Authors: Alejandro A. Dos‐Reis‐Delgado, Andrea Carmona‐Dominguez, Gerardo Sosa‐Avalos, Ivan H. Jimenez‐Saaib, Karen E. Villegas‐Cantu, Roberto C. Gallo‐Villanueva, Víctor H. Perez‐Gonzalez
Categories: Miniaturization, heating, lab‐on‐a‐chip, microfluidics, sensors, temperature, Review
Source: Electrophoresis
Temperature is a critical—yet sometimes overlooked—parameter in microfluidics. Microfluidic devices can experience heating inside their channels during operation due to underlying physicochemical phenomena occurring therein. Such heating, whether required or not, must be monitored to ensure adequate device operation. Therefore, different techniques have been developed to measure and control temperature in microfluidic devices. In this contribution, the operating principles and applications of these techniques are reviewed. Temperature‐monitoring instruments revised herein include thermocouples, thermistors, and custom‐built temperature sensors. Of these, thermocouples exhibit the widest operating range; thermistors feature the highest accuracy; and custom‐built temperature sensors demonstrate the best transduction. On the other hand, temperature control methods can be classified as external‐ or integrated‐methods. Within the external methods, microheaters are shown to be the most adequate when working with biological samples, whereas Peltier elements are most useful in applications that require the development of temperature gradients. In contrast, integrated methods are based on chemical and physical properties, structural arrangements, which are characterized by their low fabrication cost and a wide range of applications. The potential integration of these platforms with the Internet of Things technology is discussed as a potential new trend in the field.
Keywords: heating, lab‐on‐a‐chip, microfluidics, sensors, temperature
Temperature is an environmental parameter that regulates most physical, chemical, and biological processes. It will influence the outcome of any experiment where any (or many) of these processes take place. One scenario in which such processes occur simultaneously is that found in microfluidic devices, where the temperature of a liquid contained in a channel or chamber might change—because of a myriad of reasons—with time and position during an experiment. For instance, microheaters can be used to purposely increase the temperature of the liquid at a given time and at a specific location in the microfluidic system [1]. Phenomena as diverse as Joule heating, microwave heating, and heating by exothermic chemical reactions can achieve the same outcome [2, 3, 4]. Nonetheless, the change in temperature is sometimes unwanted and may adversely affect the outcome of an experiment [5, 6, 7]. Examples of scenarios where an uncontrolled temperature profile deviates an experiment from its expected outcome include the denaturing of proteins and nucleic acids [8, 9, 10], the lysis of cells [11], and bubble generation (bubbles create very significant pressure gradients that alter fluid dynamics) [12]. Consequently, heating effects have evinced the necessity of developing accurate temperature monitoring and control systems in microfluidics [13, 14, 15, 16, 17, 18].
To achieve temperature monitoring, several types of sensors have been integrated into microfluidic devices. For example, thermocouples were used in applications including polymerase chain reaction (PCR) [19], recombinase polymerase amplification (RPA) [20], atmospheric ice‐nucleating particles analysis [21], and imaging and nucleic acid analysis [22], among others [23]. On the other hand, thermistors were used to track temperature in applications as phenotypic bacterial classifications and antibiotic susceptibility tests [24], heat transfer analysis [25], and inside an in‐line heater [26]. Moreover, thermistors were also used for temperature monitoring in an on‐skin platform for sweat analysis [27]. Other temperature monitoring elements include custom‐built temperature sensors (that exploit the temperature dependency of various physical properties of different materials, including electrical conductivity [28], refraction index [29], and capacitance [30]), which have been implemented in lab‐on‐a‐chip (LOC) systems for real‐time amplification reaction assays [31], real‐time reverse‐transcription loop‐mediated isothermal amplification (RT‐LAMP) assay [32], interfacial tension analysis of two immiscible liquids [33], and cell analysis [34]. All these types of sensors have enabled scientists to keep track of the temperature evolution during an experiment, either by the scientist being present during the experiment and observing the real‐time temperature reading given by the sensor, or by checking the log (if the system has been programed and configured to automatically produce one) at the end of the experiment.
Additional advantageous features can be provided to temperature‐monitoring elements when they are coupled with the Internet of Things (IoT). This becomes critical in scenarios where real‐time monitoring is necessary but the specialist cannot be present during the experiment. These cases, which include remote diagnostics and sensing of water pollutants, have been discussed at length in recent excellent reviews [35, 36, 37, 38, 39]. For instance, a thermocouple was used together with an IoT platform to monitor temperature in real time in a microfluidic device used for nanoparticle (NP) synthesis [19]. Moreover, digital temperature sensors have been used with IoT modules in wearable microfluidic devices [40] and PCR and LAMP analysis [41]. The integration of IoT in microfluidic devices allowed data visualization in real time from the cloud at a distant location from the experiment. Additionally, temperature sensors coupled to IoT technology were recently used to automate the operation of a microfluidic system [42]. Therefore, temperature sensing and the IoT are potential enablers of smart microfluidics.
Temperature‐sensing elements are necessary for the accurate interpretation of experimental observations in microfluidics, albeit sometimes they are not sufficient to ensure optimal experimental conditions. To achieve this, microfluidic devices also require temperature control capabilities, which can be obtained through heating and cooling instruments [43]. Temperature control—described in terms of field profile (i.e., temporal and spatial distribution) or range (i.e., all temperature values between the minimum and maximum exhibited by the field)—is mostly necessary in three specific (1) to lower the temperature of the liquid when it rises unwantedly as a consequence of the phenomena taking place during a microfluidic experiment [7], (2) when the application requires to cool down the sample [44], and (3) when the application requires to heat up the sample [45]. The control of temperature gradients is also crucial for the performance of microfluidic devices [46, 47]. Moreover, the integration of heating/cooling elements into a point‐of‐care (POC) microfluidic system reduces the cost of the experimental setup and improves its portability as no expensive and bulky external heaters/coolers are required, enabling device self‐sufficiency [48].
Temperature control in microfluidics has been explored in applications, including cell culture [49, 50, 51, 52], cell imaging [53], cell analysis [54, 55, 56], PCR [22, 57–60], nucleic acid amplification test assays [61], microreactors [62], transparent electrodes [63, 64], enzyme‐linked immunosorbent assay [44], biochemical synthesis [65], particle trapping [66], and microelectronic device cooling [67], among others [68, 69]. For this, the use of microheaters [52, 55–57, 62, 70, 71], Peltier elements [25, 58, 59, 72], two‐phase materials [73], laser beams [74], and cement power resistors [60] has been reported.
Excellent reviews are readily available in the literature focused on heating [43] and cooling [75] methods in microfluidic devices. In contrast, to the best of our knowledge, no review on temperature monitoring methods in microfluidics has been written. Therefore, a comprehensive review with up‐to‐date information is necessary to provide a complete overview of the field and to highlight the many interrelated challenges that currently exist. This review aims at doing that while discussing contributions published during the last 5 years. To do this, we first describe the physics of heating sources in microfluidics. Then, we review the most relevant contributions in the field, classifying the literature into three main monitoring systems, external temperature control, and integrated temperature control. Finally, advantages and disadvantages, as well as future perspectives of temperature monitoring and control systems, are postulated in Section 4.
Contributions reviewed herein exploit several underlying physical and chemical mechanisms for temperature monitoring and control. To provide a rich discussion about those contributions in Sections 3 and 4, we briefly describe important concepts in this section.
A chemical reaction is classified as exothermic if heat is released from it or endothermic if heat is absorbed by it. By convention, for exothermic reactions, the change of enthalpy is negative (ΔH<0), whereas, for endothermic reactions, it is positive (ΔH>0) [76]. The change of enthalpy is equal to the change in energy (ΔE) plus the mechanical work (W=PΔV) applied to the system, as shown in Figure 1A. At constant pressure (P), this can be expressed as
with ΔV being a small volume. This represents the quantity of energy emitted or absorbed by a system during a change of state, and it is called the heat of the reaction [77]. The heat of the reaction is characteristic of a chemical reaction and, as it depends on the amount of substance used, standardized to 1 mol. The heat of the reaction originating from 1 mol is called the enthalpy, H, of the reaction [78].
FIGURE 1 Mechanisms involved in temperature monitoring and control in microfluidic devices. (A) Graphic representation of the energy released/absorbed during an exothermic/endothermic reaction. (B) Schematic representation of a linear temperature gradient across a microfluidic system. The microfluidic channel between a hot source on the left and a cold sink on the right. Heat flux informs about the direction of the heat flow. The temperature line varies linearly between the hot fluid and the cold one. (C) Schematic representation of the Seebeck effect. An increment of temperature at the hot point where lead a and lead b are connected creates a voltage at the right side where the cables are not connected and at room temperature. (D) Schematic representation of the Peltier effect. p‐Type and n‐type semiconductors are connected at the top by a conductor material, at the bottom, each semiconductor is connected to another conductor material, and a voltage is applied between the two conducting terminals, developing a cooling zone at the top. (E) Schematic representation of microwave heating by the rotation of water molecules induced by the electromagnetic microwave. (F) Schematic representation of Joule heating in an insulator‐based microfluidic system, where heat is mainly generated at regions of high electric field intensity. Source: (C) Reprinted with permission from Ref. [80], © (1986) Elsevier
A gradient of temperature is generated by the temperature difference between two regions of space, as shown in Figure 1B. In microfluidics, this gradient can cause a movement of the liquid due to the different densities resulting from it—it is important to stress that density is not the only physical property of the fluid that changes with temperature; thermal conductivity, electrical conductivity, viscosity, and many others change as well. The Fourier heat diffusion equation with heat flux restricted to one direction (z) along a two‐dimensional planar surface (xz) describes a temperature gradient in the z direction that does not change longitudinally (x) through the microfluidic channel [79] as
where T is temperature, and k is the thermal conductivity of the medium in which the heat is flowing.
One of the many temperature‐monitoring systems used in microfluidic devices is the thermocouple, capable of detecting temperature changes as a result of the Seebeck effect. To understand this effect, consider that one end of a rod made of a p‐type semiconductor is connected with one end of another rod made of an n‐type semiconductor. At this pn junction, there is a temperature T higher than the room temperature (heat source inside the microchannel). The free end points of both rods are close together—but not in contact—and at room temperature (cold point). If the temperature difference is maintained between the two ends, an open circuit voltage (Δφ) is developed at the cold point, as shown in Figure 1C, and it can be expressed as
where αs is known as the Seebeck coefficient (V/K), which is proportional to the slope of the Fermi energy (EF) against temperature as
with e− representing the electron charge. Thus, this coefficient depends on the chemical composition of the material and the temperature [80]. The Seebeck effect allows monitoring the temperature in a microfluidic device by measuring Δφ.
In order to achieve temperature control in microfluidic devices, cooling and heating methods must be used. Peltier modules placed outside the microfluidic channel function as external cooling systems by taking advantage of the Peltier effect. At the subatomic scale, electrons can jump from lower to higher energy states within the atom if enough energy is provided to them. This phenomenon can also be reversed, resulting in the dissipation of an excess of energy through electromagnetic waves or heat. To be precise, this second form of energy loss is known as the Peltier effect [81].
Suppose once again that an n‐type semiconductor is connected to a p‐type semiconductor (as we did in the explanation of the Seebeck effect). Only this time, a voltage is applied across the free terminals, resulting in an electric current. If the electrons go from n‐ to p‐type semiconductor, electrons will be going from higher to lower energy states expelling heat to the surroundings, producing a heating effect at the joint. On the other hand, if the electrons go from p‐ to n‐type semiconductor, electrons will need extra energy, which will be taken from the surroundings, producing a cooling effect at the joint, as shown in Figure 1D. The relation between absorbed or released heat and current applied is given by the Peltier coefficient πab:
where ΔQab is released heat, and I is the electric current. Importantly, if temperature is needed to be uniform in spatial distribution, a heat distribution plate (made from a high thermal conductivity material, e.g., aluminum) may be used between the Peltier element and the sample [82].
Rapid generation of heat at the microscale can be generated by the use of electromagnetic microwaves because of their interaction with matter. For an absorptive material, the electromagnetic field passing through it produces a constant movement of positively and negatively charged species because of the oscillating electric field induced by the electric component of the electromagnetic field, as shown in Figure 1E [83]. This phenomenon can be described mathematically as
where QE is the microwave heat source, ω is the angular frequency of the microwave resonator, ε0 is free space permittivity, ε′′ is a loss factor, and finally, |E| is the magnitude of the electric field [84].
Joule heating refers to the temperature rise of a conducting material as a consequence of the passage of an electrical current through it as shown in Figure 1F [85]. In microfluidics, Joule heating takes place when an electric field is applied to a channel to produce an electrokinetic (EK) response. Such an electric field also produces an ionic current [86]. This heating generates a nonuniform increase in the temperature of the fluid (i.e., a temperature gradient), causing an alteration in the analyzed sample properties [87]. In Joule heating, the heating power per unit volume is a function of the electrical conductivity and the square of the electric field intensity [88, 89]:
where σ refers to the electrical conductivity, Qheat to the heating power, qJH to the heat flux produced per unit volume, V to the volume, E to the electric field, and φ to the voltage. Similarly, rms alludes to the root mean square value.
Up to this point, we have briefly reviewed effects and mechanisms that can be used to either measure or control temperature in a microfluidic device. Nonetheless, to fully comprehend how those effects and mechanisms play a role during an experiment in microfluidics, attention must still be given to the fundamental equations that describe heat transfer phenomena. The following equation is known as the differential form of the energy
where ρ, Cp, and k represent the density, specific heat capacity, and thermal conductivity of the material being studied, respectively. P and T represent the scalar pressure and temperature fields, and U is the vector velocity field. Finally, Q represents any external heat source (e.g., microwave or Joule heating) present in the system. Equation (8) describes the relationship between local temperature changes over time with the presence of fluid velocity and heat sources or sinks within the system [90].
Depending on the system being analyzed, different boundary conditions must be taken into consideration when solving Equation (8) for the temperature field. Among these, constant temperature (T=T0, with T
0 a fixed temperature value), thermal insulation (−n·q=0, with n a normal unit vector and q the heat flux vector), and heat flux (−n·q=q0, with q
0 a general inward heat flux) are the most commonly found in microfluidic works [91].
From Equation (8), it is evident that thermal and electrical properties of materials used for fabricating microfluidic devices play a critical role in determining the temperature temporal and spatial profile experienced by samples being analyzed and the device itself. Because of this, Table 1 summarizes the relevant thermal and electrical properties of the most commonly used materials in microfluidics.
Regularly, microfluidic devices that incur in significant heating during operation are presented in the literature and, very often, these works do not implement temperature control or monitoring systems. As an example, consider an electrokinetically‐driven microfluidic device stimulated with a direct current (DC) voltage, which was developed to characterize the trapping voltage of PEGylated proteins. When operating, particles were trapped at cross‐sectional gaps in an array of insulating posts. However, after a few seconds of operation at 4000 V, bubbles began to show at the trapping regions [102]. In devices of this type, bubble formation may be due to two different (1) electrolysis and (2) heating. Electrolysis takes place at the electrode–liquid interface in the reservoir containing the positive electrode. Heating takes place mainly at the cross‐sectional gaps in the array of insulating posts. The computational model used to predict the outcome of the experiment did not take heating phenomena into consideration and did not prevent bubble formation and potential protein denaturation. This section reviews the different roles temperature may take during a microfluidic experiment, as well as the most recent contributions in microfluidic devices that incorporate temperature sensing and control capabilities. It is important to state that some commercial solutions are available for temperature control and monitoring (e.g., thermocouples or microheaters) for microfluidic devices. Nonetheless, the requirements set by the many different potential applications of microfluidic devices still motivates research efforts in developing custom temperature control and monitoring tools.
Through the years, the heating of a liquid sample within a microfluidic device has been a concern for the scientific community because of the many problems that may arise from it (e.g., cell lysis and protein denaturation). In contrast, heating may be required in some applications (e.g., PCR assays) to obtain the desired outcome of a process. Therefore, depending on the intended application of a microfluidic device, temperature change may either represent one of the main characters or the villain of the story. In this section, the main benefits and drawbacks of heating in microfluidics are presented.
An interesting application of Joule heating can be found in the manipulation of particles in microfluidic channels. As Joule heating produces a nonuniform temperature rise in the fluid sample, a temperature gradient develops. This gradient gives rise to thermophoresis—the movement of a particle suspended in a liquid in the presence of a temperature gradient, which occurs as a consequence of molecules impacting the particle on opposite sides with different average velocities due to their difference in temperature—which enables controlled particle motion [2]. Insulator‐based electrokinetically‐driven microfluidic devices for NP trapping were reported where thermophoresis was expected to actively participate in particle flow control. Thermophoresis is sensitive to the composition of a continuous phase and a given particle, and it has been reported to increase the flow through the entrapment zones [2, 103, 104]. However, thermophoresis is not the only means by which Joule heating can support particle manipulation efforts.
Biological cell entrapment techniques were presented by Aghilinejad et al. for the separation of MDA‐MB‐231 cells—epithelial human breast cancer cells—from Granulocytes, and in Ge et al. for the rapid concentration of deoxyribonucleic acid (DNA) in a sample [105, 106]. In both works, the electrothermal flow generated by Joule heating–induced liquid motion dragged the respective bioparticles to the hot spots in the channel. This motion enhanced the separation rate of the devices. Similarly, in 2019, Kunti et al. presented a technique that combined the use of DC and alternating current (AC) fields, generating temperature gradients that led to a nonuniform electrothermal flow. The technique was called induced temperature gradient focusing (TGF) and produces changes in the electrothermal properties of a liquid solution. These changes generate a strong electrothermal vortex, which brings colloidal particles to the center of the hot spot, enabling particle aggregation in a focused area [107]. Furthermore, Dutta et al. proposed a microdevice that also used a combination of AC and DC fields for enhancing TGF of sample solutes [108]. These devices represent a completely new and energy‐efficient technique of particle entrapment driven by Joule heating effects.
Thermophoresis and electrothermal flow can also interact with several other mechanisms (e.g., electrophoresis) present in microfluidic devices. Cornejo et al. and Vlassakis et al. presented the handling of a thermal gel to create reversible barriers for cell enrichment [109, 110]. In these works, a gel matrix was used because it could be reversibly converted between liquid and solid phases as a function of temperature. Particle enrichment was achieved by selectively inducing localized Joule heating, therefore producing regions of liquid‐phase gel and others of solid‐phase gel. With this, the cells could migrate through the liquid gel but could not enter the solid gel—accumulating at the boundary between liquid–solid gel. This technique enabled the possibility to enrich cell lysates by regulating the phase of thermal gels using voltage‐controlled Joule heating. Moreover, a variation in the equilibrium position of analyte molecules as a function of the dependency of liquid viscosity and analyte diffusivity on temperature was presented by Dutta et al. [111]. In this work, the Joule heating effect is caused by free‐flow zone electrophoresis, an assay that typically relies on the continuous pressure‐driven flow of a sample stream through a separation chamber in the presence of an electric field applied perpendicularly to this flow direction. These works thus evinced that Joule heating, a consequence of EK phenomena, can actively participate in the control of particle and liquid motion in microfluidics in a positive way. Nonetheless, as previously stated, Joule heating is not the only mechanism available to create temperature gradients within a microfluidic device.
The use of exothermic chemical reactions as a heat source in microdevices is an approach to heating that features beneficial effects (e.g., achieving passive thermal control without the need of an external power source). In this line of research, Liao et al. described a procedure to achieve passive heating via an exothermic reaction of the galvanic corrosion of MgFe, where the MgFe simply reacted immediately with the water [3]. Heating zones created at specific points by a reaction can generate temperature profiles that have the effect of separating molecules. This was presented by Goertz et al. and Huang et al. [112, 113]. In these works, it was described that heat generated in thermos by the galvanic corrosion of MgFe alloys in the presence of NaCl allowed separating the ligation and polymerization reagents for the rolling circle amplification of nucleic acids.
On an application different from temperature‐induced particle and liquid control for particle trapping or separation, thermal gradients have also been exploited to act as standalone liquid pumps, to improve the performance of other pumping mechanisms, or even to drive mixing processes in microfluidics. Regularly, pressure‐driven pumps and electric fields are used to generate liquid flow in microfluidic devices. Nevertheless, these approaches have room for improvement as pressure‐driven pumps are regularly bulky and electric‐field‐driven flows regularly feature low flowrates [114]. Exploiting temperature changes to drive fluid motion can enable greater flowrates. Saleel et al. reported a numerical analysis to study the effect of Joule heating on the dynamic viscosity of a liquid [115]. The model of study was a horizontal rectangle with a height of 100 µm and an infinite length, without any obstacle inside, aiming to study the electroosmotic flow (EOF) in the presence of temperature changes. Streamwise velocity changes were analyzed as a function of viscosity changes produced by changes in temperature. As the ratio of viscosity with respect to temperature increased, the streamwise velocity profile changed its form from the normal plug‐shaped profile to a more parabolic profile; thus, the velocity of the liquid at the center of the channel was increased. This fluid control via temperature‐dependent properties has been studied under other scenarios as well. For example, Ranjit et al. proposed a computational model with induced fluid movement in a porous microchannel through a peristaltic wave to transport ionic liquids, which led to Joule heating varying the viscosity of the liquid by increasing and decreasing temperature inside the microfluidic device. On top of that, an alternating electric field (externally applied) induces an EOF. The combination of the two induced fluid movements was studied, mainly showing EOF velocity changes [116]. The main benefits of these velocity changes can be seen in liquid mixing where the generation of different velocity profiles is necessary for an efficient operation [115].
Following a route different to Joule heating and chemical reactions to produce a temperature change, microfluidic mixing exploiting microwaves as a heat source has been reported in the literature. For instance, two channels designed for microdroplet formation were complemented with a mixing stage driven by microwaves [65]. The mixing process started with a so‐called outer coplanar transmission line loop, which produced a magnetic field that induced a strong electric field in two concentric copper loops (i.e., the microwave resonator, which stores electric and magnetic energy). Droplets traveling over the resonator were heated by induction, achieving a nonuniform temperature distribution that led to a nonuniform density, viscosity, interfacial tension, and diffusion coefficient, producing droplet mixing. Because of perturbations in the permittivity of the medium, droplets can be detected, and heating can be automated with this approach. In the literature, other microwave heaters have been studied [4, 117], and acoustic heaters have also been reported [118]. The reader is referred to Refs. [119, 120, 121, 122] for further information on flow control and mixing.
As we have discussed, a heat source is sometimes required in microfluidics and Joule heating, exothermic reactions, and microwave heaters, among others, can be exploited to our benefit. Unfortunately, this is not always the case. Chang et al. developed a microdevice for the enhancement of DNA expression in cell therapy using a nanosecond pulse electroporation system [123]. This system featured two main (1) electrolysis, leading to formation of gas bubbles at the walls of the microchannel, and (2) Joule heating effects, increasing the gas bubbles’ size, which can further lead to cell damage and even cell apoptosis. Joule heating has also been demonstrated to alter the EOF normal “plug‐like” profile, resulting in a concave‐shaped profile, deviating fluid kinetics inside a microchannel from the expected behavior [120]. In addition, in contrast to the studies reporting temperature‐driven enhancements in electrokinetically‐driven flow dynamics discussed in Section 3.1.1, it has also been reported that a negative effect of temperature rise and asymmetric heat fluxes in EOF applications is the decrease in the velocity magnitude, as reported in two different studies [124, 125]. Unfortunately, unwanted modifications on fluid and particle dynamics are not the only adverse effects brought to microfluidic experiments by modifications of the temperature profile therein.
High temperatures are dangerous for bioparticles. They can lead to cell death or protein denaturation. In EK‐driven microfluidic devices, the goal is to manipulate particles efficiently but, as Joule heating is present during an experiment, particles can be damaged. To assess this, an experimental and numerical analysis showed how an internal insulator post array inside a microfluidic channel can affect the temperature distribution when a conductive liquid is used to suspend the particles of interest [28]. For a 90‐s‐long experiment, temperatures could be increased from 21 to 35°C when a suspending medium with a conductivity of 150 µS/cm was used, and from 21 to almost 40°C when the initial conductivity of the buffer was 200 µS/cm. Numerical predictions of maximum temperature reached in the channel as a function of post array design indicated that the geometry and density of the posts played only a small role in the heating phenomena with arrays featuring a small density of large posts reaching slightly higher temperatures than arrays featuring a large density of small posts. In contrast, it was evident that the electrical conductivity of the solution chiefly determines the rate at which temperature rises within the microfluidic channel, with higher conductivities reaching exceedingly large temperatures in very little time.
In addition to the drawbacks described before, which pertain to the experiments, practical drawbacks are also exhibited by heating. Fabrication of a microfluidic device devoted to sort particles from a heterogeneous sample, for example, is a fairly “simple” task [126]. However, if the mechanism by which the sorting takes place in the device has the secondary effect of changing the temperature profile therein (and if this has adverse effects for the experiment), the device must be redesigned to include heat sources or sinks. Fabrication of such a sophisticated device might require additional alignment and bonding stages, likely increasing the cost of the device and its peripheral instrumentation requirements.
By now, it should be clear that one major concern in the field of microfluidics is the fluctuation of temperature generated by diverse heat sources (e.g., Joule heating, microwaves, and chemical reactions, among others). Therefore, temperature monitoring with maximum sensitivity is of utmost importance to enable a wide range of samples to be analyzed in microfluidic devices with a reduced risk of damage. In this section, temperature‐monitoring systems are discussed according to the type of equipment used, localization (attached/external) on the system, and the possible integration with IoT technologies to enable remote real‐time monitoring. A summary of some major contributions in each of these sections is shown in Table 2.
Several authors have proposed the use of thermocouples to keep track of temperature variations of targeted samples under various conditions in a microfluidic device. In particular, biological species for medical analysis require maximum monitoring of temperature as their viability depends on this critical parameter. Several microfluidic devices with applications associated with reverse osmosis, PCR, RPA, atmospheric ice‐nucleating particles, nucleic acid, and other types of bioparticle analysis have monitored temperature to indirectly assess the viability of the sample and enable an accurate interpretation of experimental observations [19–23, 45, 68, 129].
A simple methodology to enable temperature monitoring in a microfluidic device was implemented by Kong et al. who inserted thermocouples into a polydimethylsiloxane device. The principal function of this system was monitoring the temperature of the RPA solution. The specific measurements of temperature were evaluated at 4, 22, and 37°C, and the accuracy of the equipment was reported as ±0.3 up to ±1°C [20]. Similar results were obtained by Zhu et al. with a multilayered microfluidic system integrated with polymer tubes and the addition of a K‐type thermocouple. Here, the hot/cold water was stored in an insulated flask and then injected to the microfluidic device through a syringe pump. The K‐type thermocouple was integrated to the flask to monitor the temperature of the water, as shown in Figure 2A. Specifically, the range of temperature of the equipment was reported between 24 and 37°C, and the measurement margin of error was found to be less than ± 0.5°C [68].
FIGURE 2 (A) Schematic representation of an experimental setup consisting of a microfluidic structure with embedded Tygon tubes for convective heating/cooling of the sample, a syringe pump for infusing the sample through the microfluidic channel, a syringe pump for pulling hot/cold water through the Tygon tubes, a water flask for storing the hot/cold water, a bottle for collection of the waste sample, and an infrared camera interfaced with an analysis software for real‐time monitoring of temperature across the glass coverslip. (B) The image of the microfluidic channel integrated with Pt thermo‐sensor. (C) Illustration of an optical‐based temperature monitoring approach implemented in a microfluidic device. The channel extends to ensure a fully developed flow profile. Black arrows depict the flow direction, red spheres depict dye containing droplets. (D) Overview of a spectroscopy setup used in microfluidic applications. An optical fiber probe excited nanoparticles (NPs) within the microfluidic device using a 980 nm laser coupled into the fiber using a collimator. The light is focused using a lens and is collected with the sample fiber probe. The collected light is monitored with a charge‐coupled device (CCD) detector after passing through a short pass filter. Source: (A) Reprinted with permission from Ref. [68], © (2018) American Chemical Society; (B) reprinted with permission from Ref. [66], © (2021) Elsevier; (C) reprinted with permission from Ref. [25], © (2022) Elsevier; and (D) reprinted with permission from Ref. [128], © (2019) Royal Society of Chemistry
Alternatively, another approach has been proposed by several researchers and is based on the use of custom‐built temperature sensors. This type of sensor provides temperature measurement in a readable form through an electrical, optical, or chemical signal. Specifically, various microfluidic systems with incorporated sensors have been used to analyze the temperature of liquid samples in applications related to DNA/RNA reactions, RT‐LAMP assays, immiscible liquids interfacial tension, and cell analysis [1, 31–34, 40–42, 66, 130]. For example, Zhao et al. created a microfluidic channel equipped with a micro thermo‐sensor consisting of a thin‐film of Pt and a thin film of Si3N4 as an insulation layer, as shown in Figure 2B. The device was used to monitor small‐scale changes of temperature in physiological processes of live tumor cells with an accuracy of the system reported at a minimum of 0.008°C [66]. Gallo‐Villanueva et al. developed a copper resistance temperature detector (RTD) sensor to monitor the average temperature in several DC‐stimulated electrokinetically‐driven devices [28]. The sensor was placed below a microfluidic channel, which heated up due to Joule heating each time a DC voltage was established between its inlet and outlet. Heat was then transferred from the microfluidic channel to the RTD sensor and, because the electrical conductivity of copper is a temperature‐dependent parameter, its electrical resistivity changed. As a consequence of this change, a change in electric current flowing through the RTD sensor could be detected. This approach has the benefit of having the sample insulated from the sensor, preventing sample contamination by corrosion or electrolysis. In another work, Jhou et al. monitored temperature in an assay for the detection and quantification of three specific genes of SARS‐CoV‐2 [32]. Here, the microfluidic platform integrated a temperature control module with the addition of sensors to monitor the changes in temperature of viral lysis, RNA extraction, and RT‐LAMP processes. Respectively, the variations generated by a microcontroller were less than 0.5°C at target values of 95, 45, and 60°C. These works perfectly exemplify the flexibility provided by custom‐built sensors in applications where temperature monitoring is required.
Following a different route to the use of thermocouples or custom‐built sensors, several authors have used thermistors (thermally sensitive resistor) to monitor the fluctuations of heat in microfluidic systems. Particularly, a series of microfluidic designs have focused on integrating temperature measuring capabilities for applications associated with bacteria, antibiotic, and chemical compounds, heat transfer analysis, and organic solutions monitoring [24–27, 70]. For example, Bürkle et al. produced a microfluidic device based on laser Doppler velocimetry and laser‐induced fluorescence. This system monitored the temperature and the dependent intensity of the fluorescent light of organic dyes [25]. Specifically, the wall temperature of the polymethyl methacrylate (PMMA) microfluidic system was monitored with a Pt100 thermistor (shown in Figure 2C) for different input power values, and the temperature values reported were in the range of 21–34°C. A similar approach was implemented by Nightingale et al., with a droplet microfluidic‐based chemical sensor integrated with an in‐line heater surrounding a resistive heater included to accelerate the nitrate and nitrite reactions generated [26]. The temperature inside the heater was monitored by an internal thermistor with an accuracy of less than 0.1°C, and heating values were evaluated at room temperature and up to 80°C, demonstrating that these types of sensors can be highly accurate within a wide window of operational temperatures.
Despite the usefulness of classic temperature measurement systems, several researchers have recently proposed new approaches (e.g., the use of NPs or optic fibers) to temperature monitoring aiming at improving the performance of microfluidic devices for several applications [127, 128]. For example, Geitenbeek et al. developed a technique they called “ratiometric bandshape luminescence thermometry” in which thermally coupled levels of Er^3 +^ in NaYF4 NPs were used for in situ mapping and monitoring of temperature in microfluidic systems [128]. Using simple fiber optics, this luminescence technique (shown in Figure 2D) exploited the changes in the intensity ratio of two or more emission peaks with temperature. Specifically, the monitoring measurements were reported at room temperature up to at least 120°C, and the accuracy inside the microfluidic channel was stated at 0.34°C. Also, by innovating the field of temperature monitoring, Garcia‐Manrique et al. [127] proposed a different approach based on the use of a digital sensor probe. Here, a thermostatic microreactor platform required different transition temperatures to produce niosomes. The system temperature inside the thermostatic chamber was measured and monitored with the digital sensor probe. The effect of temperature was sensed in a range between 30 and 60°C with the principal function of controlling the size of produced niosomes.
Additionally, most applications of microfluidic devices require real‐time access to temperature measurements to ensure the validity of experimental observations. This is critical, for example, in diagnostic applications where an undetected change of temperature above or below a certain threshold might lead to a misdiagnosis [134]. Distant communities with no access to healthcare settings might benefit from diagnostic devices that can be monitored and controlled remotely in real time [35, 39]. Because of this, the integration of microfluidics and IoT technologies to improve temperature monitoring and control has become a current “hot topic” in the field [19, 22, 27, 34, 40–42, 45, 70]. In this line, Kulkarni et al. developed a temperature‐monitoring platform for PCR‐based DNA amplification where an Arduino Nano and IoT ESP826601 module enabled the capabilities of accessing and storing temperature values directly onto a smartphone in real time through ThingSpeak analytics [133]. Similarly, using an Arduino Nano, Cruz et al. were able to enable remote real‐time readings of temperature on a smartphone during thermocycle experiments [22].
Through the use of external heating methods, heat is either transferred through a direct contact of the heating element with the device (i.e., contact heat transfer) or through energy radiation or ultrasound radiation (i.e., noncontact heat transfer) to the region of the microfluidic device that needs to be heated [109]. The same can be said about external cooling methods. Here we present different methods used for external temperature control, together with its applications shown in Table 3. It is important to emphasize that, although specs like temperature range of operation, ramp rates, and time of operation are important to assess the quality of a temperature control device, no objective assessment about the performance of the devices can be drawn from these data alone, as performance depends also on the requirements set by the application at hand.
Microheaters have been used in microfluidic devices as they can provide a wide range of temperatures (up to 220°C [62]) with an adequate control. Moreover, they offer a straightforward approach for their integration in microdevices as their thickness ranges from 100 nm to 100 µm [141]. In applications exploiting external heating, microheaters can be found on the surface of the device, as shown in Figure 3A. Commonly, these heating elements are made up of materials such as Pt [57, 62], Au [1], and Ag [64]. Hence, they are highly conductive and feature a low specific heat capacity. Other materials for resistive heaters include conductive paint (for the construction of a heater pattern through screen printing) [56], and adhesive polyamide (to fabricate flexible resistances) [22], among others [112, 120–122]. The material used to fabricate a given microheater is selected according to its intended application.
FIGURE 3 (A) A schematic of a polymerase chain reaction (PCR) chip fabricated in SU‐8 on a glass substrate. A thin‐film platinum microheater and a thermometer are placed under the SU‐8‐based PCR chamber. (B) Peltier elements placed under a reactor. Schematics of a microreactor with a thermal control composed of a 61‐W Peltier cells and a thermal block. (C) A schematic of a passive heating system for microfluidic devices composed of three microfluidic chambers composed of supercooled sodium acetate trihydrate (SAT) as a heat source placed at the bottom of an organic phase change material (PCM) for thermal regulation and a reagent chamber over the PCM. Source: (A) Reprinted with permission from Ref. [135], © (2004) Elsevier; (B) reprinted with permission from Ref. [72], © (2019) Elsevier; and (C) reprinted with permission from Ref. [73], © (2020) Elsevier
Recently, Peng et al. fabricated an Au wire microheater under a microfluidic device and used a saturated calcium chloride solution for cooling at the top of the device. The temperature control system generated temperatures ranging from 2 to 37°C, which were maintained for 2 min. The Au wire generated heat through Joule heating, which was transferred to the device through convective heating. During simulations, it was found that temperature varied from the bottom to the top of the microdevice (from 2.2 to 0.8°C). However, considering the dimension of the region of interest (10 µm) and the calculated temperature gradient (less than 0.024°C/µm), it was assumed that temperature was uniformly distributed, as the temperature would vary 0.24°C in the region of interest [1].
Copper provides higher thermal and electrical conductivities than gold. Therefore, thin‐film Cu heaters were placed under a microfluidic chip and maintained at a constant temperature of 37°C for 10 days. However, when imaging with a water objective was carried out, the heater could not overcome the cooling produced by the objective, therefore yielding a nonuniform temperature distribution [52]. In another study, Bobinger et al. developed a transparent heater containing Cu nanowires (CuNWs), which enabled a uniform heat distribution over an area of 3.5 × 5.0 cm^2^. Through an electrical input of 2.7 W, temperature increased from ambient temperature to 70°C in 5 s. Heat was generated by Joule heating and, according to the resistance of the CuNWs, it was possible to generate this temperature range [63]. An unfortunate disadvantage of Cu heaters is that they are opaque. Many microfluidic devices are transparent to facilitate observation of the experiment in the lab setting, and an opaque microheater might interfere with this.
Finally, an optically transparent standout material regularly used for heating through Joule heating is indium tin oxide (ITO). According to Lei et al., a temperature control system was built placing an ITO microheater (attached to a glass substrate) under a microfluidic device, and below it, a water bath was placed [55]. Considering the linear relationship between resistance and temperature in ITO films, and the serpentine shape of the heater to diminish heat loss, temperatures generated ranged from 29 to 37°C in a microfluidic platform and were maintained for 30 s. Several microfluidic devices that use a similar heating source have been recently reported [49, 50, 53].
Thermoelectric modules can produce heating or cooling in a microfluidic device depending on to the direction in which DC current flows. According to Nasser et al., a Peltier element was placed on the surface of a microfluidic chip and performed thermal cycles of temperatures ranging from 94 to 60°C (with a cycle duration of less than 10 s) with cooling and heating rates of 8 and 8.3°C/s, respectively. Interestingly, the Peltier element reported in that work did not include the heat distribution plate commonly found in these elements acting as a contact thermal resistance to improve the spatial uniformity of the temperature field and prevent abrupt temperature changes. Therefore, heating and cooling rates increased due to the elimination of this feature [59]. In a different contribution, a Peltier element was attached to one side of a PMMA tube wall, covering an area of 10 × 30 mm^2^ where the temperature varied from 22 to 36°C. It was possible to generate a constant temperature gradient therein by fixing a specific power value for the Peltier element and ensuring a constant air flow (inside the PMMA tube) through a pressure reducer [25]. Other experiments were made aiming at a similar temperature range (up to 32°C) [51] or presenting even wider ranges (from 12 to 98°C) [58]. Moreover, the use of Peltier cells together with a cooling system (fans and a recirculating water–glycol solution for heat dissipation) was recently reported [72]. In that contribution, the authors placed the Peltier cells under the microreactor, as shown in Figure 3B, generating temperatures that ranged from −18.2 to 25.8°C.
In a similar fashion to temperature monitoring applications, nonconventional methods for heating and cooling have also been devised and implemented in temperature control applications. For example, the use of bipolar junction transistors (BJTs) as heaters was reported. A BJT is able to maintain a constant temperature as it employs a switch to turn off the applied voltage when the desired temperature is reached. It is considered a heater because it can release the heat generated according to the power applied. In an experiment with a BJT heater, temperatures ranged from 23 to 60°C, with a heating ramp rate of 11.6°C/s, being able to maintain a temperature for 164 s [137]. The heat transfer process for the BJT occurs when the energy is transferred through convection and thermal radiation to the region of interest in the microfluidic channel.
A different novel approach to heating and cooling consists in using phase change materials. An advantage of using these materials is that an external power supply is not required, improving the portability of the system. According to Vloemans et al., a passive heating system was built through the attachment of organic paraffin under a reagent chamber and the use of a sodium acetate trihydrate (SAT) solution as a heat source (as shown in Figure 3C) [73]. This passive heating system can generate temperatures in the range of 25–37°C and maintain a specific temperature for 15 min. Heat is transferred to the reagent chamber through convection, where energy is generated by the addition of salt in the SAT solution. The energy is then transferred to the paraffin layer, which regulates the amount of heat transferred as this material has a large latent energy storage capacity. Therefore, it is possible to generate a constant and homogenous temperature distribution in the microfluidic device through this heating system. Other alternative methods employed for temperature control involve the use of a water bath [138], a laser [74], and a cement power resistor wrapped with conductive copper tape and aluminum foil, both using a fan as a cooling element.
An alternative to temperature control by external means is to use integrated control methods. Integrated heating or cooling alludes to different methodologies to manipulate the temperature in a microfluidic system, which can be altered by a heat source or sink that has an active and direct participation with the sample. These strategies can be classified in three categories, depending on the heating or cooling caused by a reaction, a structure arrangement, or a material. In the following section, these strategies will be described along with their corresponding specifications.
The first approach to internal temperature control in microfluidic systems is based on the use of chemical reactions and properties. The pioneer of this method was Guijt et al., who took advantage of both, an endothermic reaction, by coupling an acetone flow (Reagent 1) and an airflow (Reagent 2) that produced the evaporation of acetone, inducing cooling; and an exothermic reaction, by mixing a 97 wt% H2SO4 dissolution (Reagent 1) with water (Reagent 2) [67]. The reactions were evaluated in a 54 µm wide and 19 µm deep channel and achieved temperatures from −3 to 76°C, which can be controlled by the injection of the reagents into their respective channels. The configuration of the device is shown in Figure 4A. A similar method was used by Vloemans et al., by exploiting the exothermic crystallization reaction caused by a mixture of supercooled SAT with water to modulate the temperature between 37 and 42°C [73]. Furthermore, Gaiteri et al. utilized femtomolar concentrations from Pseudomonas syringae (PSy) proteins to control the temperature in a freeze–thaw valves microfluidic system [44]. The PSy produces ice‐nucleating proteins (INPs) that mimic the lattice structure of ice and, therefore, trigger freezing at higher temperatures. The study exhibits that an Escherichia coli sample with a 5 pM buffer concentration of INP freezes in half the time (19 s) than a sample without it at −45°C (37 s). Additionally, the results showed that E. coli could freeze at −25°C in 31 s with PSy proteins when an unmodified sample does not show apparent freezing. Thus, the use of this property allows the cooling of the microfluidic system in an interval between −45 and −25°C. Overall, temperature control based on chemical reactions is mostly employed in POC diagnostics or bioassays due to their operating range of temperature, accuracy, and no requirement of external power [73, 142].
FIGURE 4 (A) A schematic of the arrangement of a microfluidic device, in which its sample channel temperature is controlled by the endothermic and exothermic reaction driven by the mixture of two reagents. (B) Heat distribution along a Tygon tubes temperature controlled microfluidic device. (C) A schematic of the fundamental elements of a microwave heater. (D) Configuration of a solenoid embedded microheater in a microchannel. Source: (A) Reprinted with permission from Ref. [147], © (2021) Elsevier; (B) reprinted with permission from Ref. [68], © (2018) American Chemical Society; (C) reprinted with permission from Ref. [148], © (2013) Royal Society of Chemistry; and (D) reprinted with permission from Ref. [69], © (2017) IEEE
Specific structural arrangements in microfluidic channels have been used as temperature‐control methods and have received wide attention from the scientific community. For instance, the integration of parallel/cross Tygon tubes, along the microfluidic channel (configuration shown in Figure 4B), allows the convective heating/cooling of the sample. This arrangement could be used in cellular assays, as demonstrated by Zhu et al. with calcium signaling applications, who reported that a 10 µM Yoda‐1 cell medium applied at a 60 µl/min rate, with a cross Tygon tubes configuration, increased the temperature from 32 to 37°C in 8 min with a maximum deviation of 0.27°C, allowing a 1.41 ± 0.72‐fold increase of [Ca^2+^]. Based on the same principle and configuration, a temperature control enhancement was obtained by wrapping Tygon tubes in a solenoid form along a plastic pole and placing it perpendicularly to the microchannel containing the sample [68, 143]. This modification allowed a wider range of temperature control, from 24 to 37°C, in the same calcium signaling application.
Following a different approach to the use of Tygon tubes, Juelg et al. proposed a structural arrangement that permits the modulation of temperature in gradients going from room temperature to 50°C, and dropping to 40°C, for droplet generation applications [144]. The arrangement consists in the coupling of a temperature change rate valving, a venting resistor, and a siphon in the sample droplets. Although the venting resistor heats the sample to a programmed temperature, the temperature change valving induces a fast cooling by the introduction of an air contraction, the flow resistance of which at the air vent produces a local low pressure, optimal for their application. The temperature and pressure specifications in this approach are well adequate for mixing chambers in real‐time PCR assays. Furthermore, because of its temperature range, similar approaches have been used for different biological applications [145, 146].
This approach of using structural arrangements to achieve temperature control also involves exploiting the physical and chemical properties of materials in the microfluidic system through, for example, Joule heating and microwaves. The heat distribution generated by a current (Joule heating) is given by gradients, where higher temperatures are accumulated in spaces with higher electric fields. This scenario is very common in electrokinetically‐driven microfluidics, where the inclusion of insulating structures, designed for particle manipulation, create regions of high electric field magnitudes within the main channel. These areas are denominated “hot spots” as they present the highest temperatures, and the temperature is gradually reduced away from it through the medium until another heat source is reached [149, 150]. Gallo‐Villanueva et al. evaluated this effect in a buffer solution of dissolved K2HPO4 in deionized water, under the application of 2500 V (DC) in microdevices with different post geometries of insulators for EK trapping of microparticles [28]. The sample developed gradients of temperature featuring up to 27–30°C in the hot spots, depending on the post geometry; and down to 22°C, in the coolest areas. The authors demonstrated, both mathematically and experimentally, that designs with higher post numbers, even though they increase the number of hot spots, achieve greater heat dissipation. The suspending medium's electrical conductivity and applied voltage can also be used to control the temperature inside the device (with temperature increasing with higher electrical conductivity or applied voltages) [151, 152].
Microwave heaters are another type of structural arrangements that can deliver energy selectively to each component of the sample by exploiting their different dielectric properties [65]. This approach consists in the active work of a microwave signal generator and a transmission line on a microwave resonator, which acts as a heater based on the constructive interference of propagating waves. Boybay et al. evaluated this microwave heating method for a droplet‐based microfluidic device [148]. The microwave components were covered by copper, and its configuration is shown in Figure 4C. Results indicate a temperature control between 20 and 80°C for a single water phase sample, depending on the excitation frequency, input power, and permittivity of the fluid. This temperature versatility makes it suitable for biological and mixing applications [65, 148, 153–155].
The use of integrated microheaters in the structure of microfluidic devices is another approach to controlling temperature. Unlike the microheaters mentioned in the section on external temperature control methods, these microheaters are not found at the surface of the device, but in a specific configuration within the microfluidic channel. The temperature range in which they operate depends on the material they are made of and their size [33]. For instance, Lee et al. worked with microheaters in serpentine patterns to achieve a temperature control between 25 and 70 ± 0.0755°C. Moreover, Bian et al. used a microheater in a solenoid shape surrounding the microfluidic channel, as shown in Figure 4D, which results in a modulation of temperature between 0 and 130 ± 1°C with rapid heating and cooling processes of 16 and 15.8°C/s, respectively [69]. The ramp rates for heating and cooling suggest that this methodology might be suitable for PCR applications.
Thin film microheaters are a variation of these temperature control elements that is widely used in microfluidic systems to leverage electrothermal convection for their application. Liu et al. embedded two of these heaters with a length of 25 µm at a distance of 500 and 700 µm from the anodic interface of their device [156]. This configuration enabled an electrothermal fluidic roll that went from 21 to 29°C, which can be used in LOC applications. Common rectangle heaters have also been used to work with biological samples as human DNA, animal cells, and protein cells because of their appreciable accuracy of 0.1°C in the range of 25–65°C [142, 157]. Finally, another variation involves the use of a heater chuck with three heating cartridges, as established by de Haas et al., that permits the characterization of microfluidic properties of materials in high temperatures, ranging from 130 to 190°C [158]. A summary of some major contributions in each of these sections is shown in Table 4.
Keeping track of the distribution of temperature within a microfluidic device during the time frame of any given experiment is of critical importance for many reasons. From the practical perspective, a temperature change may adversely affect the sample being analyzed [171]. From the basic science perspective, a temperature change may modify the properties of the suspending solution (e.g., viscosity, thermal conductivity, and electrical conductivity), which can deviate the observed outcome of an experiment from what is expected [7]. From both perspectives, if the microfluidic device is unable to perform temperature monitoring, it will be unlikely for the experimentalist to accurately determine the reasons behind the damaged sample or the unexpected observation. In this review, we have described works published recently that address the topic of temperature monitoring in microfluidics. Unfortunately, there are still many challenges in the field that prevent widespread adoption and implementation of these temperature monitoring schemes in every microfluidic device. Some of these challenges are self‐evident from the content of Tables 2, 3, 4 (e.g., limited ranges of operation and little control over the heating/cooling ramp rates) and from the revision of the literature provided in Section 3 (e.g., integrability of microwave heating technology to microfluidic devices). The rest merits a dedicated discussion.
Thermocouples, thermistors, and custom‐built sensors require, for the most part, simple fabrication processes (although fabricating devices requiring pn junctions as needed to exploit the Seebeck and Peltier effects cannot be said to be a simple task). Nonetheless, compatibility issues may arise while aiming at integrating that simple fabrication process with the fabrication process of the microfluidic device. For example, integrating platinum microheaters like those shown in Figure 3A within microfluidic devices built through a simple soft‐lithography approach [172, 173] or through computer numerical control micro‐milling of glass and PMMA [174] will be a very challenging task. Also, even if the fabrication and integration of the sensor is successful, it might still lead to leakage by preventing an adequate bonding of the microfluidic channel itself, or it might lead to sample contamination by being in direct contact with the analyte if the sensor requires an electric current to function (RTD sensor) or to deposition of a passivation material layer on top of the sensor, affecting its performance [175]. If the temperature sensor is built or integrated on an external surface of the microfluidic channel or chamber where temperature is to be monitored, the problems of leakage and contamination will be avoided. Nonetheless, consideration must be given to the thermal properties of the materials separating the sensor from the microfluidic channel or chamber as heat dissipation will have to be considered to accurately map the measured external temperature to the actual internal temperature [28]. These challenges must point our attention toward the development of novel temperature monitoring approaches that are easy to implement, that feature high sensitivity, and that do not interfere with the performance of an experiment.
Temperature monitoring is necessary, but not sufficient, to warrant adequate performance of any microfluidic device. Temperature control strategies must take the information provided by temperature sensors and act to produce the desired temperature profile within the microfluidic channels and chambers. Challenges similar to those discussed for temperature sensors also affect temperature control elements. For example, the use of microheaters inside a microfluidic channel (assuming its fabrication is compatible with the fabrication process of the microfluidic device) might modify flow patterns, induce unwanted chemical reactions, and even damage the device itself [141, 176, 177]. Also, the use of external microheaters must take into consideration the thermal properties of the materials separating the heater from the microfluidic channel or chamber as heat dissipation will play an important role in the heating process. Structural arrangements can overcome most obstacles related to fabrication process incompatibility discussed herein and are therefore a very attractive alternative to temperature control. However, the complexity of fabrication of the microfluidic device as a whole—incorporating the structural arrangements for temperature control—might increase significantly in some cases because of required multilayer lithography and soft lithography steps [143].
Microfluidic devices have perhaps found its greatest niche of opportunity in LOC technology [178, 179, 180]. The concept of LOC implies that all the processing steps required to analyze a sample in a given laboratory test are integrated into a single chip. Most contributions in the field of microfluidics discussed in this and other reviews, however, are capable of performing only one or a few processing steps [181, 182, 183, 184]. The envisioned application of the device dictates if this is acceptable or not. If a microfluidic LOC device is intended to be an additional tool at an analytical chemistry lab, then miniaturization provides more than enough benefits (e.g., reduced costs, smaller response times, and improved efficiency) to support their implementation [185]. In contrast, if the device is intended for POC applications, the benefits of miniaturization are not sufficient to support their implementation. Portability, ease of operation, and integration of several sample processing steps into a single chip become critical aspects of the device [48, 186]. In POC settings, bulky peripheral devices are unwanted as they limit portability, and significant manual input from a human operator (with limited access to precision tools regularly available at the lab setting) is likely to reduce reproducibility of tests. Automation of operation requires electronic instrumentation; therefore, we need to discuss the challenges associated to electronic instrumentation in temperature sensing and control applications.
For some applications (e.g., PCR), a very finely tuned temperature control is required [67]. This implies that the temperature sensor used in the microfluidic device must be able to detect slight deviations from the expected temperature in a channel or chamber. Most sensors deliver the temperature information in electric format (i.e., as a voltage or current). Therefore, a slight temperature deviation will produce a slight change in voltage or current that can be mistaken for noise if the system is not capable of accurately resolving that difference. The exact opposite situation can be found in temperature control approaches, where a voltage or current input indicates the onset of the heating or cooling element and where a slight change in voltage or current can be mistaken for noise, preventing adequate temperature control implementations. Microcontrollers have been extensively used in most contributions reviewed herein to read data from sensors and to deliver instructions to heating and cooling elements [22, 52, 56, 71, 137]. Microcontrollers are powerful tools that enable a wide range of operations to be coordinated in a timely fashion to achieve a desired operation of a microfluidic device. However, in cases similar to those described before, where high resolution is needed, microcontrollers must be accompanied by additional electronic elements (e.g., low noise amplifiers), increasing the complexity, size, cost, and power requirements of the overall system [187]. This, of course, should not demotivate our efforts to integrate electronic elements into our microfluidic devices. After all, microfluidic devices are not limited to portable LOC applications. Microfluidic devices have found application in the development of benchtop equipment as well [185]. Being aware of these many challenges, however, will allow us to decide if the technology we are designing is better suited for a lab setting (benchtop equipment) or for field deployment (LOC).
Finally, it is easy to recognize the great potential IoT has to make devices more efficient as evidenced by recent excellent reviews centered on remote diagnostics and sensing applications [35, 36, 38, 39]. The type of technology that has been discussed in this review is not exempt from this. IoT and big data can provide many positive features to microfluidic devices in terms of temperature sensing and control. For example, they can allow remote monitoring or control of experimental conditions of several devices at once [188]. Also, they can enable statistical analysis of the efficiency in temperature monitoring and control with data coming in from a wide range of devices simultaneously [189]. Therefore, we must seek to integrate connectivity to IoT to microfluidic technology in years to come.
In this review, we have described and discussed the most recent developments in temperature monitoring and control schemes available for microfluidic devices. Without a doubt, temperature is one of the more critical parameters in microfluidic research, given its potential to modify the physical and chemical properties of the suspending solution and to severely damage the analyte of interest. Moreover, temperature sometimes changes as a response to the mechanisms used to analyze or manipulate the liquid contained in the microfluidic device. Despite this, temperature spatial and temporal distributions are not monitored nor controlled in several microfluidics‐related works, sometimes preventing researchers from making accurate interpretations of their experimental observations. Modeling efforts are always welcome in microfluidics because they allow making predictions about the evolution of the environmental conditions of an experiment when heat transport conditions are taken into consideration [164, 190–196]. Nonetheless, the usefulness of analytical and numerical models would be very limited if they were not validated first against experimental evidence, and to provide experimental evidence, sensors are needed.
Herein, we reviewed temperature sensors for microfluidics, most of which consisted of either thermocouples, thermistors, or custom‐made temperature sensors. Their benefits were summarized in Table 2. It is clear that no ideal temperature sensor exists that fits the requirements of every application. Instead, it is more likely to find an ideal temperature sensor for a given application that will not perform as well if used in a different application (e.g., some might require more resolution over a narrow temperature range, whereas others might require less resolution but a wide temperature range).
Once temperature can be monitored, our attention must focus on methods to control the temperature of the liquid in microfluidic devices. Herein we reviewed both internal and external temperature control schemes than can be implemented to keep temperature at its optimal level during the time course of a process. Again, no optimal approach exists that fits all requirements and there are many different and highly efficient approaches available to implement. Exothermic and endothermic reactions, structural arrangements, and microheaters are some of the methods reviewed herein. Benefits of temperature control systems were summarized in Tables 3 and 4.
It is our hope that this review can provide a complete up‐to‐date overview of the state of the art of the research field around temperature monitoring and control in microfluidics. We foresee that these technologies will be further developed and soon implemented in most microfluidic devices. Such implementation, together with IoT and Big Data capabilities, will bring us yet closer to the development of smart microfluidics for both LOC and benchtop microfluidic solutions that can be adopted by analytical and clinical settings around the globe.
The authors have declared no conflict of interest.
This work was financially supported by the Nano‐Sensors & Devices Research Group (0020209I06) and the Federico Baur Endowed Chair in Nanotechnology (0020240I03) at Tecnologico de Monterrey.
Dos‐Reis‐Delgado AA, Carmona‐Dominguez A, Sosa‐Avalos G, Jimenez‐Saaib IH, Villegas‐Cantu KE, Gallo‐Villanueva RC, et al. Recent advances and challenges in temperature monitoring and control in microfluidic devices. Electrophoresis. 2023;44:268–297. 10.1002/elps.202200162
The data that support the findings of this study are available from the corresponding author upon reasonable request.
The data that support the findings of this study are available from the corresponding author upon reasonable request.