3i, cat. no. 20200000E).

• Digital camera (Dino-Lite Digital Microscope).

Supplies.

• 0.24-mm thick cyclic olefin copolymer (COC) sheets (Röhm, PLEX-IGLAS^®/EUROPLEX^®, cat. no. 0F304).

• Glass microscope slides, 75 × 50 × 1.1 mm (Corning Inc., cat. no. 2947–75X50).

• Waterproof abrasive paper, 3000–5000 grit (TOA).

• Fused-silica capillary, 50 μm i.d. and 150 μm o.d. (Polymicro Technologies, cat. no. TSP050150).

• Cleaving stone (Polymicro Technologies, cat. no. 1068680064).

• MicroTight sleeve yellow 0.025-inch o.d. × 0.007-inch i.d. (Idex Health & Science, cat. no. F-181)

• Union assembly, MicroTight PEEK (Idex Health & Science, cat. no. P-720).

• Kimwipes^™ Delicate Task Wipes, 1-ply (Kimberly-Clark Professional^™, cat. no. 34155).

• Liquid blocker pen (Electron Microscopy Sciences, cat. no. 71312)

3. Procedures

Previously, we reported the fabrication of COC probes by bending a 100 μm-thick wire into a specific pattern for channel imprinting [17]. To improve robustness and enable the fabrication of smaller channels, we explored the use of molds to facilitate the imprinting process. A typical mold design, shown in Fig. 1a–c, consists of a patterned structure fixed on a base. Fig. 1a shows the overall structure of the mold. The mold contains a rectangular base and a patterned channel configuration. This pattern, used to generate microfluidic channels, comprises the fluidic inlet and solvent channel. Fig. 1b shows the enlarged schematic of the pattern. The fluidic inlet consists of a cuboid (20 × 160 μm²) topped with a half-cylinder (radius 80 μm) as shown in the cross-section view. The rounded shape of the wider section of the pattern is designed to fit the shape of the external glass capillary, thereby minimizing dead volume between the external glass capillary and the fluidic inlet channel. The solvent channel is a cuboid with a 50 × 40 μm² cross-section. Fig. 1c shows the 3D view of this pattern. The different heights of the fluidic inlet and solvent channel are critical for the fabrication of COC probes. If the heights of the fluidic inlet and solvent channel are the same, it is very difficult to insert the external glass capillary into the COC probe.

Fig. 1.

Design of two MFP molds fabricated by selective laser-assisted etching (SLE): the regular MFP mold (a-c) and the V-shaped MFP mold (d-f). (a) and (d) show the front view (left) and side view (right) of the mold layout: grey rectangle indicates the base; orange lines show the microfluidic channel pattern comprising the fluidic inlet and solvent channel. Length dimensions are labelled. (b) and (e) are enlarged views of the channel pattern showing its width and height; the wider section of the fluidic channel is interfaced with the external capillary. (c) and (f) are 3D views of the pattern.

We have also developed a mold for fabricating a V-shaped MFP. In this geometry, the sampling port and the spray emitter are physically decoupled, allowing solvent extraction and electrospray ionization processes to be optimized independently. For example, the spray emitter can be pulled to different degrees to optimize the stability and magnitude of MS signal, while the sampling port geometry can be tailored to achieve a stable, small liquid bridge. The V-shaped design is also well suited for pneumatically assisted MFPs, in which a nebulizing gas can be introduced near the secondary capillary to stabilize solvent flow through the probe, as described in our recent study [21]. Fig. 1d–f show a typical V-shaped mold design which can be used to fabricate V-shaped MFPs. This V-shaped MFP mold has a design similar to the regular MFP mold, except for the microfluidic channel pattern. The fluidic inlet is a cuboid (20 × 160 μm²) topped with a half-cylinder (radius 80 μm), while the solvent channel is a cuboid with a 50 × 40 μm² cross-section. The V-shaped MFP mold is designed to interface with two glass capillaries with the O.D. of 150 μm, one of which delivers a solvent and another one generates an electrospray plume at a mass spectrometer inlet.

The mold can be fabricated from glass or metal. Glass molds can be fabricated by photolithography or selective laser-assisted etching (SLE), whereas metal molds can be manufactured by machining or 3D printing. Fig. 2 shows photographs of molds fabricated using these methods, while Table 1 summarizes their performance. CNC milling is cost-effective but produces a rounded MFP apex (Fig. 2a), because sharp corners are difficult to machine. The rounded apex yields a larger liquid bridge, so this mold is suitable only for MFPs with lower spatial resolution (>100 μm). 3D metal printing is able to produce a sharper apex, but the feature size is limited to 0.4 mm (Fig. 2b). The surface of the 3D metal–printed mold is rough, which makes the MFP channel size non-uniform. Photolithography can produce very small features (<20 μm). However, it is difficult to produce 3D features (different heights or depths) using the photolithography method. The mold shown in Fig. 2c was fabricated in a cleanroom. The equipment usage is expensive, but a large volume of molds can be fabricated at once. In contrast, the SLE mold (Fig. 2d) offers superior performance, enabling small features, a sharp apex, and full 3D structures, albeit at higher cost. In this protocol, we use an SLE-fabricated mold as an example.

Fig. 2.

Photographs of molds fabricated using different methods: (a) CNC machining, (b) Direct metal laser sintering 3D printing, (c) Photolithography, and (d) SLE.

Table 1.

The performance of molds fabricated using different methods.

Mold Material	Fabrication Method	Smallest Feature	Cost	Robustness	Notes
Metal	CNC machining	100 μm	<$100	High	No sharp corners; apex is rounded
	3D printing	400 μm	$200	High	Channel size too large; surface is rough
Glass	Photolithography	<20 μm	Varies	Moderate	Hard to produce 3D features
	SLE	<20 μm	$200	Moderate	Glass may crack during heating and cooling

3.1. Fabrication of the mold

• 1Use Autodesk Inventor CAD software to design the mold as shown in Fig. 1. This design was developed for high-precision 3D fabrication methods such as selective laser-assisted etching and 3D printing. For CNC machining, the solvent channel dimension may need to be increased to 100 μm due to technical limits. Convert the CAD file to STEP format for outsourced fabrication.
• 2The mold fabricated by FEMTIKA UAB (Vilnius, Lithuania) using SLE is shown in Fig. 2d. This mold was used for the fabrication of the COC probe described below.

3.2. Fabrication of the regular and V-shaped COC nano-DESI MFPs

Since the fabrication procedures for the two types of MFPs are similar, the protocol below primarily describes the fabrication of COC nano-DESI MFPs. Differences specific to V-shaped COC nano-DESI MFPs are indicated where applicable.

• 3Clean the mold by sonicating it in 15 mL deionized water for 3 min in a 25-mL beaker. Dry with nitrogen.
• 4Place the cleaned mold at the center of a glass microscope slide with the patterned side facing upward as illustrated in Fig. 3.
• 5Use scissors to cut two 2 × 2 cm COC sheets. Remove the protective film from the first sheet and place it over the mold (see Note 1).
• 6Cover the mold and COC sheet with another glass microscope slide, lowering it vertically without sliding to maintain alignment.
• 7Preheat the hot press to 130 °C. Using tweezers, carefully position the two glass slides and place the entire assembly between the heated plates. Press the hydraulic handle until the heated plate contacts the glass slide. Then apply ~50 psi of pressure. Maintain this pressure for 5 min to imprint the mold pattern.
• 8Use tweezers to carefully remove the assembly out of the heated plate. Remove the glass slide and inspect the imprinted structure under a microscope to confirm a uniform channel imprint.
• 9Place the imprinted COC sheet with the channel side up back onto the glass slide. Remove both protective films from the second COC sheet and align it on top of the first sheet with the blue-film side contacting the other sheet. Cover with a second glass slide (See Note 2).
• 10Place the assembly back into the heat-press machine and position it at the edge of the heated plate, with the channel apex facing toward the center of the heated plate. This positioning ensures that only the lower portion of the channel is heated and bonded while the upper portion remains unbonded to allow subsequent capillary insertion. Press the hydraulic handle until the heated plate contacts the glass slide, then apply ~100 psi of pressure. Maintain the pressure at 130 °C for 30 s.
• 11Carefully take the assembly out of the heat-press machine. Trim the upper part of the chip with scissors to expose the end of the 160 μm diameter channel.
• 12Use a cleaving stone to cut one ~30 cm fused-silica capillary (i.d. 50 μm, o.d. 150 μm) and insert it into the open end of the channel until the tip reaches the end of the 160 μm diameter opening. For V-shaped MFPs: cut two capillaries (~30 cm primary capillary and ~1 cm secondary capillary).
• 13Place the probe with the inserted capillary between the glass slides and press at 130 °C for 2 min under 100 psi pressure to secure the capillary (see Note 3), as shown in Fig. 4a and b for a regular MFP and Fig. 4c and d for V-shaped MFP.
• 14Inspect the probe to ensure complete bonding with no internal air bubbles. The channel should remain clear and the capillary should fit properly without any gap between the end of capillary and the end of the 160 μm-diameter channel (See Note 4).
• 15Mix epoxy glue until uniform white color with bubbles appear. Quickly apply glue around the capillary entrance to seal the junction (See Note 5).
• 16Leave the probe undisturbed to cure at room temperature for at least 24 h.

Fig. 3.

Process of imprinting the channel pattern onto the COC sheet described in steps 4–7. (a) shows the process of imprinting regular COC nano-DESI MFPs and (b) shows the process of imprinting V-shaped COC nano-DESI MFPs.

Fig. 4.

(a) and (c) show schematic illustrations of the bonding procedure for two COC sheets imprinted with regular and V-shaped molds, respectively, with the external capillary. (b) and (d) show the bonded COC assemblies after thermal bonding. The apex of the channel serves as the sampling port. In (b), the end of the channel will become the spray emitter. In (d), a secondary capillary is used as the spray emitter in the V-shaped MFP.

3.3. Trimming of the COC probe

• 17Use scissors to cut the probe vertically to expose the channel at the spraying end, as shown in Fig. 5a. This step is not required for V-shaped MFPs.
• 18Connect the fused silica capillary attached to the probe to the MicroTight PEEK union using a yellow sleeve (Fig. 6), ensuring the capillary extends 1–2 mm past the sleeve. Tighten both ends securely.
• 19Load ~1 mL of the working solvent into a syringe and insert the syringe needle into the other end of the union (Fig. 6). Rotate gently to ensure a tight seal.
• 20Gently push solvent through the channel and confirm continuous flow from the spraying emitter (see Note 6).
• 21
  Trim of the spray emitter and sampling port (see Note 7).

  1. Use scissors to remove excess COC material as shown in Fig. 5b to obtain the shape of the probe shown in Fig. 5c. For V-shaped MFPs, remove excess COC material as shown in Fig. 5d to obtain the shape of the probe shown in Fig. 5e.
  2. Cut at ~45° towards the sampling port to form an ~ 90° tip. For regular MFPs, the spray emitter should also be trimmed to ~60°.
  3. Under a microscope, use a blade to sharpen the tip and ensure the channel opening is visible as a small white dot.
• 22Repeat the flow test (step 18–20) by gently pushing solvent through the channel and monitoring the sampling port. A small and continuous droplet should form at the sampling port (see Note 8).

Fig. 5.

The process of trimming the COC probe. (a)-(c) show the trimming process of regular COC nano-DESI MFPs. After trimming, the apex of the channel forms the sampling port, while the outlet of the channel functions as the spraying emitter. (d)-(e) show the trimming process of V-shaped COC nano-DESI MFPs. After trimming, the apex of the channel forms the sampling port.

Fig. 6.

A photograph of the assembly of the syringe, PEEK union, and the fused silica capillary of the COC probe.

3.4. Setup of the Nano-DESI experiment

• 23Prepare a 9:1 (v/v) methanol/water mixture with 2 μM LPC 17:1 as internal standard. This solvent efficiently extracts lipids and metabolites from biological samples and is fully compatible with COC. Other compatible solvents include water, short-chain alcohols, acetone, short-chain ketones, acids and bases. Nonpolar solvents such as toluene, chloroform, heptane and some polar solvents such as acetonitrile are incompatible with COC.
• 24Position two Dino-Lite digital cameras, one overhead and one from the side, to monitor probe operation and liquid bridge formation.
• 25Mount the COC probe onto an XYZ micropositioner using a clamp or tape as needed.
• 26Align the spray emitter with the centerline of the mass spectrometer inlet. For V-shaped MFPs, align the end of the secondary capillary with the inlet.
• 27Load the extraction solvent into a syringe and mount it onto the syringe pump. Ground the syringe needle (see Note 9). Repeat steps 19–20 to connect the glass capillary of the COC probe to the syringe needle.
• 28Set the solvent flow rate between 0.5 and 1.5 μL/min. Adjust the probe position to form a small and stable liquid bridge on the sample surface while monitoring the MS signal. If the liquid bridge is too big or the signal is unstable, repeat step 21 to achieve an approximately 50 μm diameter liquid bridge and a stable signal.
• 29The procedure for nano-DESI MSI experiments using the COC probe follows the same method as for glass MFPs described in detail in our previous work [16].

3.5. Notes and troubleshooting

1. Ensure that the blue-protective-film side contacts the mold to prevent contamination. Avoid exposing COC sheets to air for prolonged periods after film removal.

2. The mold is no longer present at this stage: the imprinted COC sheet now forms the lower layer (channel-side up) and the second, unpatterned sheet forms the upper layer (Fig. 4). This arrangement differs from the imprinting step (Fig. 3), where the single COC sheet is placed on top of the mold.

3. Keep the capillary away from the hot plate as direct contact with the heated surface can melt and break the capillary.

4. If air bubbles are present between the COC sheets adjacent to the channel, place the region containing the bubbles at the center of the heat plate and press again at 130 °C and 100 psi for approximately 1 min to remove the bubbles.

5. Epoxy sets in around 60 s, so seal the probe rapidly and do not seal more than three probes at one time.

6. If no flow is observed, examine whether the channel has collapsed or melted during thermal bonding.

7. The sampling port and spray emitter must be precisely shaped: the trimming of the sampling port directly determines the liquid bridge size, while irregularities in the spray emitter shape lead to unstable spray and reduced sensitivity.

8. When both ends are open, flow should come only from the sampling port when pushing the solvent through. If flow comes from both ends or only from the spray emitter, it indicates that the sampling port opening is too small.

9. For certain instruments (e.g. Orbitrap), apply the electrospray high voltage to the syringe needle instead of grounding it.

4. Representative results

We typically evaluate the performance of the fabricated COC nano-DESI MFPs by analyzing a standard drop-casted onto a glass slide. Here, we used a 100 μM solution of lysophosphatidylethanolamine (LPE) 17:1 in methanol as a standard. Three 2 μL droplets were drop-casted onto a glass slide in a 5 mm diameter circular boundaries drawn with a liquid blocker pen to prevent solvent spreading. The center-to-center distance between adjacent spots was 2 cm.

Nano-DESI analysis using the COC MFP was performed using a Bruker timsTOF mass spectrometer in positive ion mode. A 9:1 (v/v) methanol/water mixture was used as the extraction solvent. Mass spectra were acquired in a continuous line scan at 1 mm/s scan speed across the three consecutive lipid spots (Fig. 7a). The extracted ion chronogram (EIC) of m/z 488.27 corresponding to the sodium adduct of LPE 17:1 is shown in Fig. 7b. Three distinct peaks are observed, corresponding to the three LPE spots. The separation between the peaks demonstrates minimal carryover between the samples. Some differences in signals observed for different spots are likely due to minor variations in spot size and positional misalignment during drop-casting, which results in sampling different parts of the spots during the line scan.

Fig. 7.

(a) The experimental setup for evaluating the fabricated COC MFP. Three 2 μL droplets of 100 μM LPE 17:1 were drop-casted within 5 mm diameter circular boundaries, spaced 2 cm apart (center-to-center) drawn with a liquid blocker pen. (b) Extracted ion chronogram (EIC) of the sodium adduct of LPE 17:1 at m/z 488.27 obtained during a line scan across the three spots at 1mm/s, showing three well-resolved signals from three different spots outlined by the liquid blocker pen. (c) Mass spectrum outside of the sample spot showing signals of the solvent and COC probe. (d) Mass spectrum from the LPE 17:1 standard spot. The sodium adducts of LPE 17:1 at m/z 488.27 is clearly observed with minimal background interference. Some polymer signals observed in the spectrum are likely due to the ink of the liquid blocker pen.

Mass spectra from the background region (solvent only) and from the LPE standard spot are shown in Fig. 7c and d. The background spectrum (Fig. 7c) exhibits low chemical noise and no significant interference at m/z 488.27. When the probe reaches the LPE spot (Fig. 7d), a strong signal at m/z 488.27 is observed, confirming efficient extraction and ionization of the analyte and indicating that the thermally bonded COC MFP does not introduce significant contamination or chemical background.

5. Conclusions

In summary, we developed a mold-based method for COC MFP fabrication to replace the traditional metal wire hot embossing approach. We evaluated four mold fabrication techniques and found that the SLE mold provides the best performance. The fabrication process for a single probe takes within 10 min (excluding glue drying time), and currently one probe can be fabricated at a time. Detailed fabrication procedures for the COC MFP using the SLE mold are provided in this protocol. Performance evaluation using the LPE 17:1 standard demonstrated stable liquid bridge formation, reproducible signal response, clear spatial separation between adjacent sampling regions, and low background interference, confirming the reliability of the fabricated probes for nano-DESI analysis. The lifespan of a COC probe depends strongly on the analytes and sample matrix, but a single probe can typically be used for 10–20 h of continuous data acquisition. Performance is generally consistent from probe to probe and across different fabrication batches; after normalization to the internal standard, the signal response shows good reproducibility between probes. Although these probes are well suited for nano-DESI MSI experiments, the spatial resolution currently achievable with COC MFPs is around 100–150 μm. The spatial resolution is primarily limited by the size of the liquid bridge, which is affected by the channel dimensions and the geometry of the trimmed sampling port. Further improvement in spatial resolution will therefore require probes with narrower channels, which is a direction of our ongoing work. In our laboratory, COC MFPs are used to analyze complex biological and environmental samples, including direct detection of microbial colonies on agar and environmental biofilms on a variety of surfaces such as aluminum, chemical agent resistant coating (CARC), and silicon wafers. This protocol of low-cost COC MFPs will make nano-DESI more accessible to the community.

Data availability

Data will be made available on request.