A physicochemically compatible ferrofluid droplet robotic system for automated bioanalytical assays

Abstract

Droplet robotics is an emerging area of research focused on harnessing externally programmable physical fields to drive liquid droplet motion and automate complex fluidic operations. One approach for driving droplet robotic systems utilizes magnetic attraction between droplets and magnetic actuators to enable programmable automated droplet manipulation through the introduction of magnetic components, such as nanoparticles, into the droplet. Compared to other droplet actuation mechanisms, magnetic actuation offers notable advantages including simple system design, high tolerance to liquid properties and flexible system control. However, the incorporation of magnetic ferrofluid nanoparticles introduces challenges related to their intrinsic physical colloidal stability and chemical catalytic characteristics, resulting in physicochemical incompatibility issues, restricting broader utilization in bioanalytical applications. In this work, the physicochemical incompatibilities of ferrofluid nanoparticles are investigated and resolved through surface modifications to the ferrofluid nanoparticles, enabling the development of a physicochemically compatible ferrofluid droplet robotic system. The system addresses compatibility issues including low colloidal stability and compromised chemical catalytic activity in HRP-based enzymatic assays. As a result, the enhanced actuation robustness and efficiency, as well as chemical quantification sensitivity and reliability, enable automated assays to be conducted. The enhanced physicochemical compatibility of the ferrofluid droplet robotic system facilitates the use of ferrofluid for highly efficient magnetically driven automated bioanalytical processes.

---

Introduction

The use of droplets as microrobots is a promising approach to enhance liquid processing efficiency in biotechnological applications. In a similar manner to mechanical robots, droplet robots respond to commands and driving forces to translocate substances localized within the droplet through dissolution or suspension.^1–3 Through highly precise droplet manipulation, complex chemical reactions and biological assays can be conducted with high efficiency and reliability using droplet robotics. The ability for parallel operation of multiple droplet robots improves the analysis throughput, the flexibility in adaptable fluidic paths enables on-demand customization and the integration of automated processing reduces manual labour and processing times. Droplet robotic systems are valuable tools for a wide range of applications in biochemical research,^4–6 chemical analysis,⁷ drug discovery,⁸ diagnostics,^9–11 and other healthcare-related fields.^12,13

The use of magnetic force as the actuation mechanism enables both manual mechanical control and powered automated system control,^2,14,15 and is especially suited for applications in low-resource settings. Magnetic droplet robotic systems utilizing magnetic particles enable controllable and contactless actuation, with few limitations on liquid properties.¹⁴ In particular, the use of magnetic nanomaterials to achieve magnetic droplet actuation provides microrobots with additional functionalities such as a solid surface for functionalization as well as a high surface area for enhanced interaction with the liquid component.^16–18 Compared to magnetic beads on the micrometre scale,^9,10,19–21 the use of magnetic nanoparticles such as ferrofluids in droplet robotic systems offers unique advantages, including higher magnetic susceptibility and actuation efficiency,²² higher mixing efficiency,^23,24 and greater actuation reliability with reduced risk of particle extraction from the droplet.²⁵ Ferrofluids, consisting of magnetic iron oxide nanoparticles (IONPs) in uniform aqueous colloidal suspensions, have been extensively used for a variety of fluid motion control and associated functions,²⁶ including channel pumps^27,28 and valves,²⁹ and enabling control over fluid viscosity,³⁰ deformation and wetting.^31–33 While the efficient and dynamic droplet control for ferrofluid actuation appears well-suited for carrying out complex bioanalytical processes in droplet robotic systems, the extensive utilization of ferrofluids in biochemical aqueous environments is hindered by two significant physicochemical compatibility issues (Fig. 1a): 1) the colloidal instability of ferrofluidic nanoparticles under varying conditions (e.g. pH, applied magnetic field); and 2) the intrinsic enzyme-like activity of ferrofluid nanoparticles. These compatibility issues in turn directly impact the performance of ferrofluid-based magnetic droplet robotic systems, compromising droplet actuation robustness and the sensitivity of biochemical analyses, respectively. To unlock the full potential of ferrofluids for effective bioanalysis in droplet robotic systems, the physicochemical incompatibilities must be addressed and overcome.

Fig. 1 A physicochemically compatible ferrofluid droplet robotic system with enhanced physical ferrofluid colloidal stability and suppressed chemical activity for robust and sensitive automated assays. (a) Compatibility issues including ferrofluid iron oxide nanoparticle (IONP) aggregation and interfering catalytic activity leading to unstable actuation and low sensitivity in a physicochemically incompatible ferrofluid droplet robotic system. (b) Enhanced colloidal stability and suppressed catalytic activity through ferrofluid IONP surface modification for robust actuation and high signal sensitivity. (c) Schematic diagram of the workflow of the ferrofluid droplet robotic system.

In this work, we systematically investigate the impact of the physical and chemical properties of ferrofluid nanoparticles on the performance of a ferrofluid droplet robotic system. We outline an approach to enhance physical colloidal stability and modify chemical reactivity to ensure robust actuation, high quantification sensitivity and reliability, respectively (Fig. 1b). Through enhancing the physicochemical compatibility of the ferrofluid nanoparticles specialized for different fluidic operations and HRP-based applications, we develop an automated ferrofluid-based droplet robotic analysis system (Fig. 1c). Magnetic actuators in the system respond to localized electromagnetic fields generated through computer commands, performing automated sample processing through manipulation of reagent and sample-loaded ferrofluid droplets, and finally generating a signal for quantitative analyses. The surface-modified ferrofluid IONPs ensure high colloidal stability and chemical inertness, and an automated enzymatic assay for glutathione detection and quantification was performed to validate the improved physicochemical compatibility of the system. Such a physicochemically compatible droplet robotic system facilitates the integration of magnetic ferrofluids into automated fluidic platforms in the future, offering an alternative tool for accessible automated analytical systems.

Results and discussion

Enhancing ferrofluid physical colloidal stability for robust droplet actuation

Maintaining robust and reliable actuation despite varying environmental conditions is of fundamental importance for any robotic system. The same applies to a droplet robot performing different fluidic operations, as changes in fluid properties are often required. For example, many biochemical reactions that assays and diagnostic systems are based on require changes in conditions to ensure optimal performance for each stage of a reaction.^34–36 For ferrofluid droplet robotics, the colloidal stability of the ferrofluid droplet, consisting of IONPs in suspension, is an important physical property to consider for robust performance.

For the ferrofluid droplet robotic system to perform diverse fluidic operations and assays, it is important to maintain the colloidal stability of the ferrofluid under different fluid properties, changing bioassay environments (e.g. pH),^37–41 and applied magnetic fields.^40,42,43 As IONPs are subjected to changing pH from the reaction reagents or an applied magnetic field for magnetic droplet actuation, the balance of forces between nanoparticles may change to favour attraction between nanoparticles overall, increasing their propensity to aggregate into large clusters on the micrometre scale (Fig. 2a). A direct consequence of low colloidal stability and IONP aggregation is the hindrance of robust droplet actuation. As IONPs aggregate, the maximum achievable speed reduces (Fig. 2b), as aggregates increase localized particle loading, which tends to cause droplet detachment at increased actuation speeds.²⁵ Similarly, under the same magnetic field and particle loading, the maximum actuatable droplet volume for droplets containing colloidal IONPs is higher compared to aggregated IONPs (Fig. S1). IONP aggregation also impedes droplet motion on superhydrophobic surfaces (Fig. 2c, and S2), as the larger IONP aggregates are more easily trapped on uneven terrains, detaching from the magnetic actuator and hence becomes unavailable as magnetic droplet carriers. For colloidal IONP, actuation remains robust for multiple cycles of motion. While using large particles to transport droplets can be beneficial for heterogeneous applications that involve extracting solid particles from a liquid environment, the need for stable and monodisperse particles remains crucial to ensure uniform magnetic properties.⁴⁴ Therefore, the maintenance of colloidal stability and robust magnetic actuation for droplets with different properties is a key challenge to be addressed to ensure high versatility and adaptability of the system for a wide range of bioanalytical applications.

Fig. 2 Optimization of IONP physical colloidal stability for enhanced droplet robotic actuation robustness. (a) Schematic diagram of the aggregation of IONP from changes in pH and an applied magnetic field (top) and corresponding microscopic images (bottom). (b) The maximum achievable speed and corresponding images depicting magnetic actuation of colloidal (left) and aggregated (right) IONP (error bars: 95% confidence interval, n = 8). (c) Robustness of magnetic actuation on a superhydrophobic surface for colloidal (left) and aggregated (right) IONP, at actuation speeds lower than their respective maximum achievable speeds. (d) Transmission electron microscope images showing increasing aggregate particle size with increasing IONP aggregation. (e) Aggregation kinetics of unfunctionalized IONP and IONPs with PEG and NH2 surface modifications in different pH and (f) in the presence (+) and absence (−) of an applied magnetic field (error bars: S.D., n = 3).

In this work, the colloidal stability of IONP is enhanced through different surface functional groups, including polyethylene glycol (PEG) and amine (NH₂). The size and morphology of IONPs are characterized by scanning electron microscopy (SEM) and dynamic light scattering (DLS), the mixed oxidation states of the IONPs are characterized by X-ray photoelectron spectroscopy (XPS), and the functional groups are characterized by FT-IR spectroscopy (Fig. S3 and S4). The change in aggregate size (Fig. 2d) and hence colloidal stability of IONP with different surface modifications under varying pH conditions is monitored over time (Fig. 2e, and S4–S6). In particular, IONPs with amine functionalization (IONP-NH₂) show high colloidal stability across different pH, with no significant increase in size over time. In addition, for pH 7 and 9, where all IONPs show relatively high stability, the change in size after a magnetic field (permanent magnet with a surface magnetic field strength of 3498 Gauss) is brought near the ferrofluid is further monitored (Fig. 2f and S7). Among the ferrofluids tested, only IONP-NH₂ did not exhibit a significant change in particle size after introduction of the magnetic field, indicating that the physical colloidal stability was maintained. Considering that long-term storage of the ferrofluid for the ferrofluid droplet robotic system may be necessary, the long-term colloidal stability of IONP-NH₂ was monitored over a period of 7 days (Fig. S8a). In addition, as various applications might require specific reagent conditions which may destabilize nanoparticles, such as varying temperatures, ionic strength, and particle concentration,^45,46 the colloidal stability of IONP-NH₂ nanoparticles is assessed under additional environmental conditions (Fig. S8b–g). Despite small variations in initial hydrodynamic size as a result of the different aqueous conditions,⁴⁷ in all cases, the long-term colloidal stability of IONP-NH₂ is preserved, with no significant increase in hydrodynamic size.

The Derjaguin–Landau–Verwey–Overbeek (DLVO) theory provides a framework for understanding the thermodynamic origins of the physical colloidal stability of IONPs, where the main contributors towards nanoparticle colloidal stability are steric and electrostatic interactions.^38–40 Thicker coatings and stronger particle charges generally lead to a more stable solution by increasing the steric and electrostatic repulsion between nanoparticles, respectively. In this instance, given that the IONPs used do not differ significantly in size and hydrodynamic radii (Fig. S4), and no surfactant is present in the ferrofluid which could stabilize the nanoparticles⁴⁸ (Fig. S9), electrostatic interactions are likely to be the main contributor to IONP-NH₂ stability. The positive surface charge of IONP-NH₂ provided by the amine group⁴⁹ could promote stronger electrostatic repulsion between nanoparticles. The additional charge stabilization in conjunction with the existing steric stabilization in aqueous solution further elevates the energetic barrier to nanoparticle aggregation,⁵⁰ counteracting the destabilizing effect of changing electrolyte concentration as pH changes and the presence of magnetic fields.⁵¹ The physical anti-aggregation properties of IONP-NH₂ indicate its compatibility for use on the magnetic droplet robotic platform.

Suppressing IONP chemical peroxidase activity for sensitive signal quantification

Magnetic droplet actuation offers significant advantages due to its versatility and adaptability for use in various liquid-based applications. To function effectively as magnetic carriers in different liquid environments for different applications without actively participating in reactions, IONPs must remain chemically inert. However, the peroxidase activity of IONPs⁵² poses a significant obstacle, limiting the use of IONPs as magnetic carriers in the magnetic droplet robotic system. The peroxidase activity of IONPs originates from the iron cations, which mimic the catalytic activity of enzymes with a porphyrin heme redox cofactor in the active site.^53,54 This particularly interferes with reactions based on horseradish peroxidase (HRP), a heme-containing enzyme,⁵⁵ which is popularly used as a convenient detection label in biochemical assays.^56,57 Typically, an analyte would be tagged specifically with HRP through a conjugate for reaction with an HRP substrate, generating optical signals proportional to the HRP concentration, and hence analyte concentration. However, when IONP is added to HRP-based assays, IONP also reacts with HRP substrates, producing a high background signal in addition to the analyte signal, leading to reduced assay sensitivity and inaccurate quantifications at low analyte concentrations (Fig. 3a). As the optimal reaction conditions for IONP nanozyme and peroxidase enzyme activity overlap substantially,^52,54 it is difficult to adjust the conditions to favour one reaction over another. In fact, IONPs have been reported to show activity levels up to 40 times higher than HRP,⁵² presenting a challenge in preserving HRP activity while simultaneously inhibiting IONP activity.

Fig. 3 Characterization of the effect of ferrofluid catalytic activity on HRP-based assay sensitivity and optimization of IONP for low limits of detection in ferrofluid droplet robotic systems. (a) Schematic diagram illustrating the effect of adding IONP on chemiluminescence (CL) signal. (b) Reaction kinetics and (c) the average reaction rate of IONPs with different surface modifications with luminol in the first 5 min (error bars: S.D., n = 8). (d) HRP calibration curves obtained after addition of different surface-modified IONPs, and the respective limits of detection (LOD) and slope factors (SF) (dashed lines correspond to 3 S.D. above background) (error bars: S.D., n = 3). (e) The LOD for each of the HRP calibration curves (error bars: 95% confidence interval, n = 3).

Surface modifications on the IONP surface could provide a solution to suppress the IONP peroxidase activity. As the catalytic activity of IONP is dependent on Fenton-like reactions on the nanoparticle surface,⁵³ surface modifications could alter the chemical properties of the IONP without significantly changing the physical magnetic properties.^58,59 As reaction conditions remain unchanged, HRP activity could be preserved. In Fig. 3b, the catalysis kinetics of the IONPs with different surface modifications with the HRP substrate luminol are compared against the background signal. Luminol reacts with HRP to produce a chemiluminescence (CL) signal measured at 425 nm, yielding more sensitive results compared to absorbance and fluorescence measurement in the ferrofluid system with the highest signal-to-noise ratio (Note S1, Fig. S10–S14). The rate of signal production for the first 5 minutes of the reaction, which is when chemiluminescence is optimum for detection and measurement, is extracted and assessed (Fig. 3c). IONP-NH₂ shows limited peroxidase activity, and the measured reaction rate does not show a statistically significant different to the control background signal with no IONP added. Contrastingly, the unfunctionalized IONP and IONP-PEG both show significantly escalated activity compared to the control, with an increase in reaction rate by factors of 109 and 29, respectively. The long-term inertness of IONPs is also considered, and the kinetic profiles are measured for up to 4 h (Fig. S15). The results further validate the chemical inertness of IONP-NH₂.

To confirm that the IONP-NH₂, with its low peroxidase activity compared to other IONPs, limits its chemical interference in HRP-based assays and provides the highest assay sensitivity, the different IONPs are added to an HRP assay, where the luminol substrate reacts with HRP to produce signals proportional to the HRP concentration. The slope factors (SF) obtained for each curve verify that IONP-NH₂ has the highest assay sensitivity (Fig. 3d). Furthermore, by comparing the limits of detection (LOD) for assays with each IONP added with the control in which no IONP is added, again only IONP-NH₂ does not show statistically significant difference with the control, indicating a higher sensitivity for low concentration analyte detection (Fig. 3e). The unfunctionalized IONP and IONP-PEG show a statistically significant increase in LOD from the control by factors of 11.6 and 2.94 respectively. The differences in activity for IONPs with different surface modifications could potentially be attributed to their different surface charges. Whereas PEG is a neutrally charged surface functionalization, NH₂ functionalization imparts a positive charge to the IONP. As the surface potential of nanoparticles correlates with their catalytic constants,⁶⁰ and as IONP-NH₂ has a much more positive zeta potential compared to the other IONPs in water, as well as in high pH for the optimum oxidation of luminol³⁵ (Fig. S5 and S16), this could contribute to its inhibited peroxidase activity. At pH 9, an inverse relationship between the measured zeta potentials of the IONPs tested and their respective reaction rates can be observed (Fig. S16). Chemical quenching originating from the NH₂ groups could also be a potential cause for low catalytic chemiluminescence reactions near the IONPs.⁶¹ In any case, the chemical inertness of the IONP-NH₂ ensures compatibility for HRP-based bioanalytical processes and justifies its use on the ferrofluid droplet robotic platform.

An automated bioanalytical assay on the ferrofluid droplet robotic system for antioxidant detection

Based on the optimizations made for both the physical and chemical properties of the ferrofluid IONP, summarized in Table S1, we demonstrate that the physicochemically compatible ferrofluid can be used in a droplet robotic system for automated HRP-based bioanalytical assays. Fig. 4a shows an overview of the ferrofluid droplet robotic system. Automation of the system is achieved using an electromagnetic navigation floor. This consists of an array of electromagnetic coils which can generate magnetic fields sequentially at individual coils. As these electromagnetic coils are turned on and off in succession, a permanent magnet actuator can be moved across the surface, enabling automated addressable actuation.^2,62 Accordingly, the ferrofluid droplet in the microfluidic chip is transported by magnetic attraction to the position of the permanent magnet actuator. For signal detection after performing all fluidic operations, the integrated ferrofluid droplet robotic device is designed such that the generated signals can be read using a standard microplate reader, ensuring ease-of-use and accessibility for most analytical laboratories. The programmable nature of droplet motion on the integrated ferrofluid droplet robotic device enables highly flexible system design, facilitating the automation of numerous laboratory fluidic analyses such as medical diagnostics and biochemical analyses.

Fig. 4 Demonstration of automated GSH quantification on the optimized physicochemically compatible ferrofluid droplet robotic system. (a) Working principle of the ferrofluid droplet robotic system. (b) The general workflow for a bioanalytical assay on the ferrofluid droplet robotic system. (c) The mechanistic principle for GSH detection. (d) The calibration curve for GSH obtained on the ferrofluid droplet robotic system. (e) Correlation between the GSH concentration result obtained on the ferrofluid droplet robotic system and in a reference microplate assay.

To fully demonstrate the physicochemical compatibility of the ferrofluid IONP, as well as the robustness of actuation and sensitivity of bioanalytical assays on the droplet robotic system (Table S1), a quantification assay for reduced glutathione (GSH) is performed. GSH is an antioxidant, and its levels indicate the level of oxidative stress in cells.^63,64 GSH is also linked to many pathological conditions and diseases, such as cancer,^65,66 viral infections including HIV,^67,68 and tuberculosis.^69,70 The general fluid manipulation workflow for conducting the GSH assay on the ferrofluid droplet robotic system is shown in Fig. 4b and Video S1. The principle of the assay is based on the inhibition of GSH on HRP activity^71,72 (Fig. 4c). HRP reacts with the substrate luminol to produce an oxidized product which emits light through chemiluminescence. GSH quenches the chemiluminescence signal produced by reducing the oxidized chemiluminescent product to luminol. As GSH concentration increases, there is an inhibition on the activity of HRP, and hence a lower signal is generated. A standard curve can be obtained by conducting the GSH assay on the ferrofluid droplet robotic system (Fig. 4d), which is designed for quantification at its concentration range in mammalian cells.⁷³ The validity and reliability of the assay could be demonstrated by comparing with the results obtained from a commercial GSH detection kit, based on a different reaction mechanism (Fig. 4e). Strong correlation is observed between the results from the ferrofluid droplet robotic system and the reference kit, validating the assay design, the performance of the physicochemically compatible ferrofluid droplet robotic system and the applicability of the ferrofluid droplet robotic system for automated bioassay operations.

The accurate results on the ferrofluid droplet robotic platform demonstrate that each of the physicochemical compatibility issues has been addressed and resolved, which would not have been possible without ferrofluid optimization. For the assay to generate accurate results on the ferrofluid droplet robotic platform, this work has achieved 1) robust ferrofluid actuation and maintenance of physical colloidal stability, 2) highly sensitive optical chemiluminescence detection and 3) reliable signal quantification through HRP catalysis. Overall, the development of the ferrofluid droplet robotic platform validates nanoparticle surface modification as a facile approach to address the main issues of ferrofluid-based magnetic droplet actuation previously inhibiting its use in aqueous analytical systems. The enhanced physicochemical compatibility of ferrofluids also provides a basis for the implementation of additional functionalities of the nanoparticles into ferrofluid-based droplet robotic platforms, such as signal amplification through the high surface area of the nanoparticles. Furthermore, this work introduces the use of physicochemically compatible ferrofluids as an alternative strategy for developing highly accessible automated fluid processing capabilities. As many applications require different optimal nanoparticle properties, such as a non-positive zeta potential to prevent electrostatic interactions in nucleic acids-based assays,^74,75 the approach outlined in this work could be further adapted for the selection of physicochemically compatible ferrofluid droplet carriers for other assays. Coupled with high robustness and accuracy of analytical processes, this approach is particularly relevant for various biotechnological applications aimed at improving liquid processing efficiency.

Conclusions

In this work, we first outline the physicochemical compatibility challenges of a ferrofluid-based magnetic droplet robotic platform, related to the physical colloidal stability and the chemical activity of the ferrofluid. Modulating these physicochemical properties are essential for the performance of ferrofluid-based droplet robotic systems. We introduce a surface modification-based strategy addressing each challenge for the droplet robotic platform design, demonstrating both high actuation robustness and bioanalytical sensitivity for HRP-based assays in the final integrated system. Finally, to simultaneously showcase both the high colloidal stability and chemical inertness of the physicochemically compatible ferrofluid system, an automated HRP assay quantifying GSH concentration was performed on the optimized ferrofluid droplet robotics system.

By overcoming the challenges with the use of ferrofluid in droplet robotic systems, this work encourages more extensive incorporation of ferrofluid into previously incompatible applications, and we foresee that ferrofluid-based actuation could become more widely adopted in various settings. Though the ferrofluid-based droplet robotic system demonstrates is currently limited to HRP-based enzymatic assays, the same strategy could be used to optimize a different compatible ferrofluid for other bioanalytical applications. The high adaptability of ferrofluid-based droplet actuation could pave the way for flexible and accessible solutions for automating fluidic operations, enabling additional functionalities exploiting the magnetic and high surface area properties of magnetic nanoparticles. In addition to enhancing the efficiency of small-scale specialized operations in laboratory settings, the developed ferrofluid droplet robotic system also enables robust and rapid processing of large amounts of samples in varying physicochemical environments, demonstrating great potential for large-scale applications involving biochemical testing and analyses. Its capabilities support high-throughput and high-content processes, including compound screening in drug discovery, formulation optimization in pharmaceutical development, medical diagnostics, and environmental monitoring. This automated ferrofluid droplet robotic system could substantially minimize manual labour while enhancing the efficiency for new scientific observations and discoveries.

Materials and methods

Materials

The functionalized IONPs used in this work were 30 nm in diameter and were obtained from OceanNanotech. The obtained IONPs were characterized by scanning electron microscopy (Hitachi S4800 FEG SEM), FT-IR spectroscopy (PerkinElmer Spectrum Two UATR), dynamic light scattering (Malvern Zetasizer Lab), and X-ray photoelectron spectroscopy (Thermo Scientific Escalab QXi XPS). The surface tension of the ferrofluid solution was measured using the pendant drop method, and the images were analysed and fitted using OpenDrop. No surfactant is present in the IONP solutions. The stock IONP solutions were stored at 4 °C before use. The stock solutions were diluted in distilled water to adjust particle concentration, in buffer solutions (Certipur, Sigma-Aldrich) to adjust pH, or in varying concentrations of salt solutions to adjust ionic strength. Reduced glutathione (GSH), glutathione fluorescent detection kit, horseradish peroxidase (HRP) and HRP substrates were obtained from Thermo Fisher Scientific. Coloured food dyes were used for modelling IONP absorbance at different wavelengths. Hydrofluoroether oil HFE-7500 (Novec Engineered Fluid, 3 M) was used as the oil phase in microfluidic chips where super-hydrophobic coating was applied, and HFE-7500 with Pico-surf 0.01% (w/w) was used where no superhydrophobic coating was applied.

Characterization of ferrofluid physicochemical properties

The hydrodynamic radii and zeta potential of IONPs were measured by dynamic light scattering (Malvern Zetasizer Lab). A permanent magnet (DH101, K&J Magnetics) was placed underneath the solution for 5 minutes before hydrodynamic radii measurement for assessing the effect of the applied magnetic field on IONP aggregation. The motion of droplets in microfluidic chips was recorded and analyzed using tracker. Absorbance, chemiluminescence and fluorescence were measured with a microplate reader (SpectraMax iD5). To model IONP solution absorbance, dyes were diluted with distilled water such that the final absorbance values of the solution matched those of the IONP solution at different concentrations. 1.5 μL HRP at 25 or 1.56 ng mL⁻¹, 13.5 μL dye solution for each equivalent IONP concentration and 135 μL of substrate were added into each well on a 96-well plate. The kinetics of IONP reaction with luminol was measured by mixing 15 μL 0.1 mg mL⁻¹ IONP with 135 μL luminol in 96-well plates, and measuring luminescence every minute for 20 minutes, or every 2 minutes for 4 hours.

HRP assays

To determine the effect of IONP on the HRP assay, 7.5 μL HRP and 7.5 μL 2 mg mL⁻¹ IONP solution or distilled water were mixed, and 135 μL luminol was added for each well. Standard curves were fitted with a four-parameter logistic regression model. The limits of detection for HRP-related assays were determined using the mean value of blanks +3 times its standard deviation.

Fabrication of the integrated ferrofluid droplet robotic system

The microfluidic chip in the integrated ferrofluid droplet robotic system was fabricated with a custom 3D printed structure, sandwiched by two polyethylene terephthalate (PET) sheets using double-sided tape (9474LE, 3 M). Patterns (holes and channels) on PET sheets and double-sided tape were fabricated by laser cutting (GCC LaserPro X500III). The channels were treated with an anti-wetting super-hydrophobic spray (NeverWet) before assembly. The super-hydrophobic coating was characterized by a scanning electron microscope (Hitachi S3400N VP SEM). The microfluidic chip was designed to fit inside a black 3D printed case matching the size of commercial 96-well plates, compatible with standard microplate readers. The 3D printed case has either a hollow base for absorbance measurement or an opaque base for luminescence and fluorescence measurements.

Automated GSH assay on the ferrofluid droplet robotic system

Automation on the ferrofluid droplet robotic system was enabled by a printed circuit board (PCB) based on previous work.^2,3 The PCB consists of an array of 32 × 32 addressable coils to generate a localized electromagnetic field. The PCB is connected to an external power supply, and the coils are controlled by four row switches (Maxim Integrated, MAX14662) and two column switches (NXP Semiconductors, MC33996), which is in turn controlled by an Arduino Uno receiving commands from a computer. Permanent magnetic actuators (DH101, K&J Magnetics) with surface magnetic field strength of 3498 Gauss, respond to the generated localized electromagnetic field to move across the PCB surface as specific coils are turned on and off. The GSH assay performed on the ferrofluid droplet robotic platform was based on the inhibition of GSH on HRP activity. The standard curve was obtained by mixing 2 μL solution containing 2 mg mL⁻¹ ferrofluid and 1.5 μg mL⁻¹ HRP with 2 μL GSH of each concentration, and finally mixing with 40 μL luminol substrate on the microfluidic chip. Chemiluminescence signal detection was performed with a microplate reader (SpectraMax iD5) after 5 minutes. Fresh solutions of GSH of varying concentrations were prepared, and results from the commercial GSH assay kit were compared to results obtained from the standard curve on the ferrofluid droplet robotic platform.

Author contributions

CCKAY and HL conceived the idea and designed the experiments; RZ conducted droplet actuation experiments with CCKAY. CCKAY, CZ and XF fabricated the microfluidic devices. CCKAY and CS characterized the nanoparticles. CCKAY, RZ, CZ, XF, YC and CS contributed to data analysis and interpretation. CCKAY drafted the manuscript, and all the authors provided feedback. HCS and HL supervised the study.

Conflicts of interest

The authors declare the following competing interests: Ho Cheung Shum holds shares in, or acts as advisor of MicroDiagnostics Limited, PharmaEase Tech Limited, Upgrade Biopolymers Limited, Monexus Innovation Limited, Multera Inc, EN Technology Limited, and Capsum, and he is the Founding Centre Director & Co-Director of Advanced Biomedical Instrumentation Centre Limited. He is also a member of the Board of Directors of Advanced Biomedical Instrumentation Centre Limited, Hong Kong Institution of Sciences Limited, and PharmaEase Tech Limited. The works in this paper are, however, not directly related to the works of these entities, as far as we know. The authors declare no other competing interests.

Data availability

The data that support the findings of this study are available in this article and the electronic supplementary information (SI). The SI files include additional supporting details and video demonstrations. Supplementary information is available. See DOI: https://doi.org/10.1039/d5lc00926j.

Acknowledgements

The authors acknowledge the National Natural Science Foundation of China (No. 32201181), as well as Collaborative Research Fund (C7165-20GF), Research Impact Fund (R4015-21) and General Research Fund (17307919, 17303123, 17208623) of the Research Grants Council of Hong Kong, Hong Kong. This study was supported by the Health@InnoHK initiative of the Innovation and Technology Commission of the Hong Kong Special Administrative Region Government. H. C. S. was funded in part by the RGC Senior Research Fellow (SRFS2425-7S04) by the RGC.

References

1. A. Li, H. Li, Z. Li, Z. Zhao, K. Li, M. Li and Y. Song, Sci. Adv., 2020, 6, eaay5808.
2. W. Yu, H. Lin, Y. Wang, X. He, N. Chen, K. Sun, D. Lo, B. Cheng, C. Yeung, J. Tan, D. Di Carlo and S. Emaminejad, Sci. Robot., 2020, 5, eaba4411.
3. R. Zhang, C. Zhang, X. Fan, C. C. K. Au Yeung, H. Li, H. Lin and H. C. Shum, Nat. Commun., 2024, 15, 6220.
4. Q. Ruan, W. Ruan, X. Lin, Y. Wang, F. Zou, L. Zhou, Z. Zhu and C. Yang, Sci. Adv., 2020, 6, eabd6454.
5. E. M. Miller, A. H. C. Ng, U. Uddayasankar and A. R. Wheeler, Anal. Bioanal. Chem., 2011, 399, 337–345.
6. I. Barbulovic-Nad, H. Yang, P. S. Park and A. R. Wheeler, Lab Chip, 2008, 8, 519–526.
7. F. Lapierre, G. Piret, H. Drobecq, O. Melnyk, Y. Coffinier, V. Thomy and R. Boukherroub, Lab Chip, 2011, 11, 1620–1628.
8. J. Zhai, Y. Liu, W. Ji, X. Huang, P. Wang, Y. Li, H. Li, A. H.-H. Wong, X. Zhou, P. Chen, L. Wang, N. Yang, C. Chen, H. Chen, P.-I. Mak, C.-X. Deng, R. Martins, M. Yang, T.-Y. Ho, S. Yi, H. Yao and Y. Jia, Nat. Commun., 2024, 15, 4363.
9. A. H. C. Ng, R. Fobel, C. Fobel, J. Lamanna, D. G. Rackus, A. Summers, C. Dixon, M. D. M. Dryden, C. Lam, M. Ho, N. S. Mufti, V. Lee, M. A. M. Asri, E. A. Sykes, M. D. Chamberlain, R. Joseph, M. Ope, H. M. Scobie, A. Knipes, P. A. Rota, N. Marano, P. M. Chege, M. Njuguna, R. Nzunza, N. Kisangau, J. Kiogora, M. Karuingi, J. W. Burton, P. Borus, E. Lam and A. R. Wheeler, Sci. Transl. Med., 2018, 10, eaar6076.
10. X. Hu, X. Gao, S. Chen, J. Guo and Y. Zhang, Microsyst. Nanoeng., 2023, 9, 1–12.
11. H. Norian, R. M. Field, I. Kymissis and K. L. Shepard, Lab Chip, 2014, 14, 4076–4084.
12. E. Samiei, M. Tabrizian and M. Hoorfar, Lab Chip, 2016, 16, 2376–2396.
13. D. J. Boles, J. L. Benton, G. J. Siew, M. H. Levy, P. K. Thwar, M. A. Sandahl, J. L. Rouse, L. C. Perkins, A. P. Sudarsan, R. Jalili, V. K. Pamula, V. Srinivasan, R. B. Fair, P. B. Griffin, A. E. Eckhardt and M. G. Pollack, Anal. Chem., 2011, 83, 8439–8447.
14. Y. Zhang and N.-T. Nguyen, Lab Chip, 2017, 17, 994–1008.
15. K. Choi, A. H. C. Ng, R. Fobel and A. R. Wheeler, Annu. Rev. Anal. Chem., 2012, 5, 413–440.
16. R. Hao, R. Xing, Z. Xu, Y. Hou, S. Gao and S. Sun, Adv. Mater., 2010, 22, 2729–2742.
17. K. Nejati, M. Dadashpour, T. Gharibi, H. Mellatyar and A. Akbarzadeh, J. Cluster Sci., 2022, 33, 1–16.
18. T. T. Hoang Thi, V. D. Cao, T. N. Q. Nguyen, D. T. Hoang, V. C. Ngo and D. H. Nguyen, Mater. Sci. Eng., C, 2019, 99, 631–656.
19. P. Kanitthamniyom, A. Zhou, S. Feng, A. Liu, S. Vasoo and Y. Zhang, Microsyst. Nanoeng., 2020, 6, 1–11.
20. P.-H. Lu, Y.-D. Ma, C.-Y. Fu and G.-B. Lee, Lab Chip, 2020, 20, 789–797.
21. J. Guo, L. Lin, K. Zhao, Y. Song, M. Huang, Z. Zhu, L. Zhou and C. Yang, Lab Chip, 2020, 20, 1577–1585.
22. R. Abedini-Nassab, M. Pouryosef Miandoab and M. Şaşmaz, Micromachines, 2021, 12, 768.
23. G.-P. Zhu and N.-T. Nguyen, Lab Chip, 2012, 12, 4772–4780.
24. M. A. Maleki, M. Soltani, N. Kashaninejad and N.-T. Nguyen, Phys. Fluids, 2019, 31, 032001.
25. Z. Long, A. M. Shetty, M. J. Solomon and R. G. Larson, Lab Chip, 2009, 9, 1567–1575.
26. N.-T. Nguyen, Microfluid. Nanofluid., 2012, 12, 1–16.
27. E. Kurtoğlu, A. Bilgin, M. Şeşen, B. Mısırlıoğlu, M. Yıldız, H. F. Y. Acar and A. Koşar, Microfluid. Nanofluid., 2012, 13, 683–694.
28. L. J. Love, J. F. Jansen, T. E. McKnight, Y. Roh and T. J. Phelps, IEEE Trans. NanoBiosci., 2004, 3, 101–110.
29. H. Hartshorne, C. J. Backhouse and W. E. Lee, Sens. Actuators, B, 2004, 99, 592–600.
30. W. Li, Z. Li, W. Han, Y. Li, S. Yan, Q. Zhao and F. Chen, Phys. Fluids, 2023, 35, 012002.
31. S. Manukyan and M. Schneider, Langmuir, 2016, 32, 5135–5140.
32. N.-T. Nguyen, G. Zhu, Y.-C. Chua, V.-N. Phan and S.-H. Tan, Langmuir, 2010, 26, 12553–12559.
33. N.-T. Nguyen, A. Beyzavi, K. M. Ng and X. Huang, Microfluid. Nanofluid., 2007, 3, 571–579.
34. A. Frey, B. Meckelein, D. Externest and M. A. Schmidt, J. Immunol. Methods, 2000, 233, 47–56.
35. P. Khan, D. Idrees, M. A. Moxley, J. A. Corbett, F. Ahmad, G. von Figura, W. S. Sly, A. Waheed and Md. I. Hassan, Appl. Biochem. Biotechnol., 2014, 173, 333–355.
36. E. Van Pelt-Verkuil, A. Van Belkum and J. P. Hays, Principles and Technical Aspects of PCR Amplification, Springer, Netherlands, Dordrecht, 2008.
37. D. Arndt, T. M. Gesing and M. Bäumer, ChemPlusChem, 2012, 77, 576–583.
38. M. Di Marco, I. Guilbert, M. Port, C. Robic, P. Couvreur and C. Dubernet, Int. J. Pharm., 2007, 331, 197–203.
39. F. Gambinossi, S. E. Mylon and J. K. Ferri, Adv. Colloid Interface Sci., 2015, 222, 332–349.
40. T. L. Moore, L. Rodriguez-Lorenzo, V. Hirsch, S. Balog, D. Urban, C. Jud, B. Rothen-Rutishauser, M. Lattuada and A. Petri-Fink, Chem. Soc. Rev., 2015, 44, 6287–6305.
41. M. Baalousha, Sci. Total Environ., 2009, 407, 2093–2101.
42. L. Gutiérrez, L. de la Cueva, M. Moros, E. Mazarío, S. de Bernardo, J. M. de la Fuente, M. P. Morales and G. Salas, Nanotechnology, 2019, 30, 112001.
43. E. G. Karvelas, N. K. Lampropoulos, L. T. Benos, T. Karakasidis and I. E. Sarris, Comput. Methods Programs Biomed., 2021, 198, 105778.
44. Q. Li, C. W. Kartikowati, S. Horie, T. Ogi, T. Iwaki and K. Okuyama, Sci. Rep., 2017, 7, 9894.
45. H. T. Phan and A. J. Haes, J. Phys. Chem. C, 2019, 123, 16495–16507.
46. F. Gambinossi, S. E. Mylon and J. K. Ferri, Adv. Colloid Interface Sci., 2015, 222, 332–349.
47. S. Bera, J. Padhee, J. Pal and D. Dhara, ACS Appl. Polym. Mater., 2025, 7, 3368–3377.
48. L. Kvítek, A. Panáček, J. Soukupová, M. Kolář, R. Večeřová, R. Prucek, M. Holecová and R. Zbořil, J. Phys. Chem. C, 2008, 112, 5825–5834.
49. M. P. Calatayud, B. Sanz, V. Raffa, C. Riggio, M. R. Ibarra and G. F. Goya, Biomaterials, 2014, 35, 6389–6399.
50. S. Lin, Y. Cheng, J. Liu and M. R. Wiesner, Langmuir, 2012, 28, 4178–4186.
51. P. C. Hiemenz and R. Rajagopalan, Principles of colloid and surface chemistry, CRC, Taylor & Francis, Boca Raton, Fla., 3rd edn, ev.expanded., 1997.
52. L. Gao, J. Zhuang, L. Nie, J. Zhang, Y. Zhang, N. Gu, T. Wang, J. Feng, D. Yang, S. Perrett and X. Yan, Nat. Nanotechnol., 2007, 2, 577–583.
53. H. Dong, W. Du, J. Dong, R. Che, F. Kong, W. Cheng, M. Ma, N. Gu and Y. Zhang, Nat. Commun., 2022, 13, 5365.
54. L. Gao, K. Fan and X. Yan, Theranostics, 2017, 7, 3207–3227.
55. J. N. Rodríguez-López, D. J. Lowe, J. Hernández-Ruiz, A. N. P. Hiner, F. García-Cánovas and R. N. F. Thorneley, J. Am. Chem. Soc., 2001, 123, 11838–11847.
56. X. Zhang, Q. Yang, Y. Lang, X. Jiang and P. Wu, Anal. Chem., 2020, 92, 12400–12406.
57. A. Thiha and F. Ibrahim, Sensors, 2015, 15, 11431–11441.
58. B. Bittova, J. Poltierova-Vejpravova, A. G. Roca, M. P. Morales and V. Tyrpekl, J. Phys.:Conf. Ser., 2010, 200, 072012.
59. A. Nasser, A. Qdemat, H. Unterweger, R. Tietze, X. Sun, J. Landers, J. Kopp, B. Wu, M.-S. Appavou, A. Murmiliuk, E. P. Gilbert, O. Petracic and A. Feoktystov, Phys. Chem. Chem. Phys., 2024, 26, 24912–24923.
60. F. Yu, Y. Huang, A. J. Cole and V. C. Yang, Biomaterials, 2009, 30, 4716–4722.
61. Z.-F. Zhang, H. Cui, C.-Z. Lai and L.-J. Liu, Anal. Chem., 2005, 77, 3324–3329.
62. H. Lin, W. Yu, K. A. Sabet, M. Bogumil, Y. Zhao, J. Hambalek, S. Lin, S. Chandrasekaran, O. Garner, D. Di Carlo and S. Emaminejad, Nature, 2022, 611, 570–577.
63. D. P. Jones, Am. J. Physiol., 2008, 295, C849–C868.
64. C. Mytilineou, B. C. Kramer and J. A. Yabut, Parkinsonism Relat. Disord., 2002, 8, 385–387.
65. J. M. Estrela, A. Ortega and E. Obrador, Crit. Rev. Clin. Lab. Sci., 2006, 43, 143–181.
66. F. Kong, Z. Liang, D. Luan, X. Liu, K. Xu and B. Tang, Anal. Chem., 2016, 88, 6450–6456.
67. M. R. Ciriolo, A. T. Palamara, S. Incerpi, E. Lafavia, M. C. Buè, P. D. Vito, E. Garaci and G. Rotilio, J. Biol. Chem., 1997, 272, 2700–2708.
68. R. Buhl, K. J. Holroyd, A. Mastrangeli, A. M. Cantin, H. A. Jaffe, F. B. Wells, C. Saltini and R. G. Crystal, Lancet, 1989, 334, 1294–1298.
69. L. Kennedy, J. K. Sandhu, M.-E. Harper and M. Cuperlovic-Culf, Biomolecules, 2020, 10, 1429.
70. V. Venketaraman, A. Millman, M. Salman, S. Swaminathan, M. Goetz, A. Lardizabal, David Hom and N. D. Connell, Microb. Pathog., 2008, 44, 255–261.
71. C. Plieth, ACS Omega, 2018, 3, 12295–12303.
72. J. Głowacka, K. Strzelak and R. Koncki, Molecules, 2021, 26, 5630.
73. J. Vašková, L. Kočan, L. Vaško and P. Perjési, Molecules, 2023, 28, 1447.
74. B. B. Mendes, J. Conniot, A. Avital, D. Yao, X. Jiang, X. Zhou, N. Sharf-Pauker, Y. Xiao, O. Adir, H. Liang, J. Shi, A. Schroeder and J. Conde, Nat. Rev. Methods Primers, 2022, 2, 24.
75. K. Faikhruea, I. Choopara, N. Somboonna, W. Assavalapsakul, B. H. Kim and T. Vilaivan, ACS Appl. Bio Mater., 2022, 5, 789–800.

---