Droplet-based microfluidic systems are a widely used platform for biochemical analysis due to their low sample consumption and potential for high-throughput screening1,2,3. During the procedures of biochemical analysis, one solid exchange medium has to exchange substances with multiple liquid phases, including complex samples, washing buffers, and detection solutions, for the extraction, purification and detection of biomarkers to produce sensitive and reliable signals4,5,6. Multiple external mechanical control units are required to transport, mix, and separate the solid exchange mediums as well as liquid phases7,8. For example, each liquid phase relies on a separate syringe pump for its propulsion, and mechanical mixers are further used to create a vortex for the uniform dispersion of the solid phases into the liquid phases to undergo an efficient reaction9. However, these essential mechanical units along with complex external circuits hinder the simplification of microfluidic systems, burdening their applications in developing portable devices for point-of-care testing10.

Recently, researchers have strived to develop in situ microscale control units (pumps, valves, and mixers) by using smart materials, especially deformable ones, which simplify complicated microfluidic systems and advance the development of portable devices11,12,13,14,15,16,17. Smart materials directly convert external stimuli into mechanical energy to propel liquid, thus reducing the number of mechanical units. In this case, smart materials serve as both fluid channels and microfluidic control units, thus simplifying the complexity of the microfluidics16,18. Photodeformable liquid crystal polymers are promising candidates for smart materials to manipulate solid/liquid mixed phases via fast, tunable, and multitudinous anisotropic deformation by controlling the variation in mesogen alignment19,20,21,22,23,24,25,26,27,28. In previous publications, we proposed a conceptually novel strategy to propel diverse liquid phases by asymmetric photodeformation of newly designed linear liquid crystal polymers (LLCPs)29,30,31,32. LLCPs with excellent processability were easily processed into 3-dimensional tubular actuators and microfluidic chips with photodeformable channels29,30,31,32. The asymmetric photodeformation of the LLCP channels induced Laplace pressure that propelled a train of discrete droplets with controllable speed and direction. Moreover, Laplace pressure simultaneously generated a vortex inside the moving droplets for dispersion and transportation of the solid particles29, therefore providing a platform for solid exchange mediums to exchange substances with multiple liquid phases.

Herein, we present an unprecedented combined photo/magnetic control strategy for contactless manipulation of solid/liquid phases in LLCP microtubes assisted with light-driven vortices. The key component is a photo/magnet-controlled trilayer microtube (PTM) composed of a flexible supporting layer, middle photodeformable layer, and inner hydrophilic blocking layer (Fig. 1a, b). The light stimulus and magnetic field facilitate the suspending, depositing, and separating of liquid/solid mixed phases (Fig. 1c), which capitalize on a vortex generated by a nonuniform distribution of the flow velocities in the PTM. The first prototype of a portable protein analyzer (PPA) was constructed based on this strategy to specifically purify, enrich, and analyze various target proteins in complex samples, exhibiting the advantages of a short detection time (20 min), trace sample consumption (5 μL), and low detection limit (1 μg mL−1).

Fig. 1: Design of the combined photo/magnetic control strategy.
figure 1

a Schematic representation of the trilayer photo/magnet-controlled trilayer microtube (PTM) consisting of a flexible supporting layer of the ethylene-vinyl acetate (EVA) copolymer, a photodeformable layer of the linear liquid crystal polymer (LLCP), and a hydrophilic blocking layer of bovine serum albumin (BSA). b The chemical structure of the LLCP. Mn is the number-average molecular weight of the LLCP. c Schematics showing the detailed operation processes, including suspending, depositing, and separating the liquid/solid mixed phases in the PTM by combined photo/magnetic control. The black particles represent solid exchange mediums.

Materials and methods


Ethylene-vinyl acetate (EVA) copolymer microtubes (FSFR-125H, radius: 0.6 mm, thickness: 0.1 mm, inner-surface coating of EVA hot melt glue) were purchased from Wolida Ltd., Shenzhen, China. The magnetic nanoparticles (MNPs) (ϕ: 100–200 nm, Fe3O4 core, Affimag PEO-NH2) were purchased from BaseLine® Ltd., Tianjin, China. Other substances purchased for the study include D-( + )-biotin (>98%) and bovine serum albumin (BSA) (Adamas, Beijing, China), egg-white avidin (Shanghai Macklin Biochemical, Shanghai, China), streptavidin (Duly Biotech Ltd., Nanjing, China), biotin-4-fluorecence (Keshun Bio-Technique Ltd., Shanghai, China), phosphate buffer saline (PBS) tablets (Solarbio Science & Technology Ltd., Beijing, China), and C-reactive protein (CRP) (1°Ab-CRP and 2°Ab-FITC; Anyan Biology Ltd., Shanghai, China). Enzyme linked immunosorbent assay (ELISA) kits for chicken avidin and rabbit CRP were purchased from Ruixinbio, Quanzhou, China.


The 1H NMR spectra of the monomers and the LLCPs were recorded using a DMX-500 NMR spectrometer (Bruker, Billerica, MA, USA) using tetramethylsilane as the internal standard and CDCl3 as the solvent. The molecular weights of the polymers and their polydispersity indices were measured via gel permeation chromatography (LC-10ADvp, Shimadzu, Kyoto, Japan) using tetrahydrofuran as the mobile phase. The thermodynamic properties of the obtained LLCPs were determined using differential scanning calorimetry (Q2000, TA Instruments, New Castle, DE, USA) at a heating/cooling rate of 10 °C min−1 over a temperature range of −10 to 120 °C. A Dataphysics OCA20 system was used to measure the CAs of 3 μL PBS droplets at ambient temperature. To obtain a reliable value, the average CAs were obtained by measuring the same sample in at least five different positions. Adhesion force measurements were performed on a Dataphysics DCAT11. The step rate of the sample plate was set as 0.01 mm s−1, and 3 μL PBS droplets were used. The average adhesion force values were obtained by measuring the same sample in at least five different positions. Photographs of photocontrolled liquid motion were recorded by a camera (Canon 70D). Blue light (λ = 470 nm) was provided by a laser pointer (Chen’en Ltd., Shenzhen, China). An automatic microplate reader (Flex A-200, Allsheng Instruments Ltd., Hangzhou, China) was used to detect the signal of the ELISA kit. A mini-fluorometer (Fluo-100B, λEx 460 ± 20 nm, λEm 525–570 nm, Allsheng Instruments Ltd., Hangzhou, China) was used to detect the fluorescence signal.

Detection of the nonspecific adsorption rate

Droplets of 5 μL avidin solution with a concentration of 0.5 mg mL−1 were loaded and incubated in the PTMs before and after BSA modification. The droplets were then exported at intervals to detect the decrease in avidin protein content, which was nonspecifically adsorbed by the PTM inner wall. The avidin content in the supernatant was determined by standard ELISA.

Results and discussion

To manipulate liquid/solid mixed phases in the PTM, we employed a combined photo/magnetic control strategy, used MNPs as a solid exchange medium and used complex sample droplets to demonstrate the detailed processes of suspension, deposition, and separation (Fig. 1c and Supplementary Video 1). When the droplet propelled by light reaches the preloaded band of the MNPs in the PTM, the vortex generated during droplet motion is suspended and disperses the MNPs homogeneously into the droplet. The MNPs deposit rapidly from the droplet with the aid of a magnetic field. The supernatant droplet is removed and separated from the MNPs fixed by the magnetic field.

To use the combined photo/magnetic control strategy, we fabricated a PTM aimed at manipulating solid/liquid mixed phases for protein analysis. The PTM is designed to be (i) flexible; (ii) demonstrate asymmetric deformation; (iii) hydrophilic; and (iv) have ultralow nonspecific adsorption. To attain flexibility, the commercially available EVA copolymer microtube (with outer and inner diameters of 1000 μm and 800 μm, respectively) is chosen as the supporting layer for its good flexibility and comparable modulus. Our designed LLCP with azobenzene and biphenyl side chains in a 1:2 ratio (see Supplementary Fig. 1, 2 and 3 for synthetic routes) is coated on the inner wall of the EVA microtube to achieve asymmetric deformation. As shown in the cross-section photographs of the PTM, the exposed surface of the PTM is displaced by 36.9 µm toward the 470 nm light source (80 mW cm−2) due to the deformation of the LLCP layer, which is attributed to the reorientation of mesogens induced by the Weigert effect (Fig. 2a, b)33,34,35,36. Such reversible deformation upon intermittent irradiation with 470 nm light can be repeated over 100 cycles without obvious fatigue (Supplementary Fig. 4) because both the high molecular weight and lamellar structure of the LLCP provide sufficient chain entanglement and physical crosslinking29. Upon localized irradiation, the regional geometric change of the LLCP/EVA microtube from a cylinder to a conical structure generates Laplace pressure to propel the liquid inside30. The PTM is hydrophilic and has ultralow nonspecific adsorption due to modification of the LLCP layer with BSA, a commonly used blocking protein with excellent hydrophilicity in the field of biochemical analysis36. Since the reagents used in protein analysis are all aqueous solutions, the BSA layer enhances the hydrophilicity to increase the driving force for propelling the aqueous droplets by Laplace pressure. In addition, BSA plays an important role as a blocking layer to decrease the nonspecific adsorption rate, thus reducing the detection error during protein analysis.

Fig. 2: Characterization of the PTM.
figure 2

a Schematic diagram showing the mechanism of LLCP layer deformation upon irradiation with 470 nm light. b Photographs showing local radial expansion of the PTM cross section before and after irradiation with 470 nm light (80 mW cm−2). When the light is switched on, the PTM generates local expansion radially. As a consequence, the exposed surface of the PTM displaces upward 36.9 μm. After continuous illumination for 0.5 s, the displacement of the exposed surface reaches the maximum value. When the light is switched off, the displacement recovers within 0.5 s. The rapid recovery of the PTM is mainly due to the entropic restoring force of exposed regions and the elastic recovery of unexposed regions. Scale bar: 100 μm. c Photographs showing the photocontrolled motion of a protein droplet in the PTM. The intensity of the 470 nm light is 80 mW cm−2. Scale bar: 2 mm. d Velocities of various reagent solution droplets, including phosphate buffered saline (PBS), protein solution, fluorescence solution, and magnetic nanoparticle (MNP) suspension, used during protein analysis in the PTM. e Schematic representation of the force analysis during the photocontrolled motion of droplets in the PTM, where Fd is the driving force, Fr is the resistance force, Fa is the adhesion force, and CA is the contact angle. f CAs of a 3 μL PBS droplet on the inner wall of the PTM before (top) and after (bottom) BSA modification. g Adhesion force curves of a 3 μL PBS droplet on the inner wall of the PTM before and after BSA modification. The maximum force decreases from 0.339 ± 0.031 mN (black) to 0.259 ± 0.043 mN (blue).

The detailed fabrication processes of the PTM are described in Supplementary Fig. 5. The PTM exhibits the ability to manipulate aqueous droplets of protein analysis reagents by light. As an example, a 5 μL droplet of avidin protein solution retreated from the irradiation site when the PTM was exposed to localized irradiation of 470 nm light (80 mW cm−2) and the droplets direction of movement was reversed by altering the irradiation to the other side (Fig. 2c). Moreover, other kinds of protein analysis reagents, including PBS, MNP suspension, and fluorescence solution, were all manipulated by 470 nm light (80 mW cm−2) in the PTM, traveling at speeds ranging from 0.20 to 0.25 mm s−1 (Fig. 2d). The speeds of droplet motion increase gradually with increasing light intensity, and the maximum speed of various protein analysis reagents reaches 0.4 mm s−1 (Supplementary Fig. 6).

Figure 2e shows the mechanism of photocontrolled droplet motion in the PTM. The diameter of the exposed wall increases due to the photodeformation of the LLCP layer, while the unexposed part remains unchanged. The Laplace pressure caused by this asymmetric shape change propels the wetting droplets with controllable speed and direction. It should be noted that the photocontrolled motion of aqueous droplets is attributed to the increase in driving force (Fd) and the decrease in resistance (Fr) after BSA modification. The Fd of the Laplace pressure correlates positively with the wettability of the droplets on the inner wall, and the Fr is related to the adhesion force (Fa) between the droplet and the inner wall. Contact angle (CA) measurements indicate that the CA of a 3 μL PBS droplet on the PTM inner wall decreases from 90.7° ± 2.5° to 37.5° ± 3.7° (Fig. 2f); thus, Fd increases with the enhanced hydrophilicity of the inner layer after BSA modification. The Fa evaluation curve shows that the maximum Fa of the inner layer exerted on 3 μL PBS droplets decreases from 0.339 ± 0.031 mN to 0.259 ± 0.043 mN (Fig. 2g), suggesting the reduction of Fr. According to scanning electron microscope images (Supplementary Fig. 7), BSA on the inner wall forms dendritic crystals, whose microstructures reduce the contact area and, consequently, the Fa between the droplets and the PTM37. Therefore, aqueous droplets are propelled in the PTM by combining the asymmetric photodeformation of the LLCP layer and the enhanced hydrophilicity of the BSA layer.

According to the combined photo/magnetic control strategy, a homemade prototype of the PPA was constructed by the PTM, adapter, and mini-fluorometer (Supplementary Fig. 8). MNPs have gained great popularity in microfluidic systems because of their capability to specifically capture various proteins by changing the surface modifiers. Hence, we employed biotin-modified MNPs as a solid exchange medium to demonstrate the extraction and detection of target avidin protein from a complex sample in the PPA by the fluorescence attenuation method (Fig. 3a). The preparation of the biotin-modified MNPs is described in Supplementary Fig. 9, 10 and 1138. As shown in Fig. 3b, a 5 μL complex sample is loaded to suspend the biotin-modified MNPs homogeneously, which capture the target avidin protein and then are separated from the supernatant droplet with the aid of combined photo/magnetic control (i. extraction step). In sequence, a 5 μL droplet of biotin-4-fluorescence detection solution, a reported reagent for avidin detection39,40, is loaded to suspend the biotin-modified MNPs again. The fluorescent molecules are specifically bonded to the target avidin protein captured on the biotin-modified MNPs. After the deposition and separation processes, the supernatant droplet is exported to detect the attenuation degree of the fluorescence signal (ii. detection step). To detect the fluorescence signal of the supernatant droplet, the PTM is connected to an adapter through an L-shaped quartz tube, where the supernatant will be propelled to the detection window from the PTM (Fig. 3c). The adapter is inserted into a mini-fluorometer to measure the fluorescence signal (λEx is 460 nm and λEm is 525 nm). Then, a smartphone is used to convert the fluorescence signal into the concentration information of the target avidin protein in the samples.

Fig. 3: Operations and construction of the portable protein analyzer (PPA).
figure 3

a Schematics showing the mechanism of protein extraction and fluorescence attenuation detection during protein analysis. b Photographs of manipulating solid/liquid mixed phases in the PTM by combined photo/magnetic control, including the processes of loading, suspending, depositing, and separating. (i) Target avidin protein is specifically captured on the biotin-MNPs and extracted from the complex sample. Scale bar: 1 mm. (ii) The biotin-4-fluorescence molecules are bonded to the avidin captured on the biotin-MNPs, which are separated from the supernatant droplet. Scale bar: 1 mm. c Schematic representation showing the fluorescence signal measurement of the supernatant droplet, which is exported into the adapter through a quartz tube. λEx is the excitation wavelength and λEm is the emission wavelength. The detected fluorescence signal is recorded by a mini-fluorometer and transferred to a smartphone.

The combined photo/magnetic control strategy provides reliable and time-saving detection results of avidin concentration, the standard curve of which possesses a fitting constant (R2) up to 0.99998 (Supplementary Fig. 12). The reliable detection results originate from the BSA layer, whose nonspecific adsorption rate is as low as 1.6% (Supplementary Fig. 13)41. The capture efficiency of avidin by our strategy is much higher than that observed with conventional microfluidics. As shown in Fig. 4a, 100% avidin is rapidly captured from the complex sample in only 12 min by using the vortex flow, while the time for the same capture rate in the laminar flow is prolonged to 30 min. The high capture rate in the PTM is ascribed to the vortex, which is simulated by the finite element analysis method (Fig. 4b). As shown in the velocity field of the moving droplet, the flow velocity in the center of the droplet is much higher than that at the edge because the edge flow is hindered by the PTM viscous resistance. A nonuniform distribution of the flow velocities leads to the fluid in the center tending to flow to the edge of the droplet, thus forming vortex flow trajectories42. The resultant vortex suspended the biotin-MNPs homogeneously and increased the collision probability between the biotin-MNPs and avidin, accelerating the capture procedure.

Fig. 4: Finite element analysis in the droplet of the PPA.
figure 4

a A plot showing the capture rate of the target avidin protein as a function of the reaction time in the vortex flow and the laminar flow. b Finite element analysis of the velocity field (top) and flow trajectories (bottom) to show the vortex generated in the droplet.

The droplets in conventional microfluidics driven by syringe pumps exhibited laminar flow, where the solid exchange mediums were difficult to disperse uniformly due to small channel dimensions and uniform flow rates43. To generate a vortex, sophisticated channels, including herringbone, zigzag, and serpentine, have been designed to increase channel dimensions and disturb flow rates in conventional microfluidics44,45,46,47,48. However, geometric change of the PTM from column to cone spontaneously gives rise to the vortex, resulting from the orientation change of the mesogens upon exposure to 470 nm light. All the operations of liquid/solid mixed phases, including suspending, depositing, and separating, are easily manipulated at the milligram-level with the 3 cm-long PTM by a commercial laser pointer and a magnet.

The extraction and detection of the target proteins are completed in the PPA in the absence of external mechanical units of the conventional microfluidics49,50,51,52,53. We comprehensively investigated the instrument parameters of our homemade PPA and a commercial microplate reader with ELISA kits (see Supporting Information for the detailed steps of protein analysis), which are summarized in Table 1. Egg white and cream were chosen as complex samples to analyze the content of avidin by using both the PPA and microplate reader. Compared to the commercial microplate reader, the PPA features comparable detection accuracy and shortens the analysis time from 90 to 20 min. The avidin antibodies fixed on the bottom of the ELISA kit captured the avidin by free diffusion (slow), while the biotin-MNPs were suspended by vortexing in the PPA to accelerate the reaction with avidin (fast). The sample consumption of the PPA is only 5 μL, which is 10 times smaller than that of the microplate reader (50 μL), since the PPA is a droplet-based microfluidic system, where all the processes of protein analysis are performed inside microscale droplets. Moreover, the weight of the PPA is 400 g, approximately 1/40 of that of the microplate reader.

Table 1 Comparison of the parameters between the commercial microplate reader and the PPA.

In addition to protein extraction, the homemade PPA has the ability to enrich protein samples with low concentrations, which enables dilute samples beneath the detection limit to be analyzed. For example, CRP, a biomarker of inflammation and cardiovascular disease, has a very low concentration of only 0.8–8 μg mL−1 in blood49. We demonstrated a decline in the CRP detection limit after enrichment in PPA by using 1°Ab-MNPs as the solid exchange medium and 2°Ab-FITC as the detection solution. The 1°Ab-MNPs captured the CRP from different amounts of 5 μL dilute samples loaded into the PPA sequentially. In this scenario, CRP began to gradually accumulate on the 1°Ab-MNPs after enrichment. The extraction step was repeated several times, and these captured CRP were further specifically bonded to the 2°Ab-FITC in the 5 μL detection solution, whose supernatant droplet was exported to detect the attenuation degree of the fluorescence signal (Supplementary Fig. 14). As a result, the detection limit decreased from 9 to 1 μg mL−1 as we increase the number of extraction steps to concentrate the dilute CRP samples from 45 to 5 μL (Fig. 5a). The standard curve with R2 up to 0.99983 after 9 enrichment steps spans 1–9 μg mL−1, covering the concentration range of CRP in clinical samples (Fig. 5b). Therefore, our homemade PPA provides a versatile platform for fast and accurate protein analysis of various trace complex samples.

Fig. 5: Protein enrichment by PPA.
figure 5

a A plot showing the minimum C-reactive protein (CRP) detection limit as a function of enrichment time. A 45 μL complex sample was concentrated into a 5 μL droplet after enrichment 9 times. b A plot showing the standard detection curve of CRP after enrichment 9 times. The relative fluorescence intensity (y) is inversely proportional to the concentration of CRP (x), and the linear fitting constant (R2) reaches 0.99983.


In summary, we presented a strategy for contactless manipulation of solid/liquid phases and constructed a PPA prototype with PTMs, detection adapters, and mini-fluorometers for the first time. The trilayer PTM was composed of a flexible supporting EVA layer, middle photodeformable LLCP layer, and inner hydrophilic blocking BSA layer. The LLCP was of significance to generate asymmetric deformation of the PTM, generating Laplace pressure and, importantly, the vortex to propel the droplets and suspend the solid exchange mediums. BSA enhanced the hydrophilicity that increased the driving force to propel the aqueous droplets and played an important role as a blocking layer to decrease the nonspecific adsorption rate and reduce the detection error. The photodeformation and magnetic field were used to manipulate liquid/solid mixed phases for extraction, enrichment, and detection of various proteins, thus replacing complex external mechanical units. It should be emphasized that the light-driven vortex generated during droplet motion dispersed the MNPs into aqueous solutions to effectively capture proteins. Compared to a conventional microplate reader, the homemade PPA exhibited the advantages of high precision (1 μg mL−1), low consumption (5 μL; 90% reduction), short detection time (within 20 min; 70% reduction), and light weight (only 400 g). Moreover, all the operation processes in the PPA were remotely controlled by light and a magnetic field, avoiding the risk of cross contamination caused by contact manipulation. This strategy opens new opportunities for the integration of photodeformable LCPs into droplet-microfluidic systems, which can facilitate the development of powerful platforms for biomedical diagnosis, food safety applications, and point-of-care testing.