Design of interfaces with lithographically patterned adhesive pads for gluing at the microscale

International Journal of Adhesion and Adhesives 85 (2018) 88–99

Contents lists available at ScienceDirect

International Journal of Adhesion and Adhesives journal homepage: www.elsevier.com/locate/ijadhadh

Design of interfaces with lithographically patterned adhesive pads for gluing at the microscale

T

⁎

Pieter Samyna, , Jürgen Rüheb, Oswald Pruckerb, Markus Biesalskic a

Hasselt University, Institute for Materials Research (IMO-IMOMEC), Applied and Analytical Chemistry, Agoralaan Gebouw D, B-3590 Diepenbeek, Belgium University of Freiburg, Institute for Microsystems Technology (IMTEK), Laboratory for Chemistry and Physics of Interfaces, Germany c Technical University Darmstadt, Ernst-Berl-Institute for Technical and Macromolecular Chemistry, Macromolecular Chemistry and Paper Chemistry, Germany b

A R T I C LE I N FO

A B S T R A C T

Keywords: Adhesion Patterning Microscale Lithography

The creation of small adhesive pads by traditional dispensing methods is technically limited. However, the miniaturisation of micromechanical components requires the parallel development of adhesive pads with sizes in the sub-50 µm range combining good geometrical confinement and mechanical strength. Therefore, a new design of interfaces with adhesive pads of 32–8 µm are presented through local deposition of a liquid adhesive by means of “top-down” or “bottom-up” patterning. Using lithography and photochemical process, the shape of the adhesive pads is first stabilized by partial cross-linking and effective adhesive bonding with a counterface subsequently takes place during full cross-linking. The parameters for photochemical cross-linking of the adhesive pads are optimised and the mechanical performance of the patterned adhesive interfaces is evaluated. For “top-down” patterned adhesive interfaces, the geometrical stabilisation of the adhesive pads requires relatively long cross-linking times consequently resulting in low mechanical strength. For “bottom-up” patterned adhesive interfaces, the formation of adhesive pads is controlled by self-organisation of the adhesive over chemically structured substrates and requires short cross-linking times for geometrical stabilization, leading to higher mechanical strength during adhesive bonding. The fabrication of adhesive pads by a “bottom-up” approach is further discussed in relation to the influences of processing parameters on dewetting of the adhesive.

1. Introduction Gluing is a straightforward and flexible tool for assembling complex systems, integrating various micro-electronic, optical and/or fluidic components onto different substrates. However, the ongoing miniaturization of microsystem components requires a parallel development of appropriate assembly techniques. The latter presents new challenges to adhesive technologies with respect to dimensions of local adhesive pads: a tiny amount of adhesive should be applied on a targeted position with good precision and high strength. At present, high-precision bonding at the microscale is limited because of poor control over the deposition process, geometrical stability and/or viscosity of the crosslinked adhesives. Techniques for deposition of liquid adhesive are based on dropwise application onto the surface by means of dispensing methods. Conventional dispensers allow to handle volumes of 0.01 ml, while high-end microdispensers operate with volumes of 1 nl and some recent developments employ techniques that can deliver single drops with volumes as low as 5 pl. The ink-jet printing is a traditional technology to control the precise deposition of liquid droplets. For unfilled or low⁎

Corresponding author. E-mail address: [email protected] (P. Samyn).

https://doi.org/10.1016/j.ijadhadh.2018.05.021 Accepted 22 October 2017 Available online 25 May 2018 0143-7496/ © 2018 Elsevier Ltd. All rights reserved.

viscosity fluids such as inks, the fluid partially dries during ink-jetting and forms relatively confined drops on the substrate. For printing filled and/or high-viscosity materials such as adhesives, however, the droplets are applied under wet conditions and yield proportionally larger dimensions after deposition due to capillary forces and spreading on the substrate: e.g., adhesive drops in the range of 150 µm easily spread up to 1 mm [1], or the deposition of 80 µm adhesive drops by ink-jet printing of thermoset epoxies results in drop diameters of 150 to 200 µm. The reliability of dispensing tiny amounts of adhesive has been improved under industrial conditions by integrating sensors and design of specific dispending heads [2]. Other units comprise a positive-displacement dispensing system with integrated adhesive tempering, process gas supply and process optics [3]. Other techniques such as screen printing enable to form drops with a diameter of about 100 µm and line-widths of 80–150 µm. Micro-stencils with 100 µm pitch and 50 µm apertures were recently developed for printing solder pastes [4]. Whereas most printing technologies are in general optimised for one specific substrate such as paper, adhesive technologies should be more flexible. The confinement of adhesives into small geometries requires sophisticated designs with glue guiding channels [5], or stamping and

International Journal of Adhesion and Adhesives 85 (2018) 88–99

P. Samyn et al.

room temperature is 0.09 Pa.s, with a density of 1.1 g/cm3 and surface tension of 0.04 N/m. In order to optimise the viscosity and wetting properties of the adhesive, it was mixed in different concentrations between 100 and 30 vol% with ethylacetate, ethanol or acetone (Sigma Aldrich, Germany). The photochemical cross-linking was done in a UV Stratalinker at 365 nm wavelength with a power of 65 W for different times.

selective transfer [6]. A novel micro-bonding technique using was using a focused laser beam with a spot size of 30 µm for local cross-linking [7]. However, those techniques often require the initial formation of a full adhesive layer, while adhesive residues outside the joint are removed afterwards. New approaches in designing adhesive interfaces includes micro- to nanoscale patterned substrates [8], or capillarydriven self-assembly of pads in the 100–300 µm range [9]. The need of small adhesive pads has been inspired by nature, observed as arrays of fine adhesive spots on the foot pads of many insects. Among different available adhesives, photochemically cross-linkable materials are most promising for microscale glueing, because they have advantage of low shrinkage and interfacial stresses, while the cross-linking is an ecologically friendly process at room temperature. Dissimilar materials are preferentially bonded as such, by using e.g. polyester [10], benzocyclobutene [11], polyacrylate [12], or epoxy [13] glues. The adhesives are available as monomer mixtures with photo-initiators (e.g., thiol [14], benzophenone [15] or copolymerizable initiators [16]) and are cross-linked into densely compacted networks [17] or (semi-)interpenetrating networks [18]. During crosslinking, an elastic fluid converts into a visco-elastic gel by radical polymerisation. The rheological behaviour of such adhesive networks has been studied as a function of the cross-link density [19]. Using molecular dynamic simulations, the role of network connectivity and polymerisation kinetics for different cross-linker functionalities and interfacial bond densities can be followed [20]. The resulting mechanical properties depend on the type of components and entire processing: the failure stress decreases and failure strain increases at lower percentages of cross-linking, as a volume fraction of non-polymerised or ‘free’ molecules into a cross-linked network significantly alters the adhesive properties [21]. On the other hand, a small number of polymer chains with free end segments increases the adhesion if they can penetrate into the opposing surface [22]. The photo-definable adhesive films with light curing resin may allow for photo patterning, although it is often difficult to reacting high adhesiveness after patterning and it was previously used in combination with a thermal curing resin [23]. Therefore, a close control of the processing parameters will thus plays an important role in setting of the adhesive bond, its dimensional stability and final adhesive strength. Literature search reveals that flexible techniques for deposition of adhesives pads with sizes in the sub-50 µm range are scarce. In this paper, we present an alternative strategy to form small adhesive pads where the dimensions are confined through direct lithographic patterning of the adhesive layer (here defined as “top-down”) or the template-dependent self-organization of the adhesive layer onto a lithographically patterned substrate (here defined as “bottom-up”). In particular, the “bottom-up” approach first requires a lithography process to define a hydrophobic coating on the surface which then acts as a template to guide self-assembly of the adhesive layer deposited by dip coating. We present here a two-step cross-linking process to create small adhesive pads by lithographic patterning and photochemical cross-linking: (i) partial cross-linking of the adhesive will be used to generate and stabilize an adhesive pattern (‘pre-curing’), and (ii) full cross-linking of the residual photo-active groups in contact with the counterface will allow for effective bonding (‘full-curing’). The performance of the resulting adhesive interfaces will be analysed by spectroscopy and shear testing.

2.2. Substrate types and surface preparation In a first design for “top-down” patterning of the adhesive interface, standard microscope slides of polymethyl-methacrylate (PMMA) were used as substrates that were cleaned with ethanol and D.I. water prior to use. In a second design for ‘“bottom-up”’ patterning of the adhesive interface, the microscopic slides of polymethylmethacrylate (PMMA) were patterned with alternating hydrophilic and hydrophobic surface areas. Therefore, the substrates were dip-coated in a 1 mg/ml solution of a 1% benzophenone-fluoropolymer, poly [(3,3,4,4,5,5,6,6,7,7,8,8,9,9,10,10,10-hepta-decafluoro-decylacrylate)co-methacryloyloxy-benzophenone] or MABP-FP, resulting in a film with thickness of 8 to 10 nm when deposited under controlled withdrawal speed. This hydrophobic film was lithographically patterned during cross-linking in a UV Stratalinker at 250 nm wavelengths for 30 min under the application of a mask with square patterns of 32 × 32 µm2, 20 × 20 µm2, 16 × 16 µm2 or 8 × 8 µm2 and interdistances of 32, 20, 16 or 8 µm respectively. Under UV radiation, the benzophenone groups act as photo-initiators and allow for photochemical cross-linking of the fluoropolymer film together with covalent attachment to the substrate. The non-crosslinked areas covered by the mask were subsequently removed by extraction with 1,1,2-trichlorofluoroethane for 2 h in a Soxhlet apparatus. The substrates prepared by this procedure have a hydrophobic surface with hydrophilic patterns defined by the mask layout. 2.3. Characterisation 2.3.1. Adhesive characterisation The adhesive composition was evaluated at different steps of crosslinking by spectroscopy. For FTIR spectroscopy, the adhesive was coated on a silica surface and cross-linked for different times using a UV Stratalinker at 365 nm wavelength with a power of 65 W. The FTIR measurements were done on an Excalibur FTS 3000 instrument (Bio-rad Laboratories, Cambridge, MA) at wavenumbers between 4000 and 400 cm-1 with a resolution of 4 cm-1. For UV/VIS spectroscopy, the adhesive was coated onto a glass slide and cross-linked under same conditions as mentioned before. The UV/VIS measurements were done on a Cary 50 Bio (Varian Inc., Palo Alto, CA) between 420 and 300 nm wavelengths. The mechanical strength of patterned adhesive interfaces was evaluated by a lap-shear test on joint samples with dimensions 25 mm × 75 mm and an overlap area of 25 mm × 20 mm following international standards ASTM D1002. The “top-down” and “bottom-up” patterned adhesive interfaces were joint with a counterface during a second cross-linking step, so-called “full-curing” (minimum pre-curing times were chosen for stabilization of the adhesive pads, i.e. 60 min precuring of “top-down” and 15 min procuring of “bottom-up” patterns). A non-patterned PMMA substrate cleaned with ethanol and D.I. water, was placed on top of the patterned adhesive interface under given load (10 N) and the joint was cross-linked for different times. The maximum tested curing time was taken as a practical compromise after a parametric time study, where the joint reached 80% of its maximum strength after 20 min cross-linking and 100% of its maximum strength after further ageing for 6 days. As a reference adhesive joint, a continuous adhesive layer was applied in between two non-patterned PMMA substrates by dip-coating and cured under comparable

2. Materials and methods 2.1. Glue type A commercially available UV cross-linkable adhesive NOA 74 (Norland Optical Adhesive) was used: this is a solvent-free one component glue containing 15 to 35 wt% isodecyl acrylate and 4–15 wt.-% trimethylpropane polyoxypropylene triacrylate ester as monomers, together with 43–65 wt.-% thiol-ester as photo-initiator. The viscosity at 89

International Journal of Adhesion and Adhesives 85 (2018) 88–99

P. Samyn et al.

from 80° on the non-coated PMMA areas to 120° on the MABP-FP film, confirming the hydrophobic properties of the fluoropolymer film. The adhesive contact angle is about 60° on the non-coated PMMA areas and 90° on the MABP-FP film. This difference in contact angles may be a sufficient driving force for self-organization of the adhesive onto the non-coated hydrophilic surface areas. The adhesive was locally deposited with a micro-needle (dispensing process) or by dip-coating (continuous process). When the adhesive was applied dropwise, we indeed observe immediate dewetting from the hydrophobic into the hydrophilic areas. When the adhesive was applied by dip-coating of the patterned substrate with a given withdrawal speed (optimized according to paragraph 4.2.1) from a vial of adhesive, it spontaneously organizes into the hydrophilic patterned domains after certain time. The dewetting of the adhesive on patterned surfaces following dipcoating (or alternatively dispensing from a syringe) is further analysed in the discussion. The dimensions of the adhesive pads are controlled by the shape of the hydrophilic domains and the dewetting properties, and the geometry of the mask determines the organization of the adhesive pads. The results of ‘“bottom-up”’ patterned adhesive surfaces are shown in Fig. 2b to 2e, with an array of adhesive pad sizes of 32, 20, 16, 8 µm as defined by the masks used for surface patterning. The cross-linking times were experimentally optimised: the geometries of the adhesive pads could be stabilized after 15 min pre-curing. Shorter pre-curing times do not allow good reproduction of the mask pattern. The times to stabilise the “bottom-up” patterns by partial cross-linking are much lower compared to the “top-down” patterns, as the shape of the adhesive pads is also partially stabilized by pinning at the hydrophobic/ hydrophilic boundaries of the patterned substrate (see discussion).

conditions with the cross-linking during 0–60 min pre-curing followed by different times of full-curing. The samples were fixed into a mechanical tensile tester (Zwick Z2.5, Zwick GmbH, Ulm, Germany) and sheared at 5 mm/min until fracture. The initial distance between the pulling heads was 77 mm. The shear strength was calculated as the maximum shear force divided by the effective adhesive surface area of either a continuous or a patterned interface. The maximum elongation at break was recorded at complete bond rupture. Test results were averaged from three repeated tests. 2.3.2. Surface characterisation Contact angle measurements were done on a Dataphysics OCA20 equipment with the adhesive and water as a reference liquid. The static contact angles were determined by elliptical fitting and averaged from three measurements with a standard deviation of ± 2°. Optical microscopy was done on an BX51 light microscope (Olympus, Hamburg, Germany) with objective lenses of magnification 10×, 20× and 50×. 3. Test results 3.1. Design and fabrication of “top-down” patterned adhesive interface In a first approach, a continuous adhesive film was deposited on the PMMA substrate by dip-coating and lithographically patterned through a mask followed by extraction, as schematically shown in Fig. 1a. The withdrawal speed of the PMMA substrate from a vial of liquid adhesive was set at 5 mm/s to obtain a homogeneous adhesive layer on the substrate with wet layer thickness of about 5 µm after dip-coating (see discussion). Good spreading of the adhesive on the hydrophilic PMMA substrate is confirmed by contact angle measurements, as the spreading is controlled by the surface energy equilibrium: the water contact angle on PMMA is 80°, while the adhesive contact angle on PMMA is 25°. During photochemical cross-linking of the wet adhesive through a mask, the exposed parts of the adhesive film are partially cross-linked and become mechanically stable by partial network formation (‘precuring’). During a subsequent extraction step in ethanol, the noncrosslinked parts of the adhesive film are removed and the final shape of the adhesive pads is consequently determined by transfer of the mask pattern. The result of “top-down” patterned adhesive surfaces is shown in Fig. 1b to 1e, with an array of adhesive pads with sizes of 32, 20, 16, 8 µm as defined by the used masks to pattern the adhesive layer. The geometry of the masks can homogeneously be transferred into the shape of adhesive pads over large surface areas. The cross-linking and extraction times for the adhesive pads were experimentally optimized, with 60 min pre-curing and 48 h extraction. The effects of different cross-linking times have been explored (see Supplementary information S1): (i) for shorter cross-linking times, the geometry of the adhesive pads is not well defined after removal of the mask and the cross-linked adhesive pads are partly removed after extraction; while (ii) for too long cross-linking times, no further improvements in the pattern of the adhesive pads were observed while the number of residual functional groups for subsequent adhesion will decrease. Similarly, the effects of different extraction times have been explored (see Supplementary information S2): (i) for short extraction times, the solvent does not penetrate towards the centre of the adhesive pattern; while (ii) for longer extraction times, some of the weak and partially cross-linked adhesive pads are destroyed.

3.3. Chemical quality of the adhesive interface The cross-linking properties of the adhesive are evaluated by UV–vis spectra (Fig. 3) and FTIR spectra (Fig. 4) after different curing times of 0, 2, 5, 10, 15, 20, 30 and 60 min. The UV–vis spectra (Fig. 3a) indicate that the adhesive absorbs at 320–380 nm wavelengths, which activate the photo-initiators (based on which the UV wavelength of 365 nm for curing was chosen). The timedependency of the photochemical cross-linking reaction can be followed with a decrease in absorption intensities at higher cross-linking times. The photo-chemical reaction starts after 1 min and the concentration of non-reacted photo-initiator gradually decreases (Fig. 3b) to 34% (2 min), 25% (5 min), 9% (10 min), 6% (15 min) 3% (20 min), 1% (30 min) and 0% (60 min). In general, the stability of thiol radicals is better than other photo-initiators, because the thiolene chemistry is less sensitive to oxygen and atmospheric conditions. This test data confirms the efficiency of partial cross-linking during pre-curing of the “top-down” or “bottom-up” adhesive patterns: (i) after 60 min crosslinking of “top-down” patterns, all photo-initiator groups are consumed, while (ii) after 15 min cross-linking of “bottom-up” patterns, some photo-initiator groups remain present to initiate further polymerization and covalent binding with a counterface. The FTIR spectra indicate that the adhesive is a 100% polyesterpolyacrylate with following characteristic absorption bands, as shown in Fig. 4a (3700 to 2400 cm-1) and Fig. 4b (1900 to 600 cm-1). The polymerisation reaction involves a cross-linking by conversion of C=C acrylic bonds into C-C bonds. This is confirmed by a decrease in intensity of the FTIR absorption bands related to unsaturated double bonds at 3100 cm-1. Specifically the absorption bands at 1635, 1407, 984 and 810 cm-1 related to unsaturated vinyl groups in the acrylate, reduce in intensity and completely disappear as a function of ongoing curing times (Fig. 4c): the calculated conversion is 44% (2 min), 53% (5 min), 72% (10 min), 74% (15 min), 88% (20 min), 97% (30 min), 100% (60 min). These observations confirm that the adhesive progressively converts into a polymer network during pre-curing. A discontinuity in the cross-linking behaviour is noticed after 15 min, where

3.2. Design and fabrication of “bottom-up” patterned adhesive interface In a second approach, the UV-curable adhesive is deposited onto a chemically patterned substrate and the local self-organization into geometrically well-defined adhesive pads is controlled by dewetting phenomena of the adhesive over the hydrophobically modified PMMA, as schematically shown in Fig. 2a. The water contact angle increases 90

International Journal of Adhesion and Adhesives 85 (2018) 88–99

P. Samyn et al.

Fig. 1. Fabrication of micrometer adhesive pads by “top-down” patterning, (a) schematic principle sketch of the lithographic process, (b) result of a patterned adhesive interface with 32 × 32 µm2, (c) 20 × 20 µm2, (d) 16 × 16 µm2, (e) 8 × 8 µm2 adhesive pads.

as confirmed by a change in chemical composition after 15–20 min. This results in good geometrical stability, as confirmed by morphological analysis. After patterning, a minimum amount of photo-initiators (6%) and residual photoreactive moieties (25%) remains available for adhesive bonding during further cross-linking. For some polyacrylic adhesives, the acrylic double bond could not be fully converted by photochemical cross-linking [24] due to a rapid increase of the glass transition temperature and decrease in mobility of the residual acrylic double bond under fast processing. From the present conditions, almost complete conversion into a fully cross-linked network is obtained after 60 min.

the conversion of C=C bonds is somewhat delayed: this may possibly be related to the increasing viscosity within the partially cross-linked polymer network. The bands at around 1730 cm-1 due to C=O groups show an interesting shift in maximum intensity, representing the carboxyl groups either in the acrylate (1730 cm-1) or the ester (1735 cm-1) comonomer. From the spectra, the intensity of the acrylate related C=O bands decreases with increasing cross-linking times, in favour of the ester related C=O bands. The intensity of those stretching bands is highly sensitive to (re-)orientation of the carboxyl groups, e.g. under internal or mechanical stress. Based on these observations, we conclude that the acrylate groups are esterified under cross-linking, possibly inducing reorganization and local stresses of the molecular chains. The alteration of acrylate groups with ester groups improves the flexibility of adhesive molecules. In present tests, the cross-linking time of the adhesive was longer compared to other systems as we used a relatively low power. However, the slow cross-linking process was useful in monitoring the curing behavior of the adhesive system and tailoring the properties. The amount of partial cross-linking can be strictly controlled by the radiation time,

3.4. Mechanical quality of the adhesive interface An overview of some stress-strain curves recorded during lap-shear testing of different patterned interfaces and a reference continuous adhesive film are illustrated in Fig. 5. The shape of the stress-strain curves is different for the various types of adhesive interface patterns and can be divided in continuous adhesive film, “bottom-up” pattern 91

International Journal of Adhesion and Adhesives 85 (2018) 88–99

P. Samyn et al.

Fig. 2. Fabrication of micrometer adhesive pads by “bottom-up” patterning, (a) schematic principle sketch of the lithographic process, (b) result of a patterned adhesive interface with 32 × 32 µm2, (c) 20 × 20 µm2, (d) 16 × 16 µm2, (e) 8 × 8 µm2 adhesive pads.

and “top-down” patterns. The modulus of elasticity is almost similar for interfaces with adhesive pads or a continuous film. The shear modulus (G) and elasticity modulus (E) are determined from the initial slope of the stress-strain diagrams, i.e. G = 17 MPa and E = 2 G (1 - ν) = 23 MPa with a Poisson coefficient ν = 0.33. These observations agree with Lee et al. [25], who investigated the fracture mechanics of macroscopic glass/epoxy bonds: the modulus remained constant as long as 20% of the adhesive area was covered. In our design, the adhesive pads indeed cover 50% of the interface area as determined by the applied lithographic masks. The values for shear strength (MPa) and elongation at break (%) of the different adhesive joints are summarized in Table 1 – with a statistical variation of ± 0.03 MPa on shear strength and ± 0.02% on elongation – taking into account the different conditions of pre-curing and full curing:

• For continuous adhesive interfaces, there is a clear tendency that the

•

adhesive shear strength increases with full-curing times up to

92

60 min. For longer curing times, the adhesive strength does not improve significantly or decreases a little likely due to some degradation. The adhesive strength increases significantly after precuring for 15–20 min, while longer pre-curing times reduce again the film strength. The increase in adhesive strength after short precuring is most likely attributed to an additional chemical activation of the PMMA substrate under UV radiation: an increase in total surface energy from 27.46 mJ/m2 to 36 mJ/m2 was measured after UV radiation of PMMA for 15–20 min, while the surface energy levelled at around 24 mJ/m2 after longer pre-curing times due to degradation of the surface. This effect can be attributed to the transformation of ester groups into carboxylic end-groups on the surface. The elongation at break remains below 1% for most curing conditions and is also largest after 15 min pre-curing times. The elongating is represented by creep phenomena in the continuous film as observed by a plateau-value in the stress-strain diagram before complete rupture. For “top-down” patterned adhesive interfaces, the adhesive shear

International Journal of Adhesion and Adhesives 85 (2018) 88–99

P. Samyn et al.

3.5. Fracture morphologies of the adhesive interface The fracture morphologies of a continuous adhesive film and patterned adhesive interfaces with “top-down” and “bottom-up” adhesive pads show clear differences demonstrated in the failure morphologies (Fig. 6), which might be indicative for different mechanisms of failure. The continuous adhesive films are characterized by different zones of crack initiation, crack propagation and rest-fracture (Fig. 6a). We experimentally observed that initiation occurs near the sample borders likely as a result of local stress concentrations, while the propagation continues perpendicular to the shear direction. Depending on the crosslinking conditions, the propagation path shows wavy instabilities (see arrow) that can be explained as transient patterns in sheared viscoelastic films. The life-time of those instabilities increases with the polymer viscosity (influenced by UV cross-linking) and it was indeed observed that this behaviour becomes more pronounced for highly cross-linked interfaces, represented by the plateau value in the stressstrain diagram. From an energetical perspective, the instabilities can be related to a critical film thickness above which no instability was observed. So, mainly thin films in the micro-scale range are sensitive for these phenomena, and require specific measures such as patterning to reduce instable growth of the fracture. The failure of patterned adhesive interfaces is characterized by either an adhesive fracture mechanism of the “top-down” patterned adhesive pads (Fig. 6b), or a cohesive fracture mechanism of the “bottomup” adhesive pads (Fig. 6b). The distinct behaviour clearly relates to the weak adhesive strength measured for the “top-down” adhesive pads and higher strength of the “bottom-up” adhesive pads. The adhesive failure is attributed to fracture at the interface between the adhesive pad and the counterface and is characterized by smooth surfaces and no transfer of the adhesive pads onto the counterface. The relatively high degree of cross-linking for “top-down” adhesive pads obtained during pre-curing obviously decreases the amount of non-cured or ‘free’ polymer chains that does not allow for good attachment to the counterface during fullcuring. The bond consequently fails at it weakest point located at the contact with the counterface. On the other hand, the “bottom-up” adhesive pads are characterized by cohesive failure occurring within the adhesive and representing better adhesive bonding to the counterface. The cohesive failure is characterized by an irregular fracture surface and transfer of the adhesive pads to the counterface. The lower precuring times required for stabilizing “bottom-up” patterns is beneficial to maintain a required amount of non-cured or ‘free’ polymer chains that may interact with the counterface during full-curing. During failure, the crack progressively grows under 45° relatively to the shear direction along the direction of highest shear stress. The final failure at break of the entire adhesive interface is retarded by crack arrestment and re-initiation at each single adhesive pad, as the singular pads act like crack propagation barriers in contrast with a continuous adhesive film. The high elongation after the maximum shear force indicates that the patterned adhesive interface fractures progressively and the shear force reduces gradually in parallel with a decreasing adhesive contact area.

Fig. 3. UV–vis spectrum and concentration in photo-initiator during curing of the adhesive after different times ((i) 0 min, (ii) 2 min, (iii) 5 min, (iv) 10 min, (v) 15 min, (vi) 20 min, (vii) 30 min, (viii) 60 min): (a) spectral region between 300 and 420 nm, (b) calculated photo-initiator concentration from UV spectra.

•

strength and elongation is very low due to the long pre-curing times required for stabilization of the adhesive patterns. There is tendency that the adhesive strength slightly decreases for the interfaces with smaller adhesive pads, although values are roughly comparable for the patterns with 32–16 µm adhesive pads and only significantly lower for the smallest pads of 8 µm. The shear strength for the “topdown” patterned adhesive interface is little higher than for a continuous adhesive film under comparable curing conditions (e.g., 60 min pre-curing and 20 min full-curing), although it remains too low for structural applications (σ < 1 MPa). The brittle nature of the adhesive interface and almost direct rupture under loading is directly observed from the stress-strain diagrams. For “bottom-up” patterned adhesive interfaces, the adhesive shear strength is much higher than for “top-down” patterned adhesive interfaces due to the shorter required pre-curing times that are within the range for delivering optimum strength as observed for the continuous adhesive films. The adhesive strength slightly lowers only very slightly as the size of the pads decreases from 32 to 16 µm but it remains above 1 MPa for most geometries and curing conditions. The total adhesive surface area is the same for each patten (owing to the mask design) and thus implies no significant strength variation. The slight decrease in adhesive strength for smaller patterns is therefore mainly related to local contact conditions. The adhesive strength of the “bottom-up” patterned interface is also advantageously higher than a continuous adhesive interface. The elongation at break is high due to a progressive debonding of the interface after reaching the maximum debonding stress, as observed in the gradual decrease of stress towards high strains in the stressstrain diagrams.

4. Discussion The design for formation of sub-50 µm adhesive pads indicates that the geometrical stability and mechanical performance is superior for adhesive interfaces with “bottom-up” patterned adhesive pads. The “bottom-up” formation of adhesive pads happens by self-organization of the liquid adhesive over chemically patterned substrates through ‘pinning’ or ‘dewetting’ effects. In order to better control the formation of “bottom-up” patterned adhesive pads by self-organisation, some mechanisms contributing to the local deposition of adhesive pads were further analysed, by comparing discontinuous (dispensing) and continuous (dip-coating) deposition. 93

International Journal of Adhesion and Adhesives 85 (2018) 88–99

P. Samyn et al.

Fig. 4. FTIR spectra and monomer conversion during curing of the adhesive after different times, ((i) 0 min, (ii) 5 min, (iii) 15 min, (iv) 30 min, (v) 60 min): (a) spectral region between 3700 and 2400 cm-1, (b) spectral region between 1900 and 600 cm-1, (c) calculated monomer conversion from FTIR spectra.

the drop volume, the adhesive does not spontaneously spread over the hydrophobic surface but the contact angle gradually grows due to a ‘pinning’ effect of the contact line at the hydrophobic/hydrophilic interface on the patterned substrate, as illustrated in Fig. 8a: the force balance at the interface acts as a surface energy barrier that prevents spreading and allows to grow the contact angle near the contact line. The contact angle values of the adhesive deposited onto a single adhesive pad with increasing drop volume are plot in Fig. 8b, showing the growth to a maximum value of 100° at a critical volume. Only for drop volumes above a critical value, the contact line is broken and the adhesive spreads over the hydrophobic surface. A new equilibrium on the hydrophobic surface establishes with a constant contact angle of 90° according to the Young's regime. For a given pattern area and adhesive

4.1. Organization of the adhesive during dispensing process through ‘pinning’ effects The dropwise deposition of adhesive through a syringe and organization into single pads onto a patterned substrate is illustrated in Fig. 7. A side-view of 32 µm pads (Fig. 7a) and top-view of 32, 20, 16 and 8 µm pads (Fig. 7b) shows the lateral confinement of the adhesive. The deposited adhesive volume is dispensed from the needle and determined by the meniscus of the adhesive during withdrawal of the needle (see Fig. 7a). For homogeneous substrates, the static contact angle is determined by a thermodynamic equilibrium given in the Young's equation. For chemically patterned surfaces, the contact angle depends on the drop size relatively to the pattern size: while increasing

94

International Journal of Adhesion and Adhesives 85 (2018) 88–99

P. Samyn et al.

adhesive layer onto a pre-patterned substrate. This process implies that the entire volume of the adhesive film deposited during dip-coating should fit into the patterned domains. From experimental observations, the formation of the adhesive pads by dewetting takes about 10–15 min. The liquid flow of the adhesive over the patterned substrate is driven by minimisation of the surface energy. This process requires close control of the dip-coating parameters and adhesive composition in relation to film thickness and volume. First, the influences of withdrawal speed during dip-coating are experimentally considered. Second, the influence of adhesive concentration and viscosity is studied by diluting with different solvents. 4.2.1. Influence of withdrawal speed on dewetting From experimental optimization, we observed that successfull dewetting of the adhesive only took place after deposition at given dipcoating velocities: it was observed that the adhesive successfully dewets into large adhesive pads of 32 µm after dip-coating at high withdrawal speeds (0.3 m/min), while the same conditions did not allow for good dewetting into the smaller adhesive pads. Only at lower withdrawal speeds (0.03 mm/min), favourable dewetting into 8 µm adhesive pads happens. The influence of withdrawal speed can be explained, as follows:

Fig. 5. Mechanical testing of different adhesive interfaces, (i) continuous film after pre-curing for 15 min and full-curing for 20 min, (ii) “top-down” adhesive pads (32 × 32 µm2), (iii) “top-down” adhesive pads (20 × 20 µm2), (iv) “topdown” adhesive pads (16 × 16 µm2), (v) “bottom-up” adhesive pads (32 × 32 µm2), (vi) “bottom-up” adhesive pads (20 × 20 µm2), (vii) “bottom-up” adhesive pads (16 × 16 µm2). The “top-down” patterns with 60 min pre-curing and 20 min full-curing; the “bottom-up” patterns with 15 min pre-curing and 20 min full-curing.

• The volume of adhesive deposited as a continuous film by dip-

coating can be calculated from the film thickness on a 20 × 25 mm2 surface area. The (wet) thickness h of a homogeneous film can be calculated from Formula (1), with viscosity η (Pa.s), withdrawal speed v, surface tension γ (N/m), density ρ (g/cm3) and gravity constant g [26]. The equation describes a lubrication approximation for an infinite plate withdrawn vertically from a liquid reservoir, assuming that the gravitational drainage is negligible and the film thickness forms as a balance between viscous and capillary forces. According to Wilson [27], this result is valid for low capillary numbers, Ca = η v / γ « 1. In our present case, Ca = 0.011 at 0.3 m/ min and Ca = 0.0011 at 0.03 m/min « 1, confirming the applicability of the model.

volume, the pad height is geometrically controlled and stabilized by this ‘pinning’ effect. At a critical drop volume and contact angle of 100°, the adhesive pad height can be calculated as follows: 38 µm (32 µm pads), 25 µm (20 µm pads), 18 µm (16 µm pads), 9 µm (8 µm pads). As such, the constant height of all adhesive pads can be guaranteed by deposited drop volumes in order to provide a homogeneous contact with the counterface, avoiding local stress concentrations or contactless adhesive pads. 4.2. Organization of the adhesive during dip-coating process through ‘dewetting’ effects The continuous deposition of adhesive onto patterned substrates and formation of adhesive pads as illustrated before in Fig. 2, can be done by dip-coating and spontaneous dewetting of the adhesive, resulting in the simultaneous fabrication of an array with adhesive pads. The latter requires the self-organisation of a continuously deposited

h = 0.946

(ηv)2/3 γ1/6 (ρg )1/2

(1)

The relationship between film thickness h and withdrawal speed v is given in Fig. 9a: as such, the theoretical thickness of the adhesive

Table 1 Mechanical properties of the interface with continuous adhesive film or adhesive pads. Adhesive shear strength (MPa) / Elongation (%) Full curing times

15 min

20 min

30 min

60 min

120 min

Continuous adhesive interface – – – – – – –

for for for for for for for

0 min pre-cured 5 min pre-cured 10 min pre-cured 15 min pre-cured 20 min pre-cured 30 min pre-cured 60 min pre-cured

0.74 0.73 0.76 0.80 0.82 0.55 0.29

/ / / / / / /

0.64 0.62 0.66 0.76 0.74 0.55 0.39

0.89 0.92 0.94 0.95 0.72 0.65 0.32

/ / / / / / /

0.80 0.85 0.85 0.88 0.72 0.43 0.28

0.96 / 0.84 Not tested Not tested 1.04 / 0.86 0.83 / 0.79 Not tested Not tested

1.06 / 0.92 Not tested Not tested 1.15 / 1.13 1.00 / 0.79 Not tested Not tested

1.03 / 0.82 Not tested Not tested 1.10 / 0.94 0.90 / 0.75 Not tested Not tested

0.45 0.39 0.38 0.35

/ / / /

0.18 0.12 0.10 0.10

0.36 0.35 0.35 0.33

/ / / /

0.16 0.12 0.08 0.08

0.30 0.25 0.23 0.20

/ / / /

0.15 0.09 0.08 0.07

0.24 0.25 0.22 0.20

/ / / /

0.10 0.09 0.07 0.07

2.05 1.85 1.68 0.88

/ / / /

2.78 2.38 1.78 1.25

2.20 1.95 1.86 1.02

/ / / /

2.95 2.83 2.78 2.21

2.15 1.98 1.89 1.08

/ / / /

2.55 2.91 2.40 2.00

2.10 1.85 1.75 1.05

/ / / /

1.98 2.54 2.38 1.98

“Top-down” patterned adhesive interface for 60 min pre-cured – – – –

32 × 32 µm2 pattern 20 × 20 µm2 pattern 16 × 16 µm2 pattern 8 × 8 µm2 pattern

0.38 0.35 0.31 0.19

/ / / /

0.15 0.15 0.10 0.09

“Bottom-up” atterned adhesive interface for 15 min pre-cured – – – –

32 × 32 µm2 pattern 20 × 20 µm2 pattern 16 × 16 µm2 pattern 8 × 8 µm2 pattern

1.85 1.55 1.48 0.58

/ / / /

2.68 2.53 1.38 0.98

95

International Journal of Adhesion and Adhesives 85 (2018) 88–99

P. Samyn et al.

paragraph 4.1). Then, the maximum drop volume on a single 32 µm adhesive pad is 1.13 10-5 mm3, on a 20 µm adhesive pad is 0.86 10-5 mm3, on a 16 µm adhesive pad is 0.69 10-5 mm3, or on a 8 µm adhesive pad is 0.34 10-5 mm3. The dip-coated surface area (20 × 25 mm2) contains n = 122070 pads of 32 µm, n = 312500 pads of 20 µm, n = 488281 pads of 16 µm, or n = 1953125 pads of 8 µm. This results in a total ‘storable’ adhesive volume into the adhesive pads of n.Vmax = 1.38 mm3 for 32 µm adhesive pads, 0.86 mm3 for 20 µm adhesive pads, 0.69 mm3 for 16 µm adhesive pads, 0.34 mm3 for 8 µm adhesive pads. Comparing the theoretically deposited adhesive volume (V) based on film thickness h (Formula (1)), and the adhesive volume of the adhesive pads (n.Vmax) based on the geometry of the adhesive pad with critical contact angle (Formula (2)), yields a critical withdrawal speed that allows for self-organization of the deposited adhesive film onto the patterned substrate (Fig. 9b): i.e., 0.279 m/min for 32 µm pads, 0.138 m/min for 20 µm adhesive pads, 0.099 m/min for 16 µm adhesive pads, and 0.034 m/min for 8 µm adhesive pads. The calculated withdrawal speeds have been experimentally proven against the observation that the higher withdrawal speeds did not allow for self-organization by dewetting into the smaller adhesive pads. Finally, the latter calculated values have been applied for creating the adhesive pads shown in Fig. 2b. These considerations specifically prove that the experimental formation of small adhesive pads theoretically agrees with the dewetting of a given adhesive volume into sub-50 µm patterns. 4.2.2. Influence of adhesive concentration and solvent on dewetting The efficiency of dewetting was experimentally evaluated by diluting the adhesive in different solvents such as ethylacetate, ethanol or acetone with the following concentration ratios of adhesive/solvent: 100/0, 90/10, 70/30, 50/50, 30/70 and 0/100. The contact angles of the diluted adhesive on the hydrophobic surface gradually decrease from 90° for the pure adhesive towards the contact angle of the pure solvent, i.e. 64° for ethylacetate, 74° for ethanol or 76° acetone (Supplementary information S3). The contact angle of diluted adhesives on the hydrophobic domains is lower than the pure 100/0 adhesive and suggests that the dewetting of diluted adhesives becomes worse. The dewetting during dip-coating onto chemically patterned surfaces was visually evaluated (Supplementary information S4), indicating indeed that the dewetting of the adhesive over the patterned surfaces is disfavoured and complete dewetting of the adhesive took longer time (about 1 h). We observed that a change in adhesive concentration did not significantly alter the shear strength of 100/0, 90/10, 70/30 and 50/50 adhesives, irrespective of the used solvent. However, the strength of 30/70 adhesive lowers to 0.25 MPa as the amount of adhesive after solvent evaporation becomes too low. The fast evaporation of solvent immediately after dip-coating may result in deposition of a thinner film according to the discussions on the critical adhesive volume in previous paragraph. The solvent evaporation rate theoretically depends on the vapour pressure, being 30 kPa for acetone, 14 kPa for ethylacetate, 7 kPa for ethanol at 25 °C. The high vapour pressure of acetone allows for fast evaporation and qualitatively explains the observation that acetone acts as the most suitable solvent for dewetting of a diluted adhesive relatively to the other used solvents. The effect of solvent evaporation rate on the film thickness was theoretically studied by Lee et al. [28], who confirmed that the film thickness calculated from an evaporation-induced rheology model is lower than the classical prediction. The final film thickness h(x) over the length of a dip-coated substrate is controlled by solvent evaporation. For a planar substrate, the evaporation rate E varies according to E(x) = - Dv a x-1/2 with Dv = diffusion coefficient of the vapor, and a = constant. Indeed, we visually observed that the dewetting of diluted adhesives after dip-coating started at the top of the substrate and gradually proceeded over the length (Supplementary information S5), in agreement with this theoretical evaporation model. Thus, the lateral confinement of a

Fig. 6. Failure morphology of (a) continuous adhesive film (arrow indicating instable crack front), (b) “top-down” adhesive pads 32 µm, (c) “bottom-up” adhesive pads 32 µm.

•

film is between h = 2.89 µm at 0.3 m/min or h = 0.62 µm at 0.03 m/min withdrawal speed, corresponding to a theoretically deposited adhesive volume of V = 1.44 mm3 (0.3 m/min) or V = 0.31 mm3 (0.03 m/min): the lower withdrawal speeds provide thinner adhesive films. The maximum volume of liquid adhesive that can be organised into the pre-defined patterns is geometrically defined by the pattern shape and dewetting. Assuming a droplet as a spherical drop, the maximum drop volume Vmax on pattern size d is given by the Formula (2):

Vmax =

πd3 (2 − 3 cos θmax + cos3 θmax ) 3 sin3 θmax

(2)

The maximum contact angle θmax of the deposited adhesive is defined by the ‘pinning’-effect of the contact line at the hydrophobic/ hydrophilic interface on the patterned substrate, i.e. 100° (see 96

International Journal of Adhesion and Adhesives 85 (2018) 88–99

P. Samyn et al.

Fig. 7. Dewetting of adhesive for design of “bottom-up” patterned adhesive pads during dispensing process, (a) side-views, (i) adhesive contact, (ii) needle retraction and meniscus formation by dewetting, (iii) deposition of 32 µm adhesive pad; (b) top-views: (i) 32 µm pad, (ii) 20 µm pad, (iii) 16 µm pad, (iv) 8 µm pad.

continuous adhesive film into sub-50 µm adhesive pads is also influenced by dynamic processes, rather than only static contact angles. Moreover, the time-dependent local flow of a film of diluted adhesive into adhesive pads can also be influenced by the viscosity, which is lower compared to non-diluted adhesive. From previous discussion, however, it seems that the difference in contact angles are dominant factors guiding the formation of the adhesive pads.

5. Conclusion Two approaches for the formation of adhesive interfaces with small adhesive pads in the range of 32–8 µm were presented. The adhesive pads were fabricated by lithographical methods, including either (i) the direct structuring of the adhesive layer (“top-down”), or (ii) the chemically controlled self-organisation of the adhesive layer onto 97

International Journal of Adhesion and Adhesives 85 (2018) 88–99

P. Samyn et al.

Fig. 8. Contact angle of the adhesive for design of “bottom-up” patterned adhesive pads during dispensing process: drop formation at gradually increasing adhesive volume, (a) drop view on 32 µm pad, (b) contact angle values as a function of drop volume on different pad sizes of 32 µm (▲), 20 µm (○), 16 µm (▪), 8 µm (x).

patterned substrates that act as a guiding template (“bottom-up”). The photochemical cross-linking process of the adhesive was optimized in order to provide partial cross-linking and geometrical stability of the adhesive pads during a first step, followed by adhesive bonding with a counterface during full cross-linking in a second step. The performance of the “bottom-up” organized adhesive pads was superior to that of “top-down” structured adhesive pads, as observed by adhesive strength and differences in adhesive or cohesive fracture morphology. The time needed for stabilizing the geometry during partial cross-linking is a critical parameter that determines the performance of the final adhesive bond. As the latter time can be shortened for “bottom-up” organized adhesive pads, a higher bonding strength during full cross-linking can be established. The dimensional stability of the “bottom-up” adhesive pads is improved by the dewetting behavior from the hydrophobic into the hydrophilic domains, and pinning of the adhesive droplet at the hydrophobic/hydrophilic interface of patterned substrates. As such, the height of the adhesive pads can be controlled by the deposited adhesive volume. After the deposition of a continuous adhesive layer by means of a dip-coating process, the self-organisation into small pads mainly depends on the film thickness relatively to the volume of adhesive that can be stored on the pads. Some theoretical

Fig. 9. Critical volumes of adhesive deposited into adhesive pads of 32, 20, 16 and 8 µm during continuous dip-coating and selection of critical withdrawal speed (○ thickness, • volume).

98

International Journal of Adhesion and Adhesives 85 (2018) 88–99

P. Samyn et al.

considerations indicate that the withdrawal velocity during dip-coating influences the film thickness and allows self-organisation of the adhesive pads by dewetting.

[8] Burton Z, Bhushan B. Nano Lett 2005;8:1607–13. [9] Chang B, Sariola V, Aura S, Ras RHA, Klonner M, Lipsanen H, Zhou Q. Appl Phys Lett 2011;99:03414. [10] Randy B. Polym Eng Sci 1998;38:1229–43. [11] Oberhammer J, Niklaus F, Stemme G. Sens Actuators A - Phys 2003;105:297–304. [12] Czech Z, Milker R. J Appl Polym Sci 2002;87:182–91. [13] Chiang TH, Hsieh TE. Int J Adhes Adhes 2006;26:520–31. [14] Lim DH, Do HS, Kim HJ, Bang JS, Yoon GH. J Adhes Sci Technol 2007;21:589–603. [15] Swanson MJ, Opperman GW. J Adhes Sci Technol 1995;9:385–91. [16] Czech Z. Int J Adhes Adhes 2007;27:195–9. [17] Stevens MJ. Macromolecules 2001;34:1411–5. [18] Joo HS, Do HS, Park YJ, Kim HJ. J Adhes Sci Technol 2006;20:1573–94. [19] Lindner A, Lestriez B, Mariot S, Creton C, Maevis T, Luhmann B, Brummer R. J Adhes 2006;82:267–310. [20] Tsige M, Lorenz CD, Stevens MJ. Macromolecules 2004;37:8466–72. [21] Zhang Q, Archer LA. Langmuir 2007;23:7562–70. [22] Chen N, Maeda N, Tirrell M, Israelachvili J. Macromolecules 2005;38:3491–503. [23] Takahashi T, Kawata M. J Photopolym Sci Technol 2010;23:473–5. [24] Oliva C, Bellobono IR, Morelli R. Phys Chem Chem Phys 1999;1:215–7. [25] Lee KY, Case ED. J Mater Sci 1996;31:2241–51. [26] Landau L, Levich B. Acta Physicochim URSS 1942;17:43–54. [27] Wilson SDR. J Eng Math 1982;16:209–21. [28] Lee CH, Lu Y, Shen AQ. Phys Fluids 2006;18. [052105-1-11].

Appendix A. Supporting information Supplementary data associated with this article can be found in the online version at http://dx.doi.org/10.1016/j.ijadhadh.2018.05.021. References [1] [2] [3] [4]

Kolbe J. J Adhes 2009;85:381–94. Feder M; Selvey V; Haylett K; Dahlgren B. United States Patent 6318843; 2001. Reisgen U, Scheik S. Microsyst Technol 2008;14:1895–901. Durairaj R, Jackson G, Ekere N, Glinski G, Bailey C. Solder Surf Mt Tech 2002;14:11–7. [5] Huang Z, Sanders JC, Dunsmor C, Ahmadzadeh H, Landers JP. Electrophoresis 2001;22:3924–9. [6] Schlautmann S, Besselink GAJ, Radhakrishna GP, Schasfoort RBM. J Micromech Microeng 2003;13:S81–4. [7] Kim J, Kim H, Lee C. Mater Sci Forum 2008;580–582:459–62.

99