Although µTAS are rapidly maturing, ongoing efforts continue to yield improvements and innovation in fabrication technologies and modular subcomponents on-chip. In developing an analytical device, researchers begin by fabricating a prototype and testing unit operations, such as preparing, handling, and detecting small volumes of samples. This section reviews new device fabrication methods, addresses sample preparation and handling, and introduces innovative microfabricated detectors. Interestingly, these areas of continued progress represent the first and last steps in analysis. Microchip separations, which fall between sample preparation and detection, are now quite sophisticated, perhaps because many methods transfer readily from macroscale systems. Consequently, new separation methodologies on-chip are included in the section on integration and automation (Section 2). Here, we concern ourselves with the important and challenging processes that bookend any micro-analysis.
To choose a material for an application, several characteristics must be considered: cost, robustness, surface chemistry, optical and electrical properties, biocompatibility, ease of fabrication and integration, and feasibility of large-scale production. Depending on the application, devices may be fabricated from polymers, glass, silicon, paper, fabric thread, or a combination of these materials. Fabrication technologies are categorized here based on the type of material used.
Among polymers, polydimethylsiloxane (PDMS) continues to be the most popular material for µTAS applications due to its easy fabrication, physical properties, and low cost. Among 255 references in this article, 144 studies use PDMS. Replica micromolding of PDMS against a master is well established and widely used, and a recent protocol, including a tutorial video, provided detailed instructions on making PDMS microfluidics.1
On-going microfabrication efforts in PDMS focus on developing simple approaches to generate complex devices. A 3-dimensional (3D) microfluidic channel with a near-perfect circular cross section was obtained by a metal wire removal process.2
Through-holes between layers in 3D microfluidics were generated simply and on-demand by high-throughput, localized tearing of a PDMS membrane.3
The elastomeric property of PDMS is often exploited in building functional biomimetic devices. To reconstitute organ-level lung functions on a chip, a two-layer PDMS system contained two closely apposed microchannels separated by a thin, porous, flexible membrane (), and physiological breathing movements were mimicked by cyclic stretching of the flexible membrane using a vacuum applied to side microchannels.4
Although academic laboratories favor PDMS for rapid prototyping, researchers need to be aware of its limitations, such as rapid hydrophobic recovery after surface oxidation, adsorption and absorption of hydrophobic molecules, and swelling in common organic solvents.
Figure 2 Microfabricated devices. Recent advances in microfabrication include (a) improved replication of in vivo conditions, (b) development of modular units, and (c) the use of solvent-resistant materials. (a) A lung-on-a-chip microfluidic device was composed (more ...)
Plastics, including poly(methyl methacrylate), polystyrene, polycarbonate and cyclic olefin copolymer, are increasingly common alternatives to PDMS. These materials can be processed by hot embossing or injection molding for high throughput and cost-effective mass production of microfluidic devices. In academic laboratories, hot embossing is more suitable than injection molding due to the relatively low cost of embossing equipment. For example, inexpensive and robust masters were recently fabricated photolithographically from SU-8 photoresist on copper substrates, then used for hot embossing of microfluidic reactors in a range of thermoplastic polymers including cycloolefin, polycarbonate, and UV-transparent acrylic polymers.5
Polystyrene, the most commonly used material for in vitro
cell-based research, was rapidly prototyped by embossing and bonding.6
In addition to hot embossing and injection molding, other fabrication methods were used for plastic lab-on-a-chip devices, including microthermoforming,7
This casting method generated prefabricated microfluidic blocks of epoxy SU-8 from flexible silicone molds. The blocks were quickly assembled into sophisticated microfluidic devices for a wide range of applications, potentially allowing laboratories to prototype new devices from pre-made blocks without investing in fabrication infrastructure (). Recent research also explored specialty polymers for microfluidic applications. Fluorinated thermoplastics, such as Teflon, were processed by a thermal embossing method using PDMS as master to yield Teflon microfluidic chips that exhibited extreme resistance to organic solvents ().10
A photosensitive polymer formulation, SU-8 photoresist, was used for fast prototyping of monolithic 3D micro-systems by a mask-less micro-projection lithography platform.11
Plastics overcome some limitations of PDMS, but their relatively complex fabrication still presents a bottleneck for widespread use for prototyping in academic laboratories. Relative to PDMS, plastics remain more suitable for industrial mass production of microfluidic devices rather than the rapid prototyping seen in most academic laboratories.
Glass and silicon
Compared to polymers, other materials are used for a relatively small proportion of published µTAS devices, most likely due to their more complex fabrication steps. Among inorganic materials, glass is most attractive due to its chemical inertness, optical transparency, and thermal stability. While fabrication of glass-based microfluidics generally involves etching and bonding, a new fabrication method used a spatiotemporally focused femtosecond laser beam to generate hollow microfluidic channels with a circular cross-section.12
Unlike glass, silicon microfabrication is derived from the electronics industry. These mature fabrication methods enable the integration of various functionalities into a monolithic device. Complex 3D silicon microfluidic structures were produced by a single-mask, single-etch process by utilizing reactive ion etch lag, in which etching of a small trench lagged behind a large trench.13
The channel was then sealed by deposition of dielectric films, avoiding the need for wafer to wafer bonding. While their fabrication can be complex, glass and silicon devices are particularly suitable for electrowetting-based microfluidics, highlighted in Section 1.3.
Paper and thread
Paper and thread have been exploited as the new substrates for low cost, disposable analytical devices. Inexpensive microfluidic platforms can be fabricated from paper by a number of methods; one notable strategy simply treated hydrophobic paper with a computer-controlled CO2
Fabric thread was used to fabricate 3D and semiquantitative analytical devices by sewing it onto other materials, such as a polymer film.15
Silk yarns were treated with the appropriate reagent solutions, dried and handloom-woven in one step into integrated fabric chips.16
The wicking rate and the absorptive capacity were conveniently adjusted by varying yarn parameters such as the twist frequency and weaving coverage. Cotton yarn and knots were used to build low cost passive microfluidic circuits which were capable of combining, mixing, splitting, and serially diluting streams of liquid.17
Recent progress in fluid handling and detection on paper devices is highlighted in Section 2.1.
While most devices are manufactured by the microfabrication technologies mentioned above, many devices require components made from other materials to be integrated into the device to offer mechanical, electrical, optical, acoustic or magnetic functionalities. These components can be co-fabricated with the device or integrated on-chip after primary fabrication is complete. Often multi-component fabrication involves embedding a secondary material in a channel; for example, monolithic 3D porous silica microstructures in an SU-8 microchannel resulted in a four-fold increase in mixing efficiency,18
micropatterned carbon nanotube forests confined inside PDMS channels captured particles ranging over three orders of magnitude in size,19
and free-standing lipid membranes formed in a microfluidic chamber array provided a unique platform for studying membrane transport.20
To build a portable, inexpensive and low-power heater, silver-filled epoxy was injected into and solidified in a PDMS microfluidic chip.21
Optical readout has become the predominant detection method in lab-on-a-chip, driving efforts to merge microfluidics with photonic elements. An integrated multiple internal reflection system was built on a PDMS microfluidic chip for cell screening.22
This device included self-alignment of microchannels, microlenses, and air mirrors. PDMS was dyed with ink for integration of robust, low cost filters for optical sensing in disposable labs-on-a-chip.23
For electrical control, PDMS was mixed with multi-walled carbon nanotubes and processed by standard soft lithography into flexible 3D electrodes.24
As evidenced by these devices, a functional, integrated, multicomponent device requires both creative design and careful fabrication.
Microfluidic systems require precisely tailored surface properties for many applications, and a variety of surface modification methods are now available, including plasma treatment, formation of self-assembled monolayers, application of dynamic or semi-permanent coatings, or covalent grafting of synthetic and biological molecules. Surface modification was used to aid the irreversible bonding of various plastic substrates with PDMS by forming a chemically robust amine-epoxy bond at the interfaces.25
Since one of the limitations of PDMS is its unstable surface properties after modification, surface modification methods that generate long-term stable surface properties are needed. Protein-reinforced supported bilayer membranes offered long term stability of chip performance, even when stored in a dehydrated state for up to one month.26
1.2 Sample preparation
Sample preparation is often the most challenging step for an integrated microdevice. Since raw biological samples contain a complex mixture of compounds, a preliminary purification and/or concentration step is often essential. Extensive sample preparation “off-chip” greatly reduces the utility of µTAS, so these processes should take place on-chip whenever possible. On-chip processing is particularly important for low-volume, rare or valuable samples and is critical to realizing true sample-to-answer µTAS. However, the diversity of sample preparation methods has required that each technique be optimized for on-chip use independently. Consequently, the miniaturization of conventional sample preparation has lagged behind the development of analysis and detection techniques. Nevertheless, recent progress in microscale sample preparation techniques is evident.
Extraction and purification
Raw biological samples (e.g. blood, sputum, tissue, soil and food) often consist of a variety of components, so the analytes (e.g. nucleic acid, proteins, plasma, cells) must be extracted or purified from the raw samples prior to analysis. On-chip extraction and purification lead to substantial reduction of workloads and sample volumes. Separation of plasma from whole blood is highly desirable for minimizing the noise associated with blood cells during analysis. On-chip plasma separation was realized by utilizing either capillary force through a disposable bead-packed microchannel,27
or hydrodynamic force in a microfluidic network featuring a series of constrictions and bifurcations.28
On-chip extraction and purification of rare cells from blood has become a promising diagnostic tool and will be covered in Section 3.2.
Deoxyribonucleic acid (DNA) extraction involves the removal of DNA from the cells or viruses so that further analysis can be performed on the genetic material. DNA was extracted from whole blood samples using a microchannel packed with a silica solid phase and a standard syringe pump as a single pressure source driving the extraction process in a total column volume of 1.2 µL.29
Continuous DNA extraction and purification from cell lysate was realized on a microfluidic chip based on phase-transfer magnetophoresis using superparamagnetic beads.30
Following DNA extraction, polymerase chain reaction (PCR) was used to amplify DNA molecules to an appropriate concentration for analysis (see Section 2.4).
Disease biomarkers, an important class of diagnostic tools, are often present in serum, saliva, or tissue in minute concentrations. Thus detection of these early disease indicators frequently requires extraction and purification. Using a microfluidic purification chip, protein biomarkers such as cancer antigens were extracted from a 10-µL sample of whole blood via surface binding with their cognate antibodies.31
They were then released from the chip in concentrated form by photocleavage for label-free detection. Using a micromagnetic separation chip, low-femtomolar concentrations of target proteins were extracted from serum by trapping magnetic beads coated with capture antibody and aptamers.32
After removal of the external magnetic field, the bead-bound target complexes were eluted from the chip for detection. This process can potentially be expanded for multiplexed detection of other protein biomarkers and biomolecular targets through appropriate design of the aptamers.
Concentration and dilution
After purification, the concentration of analytes can be adjusted on the chip by concentration, dilution, or gradient formation. Numerous on-chip methods for concentrating samples bring them within the limit of detection, allowing researchers to choose the best option from a wide variety of techniques. For example, analytes were preconcentrated on microfluidic devices to detectable levels using a highly ion-conductive charge-selective polymer structure,33
a thermoswitchable poly(N-isopropylacrylamide) hydrogel plug,15
microscale isoelectric fractionation (µIF) membranes,34
a porous polypropylene membrane,35
or a massively parallel nanofluidic device.36
On the other hand, analytes were diluted on-chip to generate a large number of universal stepwise monotonic concentrations with a wide range of logarithmic and linear scales, which will be useful for high throughput screening.37
Gradient formation on microfluidic chips is particularly useful for controlling the cellular microenvironment, covered in Section 3.4.
Eliminating sample preparation
The inability to detect analytes in complex mixtures (e.g. blood, tissue biopsies, etc.) has been a major stumbling block in the development of µTAS. While many important efforts focus on on-chip purification and concentration to increase selectivity and sensitivity, µTAS capable of performing these operations can be complex to the point of impracticality. Additionally, µTAS devices that directly analyze samples without extensive sample preparation will be useful in developing countries and other settings lacking extensive laboratory infrastructure. The most robust platforms may be those in which sample pretreatment is not necessary; either because the complex sample is examined in its entirety or because off-target substances do not significantly affect detection of the target molecule. A paper spray method offered a novel, inexpensive, rapid method of direct analysis of a complex mixture such as whole blood for mass spectrometry.38
Analyte traveled through the porous paper medium by capillary action while blood cells were retained, so nearly no sample preparation was required prior to analysis. Chromatographic separations were also performed prior to paper spray, and this technique is likely to have a major impact on the paper microfluidics sector (Section 2.1). A functionalized gold-nanoparticles (GNPs) sensor was developed for detection of volatile organic compounds (VOCs) from exhaled breath.39
The array may offer promise in differentiating between ‘healthy’ and ‘cancerous’ breath for some patients. Since only exhaled breath is collected for direct analysis, absolutely no sample preparation is needed for this technology.
1.3 Sampling handling
Sampling handling has always been a strength of microfluidic systems, which provide simple and precise control of small volumes of fluids. Impressive new approaches continue to be reported for precise positioning, mixing, and splitting of samples in µTAS applications.
Droplet microfluidic systems continue to advance sample containment and sample handling. While passive droplet generation and mixing are well-understood, on-demand capabilities and more complex manipulations are still emerging. Notable recent work in this area includes novel droplet generation schemes, as well as new sorting, reagent introduction, and sampling techniques. Passive droplet generation systems achieve high throughput and low volume variability at equilibrium, but droplets cannot be produced on-demand and volumes cannot be rapidly modulated. An innovative laser pulse-driven droplet generation mechanism, in which a cavitation bubble ejected droplets from an aqueous channel into an adjacent oil-filled channel, produced droplets ranging from 1 to 150 pL and at kHz rates ().40
Alternatively, other strategies produced droplets at low Hz rates, taking as compensation greater control over droplet size and composition. As a culmination of earlier work, an automated microfluidic droplet-generating system called Droplab used a syringe pump to pull samples into the tapered end of a short capillary that was physically moved between samples via an automated sample presentation system.41
Such precise control over droplet size and composition has not yet been demonstrated with alternative systems.
Figure 3 Droplet microfluidics. Recent advances (a) generate droplets on-demand, (b) synthesize lipid bilayer-enclosed droplets, and (c) selectively add reagents. (a) On-demand droplet generation by laser pulse used two microfluidic channels connected by a nozzle-like (more ...)
Other recent research focuses on controlling droplet transport. A novel approach to droplet sorting used piezoelectric membrane actuation to temporarily alter the hydrodynamic resistance of two daughter channels, thereby diverting a passing droplet into a side channel.42
Alternatively, a permanent difference of hydrodynamic resistance between daughter channels was exploited as a reliable method of splitting sample plugs, providing a simple, passive mechanism of “sampling” from plugs in segmented flow.43
Droplets were also manipulated using microfabricated depressions and canals within the walls of microfluidic channels; droplets squeezed into these structures were reversibly anchored in place or guided along a distinct path.44
A number of experiments that could be performed in droplet microfluidic systems require droplets to be modified post-production—modifications include introduction of additional reagents, diversion of some or all of a droplet toward detection equipment, and selective removal of particular analytes from the droplet. In recent work, reagents were introduced into surfactant stabilized droplets by bringing them into close physical contact with a side channel containing a pressurized reagent. An electric field destabilized the droplet/carrier phase interface to allow transient fusion with the reagent channel contents ().45
If the electric field was switched off, fusion did not occur, allowing selective reagent addition to specific droplets. Multiple, individually controlled injectors manufactured in series permitted much more complex reagent addition schemes than were previously possible. The reverse operation – sampling from droplet contents – can also be performed. Application of an electric field extracted samples from droplets into an aqueous buffer channel running underneath a main channel. This sampling methodology allowed traditional microchip capillary electrophoresis (CE) methods to be interfaced with droplet microfluidics.46
In contrast to the previous method, in which an unmodified portion of a droplet was sent to a detector, other methods increase selectivity and sensitivity by sampling analytes in a chemically-specific fashion, for example following liquid-liquid extraction. An electrowetting-on-dielectric (EWOD) device was used to position, merge, mix, and separate aqueous and ionic liquid droplets for solute extraction.47
The nature of the carrier phase is of great interest to a biologist seeking to use droplet microfluidics. Aqueous droplets do not mix with a hydrophobic carrier phase oil to any appreciable extent—however, these aqueous droplets often contain substances which will readily partition into an exterior environment of hydrocarbon or silicone-based oil. Depletion of non-polar solutes from aqueous droplets can have very serious consequences, especially when working with media for intermediate or long-term cell culture. The use of fluorocarbon oils goes a long way toward remedying this problem, but a clever alternative exists in replacing the oil phase with an aqueous carrier phase. In this case the droplet compartment is isolated from the outside environment by a lipid bilayer. Aqueous droplets were sheathed in a thin layer of lipid from the oil phase by passing them by a “skim” before ejecting them into an aqueous channel ().48
Although these synthetic vesicles were prepared with an eye toward performing reductionist biological studies (e.g. examining membrane protein biophysics in a simple system), this method also presents an innovative way of sequestering solutes in a digital microfluidic system.
Mixing, actuating, and pumping
Mixing, moving, and pumping liquids at the microscale constitute the basic unit operations of lab-on-a-chip devices, required for a variety of chemical and physical operations on-chip. Often, the rapid, controlled mixing afforded by microfluidics facilitates otherwise impossible observations of biochemical kinetics. In microfluidic devices, hydrodynamic flow is always laminar because of low Reynolds numbers; however, rapid and efficient mixing is possible using carefully designed geometrical constraints. For example, an ultrafast microfluidic mixer using 3D flow focusing achieved mixing on a 10-µs scale.49
This type of rapid mixing in microfluidic systems allowed intimate investigations of protein folding. For example, one microfluidic mixer examined intramolecular diffusion of an unfolded protein during folding.50
Single-molecule fluorescence resonance energy transfer combined with microfluidic mixers permitted direct observation of individual biomolecule conformations on millisecond to second time scales.51
Another single-molecule microfluidic mixer enabled detailed mapping of the binding-induced folding kinetics of proteins.52
Integration of valves and pumps with microfluidic mixing extended these devices’ applicability; one integrated device automatically performed reagent titrations to screen multidimensional chemical space for conformational and enzymatic changes in biomolecules.53
Combined with small angle X-ray scattering, another system provided a customized tool to study reaction kinetics of biomolecular assembly processes, such as intermediate filaments.54
Microactuators and pumps are also useful for manipulating small sample volumes, and recent research develops mechanical alternatives to electroosmotic flow. A miniature peristaltic pump consisted of a helical bundle of microfluidic channels wrapped around a central camshaft and was operated by manually or mechanically rotating the non-cylindrical cam, which compressed the helical channels to induce peristaltic flow.55
To sort mouse embryoid bodies, electrically-driven hydrogel actuators operated at low driving voltages (<1.2 V) and in cell culture media without damaging cells.56
Biomimetic pumping methods were another promising approach for moving small sample volumes. For example, magnetically actuated artificial cilia were constructed from self-assembled chains of spherical superparamagnetic particles57
or from a flexible magnetic nanoparticle-PDMS composite.58
These artificial cilia mimicked the beating movement of airway cilia to generate fluid flow. Although neither type of cilia was integrated with microchannels to-date, the suitability of these structures for microfluidic application was evident and supported by theoretical results.59
Another example of bio-inspired pumping mimicked stomatal transpiration in plants and obtained a controllable flow rate of 0.13–3.74 µL/min in microfluidic systems.60
Free-standing micro- and nanomachines also offered new functionality, such as separation of drugs, cell sorting, and biosensing. A self-propelled catalytic Ti/Fe/Pt rolled-up microtube swam in a controllable manner within microfluidic channels.61
It easily loaded multiple cargoes and transported them to desired locations in the microfluidic chip. When the micromachine was functionalized with targeting ligands, it selectively captured cancer cells, opening a new approach for capture and isolation of rare cells from biological fluids.62
Increasingly sophisticated sample handling techniques are leading to generic devices suitable for a wide range of analyses – so-called “programmable microfluidics.” In some cases, flexibility and programmability are achieved by valving. A digital microfluidic platform composed of a 2-dimensional array of microvalves automated quantitative, multi-step biomolecular assays.63
A flexible microfluidic processor system with onboard pumps and valves performed metering, mixing, and reaction incubation in a series of molecular biology steps for messenger RNA (mRNA) amplifications.64
At the extreme end of programmable microfluidics are devices in which channels are formed on-the-fly from the assay buffer and sample themselves. A hybrid integrated circuit/microfluidic chip simultaneously controlled thousands of living cells and pL volumes of fluid, enabling a variety of chemical and biological tasks.65
A microfluidic device performed a variety of low- and high-level functions without hardware modifications; instead, each task was fully implemented by software programming.66
Programmable electrowetting manipulated droplets by application of electrostatic forces on an array of electrodes. In one study, directional channel formation, as well as splitting and merging, resulted in virtual electrowetting channels formed by application of voltage to an array of polymer posts.67
In another study, light triggered droplet transport on an open and featureless surface using a single-sided continuous optoelectrowetting mechanism.68
Continuous transport, splitting, merging, and mixing of droplets were possible. The ability to perform multiple sample processing steps on demand also benefited these digital microfluidic systems. Porous polymer monoliths were formed in situ
to carry out a digital microfluidic solid-phase extraction using microliter droplets of samples and reagents.69
Surface acoustic waves (SAW) also provided a new route to program complex fluidic functions into a microchip. For example, a disposable phononic chip executed microcentrifugation for particle and cell concentration in microliter droplets.70
Simple, rugged devices
Devices with a multitude of chambers, valves, and connections are often impressive, but this complexity can lead to devices that are difficult and expensive to produce and more prone to failure. While some applications demand highly engineered devices, other assays can be adapted to simple, rugged devices with few moving parts and the potential for “hassle-free” usage. A self-powered, self-contained microfluidic blood analysis system extracted plasma from whole-blood and performed multiple protein binding assays with high sensitivity ().71
It did not require any external pumping, connections, tethers, or tubing to deliver and analyze whole blood. A similarly simple device self-digitized samples into a large array of discrete volumes; the user simply primed the chip with oil, introduced aqueous sample, which divided itself into an array of chambers, then followed with oil again ().72
The digitized samples can be mixed with additional reagents and removed for downstream manipulation or analysis. A microfluidic pipette was developed to enable high-resolution spatial control of the chemical microenvironment of selected single cells ().73
Because of its simplicity, it has potential to be a routine research tool in pharmacological and physiological studies of isolated biological cells. A common problem with complex microfluidic geometries is the probability of trapping air bubbles. A new microfluidic design, called a phaseguide, based on a step-wise advancement of the liquid–air interface using the meniscus pinning effect, gave complete control over filling and emptying of any type of microfluidic structure, independent of the chamber and channel geometry.74
Advances like these are necessary to bring microfluidic technology to a wider audience. Currently, few life scientists use microfluidic technology in everyday laboratory practice, but rates of adoption will increase if simpler, more rugged devices are developed. A new direction in microfluidic design is needed to combine robust simplicity with functionality.
Figure 4 Simple, rugged devices. New innovations simplify sample (a) transport, (b) handling, and (c) dispensing. (a) A self-priming, self-contained, tether-free system integrated volume metering, plasma separation from whole-blood, multiple biomarker detection, (more ...)
1.4 Microfabricated Detectors
Microfabrication lends itself to the improvement of established detection methods and development of new strategies. A photothermal detector, previously demonstrated with capillaries, was recently applied to microfluidics with great success.75
The detector used laser illumination to heat the sample by nonradiative relaxation. The resulting temperature change produced a corresponding change in viscosity and therefore conductivity. Prior implementation with capillaries used contactless conductivity measurements, but the ease with which electrodes can be interfaced with microchannels permitted contact measurements in the microfluidic system, dropping the detection limit for this label-free measurement to 5 nM. Microfabrication also lends itself to mechanical detectors, including cantilevers. Microfabricated cantilevers in PDMS channels were transiently deflected by non-specific binding of bovine serum albumin (BSA), and in future iterations, cantilever arrays could be individually coated with binding agents for analyte-specific detection.76
Microfabricated cantilevers were also incorporated in a DVD platform for high-throughput bio-molecular sensing, in which optics and mechanics from a DVD player were used to handle liquid samples and to read-out cantilever deflection and resonant frequency.77
Microfabricated detectors were particularly useful for detection of a single cell, single particle, or single molecule. A microfabricated cantilever containing a microchannel responded to the buoyant mass of particles with femtogram resolution as they flowed through it. Repeated measurements of individual bacterial and mammalian cells provided single cell measurements of instantaneous growth rates,78
and a recent advance allowed the detector response to be readout piezoelectrically, rather than optically.79
Another microfabricated detector rapidly counted and characterized nanoparticles passing through a nanoconstriction at rates up to 500,000 nanoparticles per second.80