Strain-induced self-rolled-up microtubes for multifunctional on-chip microfluidic applications

Most microfluidic platforms are simply miniaturized versions of macroscale systems that can be used to translate toward studying or utilizing microscale phenomena and, therefore, over the past several years, microfluidic systems have become increasingly popular in industries and academia. Microfluidics provides the ability to integrate manual lab practices onto a single chip for a variety of applications. The ability to pattern and use micro- and nanoscale features on-chip for fluid applications offers the possibility of working with smaller reagent volumes to observe and understand macroscale phenomena and use that knowledge toward developing platforms with faster access and actuation times, enhanced chemical sensitivity, efficient particle/molecule transport, and access to higher degrees of freedom with parallel operations.^1,2 Therefore, going forward, the ultimate goal of microfluidic systems is to achieve the “lab-on-a-chip” (LOC) concept, wherein multiple aspects of modern biology, physics, and chemistry can be realized on a single microchip.

There can be several components involved in making a complete microfluidic system; however, the most important component is the fluidic channel. Different fabrication techniques such as micromachining, soft lithography, micro-molding, laser ablation, and in situ construction have been used to make channels out of many different materials ranging from polymers to glass. In terms of optimizing the physics of the microfluidic device on a macroscale, the cross-sectional shape and dimensions of the fluidic channel are of great importance. As such, cylindrical or semi-cylindrical channels that can eliminate boundary effects to a significant extent are always preferred. The above-mentioned fabrication techniques are, therefore, limiting as they can only create square, rectangular, or trapezoidal cross-sectional shapes for channels. Several other bottom-up and top-down techniques have been used to fabricate microtubular structures for sensing and micro-/nanomotor applications; however, integration with microelectronics and scalability is quite challenging and, therefore, on-chip device operations cannot be realized with such approaches rendering them to mere observational platforms. Fabrication techniques that can easily create microtubular structures along with integrated optoelectronics and microfluidics are of rising importance. Self-rolled-up membrane (S-RuM) technology is one of the most promising approaches that can be used to create near-cylindrical micro- and nanotube on-chip.

Going beyond the regime of epitaxial and high-temperature films, stress in thin films can also originate from several intrinsic effects during film growth. These stresses arise because generally films are deposited under non-equilibrium conditions. In general, any redistribution of matter will result in film stresses, since the film is constrained by the substrate.³⁹ For very thin films, surface energy plays a major role in straining the small crystallite islands. Initially, a small crystallite will have an equilibrium lattice parameter less than the thin film lattice parameter (a_f). However, as the crystallite grows, the equilibrium lattice parameter increases to a_f. At this point, since crystallites are constrained by the substrate, compressive stress is generated in the film. When these crystallites coalesce (increased film thickness) and reverse the substrate constraint, a tensile stress is generated. As crystallites grow, the gap between them decreases, until it is so small that cohesion begins to develop between the crystallites. At some points, the interactions between the crystallites are strong enough to close the gap by the elastic deformation of the crystallites. Therefore, films with smaller crystallite sizes have quite high tensile stress. Densification of films in one form or another naturally leads to tensile stresses in deposited films. A reduction in the grain boundary area provides a driving force for grain growth. On the other hand, grain growth results in increased film stress and strain energy. As the grain sizes increase, grain boundaries are eliminated leading to the densification of the film. Since grain boundaries are less dense than the grain lattice, this generates an overall tensile stress in the film.³⁹ Non-epitaxial polycrystalline films are known to have significant vacancy concentration. When a film is deposited at a low temperature, there is very little surface diffusion, and one would expect the vacancy concentration to be much larger than at equilibrium. When these excess vacancies are subsequently annihilated, the associated volume change results in stress in the film. The stress change depends on the vacancy volume and the site of annihilation. Similarly, impurities and phase changes in a film cause volume change and, thus, also contribute toward stress.³⁹ 

In comparison with other 3D and 2D microscale designs on chip,⁴⁰ S-RuM fabricated microtubes provide several inherent advantages. First, it provides a compact microfluidic chamber that can be modified for specific applications. The chamber can be made transparent for better optical imaging, which can be integrated with electrodes for sensing and electrokinetics or can be mechanically transformed to yield structures suitable for self-assembly and storage.^41,42 Second, the tubular/cylindrical nature of the cuffed-in electrodes can provide more uniform electromagnetic field distribution, resulting in improved sensitivity and throughput as compared to some of the planar and 3D variations in microfluidic devices. Third, S-RuM technology is derived from a MEMS (Microelectromechanical Systems) based planar processing technique, which can be easily scaled to the industrial level and can be used to fabricate microtubes on virtually any substrate. Wafer-scale integration of these microtubular devices with multiple channels on-chip can realize successful commercialization for more sophisticated needs in the field of molecular sensing, DNA-based data storage, on-chip micromotors, and flexible bioelectronics.^43,44 Moreover, once rolled, the tubes’ footprint can also be used for placing other microfluidic chambers on the chip for integration purposes. S-RuM microtubes are, therefore, a superior candidate for fulfilling the modern-day needs of advanced microfluidic channels. The microtubes, however, still need to be integrated with traditional microfluidics and several reports in the past have demonstrated a successful integration of S-RuM devices with PDMS (Polydimethylsiloxane) and glass-based microfluidic systems.

The S-RuM technology was first reported by Prinz et al. in 2001,¹³ and since its discovery, it has served as a powerful building block for a variety of applications like rolled-up passive electronics (miniaturized on-chip power inductors,⁹ transformers,³ filters,⁴⁵ and micro-supercapacitors^46,47 and THz antennas),⁴⁸ rolled-up photonics^49,50 (vertical micro-waveguides,⁵¹ resonators,^49,52,53 and lasers),^54,55 and rolled-up neural interface.^10,56 In this Review, we present an overview of S-RuM-based 3D microtubular devices developed for sensing, transport, and macromolecular manipulation for lab-on-a-chip applications and share some perspectives on future developments of this multifunctional 3D platform that could be integrated for improved performance for DNA data storage.

Sensor platforms based on the planar geometry not only suffer from low sensitivity levels but also tend to have a high standard deviation in sensitivity across the sensing area. A microtubular structure ensures a high surface area to volume ratio and, thus, enables us to push the limit of detection (LOD) to achieve higher sensitivity. The compact microtubular chamber provides a high level of uniformity in the adsorption/desorption of particles, the flow of fluid, and the detection of electromagnetic signals. As such, over the past decade, S-RuM microtubes have gained tremendous traction in the field of high-sensitivity sensors. Before rolling, the microtube surface can be functionalized appropriately to detect a variety of entities ranging from biomaterials such as cells, proteins, viruses, and DNAs to other non-biological materials such as inorganic/organic gas molecules, magnetic micro-/nanoparticles, and other volatile organic compounds (R6G, acetone, IPA, etc.). Due to the rolling mechanism, S-RuM fabricated microtubes are naturally cylindrical (polygonal or near-circular), thus enabling the amplification of electromagnetic waves via total internal reflection or due to the circular nature of internal electrodes. In this section, we will be reviewing some of the works focused on optical and electromagnetic sensing performance, mechanisms, and integration of S-RuM-based microfluidic integrated sensors.

Due to recent breakthroughs in photonics and microfluidics, optofluidics has emerged as a fast-growing field that allows us to manipulate optical and fluidic properties on a single chip. Optical characterization techniques such as Raman spectroscopy,⁵⁷ FTIR (Fourier transform infrared) spectroscopy,⁵⁸ and UV–visible spectroscopy⁵⁹ rely on the interaction of light with the fluid in question. With the recent advancement in detection principles, the contents and the nature of certain fluids can be easily summarized with the help of such techniques. However, on a microscale, research on improving the versatility and sensitivity of such methods is of great importance as it allows us to create on-chip devices that can be commercialized for in situ detection needs.^57–59 The S-RuM platform has emerged as a viable option for exploring such scenarios. Reports on integrated microtubular optofluidic sensors have demonstrated superior sensitivity levels along with lower LOD. In the past, sugar solution-filled SiO_x/Si based rolled-up refractive index sensors were demonstrated to have a sensitivity of ∼62 nm/RIU (Refractive Index Unit); however, the sensors were not fabricated as an integrated device.⁷ In 2012, Harazim et al. demonstrated refractive index sensing via the integration of glassy SiO₂-based rolled-up optofluidic ring resonators (RU-OFRRs) into a lab-in-a-tube system on a glass substrate using a robust SU-8 polymeric structure.⁶⁰ The integrated structure simplified the sample (deionized or DI water) delivery process and minimized the sample volume down to picolitres. They were able to obtain an average Q factor of 2900 with a minimum detection limit of 3.4 × 10⁻⁴ per RIU and a maximum sensitivity of 880 nm/RIU, which is quite comparable to the state-of-the-art planar devices.^60–62 Capable multiplexed detection via the integration of multiple sensing channels with a PDMS packaging could also be achieved. Later in 2017, Madani et al. demonstrated a passive optofluidic sensor integrated on top of an SU-8 waveguide. Prestressed TiO₂ nanomembranes were rolled up to yield microtubes with a Q factor of ∼500 and a sensitivity of 130 nm/RIU [Figs. 3(a)–3(d)].⁶³ The detection limits in optically active resonators can be several orders of magnitude lower than the optically passive resonators; however, optically passive resonators require fragile optical components and complex optical setups to excite modes in the cavities and, thus, multiplexed detection can be challenging. To overcome such limitation, Madani et al., later in 2018, demonstrated the integration of multiple rolled-up microtubes onto the waveguide array [Figs. 3(e)–3(g)].⁶⁴ Their unique configuration allowed a multimode interferometer-based beam splitter to probe the system with one single input. The multiplexed sensing system was able to achieve sensitivities up to 330 nm/RIU and, thus, opened new pathways for simultaneous sensing of various fluids on a single chip. The concept of refractive index change for fluid sensing can be translated to single-molecule sensing as well. Chemical absorption/desorption of molecules on the surface of the microtubes is easily reflected into the spectral shifts, and using the same principle, specific rolled-up microtubular sensors have been fabricated to sense a variety of entities such as water molecules (humidity sensors),^65–67 NH₃ molecules (gas sensors), DNA molecules,⁶⁸ and several other volatile organic compounds.⁶⁶ Recently, Ma et al. have demonstrated the ultrasensitive (∼10⁻¹¹ M) detection of R6G molecules using rolled-up monolayer graphene via stressed Cr metal as a rolling platform.²⁸ R6G molecules, as a widely used organic dye in textile industries, often require high-performing fluorometers for their detection in water. In the past, Yin et al. fabricated graphene/Au/SiO_x based rolled-up sensors to detect the laser-induced photodegradation of R6G molecules.⁶⁹ However, only one side of graphene was exposed to the molecule and, therefore, they were only able to achieve a LOD of 10⁻⁶ M (5 orders of magnitude lower than the most recent rolled-up graphene sensor).²⁸ The Cr-integrated rolled-up graphene sensor by Ma et al. had a much larger surface area to volume ratio for the R6G molecule to interact with graphene and, thus, was able to achieve a much higher sensitivity. Moreover, the CMOS-compatible integration of Cr metal in their device can further be extended as a platform for resistive sensing and serve as a standalone on-chip molecule sensing unit. Similarly, perturbations in the optical resonance spectrum have been utilized to detect dielectric particles (polystyrene microspheres)⁷⁰ and cells (mouse cells).⁷¹ And for fluids that do not indicate a major change in the refractive index or wavelength shifts, a change in the optomechanical response (mechanical eigenmodes) due to fluid-membrane interactions can be utilized for sensing purposes.^72–74 

While on-chip optofluidic sensors offer a simplistic model for sensing a variety of analytes in a microfluidic device setting, their use in the biomedical industry is still limited due to their inefficiency in detecting macromolecules like DNA.⁶⁸ To overcome this need, electrode-based microtubular sensors have been explored. These sensors rely on the principles of electrochemistry and microelectronics to detect macromolecules like DNA and protein, biomaterials like neuron cells, a wide range of organic solvents, and even magnetic particles. The S-RuM platform provides easy integration of electrodes into the microtubular structure and, thus, small-scale perturbations in electrical signals can be amplified for the high-sensitivity detection of such entities in a microfluidic system. Medina-Sánchez et al. fabricated S-RuM integrated microfluidic devices for attomolar level (20 aM) detection of H1N1 DNA.⁴¹ As shown in Figs. 4(a)–4(e), the interdigital configuration of Au/Cr electrodes was patterned on stressed bilayer TiO₂ thin films, which, upon release, yielded microtubular devices of diameter ∼28.5 μm. The PDMS-enclosed microfluidic devices were used to compare the change in impedimetric responses due to DNA hybridization. The rolled-up sensors showed 4 orders of magnitude higher sensitivity compared to their planar counterparts [Figs. 4(f) and 4(g)]. Similarly, Blick et al. reported high-frequency in-flow cell detection.⁷⁵ A tubular coplanar waveguide (T-CPW) was embedded inside a GaAs/InGaAs microtube [Figs. 4(h)–4(j)]. The diameter of the microtube sensor (∼8.5 μm) was carefully optimized to match the diameter of cells and, therefore, enhance the signal-to-noise ratio (SNR) of the impedance-based flow sensor. Previously in 2013, a strained GaAs/InGaAs thin film configuration was used to fabricate rolled-up metal-oxide-semiconductor (MOS) thin-film transistors.⁷⁶ The authors carefully optimized the doping concentration in the GaAs channel for the rolled-up device to obtain a more sensitive current response compared to the planar configuration. The cm-long microtubular sensors, although they were not completely integrated with a microfluidic circuit, were used to successfully demonstrate the detection of polar organic solvents such as water, acetone, and propanol. The authors optimistically indicated that such monolithically integrated capillary devices could be used for efficient fluidic cooling of next-generation high-speed computing systems. Several other reports on the use of S-RuM microtube sensors demonstrated the electronic sensing of other entities such as ionic liquids,⁷⁷ neuron cells,⁷⁸ other organic solvents, and volatile organic compounds.^76,79

Since modern diagnostic and therapeutic techniques, applied in medicine and biology, often rely on magnetic nanoparticles, in-flow detection of magnetic particles is crucial. High-throughput analysis of magnetic nanoparticles requires an in-flow detection system, wherein certain organic and/or inorganic solvents containing particles with a permanent magnetic moment could easily be detected by an integrated sensor. For a planar configuration, the detection is limited to the near-surface region; however, for a 3D tubular magnetic sensor, more degrees of freedom can be achieved as magnetic stray fields can now be sensed in virtually all directions. One such integrated microfluidic GMR (Giant Magneto Resistance) sensor has recently been demonstrated by Mönch et al.⁸⁰ A self-rolled-up membrane structure based on 30 multilayers of Py/Cu revealing GMR was rolled up with integrated electrode configuration. The sensors were arranged in a Wheatstone bridge configuration to enhance the differential sensitivity [Figs. 5(a)–5(c)]. The rolled-up tube itself was efficiently used as a fluidic channel, while the integrated magnetic sensor was used to detect ferromagnetic CrO₂ nanoparticles (GMR effect) with a sensitivity of 2.6 Ω/mT [Figs. 5(d) and 5(e)].

Although recent advances in rolled-up compact microfluidic sensors have shown superiority in demonstrating in situ optical, electrical, and magnetic responses, several other integration schemes remain worth exploring. For example, the heterogeneous integration of such rolled-up membranes with two-dimensional (2D) materials such as graphene, transition metal di-chalcogenides (TMDs), and h-BN will open doors for achieving higher sensitivity, faster response time, and lower LOD. Moreover, the compact microtubular structure will always provide a higher SNR, higher interaction volumes (per analyte volume), and easier integration with traditional microfluidics for next-generation lab-on-a-chip applications. Apart from the integration of other functional materials with the tubular configuration, creating a rolled-up metamaterial configuration instead of stacking metal layers could provide enhanced sensing performance with the added advantages of easier fabrication and a smaller footprint.⁸¹ From a detection point of view, S-RuM integrated sensors, thus, provide a great opportunity to create a functional lab-on-a-chip device. However, to create an advanced lab-on-a-chip configuration, other components such as material synthesizers, transporters, and holding chambers will also need to be integrated alongside the sensing units. The S-RuM platform provides unique solutions for creating other components as well.

Self-propelled micro-/nanomachines hold immense potential in taking LOC microfluidic devices to comprehensive functionalities by assisting in cargo delivery and sensing applications. In terms of cargo delivery, functionalized micromotors can be propelled through microfluidic channels to other specific compartments on-chip delivering essential biomolecules like DNA for further analysis (sensing or synthesis). Ease of functionalization, ease of integration with existing microfluidic platforms, and demonstrated autonomous transport make micromotors an essential component for LOC applications.

Micromotors ranging from spherical to tubular shapes have been reported for a myriad of applications.⁸² Tubular-shaped micromotors are of special interest due to several reasons.^83,84 First, tubular micromotors can reach ultrafast⁸⁵ propulsion speeds (>1000 μm/s), which is essential for transporting heavy cargo. Second, the tubular structure provides a compact carriage for holding the cargo and keeps it protected over long propulsion distances in specific environments. Third, the microscale size and tubular structure provide ample opportunities to modify the micromotor structure for unidirectional propulsion in small channels. Last, the ease of fabrication makes them viable candidates for integrated LOC devices. An array of microtubes can be fabricated on-chip that can function as a collection of micromotors with varying speeds and sizes [Fig. 6(a)]. Tubular micromotors are primarily fabricated via template-assisted techniques or using the S-RuM platform. S-RuM fabricated micromotors are easier to integrate with the microfluidic devices since the planar membranes can be strategically patterned, functionalized, and placed on specific locations inside microfluidic channels before their release [Fig. 6(b)]. Moreover, the fluid in the channels can be modified to not only release the membranes but also propel the resulting tubular micromotors along the microfluidic channels. Here, we will review some of the S-RuM micromotors and their demonstrated applications that can be translated for use in next-generation LOC applications.

First, tubular micromotors fabricated using the S-RuM technology by Mei et al. utilized the Ti/Fe/Au/Ag multilayer system.¹⁹ The ∼100 μm long micromotors were able to achieve discrete velocities of up to 720 nm/ms. The Ag lining inside the micromotors catalytically reacts with the hydrogen peroxide (H₂O₂) fuel to decompose the fuel into O₂ bubbles (and water). The O₂ bubbles then exit out of the microtube propelling the tube forward. Later in 2014, Mei et al. engineered the catalytic surface (nano-porous walls) of the rolled-up microtubes to enhance the catalytic activity and were able to achieve high-speed (∼1400 μm/s) micromotors.⁸⁵ Solovev et al. used the same Fe-integrated multilayer configuration to fabricate micromotors that could be remotely guided and controlled via an external magnetic field.⁸⁶ The frequency of a rotating magnetic field varied from 0.3 to 3.3 Hz to control both the speed and direction of the micromotors. Because of the rotating magnetic field, the micromotors aligned themselves along the circular field lines and were, therefore, able to localize within a sub-20 μm radius. With the combination of sharp-tipped conical geometry and integrated magnetic materials,⁸⁷ these micromotors were demonstrated to be used as guided micro-drillers allowing them to embed themselves within certain biomaterials such as cells⁸⁸ and tissues.⁸⁹ Such micromotors when introduced into a LOC microfluidic system would often require moving against the direction of fluid flow to reach its target chambers (cell separation). Therefore, the propulsion must be powerful enough to move against the flow. The propulsion speed can be increased either by using a higher concentration of the fuel or by external forces. Increasing the fuel concentration in LOC devices could introduce other redundant chemical changes that might affect their performance. When changing the concentration of the flow medium is out of option, magnetic field-guided micromotors could be used efficiently to overcome the flow barriers. Schmidt et al. demonstrated such control over S-RuM micromotors on-chip by utilizing magnetic fields (∼2 mT) to impart excess energy to micromotors going against the flow in a microfluidic channel.^90,91 As shown in Figs. 7(a) and 7(b), S-RuM micromotors integrated into a microfluidic chip were used to transport polystyrene microparticles inside microfluidic channels. As discussed before, the localization radius (∼500 μm) of such micromotors was also efficiently controlled via external magnetic fields. Schmidt et al. later designed a closed-loop magnetic field motion control system [Fig. 7(c)] surrounding a 3D space (containing magnetic micromotors) that was able to achieve an average region of convergence of ∼300 μm for such micromotors [Fig. 7(d)].^92,93 As shown in Fig. 8(a), the closed control loop system was also used to demonstrate point-to-point motion control (localization radius of ∼90 μm) of sperm-flagella-driven magnetic micromotors with a low magnetic field strength of 1.39 mT.⁹⁴ Such micro-bio-robot schemes can be easily translated to be used for drug delivery, sperm delivery, or other biomedical LOC applications.

Electric and magnetic field allows the S-RuM micromotors to reach higher degrees of freedom by having good control over their speed, direction, and localization within the microfluidic channel; however, from a biocompatibility point of view, the integration of certain metals such as Fe and Ni into the microtubes might not always be feasible. Micromotors in that case will need other forms of motion control mechanisms. Schmidt et al. demonstrated that the propulsion of S-RuM micromotors can also be controlled by an external thermal source. An increase in the temperature of the peroxide up to 37 °C accelerates the catalytic reaction and, therefore, increases the propulsion speed of the micromotors.^95,97 Schmidt et al. fabricated a microfluidic chip containing solutions of 10× diluted blood and controlled the temperature at 37 °C, to achieve the propulsion of S-RuM micromotors along with red blood cells (RBCs) [Figs. 8(b) and 8(c)].⁹⁵ Since the diameter of the microtube also determines the propulsion speed of the motor, temperature could also be directly used to reconfigure the shape (diameter) of the micromotor. Polymer-based rolled-up microtubes showed swelling/shrinking with small temperature changes and, thus, such micromotors could be easily de-accelerated and stalled within the solution.⁹⁸ Solovev et al. showed [Fig. 8(d)] light-controlled propulsion of micromotors wherein S-RuM Ti/Cr/Pt micromotors were brought to complete rest (from an initial speed of ∼80 μm/s) within 12 s of light illumination.⁹⁶ Light with higher energy (or shorter wavelengths) degrades the peroxide fuel faster, and since most microfluidic systems are made of transparent materials (along with transparent S-RuM microtubes), light-assisted control over the motion of micromotors could be a viable approach.

In general, the high propulsion energy, direction control, and surface functionalization of fuel-driven micromotors have made them extremely attractive for diverse microchip applications. However, self-propelled micromotors could open new avenues for concentration-sensitive LOC devices.⁹⁹ The ease of integration of other functional materials like graphene¹⁰⁰ into S-RuM micromotors could render them as special on-chip components. The first attempts at pick-up and delivery of single cells by tubular micromotors were made by Sanchez et al in 2011.¹⁰¹ Such systems could easily be extended toward micromotors carrying functionalized DNAs¹⁰² for applications related to on-chip DNA-based cryptography¹⁰³ and data-storage.^104,105 Since the S-RuM technology is based on planar processing techniques, it allows easy integration of virtually any material into the micromotor system. As we have shown that such integrations allow for better performance in speeds, directionality, control, cargo delivery, and assembly, ultimately paving the way for fabricating highly intelligent micromotor components for LOC applications.

Integrated microfluidic devices with electronic,^41,106 optoelectronic,^107,108 magnetic,^109,110 thermal,^111,112 or acoustic¹¹³ functionality are well suited for serving as lab-on-a-chip (LOC) systems that enable a multitude of applications ranging from the characterization of minuscule liquid samples to running massive parallel bioassays. For the same reason, the field of electrokinetics has seen tremendous growth, as next-generation devices are able to provide more degrees of freedom toward the assembly and manipulation of micro-/nanosized particles, live cells,^114,115 and biomacromolecules, such as proteins¹¹⁶ and even DNA.^117–119 Three major electrokinetic phenomena,¹²⁰ namely, electrophoresis,¹²¹ di-electrophoresis (DEP),¹²² and electroosmosis,¹²³ are the basis of micro-/nanoparticle manipulation in aqueous solutions. DEP is applicable with both charged and neutral particles and can be useful to distinguish their behavior in either uniform or nonuniform electric fields. Electrophoresis uses an external electric field and measures the motion of charged particles relative to the liquid medium in which they are suspended. On the other hand, electroosmosis is the motion of a liquid that contains a net charge.

Techniques based on chemical modifications,¹²⁴ optical traps,¹²⁵ acoustic traps,^126,127 spontaneous fluid flow (microfiltration¹²⁸ or inertial microfluidics),¹²⁹ and electrokinetics¹²⁰ have all been used and studied extensively for their uses in particle or molecule manipulation. However, from an integrated device point-of-view, electrical methods are superior. There are several advantages of using electric fields for the on-chip manipulation of charged particles. Many parameters, including the magnitude and frequency of the signal, electrode distance, and electrode geometry, can be tuned to precisely control the field force exerted on the particles, which can help dictate the position of particles relative to electrodes or other particles in the system. Another major advantage of using an electric field stems from the ease of fabricating 2D (planar) or 3D microelectrodes on chips. Most microfluidic chips usually adopt 2D planar electrodes as they can be easily patterned using standard lithography techniques, and the planar nature makes it easy to encapsulate and package. However, to reach higher levels of efficiency, the integration of 3D electrodes with traditional microfluidics is of utmost importance. Three-dimensional electrodes provide a better definition (uniformity), larger spatial control, better control over the gradient, and a larger magnitude of the electric field. In some cases, vertical electrodes have been extruded,¹³⁰ injected,^131,132 electrodeposited,^133–135 or patterned^136–138 on the sidewalls or inside rectangular microfluidic channels, or top-bottom electrode sandwich configurations have been used to study cell separation and particle manipulation. However, the integration of 3D electrodes with microfluidic channels is still challenging as the methods that have been explored so far are expensive, prone to error, or limited by the design of electrodes. Moreover, these vertical configurations of electrodes are not “truly” 3D as they fail to cover either the side walls or the corners in a traditional rectangular microfluidic channel. To avoid such inconsistencies in electric field distribution and to eliminate redundant boundary effects, the integration of microtubular electrodes is of rising importance.^139–141 

With rapid proliferation of data, there has been a tremendous amount of research toward improving state-of-the-art data storage devices such as HDD (Hard Disk Drives),¹⁴² SSD (Solid State Drives),¹⁴³ and RAM (Random Access Memory)¹⁴⁴ based technologies in terms of latency, power consumption, and storage density. IDC (International Data Corporation), one of the world's leading market intelligence companies, has predicted that the world's total data output can grow up to 175 ZB by the year 2024 alone. Therefore, it is imperative to explore other forms of high-density data storage systems. In the quest for ultra-dense storage systems, DNA-based macromolecular data storage¹⁴⁵ stands out as a candidate for massive storage media due to its several unique properties. However, in order to contrast the current data storage mediums, DNA-based storage needs a precisely controlled and automated on-chip platform able to have controlled random access and rewriting, a high durability, and appropriate error control. We envision that a 3D platform such as S-RuM built on 2D planar processing can be optimized to provide a much more uniform electric field and, thus, better manipulate DNA inside integrated electronic devices with nano- or microscale resolution compared to microfluidics based on nontubular electrode designs. As discussed in the introduction section, S-RuM technology provides a viable approach toward such integration as it allows virtually any electrode design and electrode material to be rolled-up into a microtubular structure. So far, there has been only one record demonstrating its use in manipulating charged microspheres and low-Mw DNA. Recently, Khandelwal et al. have fabricated AlN-based microtubes with cuffed-in 3D circular electrodes, which were energized by DC (direct current) voltage (DC DEP) to manipulate/assemble particles and to localize/trap DNA on-chip [Figs. 9(a) and 9(b)].¹⁰⁵ The compact structure not only provided precise fluid control but also reduced sample consumption; additionally, the strong confinement effects enabled efficient particle manipulation [Fig. 9(c)]. Voltages between 2 and 4 V (3 orders of magnitude lower than present DC DEP devices)¹²⁰ can be used to generate a much higher e-field (>10⁴ V/m) between the electrodes, thereby minimizing power consumption and Joule heating effect. The tubular electrode geometry provided uniform field lines between two coaxial cylindrical electrodes to facilitate DNA localization inside the channel [Fig. 9(d)]. The current design can easily be adapted for alternating current DEP-based manipulation techniques that would require an even lower peak-to-peak working voltage. This renders these devices a viable option for both LOC cell separation applications and DNA-based data storage applications.

In summary, we have reviewed some of the recent advances of S-RuM technology for LOC sensing, delivery, and manipulation of particles and molecules. The design and performance of various on-chip S-RuM-based optical, electrical, and magnetic sensors are discussed. The CMOS-compatible fabrication flow and the enhanced performance enabled by the additional degrees of freedom provided by the S-RuM curvature make S-RuM sensors highly promising for integrated LOC applications, including S-RuM micromotors for cargo delivery applications and S-RuM electrode-controlled particle and macromolecule (DNA) manipulation. Looking ahead, we believe that the field of DNA-based data storage technology could strongly benefit from the integration of the S-RuM platform with microfluidics. For example, the S-RuM technology can be used to create a multifunctional platform for several biomedical or non-biomedical applications that would require a combination of on-chip molecule sensors, transporters, and manipulators. The functionalities that the S-RuM structures can enable include delivery, separation, and manipulation, all in a self-contained, CMOS-compatible, miniaturized platform. Critical challenges lie in the packaging and integration to ensure reproducibility and reliability. Overall, the S-RuM platform offers tremendous industrial level scalability; with orders of magnitude size advantage when benchmarked with respect to the state-of-the-art, the S-RuM technology stands out as a niche technology capable of offering solutions to very specific problems in the field of on-chip passive electronics, bio-microfluidics, sensing, and other LOC applications.

This work was supported, in part, by SRC 2018-SB-2839, NSF CCF No. 18-07526, and NSF ECCS No. 2200651.

The authors declare no competing interests.

Apratim Khandelwal: Conceptualization (equal); Visualization (equal); Writing – original draft (equal); Writing – review & editing (equal). Xiuling Li: Conceptualization (equal); Supervision (equal); Visualization (equal); Writing – original draft (equal); Writing – review & editing (equal).

The data that support the findings of this study are available from the corresponding author upon reasonable request.