ABSTRACT Non-invasive and label-free calorimetry could become a disruptive technique to study single cell metabolic heat production without altering the cell behavior, but it is currently

limited by insufficient sensitivity. Here, we demonstrate microfluidic single-cell calorimetry with 0.2-nW sensitivity, representing more than ten-fold enhancement over previous record,

which is enabled by (i) a low-noise thermometry platform with ultralow long-term (10-h) temperature noise (80 μK) and (ii) a microfluidic channel-in-vacuum design allowing cell flow and

nutrient delivery while maintaining a low thermal conductance of 2.5 μW K−1. Using _Tetrahymena thermophila_ as an example, we demonstrate on-chip single-cell calorimetry measurement with

MICROCALORIMETRIC MEASUREMENT OF METABOLIC ACTIVITY AND EXOTHERMIC REACTIONS ONTO MICROFLUIDIC PLATFORMS VIA HEAT FLUX SENSOR INTEGRATION Article Open access 09 May 2023 MICROFLUIDICS FOR

UNDERSTANDING MODEL ORGANISMS Article Open access 09 June 2022 ON-CHIP DIELECTROPHORETIC SINGLE-CELL MANIPULATION Article Open access 26 August 2024 INTRODUCTION Many biological processes

are inherently tied to heat generation or absorption, which can be measured calorimetrically1. The metabolic activities of microorganisms also involve heat generation as a product of

adenosine triphosphate (ATP) synthesis and oxidative phosphorylation2. Calorimetry is a powerful tool to investigate the metabolic activities because it is a noninvasive (i.e., without the

need of cell lysis to extract and analyze the metabolites) and label-free technique, enabling monitoring of the metabolic behavior of living cells in real time without altering the nature of

the microorganisms3. Calorimetry can also be applied to diagnose bacterial disease by detecting the metabolic heat of bacteria in the blood of a patient and to detect antibiotic

resistance4. Abnormal metabolic activities of cancer cells can also be probed by calorimetry, and a large amount of research has been devoted to elucidating the relationship between the

metabolic heat of cells and cancer5. Metabolic heat has also been recently measured to reveal the energetics of embryonic cell development6. Currently, the metabolic heat of cells is

measured by averaging across the entire population of numerous cells. However, the average cell property does not necessarily represent the property of individual cells7 because the cells

are highly heterogeneous in terms of the type and amount of proteins8, respiration rate9, size and mechanical properties10, and division time11. In particular, heterogeneity of mitochondria

such as diverse membrane potentials, redox states, and calcium levels leads to various metabolic behaviors of individual cells12. Calorimetry has a unique advantage for the investigation of

these metabolic heterogeneities compared with other methods; for instance, it can provide the overall metabolic information including that for anaerobic metabolism, which cannot be detected

by oxygen consumption rate analysis. Although these heterogeneities have been widely investigated using optical13, chemical14, electrical15, and mechanical10 cytometry techniques, a method

that can measure single-cell heat production rate has yet to be developed. This is mainly because the sensitivity of state-of-the-art calorimeters is insufficient to measure the metabolic

heat from a single cell. The poor sensitivity is attributed to the inherent difficulty in measuring heat; various parasitic heat loss mechanisms such as conduction, convection, and radiation

easily dissipate the small amount of heat produced from a single cell, which is usually less than a few nW. Hence, efforts to enhance the calorimetry sensitivity (_Q_s) have been centered

on minimizing the thermal conductance (_G_) to suppress the parasitic heat loss as well as reducing the minimum detectable temperature (Δ_T_min) as _Q_s = _G_ × Δ_T_min. In particular, the

introduction of microfabrication technology to calorimetry enabled the integration of high-resolution thermometers and a microfluidic cell-delivery system on a thermally insulative thin

membrane, resulting in significant improvement of the calorimeter sensitivity16,17,18,19,20,21,22,23,24,25,26,27,28. Microfluidic calorimeters have also allowed dosing stimuli onto

microorganism or monitoring of themorphological change and population of cells in real time without disturbance of the measurement26,27. Microfluidic calorimeters have evolved over the last

few decades to measure metabolic heat production from microorganisms. Johannessen et al.26 demonstrated a chip calorimeter possessing a sensitivity of 13 nW. The calorimeter was able to

measure 16.3 nW of metabolic heat from ten brown adipocytes stimulated with noradrenaline. A major breakthrough occurred in 2009, when Lee et al.16 developed a chip calorimeter with a

sensitivity of 4.2 nW. They achieved low thermal conductance by virtue of thin parylene membrane housed in a vacuum environment. However, the device was not applied for cell measurements,

presumably because the detection limit remained higher than heat production rate by single cells, even in cells with high metabolic rate such as _Tetrahymena_ (~ 3 nW)29,30,31. For about a

decade since the pioneering work of Lee et al.16, no calorimeter has demonstrated sensitivity better than 4 nW, highlighting the challenges of improving the sensitivity of chip calorimetry

with microfluidic handling capability. Here, we present a chip calorimeter capable of single-cell metabolic heat measurement with a high sensitivity of 0.2 nW. We achieve approximately an

order of magnitude greater sensitivity by implementing a one-dimensional suspended microfluidic design in vacuum and a measurement platform with long-term stable temperature (80 μK

temperature drift in 10 h). Furthermore, we achieve single-cell metabolic measurement by magnetically trapping the cells in the microfluidic channel for reliable thermal measurement without

perturbation introduced by cell movement. The microfluidic platform and the trapping technique also allow for a continuous supply of the fluid containing nutrients and oxygen to the cells.

The high sensitivity and accurate cell control system enable us to measure the nW level of heat production from single _Tetrahymena thermophile_. Our measurement reveals a positive

correlation between the metabolic heat generation and cell size. Finally, we demonstrate real-time monitoring of metabolic heat increase induced by the injection of the mitochondrial

uncoupling agent carbonyl cyanide 4-(trifluoromethoxy)phenylhydrazone (FCCP) through the microfluidic channel, which is used to extract the spare respiratory capacity (SRC). Given the

tremendous potential of probing the metabolic rate at the individual cell level, our calorimetry platform will open up a pathway for a wide range of fundamental biological studies and

potentially biomedical applications. RESULTS DESIGN OF CALORIMETER WITH HIGH SENSITIVITY The development of a high-sensitivity calorimeter relies on minimization of the thermal conductance

and minimum detectable temperature. Previous two-dimensional membrane designs still have large thermal conductance. In our design, we utilized a one-dimensional (1D) suspended microfluidic

channel with a small cross-sectional area and large length (2_L_ = 5.0 mm) to achieve low conduction and radiation thermal conductance (Fig. 1a). The microfluidic channel is composed of

parylene, which has high mechanical strength, good visual transparency, and, most importantly, low thermal conductivity (Supplementary Fig. 2b) compared with other materials widely used for

chip calorimeters (e.g., Si3N422,24,26, glass25, Si27). The parylene backbone layer had a width of 120 μm and thickness of 12 μm and formed the microfluidic tube (Fig. 1b). The microfluidic

channel had a cross-sectional area of 35 × 50 μm2, which is compatible with most cell sizes (<35 μm in diameter). The entire tube structure was suspended on a silicon wafer with a ~ 

200-μm gap. Lastly, the entire device was placed in a high-vacuum environment (<0.1 mT)16 to eliminate convection heat loss (Fig. 1d, e). The fluids into and out of the fluidic channel

were introduced via vacuum feedthroughs. Photographs of the setup are shown in Supplementary Fig. 6. A Bi–Pt thermopile and Pt heater were microfabricated and integrated in the parylene

layer underneath the microfluidic channel (Fig. 1c). The thermopile measured the temperature difference (Δ_T_) between the measuring junctions (_T_mea) in the middle of the channel and the

reference junctions (_T_ref) on the Si wafer (Fig. 1a). The Pt heater was used to characterize the thermal conductance of the fluidic channel by applying joule heating to the channel. The

thermopile was calibrated by raising the temperature of the junction to known temperatures with the Pt heater (Fig. 1c) and measuring the voltage of the thermopile. The thermopile was

designed by considering the tradeoff between the thermal conductance and temperature sensitivity: increasing the number of thermopile junctions leads to better temperature detection

sensitivity but also increases the thermal conductance. Likewise, reducing the resistance of the thermopile decreases the electrical noise (e.g., Johnson or 1/_f_ noise) but results in

higher thermal conductance. Our optimized thermopile was composed of four pairs of Bi and Pt thin films (Fig. 1c and Supplementary Fig. 1b), and the root-mean-square (rms) voltage noise of

the measurement system was 19 nV, which includes noises from the thermopile, an operational amplifier (CS3002, Cirrus Logic), and a low-pass filter (cutoff frequency: 0.016 Hz), as shown in 

Supplementary Fig. 4a. The overall thermal conductance of our calorimeter including the aforementioned components and water in the microfluidic channel was estimated as a function of the

half channel length (_L_) using a fin equation (see Supplementary Note 1 for derivation): $$G = 2mkA_c\left( {\frac{{e^{mL} + e^{ - mL}}}{{e^{mL} - e^{ - mL}}}} \right)$$ (1) where the fin

parameter \(m = \sqrt {\frac{{hP}}{{kA_{\mathrm{c}}}}}\), _k_ is the thermal conductivity, _A__c_ is the cross-sectional area of the microfluidic channel, _P_ is the outer perimeter of the

microfluidic channel, and _h_ is the radiation heat-transfer coefficient (_h_ = 4_εσT_avg3, where _σ_ is the Stefan–Boltzmann constant, _ε_ is the emissivity, and _T_avg is the average

temperature). Equation (1) shows that the overall thermal conductance is saturated at _L_ = 2.5 mm, where radiation heat loss becomes more important than the conduction heat loss because of

the large surface area (Supplementary Fig. 3a). The estimated overall thermal conductance with _L_ of 2.5 mm is expected to be 2.48 μW K−1, including the backbone parylene layers, water

inside the microfluidic channel, and Bi–Pt thermopile. It is worth noting that ~48% of the total thermal conductance comes from the water channel (Supplementary Fig. 3b), which is needed for

the continuous nutrient and oxygen supply. We also optimized the geometry to ensure the temperature uniformity around the cell (Supplementary Fig. 3c) and mechanical integrity of the

channel–substrate junction (Supplementary Fig. 3d). The fabricated device was loaded onto our measurement platform (Fig. 1d, e), which was designed to minimize the baseline temperature drift

and provide a vacuum environment. The temperature stability is especially important in single-cell calorimetry because external agitations such as light illumination from a microscope used

to visualize the cell in real time and fluid flow are inevitable. We minimized the temperature drift by using three levels of thermal insulation and temperature-control layers (Fig. 1d) as

well as a stable and hermetically sealed fluid control system (Fig. 1e). By implementing these extensive thermal and fluidic control schemes, we were able to achieve a baseline temperature

stability of our calorimeter of within 80 μK for more than 10 h (Fig. 2a) under the condition of microscope illumination. We also showed that the thermal conductance and temperature

stability were similar when the fluid in the microchannel was flowing at a speed of 0.1 mm s−1 (Supplementary Fig. 7a, c), which was needed to provide sufficient nutrient and oxygen supply

to single _Tetrahymena_ cells (see next section). The sensitivity of our calorimeter was demonstrated by heating the suspended channel with the embedded Pt heater and monitoring the voltage

from the Bi–Pt thermopile. The heating power absorbed by the measuring junctions (_Q_abs) gradually increased from 0 to 4 μW, and the voltage output from the thermopile was collected for 5 

min at each absorbed power. The voltage output from the thermopile was converted to a temperature signal (Δ_T_) using the Seebeck coefficient (_S_ = 229 μV K−1), and the average _ΔT_ and rms

temperature noise (_T_noise) of each measurement are plotted in Fig. 2b. The measured thermal conductance of our calorimeter was 2.51 μW K−1 (_G_ = _dQ_abs/_d_Δ_T_), which is consistent

with the analytical estimation obtained using Eq. (1) (2.48 μW K−1). This thermal conductance is approximately six times lower than that of the state-of-the-art chip calorimeter (16 μW 

K−1)16. _T_noise was 80 μK for 5 min, and the corresponding sensitivity of our single-cell calorimeter was 0.2 nW (_Q__s_ = _G_ × _T_noise). This sensitivity is approximately one order of

magnitude better than that of state-of-the-art on-chip calorimeters (4.2–13 nW)16,26, as summarized in Fig. 2c. The expected metabolic rates of several cells also shown in Fig. 2c; it is

apparent that our calorimetry platform achieved sufficient sensitivity to measure the metabolic rate of single cells such as protozoa29,30,31 or stimulated mammalian cells32, which have not

been able to be measured by previous calorimeters16,17,18,19,20,21,22,23,24,25,26,27,28. SINGLE-CELL TRAPPING AND CONTINUOUS NUTRIENT SUPPLY As a demonstration of our calorimeter and

measurement platform, we measured metabolic heat generation from single _Tetrahymena thermophila_ SB210. _Tetrahymena_ is widely used as a model cell line to study mammalian cells because it

has high homology with mammalian cells not only responding to mammalian hormones but also synthesizing similar molecules (e.g., insulin)33. Therefore, their metabolism has been studied by

monitoring their oxygen consumption34,35,36,37 or metabolic heat29,30,31 or the size of their mitochondria;38 however, none of the existing measurement platforms have enabled single-cell

metabolic rate analysis. Using our high-sensitivity single-cell calorimeter, we were able to measure the metabolic rate of single _Tetrahymena_ with sub-nW resolution. A key challenge to

measure _Tetrahymena_ (and other cells) is the need to trap individual cells in the microfluidic channel and supply sufficient nutrients and oxygen to the trapped cells. Notably,

_Tetrahymena_ are difficult to trap because they move to find nutrients using cilia. To trap _Tetrahymena_, we applied the magnetic cell trapping system schematically illustrated in Fig. 3a

(see “Methods”). As previously shown39, _Tetrahymena_ can take up iron oxide nanoparticles, and their movement can then be controlled by an external magnetic field. The presence of iron

oxide nanoparticles was clearly visible in the cytoplasm in the _Tetrahymena_ (Fig. 3c), whereas the _Tetrahymena_ without the particles displayed normal cytoplasm (Fig. 3b). It is worth

noting that iron oxide nanoparticles are not known to impair any cellular activity40. With the application of an external magnetic field, the ingested iron oxide nanoparticles produced an

induced magnetic dipole and were aligned with the direction of the magnetic force, as shown in Fig. 3d. The _Tetrahymena_ with the particles can then be attracted and eventually trapped in

the presence of the external magnetic field. By aligning the middle of the calorimeter with the interface of the two magnets (Figs. 1e and 3a), we were able to place and trap individual

_Tetrahymena_ cells sitting atop of the thermopile junction in the middle of the microfluidic channel. As a single _Tetrahymena_ was trapped, nutrients and oxygen had to be supplied to

maintain their normal metabolism. Notably, without a continuous supply of oxygen, the amount of oxygen contained in the microfluidic channel and transported to _Tetrahymena_ would have been

limited because of the small volume and cross-sectional area. The oxygen consumption of _Tetrahymena_ is known to be 0.1–1.0 nLO2 cell−1 h−134; however, our analysis indicated that the

amount of oxygen in the channel was not sufficient for a single _Tetrahymena_ (Supplementary Fig. 5). Furthermore, previous studies have shown that starved protozoa have a smaller metabolic

rate than normal protozoa29,34. To supply sufficient nutrients and oxygen, we applied a 0.1 mm s−1 flow of the growth medium to the trapped _Tetrahymena_. The oxygen supply by the flow was

estimated to be ~3.4 nL O2 h−1 at this flow rate, which is a few times larger than the estimated oxygen consumption of single _Tetrahymena_. This low flow rate was insufficient to overcome

the magnetic force trapping _Tetrahymena_; therefore, the trapped _Tetrahymena_ were not moved by the flow during the measurement (Supplementary Movie 1 (cell trapping)). In addition, the

flow did not have a major effect on the calorimeter sensitivity. We determined the thermal conductance of the calorimeter with the flow using the same method described in Fig. 2b to verify

the amount of possible advection heat transfer by the flow. The result indicated that the advection heat transfer was almost negligible; thermal conductance with the flow was only increased

by ~4% compared with the no-flow condition (from 2.51 to 2.62 μW K−1, Supplementary Fig. 7a). Furthermore, this low-speed constant flow was shown to have negligible effect on the temperature

stability in the fluidic channel (Supplementary Fig. 7c). This was mainly achieved by incorporating a 2 cm-long serpentine microfluidic channel on the temperature-controlled Si substrate to

preheat the fluid and stabilize its temperature before it enters the suspended channel (Supplementary Fig. 1c). The temperature signal from the suspended calorimeter chamber before and

after injecting the fluid (at constant speed of 0.1 mm s−1, corresponding to 0.175 nL s−1 flow rate) was the same, expect for the small temporary drop for a duration of about 1 min after the

injection (Supplementary Fig. 4e). In contrast, pulsed flow with a high flow rate can cause temporary temperature reduction. For example, pulsed fluid injection with ~ 10 nL s−1 flow rate

caused ~ 0.8 mK temporary temperature drop (Supplementary Fig. 4c), because the serpentine channel is not long enough to allow a sufficient residence time for high-rate pulsed flow to reach

thermal equilibrium (about 3.5 s compared with 200 s for 0.175 nL s−1 flow). We observed this instant temperature drop due to the pulsed fluid injection when we removed the cells from the

calorimeter for baseline measurement (Fig. 3f). This temperature drop was recovered in <1 min as soon as we restored the flow rate to the baseline level of 0.1 mm s−1. The procedure and a

representative result of the metabolic rate measurement of single cells are displayed in Fig. 3e, f, respectively. First, a single _Tetrahymena_ was injected through the microfluidic

channel and trapped on the measuring junctions by a magnetic field. Then, the thermopile recorded the temperature signal of the middle of the microfluidic channel with the trapped cell

(_T_cell) while 0.1 mm s−1 flow of growth medium supplied a sufficient amount of nutrients and oxygen. After the measurement, the cell was removed from the calorimeter by a pulsed flow,

which was sufficiently strong to overcome the magnetic force. Then, the temperature signal without the cell (_T_base) was recorded as a baseline. In Fig. 3f, the sudden temperature change at

5 min was caused by the pulsed flow for cell removal. This measurement process is shown in Supplementary Movie 2 (cell injection) and Movie 3 (cell removal). SINGLE-CELL METABOLIC RATE

MEASUREMENTS We successfully measured the metabolic rate of single cells. The temperature signal measured from the thermopile when a cell was trapped in the calorimeter (_T_cell) was higher

than the baseline temperature after removing the cell (_T_base). This temperature rise (_T_cell  − _T_base) indicates the heat production by cellular metabolic activity of a single

_Tetrahymena_ and can be converted into metabolic rate (_Q_cell = _G_ × (_T_cell − _T_base)). As displayed in the inset graph of Fig. 4a, the average basal metabolic rate of _Tetrahymena_

was 2.0 ± 0.7 nW cell−1. This result falls within the range of published values (from 1 to 6 nW cell−129,30,31); however, our data provide a much more precise value at the individual cell

level. As a control experiment, we also measured the heat generation rate from dead cells as well as flowing water without cells, which indicated no temperature rise from the baseline (Fig. 

4a). In addition, the temperature rise and heat generation of two and three cells were also measured, which clearly showed an increase of the heat generation from ~2 nW for one cell to ~4 nW

for two cells and finally to ~ 6 nW for three cells. These experiments confirmed that the detected temperature rise originated from the cell metabolic heat production. With these

measurements, we further quantified the cell metabolic heat rate as a function of individual cell size (Fig. 4b); the results are summarized in Fig. 4c (with all the raw data for each

measurement presented in Supplementary Fig. 8). REAL-TIME MONITORING OF METABOLIC RATE STIMULATION We further demonstrated stimulation of the metabolic rate by injecting the uncoupling agent

FCCP. The measurement procedure was the same as that shown in Fig. 3e, with the added step of injecting FCCP at a constant flow rate of 0.1 mm s−1 into the microfluidic channel after the

initial temperature of the microchannel with cell was stabilized. After trapping a single _Tetrahymena_, the heat signal from the cell indicated a typical basal metabolic rate. When FCCP was

injected into the calorimeter, the heat generation from the cell gradually increased and eventually reached a maximum (_Q_max) in 1–2 min (Fig. 4d). This stimulated heat signal was

collected for 3 min, and then the baseline was measured again after removing the cell with the pulsed flow. The basal metabolic heat generation of _Tetrahymena_ increased by a factor of 2.1 

± 0.3 with FCCP stimulation. We also measured the change in the oxygen consumption rate of a bulk volume of _Tetrahymena_ upon FCCP stimulation using a Hansatech Oxytherm apparatus41. The

basal and stimulated oxygen consumption rate of _Tetrahymena_ were measured to be 0.20 ± 0.01 and 0.39 ± 0.04 nLO2 h−1, respectively. These measurements indicated that the oxygen consumption

rate increased by a factor of 1.9 ± 0.1 with the FCCP stimulation (inset of Fig. 4d), which is consistent with our calorimeter results. DISCUSSION The basal metabolic heat is mainly

produced by mitochondrial activity such as oxidative phosphorylation, generating a proton gradient across the inner membrane of mitochondria, and synthesis of ATP using the proton

gradient42. Therefore, the mitochondria content in cells largely affects the basal metabolic heat generation31. As this mitochondria content increases with the growth of cytoplasm38,

eventually the metabolic rate is a function of cell size during their growth phases (G1–G2 phases before division)43. In fact, this allometric scaling relationship between the metabolic rate

and body size have been extensively studied in a wide range of organisms from the smallest unicellular microorganism34 to the largest mammal44 and is known to follow a three-quarter power

relationship: $$B = B_0V^{3/4}$$ (2) where, _B_ is the basal metabolic rate, _B_0 is the normalization coefficient, and _V_ is the body size. Previous studies of the cellular allometric

scaling relationship have focused on the relationship between groups of cells34,44 because conventional metabolic rate measurement systems can only measure the average value of a bulk amount

of cells. The metabolic rate change of individual cells during growth have yet to be investigated because of the lack of single-cell measurement system and has only been assumed by studying

synchronous cultures31 or using optical method38. In our single-cell measurement, we observed that the metabolic rate of _Tetrahymena_ was correlated to the cell size (Fig. 4c) with a

correlation of _r_2 = 0.42. Our results reveal the clear increase of the metabolic rate as a function of cell size. As indicated by the representative results presented in Fig. 4b, a small

cell (9 × 103 μm3) generated metabolic heat of 0.9 nW, whereas a large cell (35 × 103 μm3) produced metabolic heat as high as 3.3 nW, with a clear difference in their heat signals.

Considering that the cell metabolic rate can also be affected by other factors (e.g., gene, protein expression, etc.), the observed correlation between the metabolic rate and cell size is

significant. It is also important to note that this correlation approximately follows the conventional three-quarter power allometric scaling relationship (inset of Fig. 4c), which suggests

that the scaling relationship is also applicable in individual cells. FCCP can uncouple the respiratory chain from ATP synthesis by direct transport of protons across the inner membrane of

mitochondria and eventually dissipate the proton gradient42. This uncoupled respiration maximizes the mitochondrial metabolic activity, which provides useful information about the cellular

metabolism45. For instance, a large SRC (ratio of _Q_max to _Q_cell) of more than 5.0 can be observed in postmitotic cells such as adipocytes, myocytes, or neurons, which means the cells can

appropriately respond to an increased ATP demand and withstand periods of stress. In contrast, most proliferating cells exhibit a SRC below 3.0 because of the high basal metabolism required

to meet the energy demands of cell division. For _Tetrahymena_, although one study reported a sixfold increase of the oxygen consumption rate with FCCP treatment35, other studies have

reported SRC values of ~1.5–3.036,37 under regular growth conditions. In addition, most proliferative unicellular as well as small multicellular organisms have SRC values of <3.0 with

FCCP treatment. For instance, the SRC has been reported to be 1.4 for protozoa (e.g., _Trypanosoma cruzi_)46, 1.5–3.0 for mammalian cells45, and 1.5–2.0 for worms (_Caenorhabditis

elegans_)27. The SRC of _Tetrahymena_ measured using our calorimeter (SRC = 2.1) was similar to the typical SRC of other proliferative cells and agrees well with our oxygen consumption rate

measurement (SRC = 1.9, Fig. 4d). These comparisons suggest that the enhanced heat signal with FCCP injection in the calorimeter was generated by the stimulated metabolic rate of

_Tetrahymena_. In addition, the SRC of 1.9–2.1 indicates the _Tetrahymena_ used in our demonstration were in a normal proliferative state at the time of the measurement. In summary, we

develop a high-sensitivity microfluidic chip calorimeter and demonstrated calorimetric measurement of heat production by single cells. The design with a one-dimensional suspended

microfluidic channel and extensive thermal shielding results in significant reduction of the thermal conductance to as low as 2.51 μW K−1 and low long-term noise of 80 μK, thus enabling a

high calorimetric sensitivity of 0.2 nW for single-cell metabolic rate measurement. We measure the basal metabolic heat of single _Tetrahymena_ and reveal a clear correlation between the

metabolic rate and size of individual cells. Finally, we stimulate metabolism of _Tetrahymena_ by dosing the mitochondrial uncoupling agent FCCP through the microfluidic channel and achieve

excellent agreement between our heat-generation and oxygen-consumption measurements. These demonstrations suggest that our sub-nanowatt microfluidic calorimetry could be utilized for

noninvasive, real-time study of single-cell behaviors, such as cell proliferation5 and embryonic development6 as well as for cell cytometry and phenotyping8,10. METHODS DEVICE FABRICATION

The fabrication process is illustrated in Supplementary Fig. 1a. First, a 2-in. Si wafer surface was treated with A-174 adhesion promoter to improve the adhesion between the Si and parylene.

A 3-μm-thick first parylene layer was deposited on the Si wafer using a PDS 2010 Parylene Coater. Next, 60-nm-thick Pt and 500-nm-thick Bi layers were sequentially sputtered on the first

parylene layer and patterned by photolithography and a lift-off process to form the thermopile. After that, the second parylene layer (3-μm thick) was deposited on the metal layers to

passivate the thermopile. Then, 35-μm-thick photoresist (PR) was patterned by photolithography to form a template of the microfluidic structure, which was later removed. The third parylene

layer (6-μm thickness) was deposited on the PR template to fabricate the microfluidic channel. Subsequently, the parylene layers where the inlet/outlet holes and the Si etching window would

be formed were etched by reactive-ion etching. A 60-μm thick SU-8 layer was then patterned to protect the parylene microfluidic channels. The device was then immersed in propylene glycol

monomethyl ether acetate to dissolve the PR microfluidic template structure. Finally, the suspended structure was released by etching the Si underneath the microfluidic tubes using XeF2 gas.

DEVICE CHARACTERIZATION The thermal conductance and sensitivity of the calorimeter were measured using a Pt thin film heater embedded in the suspended microfluidic tubes. First, the

temperature coefficient of resistance (TCR) of the Pt film was measured by changing the sample stage temperature and monitoring the resultant electrical resistance change. The measured TCR

was 0.00189 K−1. Then, heating power was applied to the Pt heater and gradually increased while the voltage generation of the Bi–Pt thermopile and resistance change of the Pt heater were

monitored at a constant stage temperature. The resistance change of the Pt heater was converted to the average temperature change along the suspended tube using the TCR. The overall thermal

conductance of the calorimeter (_G_) and Seebeck coefficient of the Bi–Pt thermopile (_S_) were calculated by COMSOL simulation using the applied power to the Pt heater and resultant average

temperature change. In addition, the absorbed heat by the measuring junctions (_Q_abs) during the Joule heating by the Pt heater, which heated the entire suspended tube, was estimated by

the COMSOL simulation. The thermal conductance was calculated based on _Q_abs and the corresponding Δ_T_mea change (_G_ = _Q_abs/Δ_T_mea). The material parameters of parylene used in the

COMSOL simulation were experimentally measured; the in-plane thermal conductivity and emissivity were measured to be 0.18 W m−1 K−1 and 0.21, respectively (Supplementary Fig. 2). In

addition, the thermal conductivity of the Pt thin film used in the simulation was ~ 54 W m−1 K−1, which was estimated from the measured electrical conductivity and the Wiedemann–Franz law

(_k_ = _σLT_, where _k_ is thermal conductivity, _σ_ is the electrical conductivity, _L_ is the Lorenz number, and _T_ is the temperature), and the thermal conductivity of Bi thin film was

assumed to be ~ 4 W m−1 K−1 based on the literature values for similar films47. MEASUREMENT PLATFORM The measurement platform was composed of three temperature-controlled layers: an outer

chamber, a vacuum chamber, and a sample stage. The temperature of the layers was kept constant by temperature controllers (PTC-10, Stanford Research Systems), thin film heaters, and a

thermoelectric device, as illustrated in Fig. 1d, e. The outer chamber and vacuum chamber were insulated by electrically neutral balsa wood to prevent electrostatic build-up, and the sample

stage was insulated by a high-vacuum environment. Thermal radiation from the immediate surroundings was blocked by a microfluidic clamp, which covered the calorimeter as a radiation shield

with only a small viewport to see the microfluidic channel. The microscope illumination was provided from a LED light source with a color temperature of 5000 K, which has no infrared

component. To avoid excessive heating by the illumination (could be as high as ~100 mK in temperature rise and a few hundred μK in temperature noise if a strong illumination was used), the

LED light intensity was adjusted to the minimal level needed for the cell observation in the microchannels, which resulted in no noticeable temperature noise (Supplementary Fig. 7c). In

addition, we used an aluminum cover with a small viewport size as a radiation shield. The viewport on the clamp was further covered by two separated pieces of IR-absorbing glass to prevent

the transmission of the thermal radiation from the light and the environment, as shown in Supplementary Fig. 4f. These thermal radiation shields substantially reduced the temperature drift

in the microfluidic channel from ~5 mK to ~80 μK (Fig. 2a, b, Supplementary Fig. 4g, h). For accurate and leakage-free fluid control, the inlet/outlet of microfluidic channel in the

calorimeter were connected to an external fluid control system by a custom-made microfluidic clamp. Precisely designed microfluidic fittings were screwed on the microfluidic clamp and

provided a vacuum-sealed and small dead-volume fluid connection between the microfluidic channel and the tubes. The small dead volume of the platform enabled accurate fluid control as well

as efficient cell and stimuli injection, and vacuum-tight sealing prevented temperature agitation by fluid evaporation. The small inner size of the tubes (50 μm in diameter) also minimized

the dead volume of the fluid system and the inlet/outlet tubes passed through the lead of the vacuum chamber with a vacuum-sealed feedthrough. A syringe pump (Pump 11 Pico Plus, Harvard

Apparatus, MA, USA) and a 100 μL syringe (Hamilton Company, NV, USA) were used to pump the fluid into the microchannel. Considering the cross-section area of the microfluidic tube (35 × 50 

μm2) and the inner diameter of the syringe (1.4 mm), a volumetric flow rate of 0.175 nL s−1 was applied to yield fluid velocity of 0.1 mm s−1 in the microfluidic tube, in order to

continuously supply the nutrient and oxygen to the cells. CELL PREPARATION _Tetrahymena thermophila_ SB210 cells (Tetrahymena Stock Center, Cornell University) were axenically cultured to

middle log phase in a liquid medium containing 0.25% yeast extract, 0.25% proteose peptone, 0.5% D-Glucose, and 1 mM FeCl3. Then, 0.1 vol% iron oxide nanoparticles (40 nm in diameter) were

added to the culture medium with _Tetrahymena_ and gently mixed. The resultant suspension was allowed to stand for 1 h to ensure that the _Tetrahymena_ took up the nanoparticles. The

magnetic trapping effect on cells was verified by placing a droplet of the cells on a glass slide and applying a magnetic field before injecting the cells in the calorimeter. A fixation

buffer for the dead cell measurement was prepared by mixing 1 mL of 37% formaldehyde solution, 1 mL of phosphate buffer solution (PBS), and 8 mL of water. _Tetrahymena_ was centrifuged to

aspirate the growth medium, and the fixation buffer was added to kill the cells. After incubation at room temperature for 10 min, the cells were washed with PBS and stored in 70% ethanol

solution. MAGNETIC CELL TRAPPING We used two NdFeB permanent magnets embedded in the sample stage to generate the magnetic field for cell trapping (Fig. 1e); the first magnet was placed with

its south pole upwards, and the second magnet was placed with its north pole upward next to the first magnet. The iron oxide nanoparticles were mixed with the culture media, and

_Tetrahymena_ took up the particles before measurement. The cells containing the iron oxide nanoparticles were trapped at the interface of the two magnets where the highest magnetic field

occurred (Supplementary Movie 1 (cell trapping)). CELL VOLUME MEASUREMENT The volume of _Tetrahymena_ (_V_) was calculated based on the morphometric analysis with the assumption that the

cells had a prolate spheroid shape48: $$V = \frac{\pi }{6}L_TB_T^2,$$ (3) where _L__T_ is the length and _B__T_ is the breadth of _Tetrahymena_. The standard deviation of this optical

estimation compared with electrical measurement (Coulter counter)48 is ~1800–5500 μm3. OXYGEN CONSUMPTION RATE MEASUREMENT The oxygen consumption rate was measured using a Hansatech Oxytherm

(Hansatech, King’s Lynn, UK). The instrument was calibrated at 28 °C per the manufacturer’s instructions. _Tetrahymena_ were grown and assayed in normal culture media as detailed above.

Then, 0.5 mL of cell culture with concentration of 280,000 _Tetrahymena_ per mL was added to the chamber under constant magnetic stirring. After the basal oxygen consumption rate was

measured, successive 0.2-μL additions of 0.2 mM FCCP were added until a maximal uncoupler-stimulated respiration rate was achieved. REPORTING SUMMARY Further information on research design

is available in the Nature Research Reporting Summary linked to this article. DATA AVAILABILITY Source data for Figs. 2a, b, 3f, 4a–d and Supplementary Figs. 4c, d, e, g, h, 7a, 7c, 8 have

been provided as a Source Data file. All other data supporting the findings of this study are available from the corresponding author upon reasonable request. Source data are provided with

this paper. REFERENCES * Ladbury, J. E. & Doyle, M. L. _Biocalorimetry 2: Applications of Calorimetry in the Biological Sciences_. John Wiley & Sons (2004). * Chance, B. &

Williams, G. The respiratory chain and oxidative phosphorylation. _Adv. Enzymol. Relat. Areas Mol. Biol._ 17, 65–134 (1956). CAS  Google Scholar  * Lee, W., Lee, J. & Koh, J. Development

and applications of chip calorimeters as novel biosensors. _Nanobiosens Dis. Diagn._ 1, 17–29 (2012). Article  Google Scholar  * Baldoni, D., Hermann, H., Frei, R., Trampuz, A. &

Steinhuber, A. Performance of microcalorimetry for early detection of methicillin resistance in clinical isolates of Staphylococcus aureus. _J. Clin. Microbiol._ 47, 774–776 (2009). Article

  PubMed  PubMed Central  Google Scholar  * Vander Heiden, M. G., Cantley, L. C. & Thompson, C. B. Understanding the Warburg effect: the metabolic requirements of cell proliferation.

_Science_ 324, 1029–1033 (2009). Article  ADS  CAS  Google Scholar  * Rodenfels, J., Neugebauer, K. M. & Howard, J. Heat oscillations driven by the embryonic cell cycle reveal the

energetic costs of signaling. _Dev. Cell_ 48, 646–658 (2019). Article  CAS  PubMed  PubMed Central  Google Scholar  * Lidstrom, M. E. & Meldrum, D. R. Life-on-a-chip. _Nat. Rev.

Microbiol._ 1, 158 (2003). Article  CAS  PubMed  Google Scholar  * Cohen, D. et al. Chemical cytometry: fluorescence-based single-cell analysis. _Annu. Rev. Anal. Chem._ 1, 165–190 (2008).

Article  CAS  Google Scholar  * Etzkorn, J. R. et al. Using micro-patterned sensors and cell self-assembly for measuring the oxygen consumption rate of single cells. _J. Micromech.

Microeng._ 20, 095017 (2010). Article  ADS  CAS  Google Scholar  * Gossett, D. R. et al. Hydrodynamic stretching of single cells for large population mechanical phenotyping. _Proc. Natl

Acad. Sci. USA_ 109, 7630–7635 (2012). Article  ADS  CAS  PubMed  Google Scholar  * Strovas, T. J., Sauter, L. M., Guo, X. & Lidstrom, M. E. Cell-to-cell heterogeneity in growth rate and

gene expression in methylobacterium extorquens AM1. _J. Bacteriol._ 189, 7127–7133 (2007). Article  CAS  PubMed  PubMed Central  Google Scholar  * Kuznetsov, A. V. & Margreiter, R.

Heterogeneity of mitochondria and mitochondrial function within cells as another level of mitochondrial complexity. _Int. J. Mol. Sci._ 10, 1911–1929 (2009). Article  CAS  PubMed  PubMed

Central  Google Scholar  * Perfetto, S. P., Chattopadhyay, P. K. & Roederer, M. Seventeen-colour flow cytometry: unravelling the immune system. _Nat. Rev. Immunol._ 4, 648 (2004).

Article  CAS  PubMed  Google Scholar  * Wu, H., Wheeler, A. & Zare, R. N. Chemical cytometry on a picoliter-scale integrated microfluidic chip. _Proc. Natl Acad. Sci. USA_ 101,

12809–12813 (2004). Article  ADS  CAS  PubMed  Google Scholar  * Werdich, A. A. et al. A microfluidic device to confine a single cardiac myocyte in a sub-nanoliter volume on planar

microelectrodes for extracellular potential recordings. _Lab. Chip_ 4, 357–362 (2004). Article  CAS  PubMed  Google Scholar  * Lee, W., Fon, W., Axelrod, B. W. & Roukes, M. L.

High-sensitivity microfluidic calorimeters for biological and chemical applications. _Proc. Natl Acad. Sci. USA_ 106, 15225–15230 (2009). Article  ADS  CAS  PubMed  Google Scholar  * Recht,

M. I. et al. Enthalpy array analysis of enzymatic and binding reactions. _Anal. Biochem._ 377, 33–39 (2008). Article  CAS  PubMed  PubMed Central  Google Scholar  * Torres, F. E. et al.

Enthalpy arrays. _Proc. Natl Acad. Sci. USA_ 101, 9517–9522 (2004). Article  ADS  CAS  PubMed  Google Scholar  * Wang, L., Sipe, D. M., Xu, Y. & Lin, Q. A. MEMS thermal biosensor for

metabolic monitoring applications. _J. Microelectromech. Syst._ 17, 318–327 (2008). Article  CAS  Google Scholar  * Hartmann, T. et al. Thermopile chip based calorimeter for the study of

aggregated biological samples in segmented flow. _Sens Actuators B_ 201, 460–468 (2014). Article  CAS  Google Scholar  * Lerchner, J. et al. Nano-calorimetry of small-sized biological

samples. _Thermochim. Acta_ 477, 48–53 (2008). Article  CAS  Google Scholar  * Zhang, Y. & Tadigadapa, S. Calorimetric biosensors with integrated microfluidic channels. _Biosens.

Bioelectron._ 19, 1733–1743 (2004). Article  CAS  PubMed  Google Scholar  * Xu, J., Reiserer, R., Tellinghuisen, J., Wikswo, J. P. & Baudenbacher, F. J. A microfabricated

nanocalorimeter: design, characterization, and chemical calibration. _Anal. Chem._ 80, 2728–2733 (2008). Article  CAS  PubMed  PubMed Central  Google Scholar  * Chancellor, E., Wikswo, J.,

Baudenbacher, F., Radparvar, M. & Osterman, D. Heat conduction calorimeter for massively parallel high throughput measurements with picoliter sample volumes. _Appl. Phys. Lett._ 85,

2408–2410 (2004). Article  ADS  CAS  Google Scholar  * Fiorino, A. et al. Parallelized, real-time, metabolic-rate measurements from individual Drosophila. _Sci. Rep._ 8, 14452 (2018).

Article  ADS  PubMed  PubMed Central  CAS  Google Scholar  * Johannessen, E. A. et al. Micromachined nanocalorimetric sensor for ultra-low-volume cell-based assays. _Anal. Chem._ 74,

2190–2197 (2002). Article  CAS  PubMed  Google Scholar  * Krenger, R., Lehnert, T. & Gijs, M. A. Dynamic microfluidic nanocalorimetry system for measuring Caenorhabditis elegans

metabolic heat. _Lab Chip_ 18, 1641–1651 (2018). Article  CAS  PubMed  Google Scholar  * Inomata, N., Toda, M. & Ono, T. Highly sensitive thermometer using a vacuum-packed Si resonator

in a microfluidic chip for the thermal measurement of single cells. _Lab Chip_ 16, 3597–3603 (2016). Article  CAS  PubMed  Google Scholar  * Chen, X., Feng, W. & Yu, Y. Studies on the

nongrowth metabolism of the different strains of Tetrahymena cells by isothermal microcalorimetry. _J. Therm. Anal. Calorim._ 115, 2145–2149 (2014). Article  CAS  Google Scholar  * Chang-Li,

X. et al. Microcalorimetric studies on Tetrahymena pyriformis: part 1. Growth metabolic power and thermal equations. _Thermochim. Acta_ 183, 117–123 (1991). Article  Google Scholar  *

Lloyd, D., Phillips, C. A. & Statham, M. Oscillations of respiration, adenine nucleotide levels and heat evolution in synchronous cultures of Tetrahymena pyriformis ST prepared by

continuous-flow selection. _Microbiology_ 106, 19–26 (1978). CAS  Google Scholar  * Hall, J. E. _Guyton and Hall Textbook of Medical Physiology_. Elsevier Health Sciences (2015). * Le Roith,

D., Shiloach, J., Roth, J. & Lesniak, M. A. Evolutionary origins of vertebrate hormones: substances similar to mammalian insulins are native to unicellular eukaryotes. _Proc. Natl Acad.

Sci. USA_ 77, 6184–6188 (1980). Article  ADS  PubMed  Google Scholar  * Fenchel, T. & Finlay, B. Respiration rates in heterotrophic, free-living protozoa. _Micro. Ecol._ 9, 99–122

(1983). Article  CAS  Google Scholar  * Prikhodko, E., Brailovskaya, I., Korotkov, S. & Mokhova, E. Features of mitochondrial energetics in living unicellular eukaryote Tetrahymena

pyriformis. A model for study of mammalian intracellular adaptation. _Biochemistry_ 74, 371–376 (2009). CAS  PubMed  Google Scholar  * Conner, R. & Cline, S. Iron deficiency and the

metabolism of Tetrahymena pyriformis. _J. Protozool._ 11, 486–491 (1964). Article  CAS  PubMed  Google Scholar  * Hamburger, K. & Zeuthen, E. Synchronous divisions in Tetrahymena

pyriformis as studied in an inorganic medium: the effect of 2, 4-dinitrophenol. _Exp. Cell Res._ 13, 443–453 (1957). Article  CAS  PubMed  Google Scholar  * Kolb-Bachofen, V. & Vogell,

W. Mitochondrial proliferation in synchronized cells of Tetrahymena pyriformis: a morphometric study by electron microscopy on the biogenesis of mitochondria during the cell cycle. _Exp.

Cell Res._ 94, 95–105 (1975). Article  CAS  PubMed  Google Scholar  * Kim, D. H., Cheang, U. K., Kőhidai, L., Byun, D. & Kim, M. J. Artificial magnetotactic motion control of Tetrahymena

pyriformis using ferromagnetic nanoparticles: a tool for fabrication of microbiorobots. _Appl. Phys. Lett._ 97, 173702 (2010). Article  ADS  CAS  Google Scholar  * Jeng, H. A. &

Swanson, J. Toxicity of metal oxide nanoparticles in mammalian cells. _J. Environ. Sci. Health, Part A: Toxic./Hazard Subst. Environ. Eng._ 41, 2699–2711 (2006). Article  CAS  Google Scholar

  * Divakaruni, A. S., Rogers, G. W. & Murphy, A. N. Measuring mitochondrial function in permeabilized cells using the seahorse XF analyzer or a clark‐type oxygen electrode. _Curr.

Protoc. Toxicol._ 60, 25.22.1–25.22.16 (2014). Article  Google Scholar  * Rolfe, D. & Brown, G. C. Cellular energy utilization and molecular origin of standard metabolic rate in mammals.

_Physiol. Rev._ 77, 731–758 (1997). Article  CAS  PubMed  Google Scholar  * Cleffmann, G., Reuter, W. & Seyfert, H. Increase in macromolecular amounts during the cell cycle of

Tetrahymena: a contribution to cell cycle control. _J. Cell Sci._ 37, 117–124 (1979). CAS  PubMed  Google Scholar  * West, G. B., Woodruff, W. H. & Brown, J. H. Allometric scaling of

metabolic rate from molecules and mitochondria to cells and mammals. _Proc. Natl Acad. Sci. USA_ 99, 2473–2478 (2002). Article  ADS  PubMed  Google Scholar  * Divakaruni, A. S., Paradyse,

A., Ferrick, D. A., Murphy, A. N. & Jastroch, M. Analysis and interpretation of microplate-based oxygen consumption and pH data. _Methods Enzymol_. 547, 309–354 (2014). Article  CAS 

PubMed  Google Scholar  * Shah-Simpson, S., Pereira, C. F., Dumoulin, P. C., Caradonna, K. L. & Burleigh, B. A. Bioenergetic profiling of Trypanosoma cruzi life stages using Seahorse

extracellular flux technology. _Mol. Biochem. Parasitol._ 208, 91–95 (2016). Article  CAS  PubMed  Google Scholar  * Abrosimov, V., Egorov, B., Lidorenko, N. & Karandashev, V.

Dimensional effect of transfer-coefficients in oriented films of bismuth grown on polymer substrates. _High Temp._ 12, 456–461 (1974). Google Scholar  * Nilsson, J. R. On the correlation of

the volume of a prolate‐spheroidal cell, tetrahymena as determined by electronic particle counters and by morphometry. _J. Protozool._ 37, 500–504 (1990). Article  Google Scholar  Download

references ACKNOWLEDGEMENTS R.C. acknowledges the startup fund from UCSD and National Science Foundation (CBET-1336428 and DMR-1508420) for the support of this work. C.M. acknowledges the

support from National Science Foundation (CAREER Award CBET-1454425). AUTHOR INFORMATION Author notes * These authors contributed equally: Edward Dechaumphai, Courtney R. Green. AUTHORS AND

AFFILIATIONS * Materials Science and Engineering Program, University of California San Diego, La Jolla, CA, 92093, USA Sahngki Hong, Ratneshwar Lal & Renkun Chen * Department of

Mechanical and Aerospace Engineering, University of California San Diego, La Jolla, CA, 92093, USA Sahngki Hong, Edward Dechaumphai, Ratneshwar Lal & Renkun Chen * Department of

Bioengineering, University of California San Diego, La Jolla, CA, 92093, USA Courtney R. Green, Ratneshwar Lal & Christian M. Metallo * Department of Pharmacology, University of

California San Diego, La Jolla, CA, 92093, USA Anne N. Murphy Authors * Sahngki Hong View author publications You can also search for this author inPubMed Google Scholar * Edward Dechaumphai

View author publications You can also search for this author inPubMed Google Scholar * Courtney R. Green View author publications You can also search for this author inPubMed Google Scholar

* Ratneshwar Lal View author publications You can also search for this author inPubMed Google Scholar * Anne N. Murphy View author publications You can also search for this author inPubMed 

Google Scholar * Christian M. Metallo View author publications You can also search for this author inPubMed Google Scholar * Renkun Chen View author publications You can also search for this

author inPubMed Google Scholar CONTRIBUTIONS R.C. and C.M. conceived the project. S.H. and E.D. designed and fabricated the devices and made the setup. S.H. did the calorimetry experiments.

C.G., A.M., C.M. contributed to cell culturing and oxygen consumption measurements. S.H., C.G., C.M., and R.C. interpreted the results. S.H. and R.C. wrote the manuscript with substantial

contributions from E.D., C.G., and C.M. All the authors discussed the data and edited the manuscript. CORRESPONDING AUTHOR Correspondence to Renkun Chen. ETHICS DECLARATIONS COMPETING

INTERESTS The authors declare no competing interests. ADDITIONAL INFORMATION PEER REVIEW INFORMATION _Nature Communications_ thanks Naoki Inomata and the other, anonymous, reviewer(s) for

their contribution to the peer review of this work. Peer reviewer reports are available. PUBLISHER’S NOTE Springer Nature remains neutral with regard to jurisdictional claims in published

maps and institutional affiliations. SUPPLEMENTARY INFORMATION PEER REVIEW FILE SUPPLEMENTARY INFORMATION SUPPLEMENTARY MOVIE 1 SUPPLEMENTARY MOVIE 2 SUPPLEMENTARY MOVIE 3 DESCRIPTION OF

ADDITIONAL SUPPLEMENTARY FILES REPORTING SUMMARY SOURCE DATA SOURCE DATA RIGHTS AND PERMISSIONS OPEN ACCESS This article is licensed under a Creative Commons Attribution 4.0 International

License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source,

provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons

license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by

statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit

http://creativecommons.org/licenses/by/4.0/. Reprints and permissions ABOUT THIS ARTICLE CITE THIS ARTICLE Hong, S., Dechaumphai, E., Green, C.R. _et al._ Sub-nanowatt microfluidic

single-cell calorimetry. _Nat Commun_ 11, 2982 (2020). https://doi.org/10.1038/s41467-020-16697-5 Download citation * Received: 06 October 2019 * Accepted: 18 May 2020 * Published: 12 June

2020 * DOI: https://doi.org/10.1038/s41467-020-16697-5 SHARE THIS ARTICLE Anyone you share the following link with will be able to read this content: Get shareable link Sorry, a shareable

link is not currently available for this article. Copy to clipboard Provided by the Springer Nature SharedIt content-sharing initiative