Microfluidic tools for multidimensional liquid chromatography Ianovska, Margaryta
IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to cite from it. Please check the document version below.
Document Version
Publisher's PDF, also known as Version of record
Publication date: 2018
Link to publication in University of Groningen/UMCG research database
Citation for published version (APA):
Ianovska, M. (2018). Microfluidic tools for multidimensional liquid chromatography. University of Groningen.
Copyright
Other than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons).
Take-down policy
If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim.
Downloaded from the University of Groningen/UMCG research database (Pure): http://www.rug.nl/research/portal. For technical reasons the number of authors shown on this cover page is limited to 10 maximum.
Chapter IV
Fabrication of a pressure-resistant
microfluidic mixer in fused silica
using Selective Laser-Induced
Etching
Margaryta A. Ianovska
1,2, Jean-Paul S.H. Mulder
1, Martin Hermans
3, Elisabeth
Verpoorte
11Pharmaceutical Analysis, Groningen Research Institute of Pharmacy, University of
Groningen, The Netherlands
2 TI-COAST, Amsterdam, The Netherlands 3 LightFab GmbH, Aachen, Germany
Abstract
We report a microfluidic mixer fabricated in a solid block of fused silica using Selective Laser-Induced Etching (SLE). The micromixer contains herringbone grooves (HG) that induce mixing based on chaotic advection, as investigated in our previous work. The chip was designed for utilization in the interface between two columns in a multidimensional liquid chromatography system, which implies the utilization of pressure-resistant devices. Our first chips were hybrid devices made from poly(dimethylsiloxane) (PDMS) and glass. These devices could not withstand pressures higher than 10 bar, due both to the elastic properties of PDMS and a lack of a robust bond between PDMS and glass. We therefore opted for utilization of a relatively new technique, namely Selective Laser-Induced Etching (SLE), as a route to making monolithic devices in a single block of a rigid material. Our material of choice was a block of fused silica. This eliminates the need to bond a structured chip to a chip acting as the lid for a microfluidic device. Moreover, fused silica is four orders of magnitude more rigid than PDMS and can thus withstand higher pressures. We aimed to fabricate in silica, for the first time, microfluidic mixers with channels up to 33 mm long containing arrays of microgrooves. Fabrication proved challenging, as removing laser-modified silica from the patterned channels by the introduction of etchant from the channel ends meant longer exposure to etchant at the beginning and end of the mixing channel. This resulted in overetching of the channel and grooves in the end regions with an accompanying loss of groove resolution. In order to solve these fabrication issues and to account for differences in etch progression in different device regions, we have made use of an adjusted design that provided improved mixing performance. The pressure tests showed that the fused-silica chips can withstand pressures of up to 85 bar and can be used in the interface between two columns of a multidimensional liquid chromatography system to facilitate the fast adjustment of mobile phase composition.
Keywords: micromixers; pressure-resistant chips; fused silica chips; Selective Laser-Induced
Introduction
The growing interest in microfluidics over the last few decades has led to the development of many different techniques and methods for fabrication of microfluidic devices in scientific settings in both academia and industry. Miniaturization imparts such advantages as reduced consumption of reagents, shortened analysis times, and the possibility to have good control over flow conditions, as well as mass and heat transfer.1 This makes microfluidics an attractive field
for flow chemistry, materials sciences and also as a means to realize components to improve state-of-the-art analytical separation techniques such as high performance liquid chromatography (HPLC). Very often such applications require utilization of high pressures. Therefore, development of different types of pressure-resistant microfluidic systems recently became a new trend in the microfluidics field.1
There is a big variety of materials available for chip fabrication, such as silicon, glass and elastomeric polymers. Polydimethylsiloxane (PDMS) has gained wide acceptance in the academic microfluidics community due to its low cost, robustness, route to simple device fabrication, optical transparency and non-toxicity.2 In previous work,3 we developed a 1.6-μL microfluidic mixer that provides good mixing within seconds at flow rates compatible with LC×LC (0.1-1 mL/min). This chip was designed to be placed in the interface between two columns of a multidimensional liquid chromatography system and so had to withstand high pressure pulses (up to 200 bar), which arise from valve switching and additional back pressure from the second column. In order to provide mixing, the device contained an array of herringbone-shaped grooves (herringbone grooves, HG) with a depth of 50 µm and a width of 110 µm. The presence of these grooves led to the generation of two counter-rotating vortices by chaotic advection.4 This mixer was fabricated in PDMS bonded to a glass plate. However, the elastomeric nature of PDMS and its low Young's modulus become a significant problem for the development of a chip that could withstand high pressures. In our experience, device failure occurred above pressures of 10 bar either in the PDMS itself (mostly) or at the interface between the PDMS and glass. Even at low flow rates, significant channel deformation can occur, which leads to alterations of the flow profile and subsequently to changes in device performance.2
Because of problems mentioned above, other materials are used in order to fabricate
Ch
ap
te
r
and silicon.8–11 Liu et al.5 developed a COC chip containing in situ photopolymerized polymethacrylate monolithic stationary phases for HPLC separations of fluorescein-labelled intact proteins. This chip was fabricated by hot embossing and can withstand 200 bar. Another chip suitable for liquid chromatography was demonstrated by Chen et al.6 with a maximum
burst pressure (the pressure that a device can withstand before failing) of 400 bar. It was also fabricated using COC but by direct microscale mechanical milling. Several examples of silicon/pyrex10 and glass/glass11 microreactors that can be used at pressures up to 300 bar have been presented. Silicon/pyrex chips have been fabricated using deep-reactive-ion etching (DRIE) followed by anodic bonding of the twp layers,10 whereas for fabrication of glass/glass microreactors11 wet etching and direct fusion bonding were used. Recently, an alternative method for inexpensive rapid prototyping based on off-stoichiometry thiolenes (OSTEs) was demonstrated by Martin et al.1 The chip, fabricated using UV-curable OSTE and bonded to glass, could withstand 200 bar, and was used to perform multiphase flow visualization studies in microchannels. However, the chip fabricated in this study had a square cross-sectional channel of 200 µm without any features inside.
Multiple methods for fabrication of microfluidic devices exist, each with unique advantages and drawbacks. Injection molding and hot embossing can be considered as fast but expensive methods for polymer prototyping due to the high initial cost of making the molds. On the other hand, glass/silicon micromachining processes by wet/dry etching create good- precision structures but are technically demanding and time consuming to fabricate.2 Direct
fabrication methods such as micromilling and laser ablation, though cost-accessible and enabling complex 3D-multilayer structures, have low resolution (around 50 µm). They also generate surface roughness, and fabrication has limited throughput due to the inherent serial nature of the fabrication process.2 All these methods suffer from one inherent drawback: they create a 2D-open channel network on one substrate surface that has to be sealed (closed or bonded) to a second chip in order to obtain a microfluidic channel. This creates a weak point which manifests itself as bond breakage in any high pressure application, as mentioned above. Developed more than a decade ago, Femtosecond Laser Irradiation followed by Chemical Etching (FLICE),12 also called Selective Laser-Induced Etching (SLE),13,14 has emerged as a novel powerful approach for direct fabrication of complex 3D structures inside a solid transparent material such as fused silica. The SLE technique consists of two steps: 1) the exposure of glass to scanning focused ultra-short (fs or ps) pulsed laser radiation, which locally
changes glass properties in the focal volume to create self-aligned nanocracks perpendicular to the laser polarisation direction;15 2) etching of the laser-modified zone by HF or an alkaline solution such as KOH in water.13,15 During the etching process, the nanocracks created by the pulsed laser act as channels through which the etching agent diffuses deeper into the fused silica. Etching takes place where the etching agent comes into contact with modified fused silica along the diffusion path.15 Being a direct fabrication technique inside a solid piece of material,
SLE provides an appealing solution to avoid microchannel-sealing or chip-bonding steps during device fabrication. It also doesn’t require complex cleanroom facilities and allows for the fabrication of complex 3D structures.15 The degree of feature resolution and the aspect ratios
possible in glass and silica are higher with SLE than with wet etching. SLE thus allows the exploitation of the unique properties of glass (transparency, rigidity, inertness etc) in devices having smaller and better defined features than previously was possible in glass.
However, the SLE approach has some limitations regarding channel length, shape and aspect ratio. As the channel is etched after patterning by starting from one end, it is necessary to continuously remove the reaction products and provide fresh etching agent to diffuse along the channel. However, as the channel length increases, the amount of fresh acid able to reach the end of the channel reduces and the etch rate gradually decreases,16 i.e. the etching process saturates at longer etching periods.17 This saturation leads to microchannels with lower aspect ratios and/or channels with conical shapes (tapered channels), geometries which become more pronounced the longer channels get. The longest dead-end channels reported were about 1.8 mm long, and had an aspect ratio (length-to-hydraulic diameter ratio4) of ~ 20.18 This effect can
be reduced by etching the microchannel simultaneously from opposite ends. Vishnubhatla et
al.19 managed to obtain a 4-mm-long microchannel with an aspect ratio of 4 by etching from both ends of the channel in an ultrasonic bath containing a 20% solution of HF in water for 4.5 hours.
Studies have shown that the depletion of the HF acid toward the center of the etched microchannel and the difficulty of replenishing it in this region often leads to self-termination of the etching process.15 Moreover, HF is an isotropic etching agent removing material laterally at a similar rate to the speed of downward etching, and the selectivity of the HF for the
4An aspect ratio in this work is calculated not as the ratio of the width to the channel height but as the channel
Ch
ap
te
r
modified region with respect to unmodified material is not sufficient.16 This results in limited channel lengths (about 1.5 - 2 mm)20 and limited length/diameter aspect ratios. In addition,
water, which is formed during the HF action on silica, dilutes the HF acid, and further impeding the etching process.21 On the other hand, these effects were not observed when aqueous KOH was used as the etching agent.16 Having a higher selectivity for modified fused silica due to the formation of a Si-rich structure16 and being an anisotropic etching agent, aqueous KOH provides slow etching with almost constant selectivity regardless of the etching period.15,16 It was reported that the selective etch rate with KOH is even higher (14 times higher) than for etching with ~2% HF.14 Thus, using prolonged 60-hour etching in KOH, Kiyama et al.16 fabricated 10-mm microchannels with less than 60 μm diameter (an aspect ratio of almost 200).
In this work, we aimed to fabricate a 30-mm-long microfluidic mixing channel having an aspect ration of more than 100 in fused silica by the Selective Laser-Induced Etching method. Besides the problems that arise when such a long channel is to be fabricated (as described above), the need to have herringbone grooves on the channel wall in order to generate mixing complicate the fabrication process. The fabrication of such chips has never been explored before using the SLE technique. In the current study, we describe several generations of chips fabricated in fused silica with different dimensions. In order to solve some fabrication issues and account for differences in etch times in different device regions, we exploit a modified micromixer design incorporating compensatin structures to counteract overetching in regions where the silica is exposed to HF etchant for longer periods. We refer to this design as a compensation design. Compensation structures incorporated into regions subjected to longer etchant exposure were designed so that longer HF treatment was required to yield the final features having the desired dimensions. The fabricated micromixer with this design showed improved mixing performance compared to previous chip generations where overetching had (partially) eroded the grooves and made channels undesirably wider. Also, pressure tests showed that fused silica chips can withstand pressures of up to 85 bar.
Material and Methods
Chip designsAll chip designs were drawn in SolidWorks© (Waltham, Massachusetts, USA) and saved as a STEP file for further fabrication. The SolidWorks design of the first fabricated chip with different relevant dimensions is given in Figure 1A. We refer to it as the 1st generation chip as it forms the basis upon which other devices were subsequently designed. All dimensions of this chip are summarized in Table 1. A schematic representation of herringbone grooves (HG) in the channel with names of different parameters is shown in Figure 1C. The 1st generation chip had the same herringbone groove dimensions as were used in our previous work.3 However, the channel dimensions are different. Moreover, the inlets and an outlet were designed as standard 10-32 female HPLC connectors. The midpoint of the chip channel cross-section is aligned dead centre with the midpoint of the 1/16” ID peek tubing fixed in the inlets and outlet.
Figure 1. The SolidWorks© design of the (A) 1st and (B) 2nd generation microfluidic mixer with herringbone
grooves; all dimensions are in mm. (C) Schematic drawing of herringbone grooves in the channel.
Ch
ap
te
r
Table 1. The channel parameters of the 1st and 2nd generation chips.
Parameter 1st generation 2nd generation
Channel length, mm 31 33.6
Length of the inlet, mm 9.3 2.09
Hydraulic diameter, mm5 0.161 0.316 Inlet width, µm 150 215 Channel height, µm 60 150 Groove depth, µm 50 100 Channel width, µm 300 430 Groove width, µm 110 260 Ridge width, µm 50 70 Volume, µL 1.6 3.6
The first fabricated chip revealed that the chosen fabrication method is not suitable for the fabrication of well-resolved grooves having dimensions smaller than 100 µm. Based on these results, all dimensions of the second chip (2nd generation chip) were increased (Table 1). We decided to increase the channel height and channel width to 150 µm and 430 µm, respectively, in order to provide better access for etching agent to penetrate into the channel. Based on these values, the other parameters were recalculated following the protocol for achieving optimized geometry22 (Table 1), which is described in our previous work.3 Moreover,
the Y-junction was replaced by the T-junction (Fig. 1B) and the length of the inlet channel was shortened significantly, for reasons which will be discussed later.
Chip fabrication
All chips were fabricated in quartz glass (fused silica) by Selective Laser-Induced Etching (SLE) using the LightFab 3D Printer at LightFab (Aachen, Germany).13,14 The thickness of the fused silica was 7 mm with optically polished surfaces. The FCPA laser (Satsuma, Amplitude Systemes, Pessac, France) provided ultrashort laser pulses having a wavelength of 1030 nm, with a pulse duration of 1000 fs, a pulse energy of about 500 nJ and a writing velocity of 200 mm/s at a repetition rate of 750 kHz. Laser radiation is focused by a 20x microscope objective with a numerical aperture of 0.45 (LCPLAN N 20x/0.45 IR, Olympus Europa GmbH; Hamburg, Germany) equipped with a collar for correction of spherical aberrations.
The chip designs that were saved as STEP files were opened in the CAM software LightFabScan. The laser parameters, the three linear axes and the three dynamic axes in the 3D Microscanner were controlled using the same software. Vectors were generated automatically by SliceLas (available from LightFab) in the CAD software Rhinoceros 3D (from Rhino3D) and transferred to the CAM software LightFab Scan.
For development of the structures by wet-chemical etching, the laser-heated silica device was immersed in an aqueous solution of 8 mol/L KOH for 10 days at 85 ̊C with ultrasonic excitation. The heavy-duty ultrasonic bath was equipped with a 99 h timer, automatic heating and cooling with temperature control, and automatic water refill to compensate for evaporation.
Characterization of channels and grooves
Photos of the fabricated grooved channels were obtained using a microscope (model “DMIL”, Leica Microsystems, The Netherlands) equipped with a 40x magnification and a CCD camera. Channel and HG dimensions were measured using ImageJ to analyse the photos (U. S. National Institutes of Health, Bethesda, Maryland, USA). These values were then compared with the dimensions of the original Solidworks design. The data so obtained was plotted in OriginPro 9.1.0 (OriginLab Corporation, Massachusetts, USA).
Mixing test
In order to visualize mixing, the commercially available solutions of two food dyes, Brilliant Blue (BB, E133, KoepoE, Indonesia) and Tartrazine (Tz, E102, KoepoE, Indonesia), were introduced from separate inlets into the channel junction with a 5-mL syringe (B.braun, The Netherlands) through a 1.59-mm (od), 0.8-mm (id), polyetheretherketone (PEEK) tubing (Kinesis Ltd, Cambridgeshire, UK) using a syringe pump. Photos were taken at different positions along the channel with a camera (Canon EOS 700D) that was mounted on the microscope (Leica S8 APO, Leica Microsystems, Germany).
To ensure that micromixers having different internal dimensions and volumes could be tested and compared under the same flow conditions, the dimensionless Péclet number (Pe) was used to calculate appropriate flow rates for mixing experiments. Pe characterizes molecular
Ch
ap
te
r
mass transport in flow conduits as a ratio of advective transport (flow) rate to diffusive transport rate, can be calculated according to the formula:
𝑃𝑒 = 𝑣𝑑ℎ
𝐷 (1),
where v is average linear velocity (mm/s), D represents diffusion coefficient (mm2/s) and dh denotes the hydraulic diameter for a rectangular duct (e.g. equivalent diameter of a
channel, mm). This latter parameter can be calculated according to Equation 2,
𝑑ℎ = 2𝑤ℎ
𝑤+ℎ (2),
where w (mm) is channel width, h is channel height (mm).
Because the hydraulic diameter of the 2nd generation chip is 0.316 mm, which is almost
twice as large as in the 1st generation, mixing was tested under constant Péclet-number
conditions rather than constant flow rates to ensure the same mass transport conditions (Table 2).
Table 2. Tested flow rates based on Pe calculation for chips with different dimensions; (d+h)1
= 110 µm; (d+h)2 = 250 µm, where d is groove depth; dh1 = 0.161 mm (w = 300 µm), dh2 =
0.316 mm (w = 430 µm), where dh is hydraulic diameter; ρ = 103 kg/m3, µ = 10-3 kg/(m*s), DBB
= 2.8×10-9 m2 /s (for BB),23 D
Tz ~10-9 m2 /s (for Tz).24
Chip generation Flow rate, mL/min Pe, 103
1st 0.12 3.5 2nd and 3rd 0.2 1st 0.6 17.5 2nd and 3rd 1.0
All experiments were performed after conditioning the channel with 0.6 mol/L NaOH (Sigma-Aldrich, Sweden, Missouri, USA) and 18 M-ohm ultrapure water (Arium 611, Sartorius Stedim Biotech, Germany).
Pressure test
To evaluate the burst pressure, the inlets of the 2nd generation chip were connected to HPLC pumps (Waters 515, Waters Corporation, Massachusetts, USA) and the outlet was connected to an HPLC column (HYPERSIL SAX 5U 4.6×150 mm, Alltech) using standard HPLC fitting (Fingertight Fitting One-Piece PEEK, Upchurch Scientific, IDEX Health & Science, CA,
USA), as shown in Figure 2. In the first set of experiments, the column was open and the tested flow rate range was set to 0.2-2.0 mL/min. Afterwards, the column was closed with threaded stopper to build up the pressure. The water was pumped at a total flow rate of 0.5 mL/min.
Figure 2. Set-up for the pressure test: A 2nd generation micromixer was connected to HPLC pumps and a sealed
column. Ch ap te r IV
Results and Discussion
Fabricated fused-silica chipsThe fabrication of long channels with uniform cross-sections using the SLE technique has to date been limited as was mentioned in the Introduction. However, the appealing idea to directly fabricate buried channels in a solid block of fused silica, thereby circumventing the need to seal channels in a bonding step, motivated us to explore this technique in more detail. In our previous work,3 we developed and investigated a microfluidic mixer with herringbone grooves, that was destined to be used in multidimensional chromatography. Our experiments showed that when fabricated in PDMS/glass, such chips cannot withstand pressures more than 10 bar, due to the low Young's modulus of PDMS (a measure of the ability of a material to withstand changes in length (elasticity) when under lengthwise tension (pressure)). The value of the Young's modulus
lies in the range between 0.57-3.7 MPa for PDMS and 73 GPa for fused silica. Therefore, we aimed to fabricate the micromixer in fused silica, which is three to four orders of magnitude less elastic. The first generation chip had the same dimensions as the PDMS/glass chip reported previously by our group.3 Because of the possible difficulty of fabricating the 50-mm-long
channel from the original design, the channel length was decreased to 30 mm. Thus, the channel aspect ratio (length to hydraulic diameter ratio) of the 1st generation chip turned out to be 186.
Figure 3. Photos of the 1st and 2nd generation chips with herringbone grooves fabricated in fused silica: (A),(B)
whole chips and (C),(D) their channels imaged using a microscope. Images have been stitched together to show the full channel; a 40×magnification objective lens was used for (C) and (D). The scale of the images in (C) and (D) is the same.
Figures 3A and 3C show the whole 1st generation chip and its magnified channel, respectively. A clear difference is observed between the laser-modified (white region) and overetched channel (black region) material observed (Fig. 3C). Unfortunately, the channel quality is not particularly good and the resolution of grooves is poor. The channel diameter varies along the channel, with a larger cross-section observed at the entrance as compared to the middle part of the channel (width 300 µm and depth 60 µm). As was discussed above, it is an inherent feature of the etching process itself. Only in the middle part of the channel (Fig. 3C, marked in red) did we observe channel parameters that correspond to the original design, with well-resolved grooves. Such results were predictable due to the typical problems associated with etching. The longer the channel is, the more etching time is needed to reach laser-modified material in the middle of the channel. This means that both ends of the channel where laser-modified material had already been removed remain in contact longer with etching solution than section towards the middle of the channel. Even though we used KOH as an etching solution, the etching selectivity of which remained almost constant regardless of the etching time, some excessive etching took place. For comparison, Kiyama et al.16 fabricated 9.2-mm microchannels with less than 60 μm diameter and with aspect ratio of 153, which is very similar to the one we were aiming for (186). The etching time was 60 hours when 10 mol/L (35.8%) aqueous KOH was used. This is 4 times shorter than the time that we used, which is not surprising given that the microchannel length and hydraulic diameter were ~4 times shorter and 5 times smaller respectively than in our case.
Nevertheless, we decided to increase the cross-section of the 1st generation chip in order to simplify the introduction of the etching solution into the mixing channel by using wider inlets (Table 1). New dimensions for the 2nd generation chip were calculated based on the protocol for optimized geometry.22 Figures 3B and 3D show the 2nd generation chip and a top-view of its channel under the microscope. This chip has almost twice as large a channel diameter, with a channel aspect ratio of 104, compared to the 1st generation chip. In general, the channel quality is better and grooves are visible along the full channel length. It is also clearly visible that the section with HG having good resolution is significantly longer for the 2nd generation chip (Fig.
3D, outlined by a red rectangular) than for the 1st generation (Fig. 3C).
Ch
ap
te
r
Figure 4. Micrographs of the (A-C) 1st and (D-F) 2nd generation channels: (A, D) inlet; (B, E) junction and (C, F)
middle part which correspond to the original design with clearly resolved HG. Red lines (B, E) indicate the border of the channel according to the original design. All photos are made with the same magnification.
In addition to increased channel dimensions, the 2nd generation chip was designed with T-junction (Fig. 3B, 4D, 4E) instead of Y-junction (Fig. 3A, 4A, 4B). Because the etching solution enters the chip through inlets, the etching time proportionally increases with their lengths. The T-junction provided a 4-fold shorter inlet length, providing easier access for KOH to enter the channel. The clear improvement in etching of the middle part of the 2nd generation chips can be also seen in Figure 4F compared to the 1st generation chip (Fig. 4C) (photos are made with the same magnification.).
It is important to mention that besides the difficulties associated with fabrication of channels longer than few mm using the SLE technique, the chip fabrication in our case is complicated by the inclusion of herringbone groove arrays on the bottom of the channel, which are essential for generation of mixing based on chaotic advection. It should be noted that the fabrication of herringbone grooves having a rectangular cross-section is impossible with conventional photolithographic patterning and HF etching in glass. This is due to the isotropic nature of the HF etching process which precludes high-aspect-ratio channels. SLE provides the possibility to fabricate channels with a high-aspect ratio that have well-defined grooves in the glass surface. However, to the best of our knowledge, no other groups have fabricated anything other than long straight or bent channels having no internal features. The only attempt to fabricate a similar fused-silica herringbone mixer by femtosecond-laser direct writing combined with wet etching using HF was proposed by Lin et al.25 Several 2D and 3D designs were fabricated, one of which contained walls that were also patterned with slanted grooves.
However, in that study, the surface of the fused silica substrate was irradiated to form the grooved channel, which was then sealed to PDMS. Lin et al.25 thus demonstrated the use of SLE to produce arbitrary patterns on the vertical side walls, but didn’t exploit the most attractive feature of SLE technique, namely, the elimination of the sealing/bonding process by fabricating the full channel inside a block of fused silica.
Figure 5. Side-view of the 2nd generation chip with regions where ridges are clearly visible (middle part) and
where they disappear (at the beginning and the end of the channel). These are photographs taken of a tilted channel/device.
Grooves in the micromixer are defined by ridges that separate grooves from each other. If ridges are “eaten away” by etching solution during fabrication, there will no longer be any grooves left over by the end of device fabrication to induce any mixing. Figure 5 presents the side view of the 2nd generation chip to illustrate this situation. Ridges have disappeared at the beginning and end of the channel, as a result of excessive etching over several days. This means that mixing based on chaotic advection will occur only in the middle part where the grooves still have good resolution, i.e. only within 3 mm out the total 33 mm of the channel length (see Section 3.3).
Compensation design and the 3rd generation chip
In order to obtain channels with constant cross-section, we decided to make a design that compensates for the excessive etching in the beginning and at the end of the channel. Mixing channels were designed to have varying widths and depths, with narrower and shallower regions at the ends to allow for more etching in these regions. Both channel width and depth increased
Ch
ap
te
r
micrographs of the 2nd generation chip channel (Fig. 6). We made two measurements every 0.4 - 0.45 mm along the channel, as shown in Figure 6. One value was obtained for laser-modified (white) regions, w1, while the other was made to include etched (dark) regions, w2, respectively. The measured data was plotted as presented in Figure 7.
Figure 6. Measurements of the channel width at different positions along the channel of the 2nd generation chip
(a) at modified (w1) and (b) etched (w2) areas at (1) junction; (2) 4.5 mm; (3) 10 mm (4) 13 mm; (5) 16 mm (6) 21 mm and (7) 27 mm along the channel. Red line across the channel shows the location of the measurement.
Figure 7 reveals five distinct regions for etching behaviour: (I) a large difference between w1 and w2 is observed here; (II) exhibits a monotonic decrease in the difference between w1 and w2 as channel distance increases, (III) w1 ≈ w2; (IV) shows a linear increase in the difference between w1 and w2 and (V) exhibits almost constant w2. Also noteworthy is the fact that w1, the width measured across the laser-modified region of the final channel, remains constant along the entire length of the channel. Only in region (III), where modified and etched areas overlap, do both channel width and groove parameters equal those in the original design. This area comprises less than 3 mm (12 %) of the total 30-mm length of the channel. Based on these data, we created a compensation design (Fig. 8A) with a channel that has five regions with different channel parameters.
The final channel design has five regions (Fig. 8A) where different changes to the original design were made, including channel width and channel depth. Table 3 summarizes dimensions in each region of the compensation design. The maximum difference between w1 and w2 was calculated to be 190 µm. Thus, the channel widths in Regions I and V have been assigned values of 240 m rather than 430 m as in the original device. This resulted in the channel with narrower region in the beginning and at the end comparing to the middle region. In Region II, the channel width increased linearly from 240 m to 430 m, whereas it was
reduced linearly from 430 m to 240 m in Region 4. In general, a channel has a narrower width of 240 µm in the beginning and at the end compared to the 430 µm in middle section, where the original design was kept unchanged. The channel was deepest in the middle section of the channel (Region IV) (150 m; Fig.8D) and was designed to become shallower towards the ends of the channel.
Figure 7. Difference between channel width of modified (red curve, w1) and etched (black curve, w2) regions based on the channel of the 2nd generation chip: (I) increase in difference between w1 and w2; (II) linear decrease
of difference between w1 and w2, (III) w1 ≈ w2; (IV) linear increase of difference between w1 and w2 and (V) constant w2.
An important consideration in the compensation design was to adjust groove dimensions (e.g. groove width, ridge width and groove depth) to counteract effectscaused by excessive etching in the channel. This is especially true in regions (II) and (IV), where channel width linearly decreases towards the middle of the channel and increases again afterwards. Channel depth increases as the middle section of the channel is approached, and decreases again once the midpoint of the channel has been passed. As with the channel, the HG design resemble the originally designed HG only in the middle section (Fig. 8B). All grooves touched the walls of the new channel. In addition, groove depth was also adjusted according to the changing channel
Ch
ap
te
r
depth. All HG dimensions for each region are summarized in Table 3. Figures 8C and 8D show the groove depth in the beginning and in the middle section of the channel. The etching process also influences groove width as can be seen on Figure 6. Therefore, we measured the difference in groove width the same way as was described above for the channel width (Fig. 6). In the original design the groove and ridge widths are 260 µm and 70 µm respectively. In order to keep a groove width of 260 µm after etching, grooves in the compensation design were narrowed to different extents in all regions except the middle section of the channel (Table 3). The same approach was taken for ridges that decrease during etching. Therefore, in order to maintain the same ratios between grooves and ridges, we kept the distance taken up by each set of groove + ridge along the channel the same, at 330 µm.
Figure 8. SolidWorks© design of the compensation chip design with herringbone grooves representing different regions of the channel with different: (A) channel width; (B) groove and ridge width (dark regions); groove depth at the (C) beginning (70 µm) and (D) middle part (100 µm) of the channel.
Table 3. Five regions of the compensation design with different parameters. Region in the channel, mm Groove width, mm Ridge width, mm
Channel width, µm Groove depth, µm
0 – 6.5 0.2 0.13 240 70 6.5 – 12.7 0.22-0.24 0.11-0.1 240-430 (linear increase) 70-100 (linear increase) 12.7 – 15.2* 0.26* 0.07* 430* 100* 15.2 – 25.2 0.23-0.21 0.10-0.12 430-240 (linear decrease) 100-70 (linear decrease) 25.2 – 42.3 0.2 0.13 240 70
Based on these adjustments, we designed the chip with compensation structures (the 3rd generation chip) which is presented in Figure 9A. We noticed improvement in terms of uniformity of the channel width compared to the previous designs. The measurements of channel widths in the laser-modified and etched parts of the 3rd generation chip (Figure 10) clearly show that the middle section having the desired width (430 µm) is ~ 6 mm long, which is larger than for the 2nd generation chip (2.5 mm). Squeezing the two sides of the channel even
more could help obtain channels with an even more uniform cross-section. On the other hand, narrowing the beginning of the channel too much is undesirable due to the associated reduction of the etching solution access and, thus, an increase in the etching time.However, etching didn’t quite proceed as we had assumed or hoped. In regions (I) and (V), the maximal difference between the modified and etched parts still remains more than 300 µm. These regions better resemble the compensated design than the desired original design with a narrower channel in the beginning and at the end of the chip.
Figure 9. Photos of the 3rd generation chip with compensation design: (A) top-view of the whole channel, 40x
magnification objective lens and (B) side-view showing the difference in channel depth.
As with channel width, groove shapes have also been altered to take compensation for longer etching times into account. compensated design (wider in the beginning and at the end of the channel). Besides, Figure 9B reveals clear variations in channel depth. Probably, the difference of 30 µm in depth between the beginning/end and the middle part of the channel was not sufficient to compensate long etching times and to obtain uniform channel depth. However, the side-view of the 3rd generation chip shows that the resolution of the HG is better compared to 2nd generation chip. Ridges are present along the whole channel length, which probably
Ch
ap
te
r
Our approach for compensation of the Solidwrks designdesign for the conical channel shape is somewhat similar to the method proposed by Vishnubhatla et al.19 That approach
consisted of an irradiation of a reverse conical-shaped channel with respect to the one normally obtained with the SLE technique. However in our approach we decided to make changes to the actual device design and not to the SLE procedure. It is also worth mentioning that the method proposed by Vishnubhatla et al. would be difficult to realize in our chip due to the presence of herringbone grooves on the bottom.
Another possible solution to decrease the influence of excessive etching could be to cycle the etching process,17 i.e. by frequently interrupting it. Acid or base would then be replenished in the narrow modified regions during each cycle to decrease the effect of excessive etching.
Figure 10. Difference between channel width of laser-modified (red curve, w1) and etched (black curve, w2) regions of the 3rd generation chip channel. The marked regions represent the five regions of the original design
(Table 3).
Evaluation of mixing performance
All chip generations were tested for their mixing performance. As the mixing is enhanced by the herringbone grooves on the bottom of the channel, the mixing quality relies on how good the resolution/quality of the obtained grooves is. In other words, how well the original design of HG was transferred into the fused-silica chip is of critical importance. We used a qualitative
approach with food dyes to visualize the mixing. Because of different channel dimensions, the mixing performance was tested at different flow rates according to the same Peclet number representing the same mass transfer conditions (Section 2.4).
Figure 11 and Figure 12 show results for mixing experiments with all three chip generations at Pe values of 3.5×103 and 17.5×103, respectively. For both Pe, the best mixing is observed in the 3rd generation mixer, as indicated by the appearance of green colour across the
entire channel much earlier. Such performance can be attributed to several factors. Though the green colour starts to appear within first few mm of the channel, this is probably caused not only by the existence of grooves. As was discussed above, the depth of the 3rd generation chip
decreases towards the middle section of the channel, which linearly increases the groove depth-to-channel height ratio (Fig. 9). Previous studies26,27 showed that mixing performance of herringbone grooves improves with an increase in the value of this ratio, which is explained in detail elsewhere.22 Thus, the mixing at the beginning of the channel can probably be attributed in the first place to the decreased width and depth of the channel at the junction where fluid streams are physically brought into the contact (Fig.11C). On the other hand, mixing based on chaotic advection occurs only in the middle section of the channel where grooves correspond to or closely resemble their original design. This can be clearly seen in Figure 12D. Similar mixer performance was observed in all studies involving similar herringbone-groove designs.3,4,28 The 2nd generation chip also has a region where chaotic advection takes place, however, the overall mixing performance is poor. The 1st generation chip (Fig. 11A, Fig. 12A) has a very small region with well-resolved grooves, which results in even poorer mixing performance.
Figure 11. Mixing performance of the (A) 1st, (B) 2nd and (C) 3rd generation fused-silica micromixers; Pe = 3.5×103
Ch
ap
te
r
However, the mixing is not complete by the end of the channel for any of chips, which is obvious from the streaks of unmixed dyes at the channel outlet (Fig.11 and 12). This relates to the insufficiently long channel region exhibiting the generation of chaotic advection. Furthermore, at higher flow rates the mixing efficiency decreases as well, as evidenced by the green-coloured area that appears further down the channel and is less pronounced at lower flow rates. This effect can be explained by the decrease in the residence time and, thus, decrease in the mixing that is governed by diffusion.
Clearly then, the fabrication process needs to be optimized to realize all the grooves initially patterned by the pulsed laser, as only then will these chips function as efficient mixers.
Figure 12. Mixing performance of the (A) 1st, (B) 2nd and (C) 3rd generation fused-silica micromixers; (D) the
middle part of the 3rd generation chip where mixing by chaotic advection is evident; Pe =17.5×103; combined
photos.
Pressure tests
Because of the original intention to place the chip in the interface of a two-dimensional liquid chromatograph, and the future plan to integrate a monolithic column into the same chip, the fabricated chips had to be tested for pressure resistance. For this test only the 2nd generation chip was chosen, as the 1st generation chip was received with a small crack in the inlet.
In the test, the column connected to the chip outlet was left open. The pressure was increased from 0 to 22 bar for flow rates of 0.2 - 2.0 mL/min. However, no changes in the chip or connected tubing were observed. Next, the column was closed off. The results of this second
experiment are summarized in Table 4. In the first two attempts, the tubing detached from different inlets as the pressure increased. In the last experiment, when the pressure reached 85 bar, a big part of the outlet region split from the rest of the chip and the chip was destroyed (Fig.13). Device fracture in the outlet region could be expected, as the used silica was not very thick and fell subject to high total back pressures. Thus, the pressure test showed that the fused-silica chip could withstand a pressure of 85 bar. This value is sufficient for utilization of this device in the interface of two-dimensional liquid chromatography as a separate mixing device. However it may not be sufficient for applications involving an integrated monolithic stationary phase on the same chip, as the channel filled with stationary phase would increase the flow resistance and, thus, the internal pressure drop in the device itself.
Table 4. Pressure test with the 2nd generation chip when the column was sealed at the total flow
rate 0.5 mL/min.
Total max pressure Observation
63,5 bar
Different inlets leakage 77.6 bar
85.15 bar The chip split into two part in the outlet region
Figure 13. (A) Top-view and (B) side-view of the two parts of the 2nd generation chip after the pressure test.
In the literature, there is limited information regarding glass-based microfluidic devices for high-pressure chemistry. In most cases, the main limitation associated with the development of such devices is their micro-to-macro interface (i.e. quality of the connections). Working pressures of 50-150 bar were achieved when glass chips were placed in different types of clamp-holders or similar support units.29,30 Szekely and Freitag29 clamped a glass chip into a Teflon holder/interface with integrated high pressure connectors and O-rings that withstood back
Ch
ap
te
r
only straight lines and crosses can be fabricated. Tiggelaar et al.11 reached working pressures of 300 bar using an in-plane, fiber-based interface with glass capillaries connected to a glass microreactor. However, in this approach epoxy resin fibers were glued to the chip, which decreased the flexibility for connection of the chip to different equipment. In our work, we also used in-plane connectors for improved robustness. In contrast to other studies, no extra holders or gluing is needed in our approach. Having female connectors with standard 10-32 thread (suitable for standard HPLC connectors) allows easy connection of the chip to conventional equipment and makes the interface user-friendly. Even though our chip didn’t reach pressures beyond 85 bar, the obtained result is compatible with other high-pressure resistant chip-based devices. Besides, it should be noted that the chip was not placed in any clamp-holder or support; doing so would most likely allow for much higher operating pressures.
Conclusions
In this work, we explored the fabrication of microfluidic mixers with integrated arrays of herringbone grooves in fused silica using the Selective Laser-Induced Etching (SLE) technique. This approach allowed us to obtain devices with complex features inside solid pieces of fused silica, eliminating the need for a bonding step. We aimed to fabricate microfluidic mixers with channels up to 33 mm long and with aspect ratios more than 100, the first time that SLE has been applied for this type of application. We managed to fabricate several generations of chips with different dimensions. Our results showed that increasing the channel diameter allows higher-resolution channels with grooves, due to faster access of etching solution. However, we didn’t succeed in the fabrication of channels with uniform cross-section. The channel still had a tapered, conical shape toward the middle of the channel, due to the etching process. To overcome this effect, we proposed a compensation design that resulted in slightly better resolutions of grooves in the channel.
All micromixers were tested for mixing performance. Tests revealed that mixing based on chaotic advection is observed only in the middle part of the channel where herringbone grooves correspond to or resemble most the original design. The best mixing performance was observed for the chip with compensation design, due to changes in channel depth/width and better resolution of herringbone grooves. However, the mixing was still not completed by the end of the channel at either of the tested flow rates (0.2 mL/min and 1.0 mL/min).
In terms of pressure, the obtained chip can withstand pressures up to 85 bar. This is within the pressure range reported for the glass chips placed in different types of clamp-holders or similar support units.29,30 However, in our tests the chip wasn’t placed in any housing, which means that the pressure resistance could be improved with both utilization of the extra support and using a thicker block of fused silica. Compared to other studies, no extra holders or gluing is needed in our approach. The device can be easily connected to any conventional equipment using standard 10-32 thread (standard HPLC connectors).
We believe that it is worthwhile to further explore the fabrication of microfluidic devices using SLE. To achieve desired structures, an optimized fabrication procedure using a compensation design to account for variation in exposure times to etching solution should be developed. Ch ap te r IV
Acknowledgements
This work was financially supported by The Netherlands Organization for Scientific Research (NWO) in the framework of the Technology Area-COAST program, project no. (053.21.102) (HYPERformance LC).
References
1. Martin, A., Teychené, S., Camy, S. & Aubin, J. Fast and inexpensive method for the fabrication of transparent pressure-resistant microfluidic chips. Microfluid. Nanofluidics 20, 1–8 (2016).
2. Sollier, E., Murray, C., Maoddi, P. & Di Carlo, D. Rapid prototyping polymers for microfluidic devices and high pressure injections. Lab Chip 11, 3752 (2011).
3. Ianovska, M. A., Mulder, P. P. M. F. A. & Verpoorte, E. Development of small-volume, microfluidic chaotic mixers for future application in two-dimensional liquid chromatography. RSC Adv. 7, 9090–9099 (2017).
4. Stroock, A. D. et al. Chaotic mixer for microchannels. Science 295, 647–651 (2002).
5. Liu, J. et al. Polymer Microchips Integrating Solid-Phase Extraction and High-Performance Liquid Chromatography Using Reversed-Phase Polymethacrylate Monoliths. 81, 2545–2554 (2009).
6. Chen, C. F. et al. High-pressure needle interface for thermoplastic microfluidics. Lab Chip 9, 50–55 (2009).
7. Mair, D. A. et al. Room-temperature bonding for plastic high-pressure microfluidic chips. Anal. Chem. 79, 5097–5102 (2007).
8. Urakawa, A., Trachsel, F., von Rohr, P. R. & Baiker, A. On-chip Raman analysis of heterogeneous catalytic reaction in supercritical CO2: phase behaviour monitoring and activity profiling. Analyst 133, 1352–1354 (2008).
9. Trachsel, F., Hutter, C. & von Rohr, P. R. Transparent silicon/glass microreactor for high-pressure and high-temperature reactions. Chem. Eng. J. 135, 309–316 (2007).
10. Marre, S., Adamo, A., Basak, S., Aymonier, C. & Jensen, K. F. Design and packaging of microreactors for high pressure and high temperature applications. Ind. Eng. Chem. Res. 49, 11310–11320 (2010). 11. Tiggelaar, R. M. et al. Fabrication, mechanical testing and application of high-pressure glass microreactor
chips. Chem. Eng. J. 131, 163–170 (2007).
12. Matsuo, S., Tabuchi, Y., Okada, T., Juodkazis, S. & Misawa, H. Femtosecond laser assisted etching of quartz: Microstructuring from inside. Appl. Phys. A Mater. Sci. Process. 84, 99–102 (2006).
13. Gottmann, J., Hermans, M., Repiev, N. & Ortmann, J. Selective laser-induced etching of 3D precision quartz glass components for microfluidic applications-up-scaling of complexity and speed.
Micromachines 8, (2017).
14. Hermans, M., Gottmann, J. & Riedel, F. Selective, laser-induced etching of fused silica at high scan-speeds using KOH. J. Laser Micro Nanoeng. 9, 126–131 (2014).
15. Osellame, R., Hoekstra, H. J. W. M., Cerullo, G. & Pollnau, M. Femtosecond laser microstructuring: An enabling tool for optofluidic lab-on-chips. Laser Photonics Rev. 5, 442–463 (2011).
16. Kiyama, S., Matsuo, S., Hashimoto, S. & Morihira, Y. Examination of etching agent and etching mechanism on femotosecond laser microfabrication of channels inside vitreous silica substrates. J. Phys.
Chem. C 113, 11560–11566 (2009).
17. Hnatovsky, C. et al. Fabrication of microchannels in glass using focused femtosecond laser radiation and selective chemical etching. Appl. Phys. A-Materials Sci. \& Process. 84, 47–61 (2006).
18. Osellame, R., Maselli, V., Martinez Vazquez, R., Laporta, P. & Cerullo, G. Integration of optical waveguides and microfluidic channels fabricated by femtosecond laser irradiation. Conf. Lasers
Electro-Optics, 2007, CLEO 2007 231118, 88–91 (2007).
19. Vishnubhatla, K. C., Bellini, N., Ramponi, R., Cerullo, G. & Osellame, R. Shape control of microchannels fabricated in fused silica by femtosecond laser irradiation and chemical etching. Opt. Express 17, 8685– 8695 (2009).
20. Maselli, V. et al. Fabrication of long microchannels with circular cross section using astigmatically shaped femtosecond laser pulses and chemical etching. Appl. Phys. Lett. 88, (2006).
21. Venturini, F. et al. Maskless, fast and highly selective etching of fused silica with gaseous fluorine and gaseous hydrogen fluoride. J. Micromechanics Microengineering 24, 025004 (2014).
22. Lynn, N. S. & Dandy, D. S. Geometrical optimization of helical flow in grooved micromixers. Lab Chip 7, 580–587 (2007).
23. Ghoreishi, S. M., Behpour, M. & Golestaneh, M. Simultaneous voltammetric determination of Brilliant Blue and Tartrazine in real samples at the surface of a multi-walled carbon nanotube paste electrode. Anal.
Methods 3, 2842 (2011).
24. DIACU, E., UNGUREANU1, E.-M., ENE, C. P. & IVANOV, A. A. Voltammetric Studies for Detection
Ch
ap
te
r
25. Lin, D. et al. Three-dimensional staggered herringbone mixer fabricated by femtosecond laser direct writing. J. Opt. 15, 025601 (2013).
26. Wang, S. et al. Highly Efficient Capture of Circulating Tumor Cells by Using Nanostructured Silicon Substrates with Integrated Chaotic Micromixers. Angew. Chemie Int. Ed. 50, 3084–3088 (2011).
27. Yang, J.-T., Huang, K.-J. & Lin, Y.-C. Geometric effects on fluid mixing in passive grooved micromixers.
Lab Chip 5, 1140–1147 (2005).
28. Stroock, A. D. et al. Chaotic mixer for microchannels. Science 295, 647–651 (2002).
29. Szekely, L. & Freitag, R. Fabrication of a versatile microanalytical system without need for clean room conditions. Anal. Chim. Acta 512, 39–47 (2004).
30. Shintani, Y. et al. Polydimethylsiloxane connection for quartz microchips in a high-pressure system. Anal.
