1 Introduction

Miniaturisation of chemical analysis systems employing microfluidic handling enables rapid and low cost analyses that generate little chemical waste (Becker and Gartner 2008). In the design of microfluidic systems, one must take into consideration, the type of material used to fabricate the device. The material has to be compatible with all the chemicals used, the temperatures applied, the biochemical and biophysical protocols and methods utilised, and the functional components involved in the miniaturised analysis. In the case of devices aimed for volume applications, the material should also be inexpensive and compatible with mass production microfabrication methods. Polymeric materials have been identified as a good alternative to conventional materials, such as glass and silicon, allowing to mass fabricate microsystems at affordable costs (Becker and Locascio 2002; Becker and Gartner 2008).

Selection of the most suitable polymer material is crucial for the fabrication approach as well as for the purpose of the chip. Some of the most common polymeric materials used to fabricate microfluidic devices are: poly(dimethylsilonxane) (PDMS) (Kee et al. 2008; Park et al. 2008b; Sia and Whitesides 2003), poly(methylmethacrylate) (PMMA) (Chen et al. 2008a; Tan et al. 2008), polycarbonate (PC) (Wang et al. 2005, 2008), polyester (Guo et al. 2005), polystyrene (PS) (Toft et al. 2008) and the negative photoresist SU-8 (Abgrall et al. 2007; Christensen et al. 2008; Sikanen et al. 2007). Amongst these, SU-8 can easily be structured using standard lithography techniques, with features ranging from a few to hundreds of micrometers (Microchem 2009). Unfortunately, SU-8 has the drawbacks of high optical propagation losses in the UV region of the spectrum and high water permeability (Bilenberg et al. 2004; Mogensen et al. 2003). PDMS is probably the most popular polymer for soft lithography due to easy fabrication and bonding. It offers higher permeability to O2 and CO2 than any other polymer (facilitating cell culturing on chip), and it is optically transparent from the UV region to the NIR which is important for optical detection methods. However, its does not resist pressure above a few bars, swells in organic solvents, and surface treatments are often unstable over time (Mcdonald et al. 2000; Sia and Whitesides 2003). Rigid polymers, such as PMMA and PC, can be used to fabricate less deformable structures. However, their applicability is limited in some lab-on-a-chip applications due to their low stability against organic solvents. Organic solvents such as methanol and acetonitrile are, for instance, used to tune chromatographic separations.

Recently, a new group of polymers, cyclic olefin polymers (COPs), has been introduced as a promising plastic substrate for microfluidic devices. COPs, in contrast to other polymers used for lab-on-a-chip applications, are highly resistant to chemicals including polar solvents (Topas Advanced Polymers 2009; Zeon Chemicals 2009). The high chemical resistance is opening the door, for the first time, to use polymeric microfluidic systems, e.g. in organic electrochemistry applications. COPs also present a high biological inertness that makes them suitable for biomedical applications (Bhattacharyya and Klapperich 2006; Choi et al. 2009), allow long-term stable surface treatments (Huang et al. 2000; Hwang et al. 2008; Steffen et al. 2007), and provide a low water absorption (Japan Synthetic Rubber 2009; Mitsui Chemicals America 2009; Topas Advanced Polymers 2009; Zeon Chemicals 2009). Low water absorption is beneficial to ensure that the dimensions of the structures do not change with the environmental conditions. Moreover, COPs are highly transparent in the visible and near ultraviolet regions of the spectrum making the polymer also very interesting for optical components (Hansen et al. 2005).

The aim of this article is to provide the reader with a review of the current state-of-the-art regarding the use of COPs for microfluidic applications. First of all, we describe the physical and chemical properties of the polymer, which are indisputably the reason for its increasing popularity. This is followed by a description of the methods available to fabricate COP devices, and a summary of the different applications of COP microfluidic systems found in the literature. Finally, we draw some conclusions and point out future possibilities for this family of materials.

2 Chemical and physical properties of COPs

In this section, we present the chemical and physical properties of the COPs, putting special emphasis on comparing the properties of the COPs with other common polymers used for lab-on-a-chip applications. Many of the properties are summarised in graphs and tables to simplify the selection of the most suitable material for various applications.

Cyclic olefin polymers are a class of polymers based on cyclic olefin monomers and ethene. Different COP materials are commercially available under various brand names: TOPAS by Topas Advanced Polymers GmbH (2009), APEL by Mitsui Chemicals America Inc. (2009), ARTON by Japan Synthetic Rubber Co. (2009) and Zeonex and Zeonor by Zeon Chemical L.P. (2009). Slightly different materials are obtained depending on the cyclic monomer and the polymerisation process used for their synthesis (Shin et al. 2005). Two different polymerisation processes are used, chain copolymerisation of cyclic monomers with ethene (Topas and Apel), and ring-opening metathesis polymerisation of cyclic monomers followed by hydrogenation (Arton, Zeonex and Zeonor), as shown in Fig. 1. Since the former polymers are made from more than one type of monomer, they are usually referred to as cyclic olefin copolymers (COCs), whilst the latter are usually called COPs. Both COP and COC are often used as general terms for the whole group. In this article, we have chosen to use COP as the collective term, since some of the polymers are not copolymers.

Fig. 1
figure 1

Typical polymerisation schemes and generalised chemical structures for COPs (Shin et al. 2005) starting from a generic norbornene

2.1 Chemical properties

Cyclic olefin polymers (COPs) have a low water absorption (<0.01%), four times less than PC and 10 times less than PMMA, even in high humidity environments (Japan Synthetic Rubber 2009; Mitsui Chemicals America 2009; Topas Advanced Polymers 2009; Zeon Chemicals 2009). Low water absorption provides excellent dimensional stability in changing environmental conditions, which is beneficial when the polymer is used for, e.g. optical purposes, since it provides a stable focal length (Appasamy et al. 2005).

The hydrophobicity of the COPs can be problematic in microfluidic devices (water contact angles around 90°, see Table 1). In order to decrease the surface hydrophobicity low-pressure plasma (Nikolova et al. 2004), oxygen plasma (Hwang et al. 2008), UV/ozone (Tsao et al. 2007) and chemical coating treatments (Stachowiak et al. 2007; Steigert et al. 2007; Tsao et al. 2007; Zhang et al. 2008) have been investigated. The results are summarised in Table 1.

Table 1 Surface treatments to reduce hydrophobicity of COP materials

Sometimes, hydrophobic patches are useful (e.g. to control liquid flow). Hydrophobicity can then be increased locally by ‘writing’ fluoropolymers onto the channels with a flit pen (Steigert et al. 2007).

In addition to the low water absorption, COPs are resistant to hydrolysis, by acids and alkaline agents, as well as to polar solvents, such as methanol. However, the material is attacked by non-polar organic solvents, such as toluene and hexane. Table 2 compares the chemical resistance of COP with other common polymers used in microfluidic applications.

Table 2 Qualitative chemical resistance comparison between COPs and other common polymer substrates used in microfluidics

2.2 Physical properties

2.2.1 Optical properties

Cyclic olefin polymers (COPs) have been identified as interesting materials for optical applications. Hwang and Yu (2005) and Nilsson et al. (2005) exploited COP as a waveguide core material. Appasamy et al. (2005) used COP, for the first time, for microlens applications, and Hansen et al. (2005) presented a COP lab-on-a-chip system, which integrates microfluidics and optics. These applications have been possible due to the fact that COPs have a high optical transparency over a wide wavelength range (from 300 to 1200 nm; Topas Advanced Polymers 2009; Zeon Chemicals 2009), a high refractive index (1.53 for Topas 5013; Khanarian and Celanese 2001), a large Abbé number (56.4 for Topas 5013; Khanarian and Celanese 2001) and a low birefringence. Moreover, in the near UV region, the transmittance is higher than for other polymers, such as PMMA, PC and PS. The transmission spectrum varies slightly in the region between 200 and 300 nm for different brands and grades, but an example is shown in Fig. 2.

Fig. 2
figure 2

Transmission comparison between Topas 8007X10, other common polymer substrates and quartz glass used in microfluidics for wavelengths ranging from 200 to 400 nm (Topas Advanced Polymers 2009)

Another crucial property is the autofluorescence of a material, which should be low when fluorescence detection is employed. The autofluorescence of COPs is higher than for Borofloat glass and PDMS, but in the same order of magnitude as for PMMA and PC (Piruska et al. 2005). It is lower for longer wavelengths and decays with illumination time. According to Hawkins and Yager (2003), it also varies between different brands of COP, with a higher observed autofluorescence for Zeonor than for Topas.

In Table 3, the optical properties of COPs are compared to those of PC and PMMA.

Table 3 Comparison of the general physical and optical properties of different brands of COP with other clear polymers

2.2.2 Thermal properties

Cyclic olefin polymers are commercially available in different grades with different glass transition temperatures (T g). Figure 3 compares the T g for different commercialised COPs with other common polymers used in microfluidics. Notice that T g increases with a higher cyclic olefin content and that there are COP grades with a higher glass transition temperature than PMMA, PC and PS (Shin et al. 2005). For example, Topas is available in grades with glass transition temperatures ranging from 33 to 180°C by changing the content of the norbornene component (a cyclohexene ring bridged with a methylene group in the para position, see Fig. 1) (Topas Advanced Polymers 2009). This makes it possible to use COPs in applications exposed to higher temperatures without the risk of deformation.

Fig. 3
figure 3

Glass transition temperatures of different COPs compared to other standard polymers used in microfluidics: polycarbonate (PC), poly(methylmethacrylate) (PMMA) and polystyrene (PS). The glass transition temperatures for the different grades of COPs are according to the fabricant specifications (Mitsui Chemicals America 2009; Topas Advanced Polymers 2009; Zeon Chemicals 2009)

On the other hand, COPs show a distinct softening at their respective T g. As will be discussed in the fabrication section, this particular property of thermoplastic materials is used for thermal bonding processes, e.g. to fabricate structures in high T g COP grades using replication methods and then seal the imprinted structures with a low T g COP grade (Bilenberg et al. 2005; Nielsen et al. 2004) in the bonding step.

2.2.3 Mechanical properties

Cyclic olefin polymers (COPs) are thermoplastics with high strength and rigidity. Their mechanical properties vary between brands and grades, but are generally similar to those of PC and PMMA. The different properties are summarised in Table 3.

2.2.4 Electrical properties

The electrical properties of the substrate material are especially interesting for electromigration and electroseparation applications, where a high voltage should be applied without causing dielectric breakdown through the substrate. COPs have very good electrically insulating properties, with a volume resistivity above 1016 Ω cm, and a dielectric breakdown voltage of 30 kV/mm (Lamonte and Mcnally 2001), as well as a dissipation factor below those of polystyrene (PS) and polypropylene (PP) (Lamonte and Mcnally 2001). The room temperature dielectric constant of COPs is about 2.35, and it is less dependent on temperature than the dielectric constants of polypropylene (PP) and other olefinic materials (Topas Advanced Polymers 2009).

2.3 Biological inertness

The substrate material of a microfluidic chip can have significant effects on the performance of biological assays. It can, for instance, affect adhesion, growth and behaviour of cells, cause adsorption of DNA and proteins, or denature proteins. These effects are important to consider when choosing a substrate material for biological applications.

Cyclic olefin polymer (COP) chips have been successfully used for analysis of several kinds of biological samples, such as blood (Choi et al. 2009; Grumann et al. 2006; Jang et al. 2006) and DNA (Bhattacharyya and Klapperich 2006; Gulliksen et al. 2005; Larsen et al. 2008). In order to reduce protein adsorption, Stachowiak et al. (2007) photografted poly(ethylene glycol) methylacrylate (PEGMA) in Topas channels. This reduced the adsorption of fluorescein-labelled bovine serum albumin (FL-BSA) by 78%. Zhang et al. (2008) used a dynamic coating of hydroxyethyl cellulose (HEC) for the same purpose, reducing the FL-BSA adsorption to an undetectable level.

Johansson et al. (2002) investigated cell growth on Topas (5513X2), Zeonex (480R), PMMA, PS and SAN (styrene–acrylonitrile copolymer). They compared cell number, spread and morphology of HeLa cervix carcinoma cells on untreated and air plasma-treated substrates to commercial tissue-culture PS. None of the untreated substrates were suitable for culturing these cells, but Zeonex, PS and SAN substrates with the right air plasma treatment were as good as the commercial substrates. This was not achieved for Topas and PMMA. The difference between Topas and Zeonex is possibly due to additives in the polymers.

Since the effect on biological molecules might vary between different brands and grades of COP, it is a good strategy to investigate this and possible surface modifications in the literature or experimentally for the intended application before choosing a substrate material.

3 Fabrication methods

This section describes the available fabrication and bonding methods for COPs. Recently, a considerable emphasis has been put on the fabrication of several COP devices by replication techniques, such as hot embossing, injection moulding and nanoimprint lithography (NIL) (Bilenberg et al. 2005; Kameoka et al. 2001; Steigert et al. 2007). These methods are efficient for low-cost production; however, they are not the most adequate methods for fast prototyping. Hot embossing and NIL can be used to produce a small number of complex structures, whereas injection moulding is more suitable for large-scale fabrication. Direct structuring methods, such as micromilling and laser ablation, are more time consuming (depending on the time taken to fabricate a single device), but perfectly suitable for COP fast prototyping (Bundgaard et al. 2006; Steigert et al. 2007).

The choice of COP type and grade is closely related to the fabrication method. COP can be purchased either in pellet form (e.g. for injection moulding), in solution (e.g. for nanoimprinting), or in sheets (e.g. for hot embossing). Pellets may require to be dissolved in a proper nonpolar solvent, such as toluene or sec-butyl benzene, except if injection moulding is used.

3.1 Replication methods

3.1.1 Injection moulding

Injection moulding is one of the most common techniques for the fabrication of macroscopic polymer parts, since it offers short fabrication cycle times (Rotting et al. 2002). The fabrication process consists of two major steps: mould fabrication (needed for each new design) and injection moulding (Attia et al. 2009). The high temperature and pressure ranges used when moulding, limits the use of silicon (Si), resists and other polymers as mould material, thus metals are commonly chosen (Kim et al. 2006). The mould inserts are typically fabricated by standard photolithography techniques [for instance, using SU-8 (Kim et al. 2006) or AZ4620 resist (Appasamy et al. 2005)], followed by electroplating (Chen et al. 2008b; Choi et al. 2009; Mccormick et al. 1997; Mela et al. 2005).

In injection moulding, pellets of COP are fed to the hopper of the injection moulding machine. The material is transported in the mould direction and simultaneously melted at a high temperature. The polymer is then injected with high pressure against the mould, which is kept at the de-moulding temperature. The injection lasts for a very short time, after which a constant packing pressure is applied to the COP. The sample is then cooled down and de-moulded. There are many parameters in the injection moulding process that have a direct effect on the COP replicas. For example, Ito et al. (2007) analysed how the injection speed, mould wall temperature and structure depth affect the COP replicas. Furthermore, Ito et al. showed that the injection moulded part has a structure distribution within itself, made of a skin–shear–core structure with different molecular orientations. Hence, surface replication properties are not only influenced by the pressure distribution inside the mould but also by the internal structural development (Ito et al. 2007). Angelov and Coulter (2008) developed COP gratings using two different injection moulding conditions showing that higher temperatures in various parts of the injection moulding machine (e.g. nozzle and mould) lead to better quality replicas. Higher temperatures typically lead to a better flow behaviour, thus to a better filling of the master cavities. Monkkonen et al. (2002) and Kalima et al. (2007) compared the filling properties of COP during injection moulding with other polymers. 600 nm wide and 200 nm deep lines were nicely filled by the different polymers analysed; however, the 100 nm wide and 200 nm deep structures were better filled by polycarbonate than by COP (Monkkonen et al. 2002). Despite that, Appasamy et al. (2005) injection moulded microlenses in COP. Here, it was claimed that apart from the favourable optical properties, COP also had excellent flow behaviour for injection moulding.

In injection moulding, as in several other replication methods, the quality of the replicated structures depends to a great extent on the master, which typically is metallic, and intrinsically on its fabrication process. A nickel mould has been used by Choi et al. (2009) to replicate low aspect ratio structures in Topas (grade 5013, Topas Advanced Polymers GmbH). Here, a high pressure of 70 MPa and a melt temperature of 280°C have been used. Kim et al. (2006) have also used an electroplated nickel mould to sustain a pressure of 30 MPa and a melt temperature of 290°C, whilst injection moulding pillars with a minimum width of 50 μm in a 1.5 cm long and 1 cm wide channel in Topas (grade 5013, Topas Advanced Polymers GmbH). Steigert et al. (2007) have succeeded in circumventing the need of a time consuming metallic master fabrication process by using a high-performance epoxy containing aluminium powder as a master. Steigert and coworkers reported that no master degradation after 200 cycles occurred when injection moulding channels in COP discs, but without specifying either the used pressure or temperature.

Standard photolithography and etching procedures can be done on a COP wafer, due to the high chemical resistance of COP. The possibility of performing standard photolithography and etching procedures is rather attractive, since the fabricated devices can, for instance, have integrated temperature sensors, thus diversifying the number of lab-on-a-chip applications. Standard photolithography and etching procedures are already used in the semiconductor industry for mass fabrication of several devices, thus a process combining injection moulding and the referred techniques would be beneficial for the fabrication of lab-on-a-chip devices. The main shortcoming of such a combined process would be the increased fabrication time, when compared to typical times for injection moulding fabricated devices. Lee et al. exploited this fact to wet etch gold using an aqueous solution of potassium iodide and iodine \( \left( {KI \cdot I_{2} } \right) \) on a low roughness flat COP wafer prepared by injection moulding using highly polished electroplated moulds. The etched gold patterns were used to fabricate microheaters, temperature sensors and microelectrodes, where the thermal stability of COP was another inherent advantage (Lee et al. 2005).

In the experiments, performed by Lee et al. (2005), it was noticed that the gold adhesion to the COP substrate was better than when using a typical pre-adhesive layer (usually Cr or Ti). The lack of an adhesion layer is an advantage especially when fabricating microelectrodes, since it eliminates the need of a passivation layer to protect the edges of the electrode that expose the adhesion layer to the solution. Angelov and Coulter (2008) similarly demonstrated the good adhesion between COP and metals, in this case, 150–250 nm thick Al. No pre-adhesive layer was used and delamination did not occur upon substrate immersion in deionised boiling water. Photoelastic tests were also performed and high stress levels were only observed around specific moulded parts, demonstrating the suitability of COP for optical applications (Angelov and Coulter 2008).

Most of the study reported so far consists in the fabrication of rather large structures (e.g. 100 × 60 μm2 microfluidic channels). However, replication of nanostructures into polymer surfaces has also received considerable attention, where the fabrication of biomedical materials is one of its several possible applications. Gadegaard et al. (2003) successfully replicated in Topas (grade 8007, Topas Advanced Polymers GmbH) collagen fibrils with 30 nm groove widths and depths up to 5 nm, where the nano-topography is assumed to influence cellular behaviour. So far, the main limitation seems to lie in the grain size of the master material, to successfully replicate even smaller structures.

3.1.2 Hot embossing

In hot embossing, the material is in wafer or sheet form (Mela et al. 2005; Tsao et al. 2008; Yang et al. 2005a). This was achieved by some groups by melting COP pellets at 140°C in a mould under a pressure of 10 MPa for 1 min (Faure et al. 2008). However, such sheets can also be acquired commercially, for instance, from Zeon Chemical L.P. (2009).

In hot embossing, a microstructured mould is pressed against a COP sheet or wafer heated above the glass transition temperature. Depending on the master material, design and type of COP to be embossed different process parameters have to be optimised. The applied pressure, time and temperature are the most important factors influencing the end quality of the embossed structures in the chip. Parameter variation during hot embossing of COP (e.g. compression rate, moulding temperature and pressure) has been extensively investigated in terms of the resulting device properties by Fredrickson et al. (2006). Channel depth was observed to be independent of the compression rate, moulding temperature and compression force in the analysed ranges. Furthermore, Fredrickson et al. (2006) showed that lower mould temperatures and cooling rates led to smaller areas with visible stress whitening.

Hot embossing is generally done using lower pressure and temperature ranges than injection moulding. Despite the typically low pressure and temperature used (0.5 MPa and 125°C), Faure et al. (2008) used a nickel/stainless steel mould to emboss fluidic channels into Zeonor (grade 1020R, Zeon Chemicals). A constant pressure of 0.5 MPa was applied for 10 min to the heated material and also whilst it cooled down to 80° C. A steel master has been used by Dhouib et al. (2009) to hot emboss Topas (grade 5013, Topas Advanced Polymers GmbH). A steel substrate was laser machined and pressed against the heated Topas (T g + 50°C) with a pressure of 2.5 MPa over a 402 mm2 square for an unspecified amount of time. De-moulding was performed at T g − 50°C and even though no anti-sticking layer (to prevent the polymer from sticking to the stamp and facilitate its release) was used a good fidelity was achieved for the rather large embossed structures (~100 μm wide channels). Similar pressure and temperatures were used by other groups with Si masters (Park et al. 2008b; Yang et al. 2005a). Kameoka et al. (2001) cut Zeonor polymer sheets into the size of the final chips before embossing them using a silicon master. The embossing occurred at 130°C and with a pressure of 1.7 MPa applied for 7 min. After embossing, the master and the substrate were cooled down to a temperature below the glass transition temperature and then de-moulded. In this way, Kameoka et al. (2001) embossed capillary electrophoresis chips in Zeonor (grade 1020, Zeon Chemicals) polymer sheets obtaining 60 μm wide and 20 μm high channels, with smooth surfaces.

SU-8 has also been used as a master material for embossing COP under conditions similar to the ones used with Si masters; however, these masters could typically only be used <10 times (Liu et al. 2007). SU-8 masters have also been used by Bhattacharyya and Klapperich (2006) to emboss Zeonor; however, in this case, an aluminium coating was used to facilitate the substrate separation from the master. Illa et al. (2009) used an anti-sticking layer on their Si master to facilitate the release of the 15.3 μm wide cylindrical pillars in a pillar array embossed in Zeonor (ZeonorFilm, ZF 14-188, Zeon Chemicals). In hot embossing, the distance of the COP has to flow towards the master is considerably smaller than in injection moulding, which may lead to reduced stress and shrinkage effects (Gerlach et al. 2002). Hence, more fragile structures with higher aspect ratios than the ones achieved with injection moulding are feasible. High aspect ratio structures, 200 nm wide and 60 μm high, have been fabricated in COP by Yoo et al. (2009) using an anodic aluminium oxidation nanostructured membrane. This membrane was later dissolved to release the embossed nanostructures. Monkkonen et al. (2000) embossed 125 nm deep cylindrical pillars with a radius of 150 nm, to be used as an antireflecting element, in Topas. Scanning electron microscopy and optical transmission characterisation showed that the initial master structure was well-replicated.

3.1.3 Nanoimprint lithography

Nanoimprint lithography (NIL) can be used to fabricate COP structures with minimum features in the nanometer range. Lab-on-a-chip systems fabricated by hot embossing are typically made using a thick polymer substrate. In hot embossing, only a small portion of the substrate is embossed compared to the initial substrate thickness. In NIL, either the entire or nearly the entire thickness of the polymer film is patterned, leading to a thin or non-existent residual layer. The advantage of a thin residual layer lies in the possibility of integrating optical components on Lab-on-a-chip devices. Moreover, in nanoimprinting, the high degree of control of the temperature and pressure distribution enables the fabrication of sub-micrometer structures at sub-wavelength resolution.

Nanoimprint lithography (NIL) requires the COP material to be spun onto a substrate that can be silicon, silicon dioxide (Bilenberg et al. 2005) or glass wafers (Gustafsson et al. 2008). COP can be commercially acquired in solution (e.g. from (Topas Advanced Polymers 2009)) or in pellet form. If pellets are used, these need to be dissolved in a proper nonpolar solvent, such as toluene or sec-butyl benzene. A substrate material with a lower refractive index than that of COP (oxidised silicon or glass wafers) enables the fabrication of waveguides and other optical components, thus taking advantage of the high optical transmission of COP materials (Hwang and Yu 2005).

A good adhesion and a homogenous distribution of the COP film on the substrate are assured by pre-cleaning the wafers in a solution of sulphuric acid and hydrogen peroxide (4:1 (v/v)), to remove any native oxide layer and organic impurities, and dehydrating. The substrates are baked after spin-coating to remove the solvent in which the COP was diluted. A higher temperature than the glass transition temperature of the polymer, but lower than the evaporation point of the solvent (e.g. 110°C for toluene and 174°C for sec-butyl) should be used to prevent bubble formation in the film upon baking.

The quality of the imprinted replica structure depends again to a great extent, like in other replication methods, on the master used. In this case, due to the low pressures used (0.2–1.9 MPa; Nilsson et al. 2005; Hansen et al. 2005) and the fairly moderate temperatures (c.a. 190°C), silicon is the preferred master material (Nilsson et al. 2005). The master consists of a negative pattern of the chip layout. These are usually fabricated on 4 inch wafers, with the possibility of being scaled up (Gourgon et al. 2005). The pattern from the mask is transferred to the wafer using UV photolithography or electron beam lithography for smaller structures. Deep-reactive ion etching (DRIE) is then used to transfer the resist pattern into the silicon wafer. The resist is then stripped off either in an O2 plasma or in acetone. Negatively sloped sidewalls and high sidewall roughness are undesirable. These may cause the imprinted structures to remain ‘trapped’ inside the cavities of the master (Gustafsson et al. 2008). Sidewall roughness can be further decreased after processing by thermally oxidising the silicon stamp (Nielsen et al. 2004; Nilsson et al. 2005). Finally, to prevent the polymer from sticking to the stamp and facilitate its release, the use of an anti-sticking layer is recommended. A layer of (1H,1H,2H,2H-perfluorodecyltrichlorosilane) deposited by chemical vapour deposition has been successfully used by Gustafsson et al. (2008) for this purpose.

Imprinting is done in a vacuum chamber with two parallel plates that can be pressed together. First, the sample is heated-up above the glass transition temperature (usually 90°C above T g; Bilenberg et al. 2005; Nilsson et al. 2005). Then, the stamp is pressed against the substrate at a pressure ranging from 0.2 to 1.9 MPa, for a period of time raging from 5 to 10 min. After the imprinting step, the sample is cooled down to a temperature lower than T g, whilst the pressure is maintained to avoid deformations. Finally, the stamp and substrate are separated at room temperature.

In most NIL processes, there is a residual polymer layer after imprinting the substrate. Removal of this residual layer is crucial in waveguiding structures. An O2 plasma in a standard reactive ion etcher system can be used to etch the remains of the residual layer (Bilenberg et al. 2005; Hansen et al. 2005). The etching process should be done with care, since it also heats the sample, thereby causing polymer reflow (Hansen et al. 2005). So far, the smallest structures imprinted in a 7 μm thick COP film, spun on 4 inch silicon wafer, are presented in Fig. 4 (Gustafsson et al. 2008). These structures are part of an electrochromatographic column, where the distance between the 4.4 μm wide and 4.9 μm high pillars is 1.6 μm.

Fig. 4
figure 4

SEM image of pillar structures of an imprinted chromatography column in Topas. The 4.4 μm wide pillars, separated by a gap of 1.6 μm, were used directly as the stationary phase (Gustafsson et al. 2008)

3.1.4 Other methods

Soft lithography is a technique used for fabricating structures using elastomeric materials (e.g. PDMS). Steigert et al. (2007) used a PDMS master to emboss channels in Topas discs (grade 5013, Topas Advanced Polymers GmbH). The PDMS master was cast on a negative SU-8 master. The PDMS properties were chosen as a compromise between a low viscosity to promote a good filling of the primary SU-8 master cavities and a low deformability, to withstand an applied force of 3 kN for 5 min at 175°C when embossing the Topas discs.

Mosaddegh and Angstadt (2008) described a technique based on replicating structures in Topas (grade 6013, Topas Advanced Polymers GmbH) by heating the mould to 241°C, without applying any force other than gravity, thus avoiding the use of expensive fabrication machinery (e.g. a hot embossing system). Here, the authors illustrated the importance of interfacial effects and viscosity in filling and replicating microstructures. A hollow metal cylinder is placed over the structured mould surface, and then filled with pellets. The mould and the hollow cylinder filled with pellets are heated to the desired temperature and after 5 min of moulding; the stack is brought to room temperature for 10 min and then de-moulded, where the features are replicated on the lower base of the cylinder. The aspect ratio of the fabricated COP structures was not as good as for PS and PMMA, due to sticking to the silicon oxide stamp (Mosaddegh and Angstadt 2008). This drawback could probably be avoided if an anti-sticking layer had been used (Gustafsson et al. 2008). The long process time (~45 min), when compared to injection moulding, makes this replication method more suitable for applications in academia than for mass-fabrication of lab-on-a-chip devices.

An embossing technique using lamination has been investigated by Paul et al. (2007) to replicate long channels (~10 cm), thus addressing the shortcomings of other replication methods in patterning long structures. The mould is fabricated by laminating a dry resist film on a substrate, e.g. glass or a thick film of poly(ethylene terephthalate), UV exposing and developing it. A Topas sheet (grade 8007, Topas Advanced Polymers GmbH) and the mould are placed in a laminator, where both rolls are heated to 130°C and a laminating speed of 0.5 m/min is applied. Early results indicate that higher laminating temperatures and lower laminating speeds should be used to have channels with vertical instead of sloped sidewalls.

3.2 Direct structuring

Laser ablation and micro milling are two techniques that can be used for rapid prototyping, due to the hardness and chemical inertness of the COP material.

3.2.1 Laser ablation

In laser ablation, the interaction of a high intensity laser beam with the material causes the latter to evaporate at the laser focal point (Malek 2006). Laser ablation has been used on other polymeric materials, such as PMMA (Johnson et al. 2001); however, so far, its use for microstructuring of COP has not been subject of any intensive study yet. Sabbert et al. (1999) presented some results using an ArF excimer laser (λ = 193 nm) and determined that the ablation depth of COP is 3.4 times lower than for PMMA. Nevertheless, Sabbert et al. claim that this lower ablation depth can be used to better control the depth of the device structures. Bundgaard et al. (2006), using a CO2 laser, consider this limitation a severe constraint, since using COP only low aspect ratio and poorly defined structures could be fabricated. Sabbert et al. further analysed their results and revealed that for features not deeper than 200 μm, no debris is redeposited on the surface. However, for deeper structures, clusters (~100 nm) of redeposited material are visible close to the ablated holes. Ultrasonic treatment in isopropanol proved inefficient to remove these clusters.

A disadvantage of the ablation technique is that it can lead to changes in the surface chemistry of the polymer in comparison to the bulk properties (Johnson et al. 2001). This can be difficult to control and in applications where surface chemistry plays a major role (e.g. in electrokinetic separations) the performance of the device can be affected in an unpredictable way.

3.2.2 Milling

Mechanical milling of polymeric structures is a commonly used method for rapid prototyping. Milling consists of patterning a substrate by moving it against a rotating tool (mill or drill). The dimensions of the tool used affect both the resolution and the wall roughness.

Until now, not many results have been presented concerning milling of COP, due to the roughness of the fabricated structures. Rough surfaces are, for instance, detrimental for optical applications, where they scatter light. Grumann et al. (2006) used computer numerical control (CNC) micromachining to fabricate a COP lab-on-a-disk. The milled V-grooves, fabricated for deflecting a light beam by 90°, had a surface roughness low enough that it did not significantly influence the overall results. The roughness of structures milled in two different grades of Topas (grade 5013 and 8007, Topas Advanced Polymers GmbH) has been investigated by Bundgaard et al. (2006). A tool with a diameter of 200 μm was used and the average roughness was measured depending on the feed speed. 50 nm was the lowest average roughness obtained. The roughness was shown to mainly dependent on the feed speed, varied between 10 and 60 mm/min, and the grade of Topas milled, with the more viscous grade giving the lower roughness.

Steigert et al. (2007) used milling as a simple and rapid way to integrate macrostructures next to microstructures that were previously imprinted in a COP substrate.

In order to provide a short overview of the fabrication section, Table 4 summarises the different COP fabrication methods in terms of their throughput, fast prototyping capabilities and ability to fabricate different structure sizes.

Table 4 Overview of the different COP fabrication methods, where (+++) means excellent; (++) means good; (+) means acceptable and (−) means less suited. Feature size refers to the adaptability of the described processes for fabricating structures ranging from nm to cm in the same process step

3.3 Bonding

Regularly, the fabrication of lab-on-a-chip systems involves a bonding step by which microfluidic channels can be sealed. Gluing is one of the most common bonding techniques, where a layer of glue is used to join two different polymer parts (Becker and Gartner 2000). However, this involves a potential risk that the glue layer flows into the microfabricated structures, e.g. leading to clogging of channels. Do and Ahn (2008) successfully used a room temperature UV adhesive to bond three distinct COP parts of a lab-on-a-chip system.

Cyclic olefin polymer (COP), as a thermoplastic polymer, is frequently bonded using thermal bonding (Bedair and Oleschuk 2006; Bhattacharyya and Klapperich 2006; Bilenberg et al. 2005; Choi et al. 2009; Gustafsson et al. 2008; Hansen et al. 2005; Kameoka et al. 2001; Kim et al. 2006; Li et al. 2005; Mair et al. 2006; Mela et al. 2005; Nilsson et al. 2005; Wallow et al. 2007; Yang et al. 2005a). Polymer chains in between the mating parts diffuse between the surfaces promoting adhesion. Using COP pieces with slightly different glass transition temperatures is a better choice, since then only one of the layers softens, avoiding deformation and obstruction of the microstructured features (Bilenberg et al. 2005; Gustafsson et al. 2008). Yang et al. (2005a, b) thermally bonded Zeonor substrates by heating them to the T g of the polymer and pressing them together (~0.5 kN) for 5 min. After this step, the applied force was removed and the substrates were annealed at T g + 10°C for 10 min. This proved to increase the bond strength, so that a flow rate of 10 μl/min could be supported in the fluidic channel (Yang et al. 2005a). A similar annealing step has also been performed by Illa et al. (2009) resulting in a good quality bonding. Nevertheless, thermal bonding typically yields relatively low bond strength due to the low surface energy of thermoplastics. Several tests have been performed on how to improve the bond strength. O2 plasma activation has been shown to be more effective compared to N2 and N2–O2 plasma activation to improve the bond strength of COP substrates (Kettner et al. 2006). UV/ozone surface treatments have increased bonding strengths of 0.003 mJ/cm2 for native (untreated) substrates up to 0.8 mJ/cm2 under specific bonding conditions (Tsao et al. 2007). Improved diffusion of polymer chains, due to surface treatments, is seen as an explanation for closer contact between the mating surfaces. In thermal bonding, non-optimised parameters, such as temperature and pressure, may cause channel deformation. This issue has been circumvented by Steigert et al. (2007) and other groups where a lamination machine was used to bond two different grades of Topas. Experiments showed that pressures of 0.4 MPa could be applied to the fabricated channels (Steigert et al. 2007). Bond strength evaluation showed that temperature is the most important parameter in lamination (Fredrickson et al. 2006). Too high temperatures would lead to channel deformation, including to bending of the device, whereas too low temperatures are insufficient for bonding (Fredrickson et al. 2006).

Solvent bonding is another bonding technique especially suited for achieving high bond strengths required for some COP microfluidic devices (Chen et al. 2008b; Faure et al. 2008; Liu et al. 2007; Ro et al. 2006; Tsao et al. 2008; Wallow et al. 2007). Polymer solubility enables a better entanglement of the polymer chains, since these become more mobile across the bonding interface. A timed immersion of Zeonor (grade 1060R, Zeon Chemicals) pieces in an ethanol/decalin solution enables a high bond strength after thermal bonding (Wallow et al. 2007). Sealed microfluidic channels withstood pressures up to 13–16 MPa, without deformation of the structures beyond the thickness of the swollen layer (Wallow et al. 2007). Mair et al. (2006) improved the bonding technique used in their earlier chips by exposing one of the bonding surfaces to solvent vapour, bringing the mating surfaces into contact and then exposing the stack to UV light (Mair et al. 2007). This was done at room temperature to only promote a higher polymer chain mobility at the mating surfaces and not in the bulk. It led to an improved burst pressure of 34.6 MPa (Mair et al. 2007) compared to previous results of 15.6 MPa (Mair et al. 2006), both regarded as the highest in channel back pressure at the time of publication. A solvent mixture of hexadecane and isopropanol has also been used to bond COP sheets in a lamination machine (Paul et al. 2007).

Localised bonding (Tsao and Devoe 2009) is done by inducing heat and softening the interface between mating surfaces using ultrasonic energy. Another way of achieving localised welding is by depositing metallic films on the mating surfaces and heating them up using microwave energy. COP pieces can be bonded using infrared laser welding, due to COP transparency at this wavelength range. However, an opaque surface has to be used for energy absorption and localised heat generation, such as absorbing particles or special dyes (Bundgaard et al. 2006; Pfleging and Baldus 2006). In general, pigments absorbing in the near infrared range can be used to bond transparent polymers, such as COP (Clearweld, Gentex Corporation, UK or Lumogen, BASF AG, Germany). Pfleging and Baldus (2006) reported that at 20°C below the T g COP was welded without any structural damage, using very thin carbon layers as an absorbing layer. The maximum bond strength was achieved for a temperature slightly under T g. Despite the existence of these techniques, their application to microfluidic chips is not common practice (Tsao and Devoe 2009).

4 Applications

This section gives examples of how COP microsystems are used for various applications. First, implementations of different chemical analysis techniques, such as electrophoresis, chromatography, solid phase extraction (SPE), isoelectric focusing (IEF) and mass spectrometry (MS) are described. Then, chips with integrated optical elements and applications within blood and DNA analysis are presented. Other examples of COP microchips include magnetic immunoassays (Do and Ahn 2008), streaming potential detection (Pu et al. 2008) and light actuated microvalves (Chen et al. 2008b).

4.1 Microchip electrophoresis

In capillary electrophoresis, a high voltage is applied between the ends of a narrow capillary, separating analytes depending on the ratio between their charge and viscous drag. Transferring the technique from capillaries to microchannels has resulted in faster separations. Moreover, integration of other functionalities, such as sample preconcentration (Sueyoshi et al. 2008), electrokinetic sample injection (Bias et al. 2008) and waveguides for fluorescence (Bliss et al. 2007) or absorbance (Mogensen et al. 2001) detection, is also possible. It is therefore not surprising that this is a common application for microfluidic COP chips as well. Several different detection methods have been employed in COP electrophoresis chips, e.g. electrochemical detection (Castano-Alvarez et al. 2005, 2006a, b, 2007), fluorescence detection (Hurth et al. 2008; Yi et al. 2008) and MS (Kameoka et al. 2001; Shinohara et al. 2008).

Commercial Topas electrophoresis chips consisting of an injection cross and a straight separation channel are already available, e.g. from Microfluidic ChipShop (Jena, Germany; Microfluidic Chipshop 2009). Chips from this company have been evaluated by Castano-Alvarez et al. (2005, 2006a, b, 2007) and compared to PMMA. The electroosmotic mobility was higher for Topas than for PMMA, but approximately halved compared to glass. The slightly more expensive Topas chips turned out to be more durable than the PMMA chips (~100 runs for Topas compared to ~50 runs for PMMA). Yi et al. (2008) compared electrophoretic separations in Zeonex and PMMA chips. In their findings, the Zeonex chips had lower background fluorescence, higher electrophoretic efficiency and better reproducibility for separation of Cy3 and Cy5 dyes. For DNA samples, the Zeonex chips displayed higher signal to noise ratio, whilst the PMMA chips provided a higher degree of separation.

4.2 Chromatography and solid phase extraction

In chromatography, analytes are separated depending on how much they are slowed down due to their affinity to a stationary phase. Solid phase extraction (SPE) is a related method where either the analytes of interest or the undesired impurities are completely retained by the stationary phase. These methods are usually performed in columns and syringes, but have also been transferred to microfluidics. The main challenge is then to create a stationary phase in the microfluidic channel. The optical and chemical properties of COP polymers have turned out useful for this purpose.

One method is to use the partial UV transparency of COPs for photo-initiated polymerisation (Stachowiak et al. 2003). The channel surface is first modified by photo-initiated graft polymerisation to make it possible to attach a stationary phase. Then, a stationary phase is created by photo-initiated polymerisation of a porous polymer in the channels (Fig. 5). This method has been used by several groups for various purposes. SPE of DNA in porous polymer containing silica microspheres was demonstrated by Bhattacharyya and Klapperich (2006). Yang et al. (2005b) also used the porous polymer for preconcentration by SPE. In this case, analytes were retained by the stationary phase and subsequently eluted and analysed by electrospray ionisation mass spectrometry (ESI-MS). Ro et al. describe a system to perform chromatography of peptides and proteins, which were then directly deposited by a pulsed electrical field onto a matrix assisted laser desorption/ionisation time-of-flight mass spectrometry (MALDI-TOF-MS) target (Liu et al. 2007; Ro et al. 2006). Reversed phase electrochromatography with fluorescence detection was demonstrated by Faure et al. (2008) with the porous polymer as stationary phase.

Fig. 5
figure 5

SEM image of chromatography channels with photo-polymerised porous polymer (Ro et al. 2006)

Another method to create a stationary phase is to increase the surface area using microfabricated structures. Gustafsson et al. (2008) used the hydrophobic COP polymer directly as stationary phase. A microfluidic chip with an injection cross and a chromatography channel was fabricated by NIL in a 7 μm thick layer of Topas spun over a glass substrate. The chromatography channel contained approximately 120,000 hexagonal pillars, as shown in Fig. 4. The higher regularity of microfabricated structures compared to traditionally packed bed columns is expected to improve the efficiency of the separation. In order to demonstrate the function of the chip, three fluorescently labelled amines were separated by reversed phase electrochromatography. Illa et al. (2009) also used a COP polymer directly as stationary phase, but this time for pressure-driven chromatography. They successfully separated coumarin dyes in a hot-embossed chip.

4.3 Isoelectric focusing

In isoelectric focusing (IEF), an electrical potential is applied through a channel containing the analytes and a pH gradient, separating the analytes depending on their isoelectric point (pI). When this is performed in COP microchips, the hydrophobicity of the surface can cause protein adsorption, whilst the electroosmotic flow may disrupt the separation. These problems can, however, be solved by applying a dynamic coating of hydroxyethyl cellulose (HEC; Zhang et al. 2008), or using the partial UV transparency of COPs to photograft a polyacrylamide coating (Li et al. 2005).

Das et al. (2007a, b) integrated IEF with multi-channel polyacrylamide gel electrophoresis (PAGE) for protein separation. IEF is first performed in one channel and PAGE is then carried out in multiple channels perpendicular to the IEF channel, as illustrated in Fig. 6. The partial UV transparency of the COP material was here utilised to photo-initiate in situ gel polymerisation, to form pseudo-valves at the inlet to the PAGE channels. They prevent mixing of the two separation media, whilst proteins can be pumped through these barriers electrokinetically.

Fig. 6
figure 6

a Optical image of a COP device for 2D protein separation. The proteins are first separated by IEF in the channel from A to B and then by polyacrylamide gel electrophoresis (PAGE) in the parallel channels from C to D. The inset shows an SEM image of channel intersections (Das et al. 2007b). b Micrograph of the channel intersections with in situ polymerised gel pseudo-valves (Das et al. 2007a)

4.4 Mass spectrometry

In mass spectrometry (MS), the analytes are ionised, accelerated and separated depending on their path in an electric and/or magnetic field. COP chips have been used as electrospray emitters, where the liquid sample is ionised and turned into an aerosol (Kameoka et al. 2002; Liu et al. 2007; Park et al. 2008a; Shinohara et al. 2008; Yang et al. 2004, 2005a, b). Bedair and Oleschuk (2006) reduced clogging and droplet spreading problems by utilising a porous polymer monolith at the end of the channel. They compared emitters fabricated in PDMS, COP and PMMA, where COP emitters showed the highest chip-to-chip reproducibility for the MS data. This is most likely because the COP is more compatible with the UV light and solvents used for the formation of the porous polymer monoliths.

An alternative to electrospray ionisation is matrix assisted laser desorption/ionisation (MALDI), where the sample is deposited onto a target, and subsequently, vaporised and ionised by a laser beam. Ro et al. (2006) performed liquid chromatography in COP chips followed by deposition onto a MALDI target. COP chips have also been used to add other functions prior to MS, such as microchip electrophoresis (MCE; Kameoka et al. 2001; Shinohara et al. 2008), SPE (Yang et al. 2005b) or enzymatic reactions (Benetton et al. 2003).

4.5 Integrated optics

Cyclic olefin polymers (COPs) are attractive for micro-optics, since it is highly transparent at shorter wavelengths, more so than many other polymers (Topas Advanced Polymers 2009). Microlenses (Appasamy et al. 2005) and photonic crystals (Bilenberg et al. 2006) have for instance been fabricated. Still there are surprisingly few examples of microfluidics with integrated optics in COP.

One example is the study of Nilsson et al. (2005) who fabricated a microfluidic dye laser coupled to waveguides. The laser was later integrated with Fresnel lenses to couple light in and out of the waveguides, a microfluidic mixer and an absorbance cell (Fig. 7a; Hansen et al. 2005). Another example of waveguides with a COP core was recently published by Okagbare et al. (2010). The waveguides were fabricated by curing COP solution in channels in a PMMA substrate. It was subsequently bonded to another PMMA substrate containing microfluidic channels in the orthogonal direction. The assembly was used for evanescent fluorescence excitation of fluorophores in the channels. However, to the best of our knowledge, no waveguides have yet been published where both the core and cladding are made from COP materials. This should, in theory, be possible, since there is a small difference in refractive index between different COP materials and grades.

Fig. 7
figure 7

a Image of a microfluidic chip where the dye laser is integrated with waveguides, Fresnel lenses, a mixer, and an absorbance cell (Hansen et al. 2005). b Picture of a blood typing chip prototype (Kim et al. 2006)

With their combination of excellent optical and chemical properties, COP polymers are likely to become increasingly popular within optofluidics. For this to happen, more research is needed, e.g. within fabrication and characterisation of COP waveguides.

4.6 Blood analysis

Many clinically relevant parameters can be measured in blood samples, making blood analysis an important area for microfluidics. Ahn et al. (2004) developed a COP chip for electrochemical measurement of oxygen, glucose and lactate in blood. Liquids on the chip are controlled by passive valves and pressurised air stored on-chip. The pressurised air is released by melting of thermoplastic membranes by localised heaters. In this way, bulky and power consuming pumps can be omitted. They list several reasons for choosing COP as substrate material instead of PC and PMMA. These include lower water absorption, excellent adhesion for most metallic films, resistance to most polar solvents and better properties for injection moulding.

Grumann et al. (2006) also developed an assay capable of measuring glucose levels in blood. In this COP “lab-on-a-disk”, sample and reagent are driven by centrifugal forces through passive hydrophobic valves into a detection cell, where the glucose level is read via an optical transmission measurement. The optical path length is increased by V-groves reflecting the light into and out of the plane of the disk.

Jang et al. (2006) fabricated a chip from COP, dry film resist, PDMS and gold where blood plasma is separated from cells by micropillars. The blood is pumped through the pillar array by an integrated micropump, consisting of a heater and a flexible membrane.

Cyclic olefin polymer (COP) chips for blood typing by agglutination have also been demonstrated. In these chips, blood and reagents are mixed in serpentine laminating micromixers. The result can be read by eye either as the degree of light transmission through a microwell (Choi et al. 2009) or by how far sample reaches in an array of microfilters with decreasing gap sizes (Fig. 7b; Kim et al. 2006).

4.7 DNA analysis

Analysis of DNA or RNA is useful for identifying a virus or bacteria, and for analysing a patients genetical predisposition for developing a certain disease or responding to a treatment. Several COP chips have been developed for this purpose. As mentioned before, Bhattacharyya and Klapperich (2006) concentrated DNA by SPE. The DNA was retained by a porous monolithic polymer column impregnated with silica particles, and subsequently eluted.

Larsen et al. (2008) detected single nucleotide polymorphisms (SNPs), important for both forensics and the detection of genetic disease factors. They attached the mutant and wild-type sequence to beads of different size, which were then separated by pinched flow fractionation. Gulliksen et al. (2005) amplified mRNA by real-time nucleic acid sequence-based amplification (NASBA), to detect markers for cervical cancer. There are also examples of MCE of DNA with both fluorescence (Hurth et al. 2008) and electrochemical detection (Castano-Alvarez et al. 2007) using COP microchips.

5 Conclusions and outlook

Recently, COP polymers have emerged as an interesting alternative to the most common polymers used in microfluidics. COP polymers have several favourable properties, such as low water absorption, high transmission in the near UV range and good chemical resistance.

There is a wide array of fabrication methods available. In order to choose the most suitable one, it is important to consider the resolution, surface roughness and manufacturing cost. Replication methods, such as injection moulding, hot embossing and NIL, are the ones most commonly used. Direct structuring of COP by laser ablation and micromilling has also been demonstrated. These methods are more convenient for fast prototyping, but have not yet been widely used and need further investigation.

COP polymers are increasingly popular for microfluidics and a multitude of chemical analysis techniques have already been implemented, e.g. electrophoresis, chromatography, SPE, IEF and MS. Surprisingly, few examples of microfluidic COP chips with integrated optics have been reported. COPs are, however, likely to have a bright future in optofluidics, considering their exceptional optical and chemical properties compared to other polymers.

In conclusion, we expect COP polymers to further increase in popularity for microfluidics, in particular optofluidics, due to their unique properties and ease of fabrication.