Compact analytical flow system for the simultaneous determination of L-lactic and L-malic in red wines

The malolactic fermentation (MLF) in winemaking converts L-malic acid primarily into L-lactic acid, shaping aroma, taste, color stability, and microbial control, especially in red wines. Continuous control over the 20–40 day MLF process is crucial for wine quality, yet current standard methods (chromatography, colorimetry, NADH absorbance assays) are centralized, require bulky equipment, skilled personnel, and involve sample collection, stabilization, transport, and storage—hindering on-time, in-situ corrective actions. Miniaturized analytical methods enable multiplexed analysis in low-cost, fast-response portable devices with low reagent volumes. Polymers, particularly PMMA, are attractive for compact and robust flow-systems fabricated by rapid prototyping. Electrochemical biosensors are well-suited for integration and on-site analysis. The authors previously developed individual amperometric biosensors for L-lactate and L-malate in batch, using thin-film electrochemical transducers modified with polypyrrole (PPy) matrices entrapping specific enzyme pairs and mediators, demonstrating long-term stability (>37 days). Building on this, the present work designs and demonstrates, for the first time, a compact bi-parametric flow-system integrating both biosensors on a single silicon chip within a PMMA flow cell, enabling simultaneous, real-time, remote determination of L-lactic and L-malic acids in red wine samples during MLF, and validating performance against standard methods.

Background methods for L-lactic and L-malic determination include chromatographic and colorimetric laboratory assays, and enzymatic NADH absorbance methods (OIV protocols), which are not deployable on-site. Polymer-based microfluidics (notably PMMA) enable low-cost, robust, and easily machined flow systems. Electrochemical biosensors are widely applied in food control and are suitable for portable flow integration. The authors previously reported long-term stable, PPy-based bienzymatic amperometric biosensors for L-lactate (LOX/HRP with ferrocyanide mediator) and L-malate (MDH/DP with HAR mediator and NAD+ cofactor) in batch. Comparative literature on bi-parametric systems for simultaneous L-lactate/L-malic in wines shows prior approaches using: (i) FIA with separate enzyme reactors and fluorimetric NADH detection; (ii) FIA with enzyme reactors and Clark electrode membranes; (iii) FIA with dialysis membranes and amperometric O2 detection; (iv) FIA with nylon multienzymatic membranes and Clark electrode; and (v) batch graphite composite transducers with ferricyanide chronoamperometry. These systems typically require separate reactors/membranes, shared reagents in solution, and FIA liquid handling, with LODs generally higher than the present work, especially for L-malate. The present system uniquely integrates both biosensors on-chip by electrogenerating PPy membranes entrapping enzymes and mediator, reducing reagent/sample consumption and improving LODs while maintaining analysis times of 3–6 minutes per sample (here ~5 minutes).

Design and fabrication: A silicon chip (11 × 9 mm²) with four parallel Pt electrodes (fabricated by standard photolithography) comprises a 2 × 2.5 mm² counter (CE), a 1 × 2.5 mm² pseudo-reference (p-RE), and two 1 × 2.5 mm² working electrodes (WE1, WE2) with 0.6 mm spacing. The chip mounts in a multi-layer PMMA flow cell (laser-machined), with a 15 µL electrochemical chamber and 1 mm × 7 mm microchannels (175 µm thick), PDMS gasket (180 µm), threaded fluidic ports to Teflon tubing, and spring-loaded electrical contacts. A separate 50 µL PMMA batch cell was used for electrode modification. A peristaltic pump provided flow. An Autolab PGSTAT-100 with NOVA v2.0 controlled measurements. Electrode cleaning/activation: Mechanical cleaning with ethanol (96%), 6 M H2SO4, and water; electrochemical activation by cyclic voltammetry in 0.1 M KNO3 (20 scans from +0.8 to −2.2 V at 100 mV s⁻¹). Biosensor construction (sequential on-chip modification): Electrogeneration of PPy membranes at +0.7 V vs Ag/AgCl in 0.05 M phosphate buffer (PB, pH 7) containing 0.4 M pyrrole and 0.1 M KCl (generation solution). WE1 (L-lactate biosensor): generation solution plus LOX (10 U) and HRP (200 U); deposition to 500 mC cm⁻². WE2 (L-malate biosensor): first PPy layer to 250 mC cm⁻² with 10 mM HAR(III); second PPy layer to 500 mC cm⁻² with MDH (45 U) and DP (7.5 U). After synthesis, chips were rinsed with PB; stored at 4 °C in PB when not in use. PPy overoxidation: To stabilize baseline, both PPy membranes were overoxidized by 60 CV cycles between 0 and +1 V at 100 mV s⁻¹ in PB; performed once post-fabrication. Reagents: PB (0.05 M, pH 7); KCl (0.5 M) in measurement solutions to reduce potential drop/hysteresis; L-lactate/L-malic standards; mediators: K4[Fe(CN)6] for L-lactate/HRP and hexaammineruthenium(III) chloride (HAR) for L-malate/DP; NAD+ (0.1 M freshly prepared daily) as cofactor for MDH. Enzymes: LOX, HRP, MDH, DP handled and stored per supplier directions. Electrochemical measurement protocol: Measurements performed in the 15 µL flow cell under stop-flow conditions with the on-chip Pt CE and p-RE; p-RE positioned upstream of biosensors. Initial CVs (20 mV s⁻¹) recorded in PB + 0.5 M KCl containing required reagents to determine operational potentials. Set potentials selected based on CVs: −0.35 V (vs Pt p-RE) for L-lactate (ferricyanide reduction), and −0.40 V for L-malate (HAR(II) re-oxidation, with NAD+ present). Chronoamperometry: For each concentration step, solutions/samples flowed for 30 s at 0.25 mL min⁻¹ to refresh the cell, then stop-flow amperometry recorded for 120 s; the mean current density over the last 30 s used as analytical signal. Calibration and performance evaluation: L-lactate range 1 × 10⁻⁷ to 1 × 10⁻³ M; L-malate range 1 × 10⁻⁷ to 1 × 10⁻⁵ M. Sensitivity, linear range, LOD (3σ, IUPAC), and fabrication reproducibility assessed using three independent chips per analyte. Selectivity evaluated against common wine constituents at 5 × 10⁻⁵ M (L-lactate) or 5 × 10⁻⁷ M (L-malate). Long-term stability tested by repeated calibrations every 2–4 days over >30 days. Wine samples and preparation: Red wine samples from IRTA-INCAVI (Tarragona, Spain), vintage 2013; MLF induced with Oenococcus oeni. Sampling during MLF: Wine 1 (10 samples over 28 days), Wine 2 (5 samples over 33 days), Wine 3 (13 samples over 45 days). Expected ranges: L-malic 0–8 × 10⁻³ M (0–1.2 g L⁻¹), L-lactic 0–6 × 10⁻³ M (0–0.5 g L⁻¹). Dilutions to fit biosensor linear ranges: typically up to 1:10,000 for L-malic (two intermediate 1:100 dilutions), and 1:50–1:20 for L-lactate depending on wine (general guideline: L-lactic 1:1000; L-malic 1:10,000). For L-lactate measurements, ferrocyanide added to diluted sample; for L-malate, NAD+ added. Each analysis comprised two consecutive determinations (lactate then malate), total ~5 minutes per sample. Reference method: Standard enzymatic colorimetric assays (OIV) measuring NADH at 340 nm were performed by IRTA-INCAVI for comparison.

- Fabrication/integration: Successful on-chip sequential electropolymerization of PPy membranes entrapping specific enzymes and mediators on two WEs without cross-contamination; robust integration in a 15 µL PMMA flow cell with on-chip Pt CE and p-RE. - Operational potentials: −0.35 V (L-lactate, ferricyanide reduction); −0.40 V (L-malate, HAR(II) re-oxidation) vs Pt p-RE. - L-lactate biosensor performance: Linear range 5 × 10⁻⁷ to 1 × 10⁻⁴ M; sensitivity (−173 ± 8) × 10² µA M⁻¹ cm⁻² (r = 0.997, n = 7); LOD 3.2 ± 0.3 × 10⁻⁶ M; saturation above 1 × 10⁻⁵ M; fabrication reproducibility RSD of sensitivity < 8%. - L-malate biosensor performance: Linear range 1 × 10⁻⁷ to 1 × 10⁻⁶ M; sensitivity (5.53 ± 0.6) × 10² mA M⁻¹ cm⁻² (r = 0.997, n = 5); LOD 6.7 ± 0.2 × 10⁻⁸ M; saturation above 1 × 10⁻⁶ M; fabrication reproducibility RSD of sensitivity < 6%. - Stability: Both biosensors retained >90% of initial response over 30–52 days (52 days for L-lactate; 37 days for L-malate in prior work). In the integrated system, after >80 measurements, retained sensitivities were ~91% (L-lactate) and ~93% (L-malate). - Selectivity: No significant responses from common wine interferents at tested concentrations for either biosensor at the set potentials. - Throughput and consumption: Total analysis time ~5 minutes per sample; reagent/sample volume per assay ~125 µL; microfluidics enabled high dilution to mitigate matrix effects at the p-RE. - Real sample validation: In three red wines monitored through MLF (10, 5, and 13 timepoints, respectively), results agreed closely with standard enzymatic methods; absolute errors < 0.15 g L⁻¹; all values within the 95% uncertainty range of the standard method. - Comparative advantage: Lower LODs than prior bi-parametric systems (notably L-malate), reduced reagent/sample consumption, and first demonstration of simultaneous on-chip biosensing in a compact automated flow platform for MLF monitoring.

The study addresses the need for on-site, real-time monitoring of L-malic and L-lactic acids during MLF by integrating two selective electrochemical biosensors within a compact flow system. By entrapping specific enzyme pairs and mediators in electrogenerated PPy membranes on a single chip, the system achieves simultaneous, low-volume analysis with rapid turnaround and minimal reagents. The operational metrics—wide linearity for L-lactate and ultra-low LOD for L-malate, high selectivity against typical wine matrix components, strong fabrication reproducibility, and long-term stability compatible with the 20–40 day MLF duration—directly support the feasibility of continuous process monitoring. Validation with real fermentation samples from three wines demonstrates accuracy comparable to standard laboratory methods, enabling reliable tracking of the malate-to-lactate conversion and objective identification of MLF endpoints. Compared with prior bi-parametric strategies relying on separate reactors or membranes and FIA infrastructure, the present chip-integrated approach reduces system complexity and consumables, while improving sensitivity, especially for L-malate. These results indicate practical advantages for winery deployment, including potential automation (daily in-line calibration, reagent handling via multi-valve manifolds), on-barrel integration, and remote operation, thereby facilitating timely corrective actions and better process control.

A compact, bi-parametric analytical flow system integrating on-chip PPy-based bienzymatic electrochemical biosensors for L-lactate and L-malate was designed, fabricated, and validated. The silicon Pt four-electrode chip embedded in a laser-machined PMMA microfluidic cell enabled precise, low-volume, and rapid determinations with excellent analytical performance (low LODs, linearity, selectivity, reproducibility) and operational stability sufficient for full MLF monitoring. Real wine fermentations were accurately tracked with strong agreement to standard enzymatic assays. Key contributions include: (1) first simultaneous on-chip integration of L-lactate and L-malate biosensors within a compact automated flow platform for MLF control; (2) reduced reagent/sample consumption and simplified workflow (total analysis ~5 min, 125 µL per assay) conducive to on-site, real-time monitoring; and (3) demonstrated robustness for long-term process tracking. Future directions include fully automated in-field deployment with multi-valve reagent handling, on-barrel integration, remote control, and real-time data processing to further simplify operation for untrained personnel and enable proactive process management.

- Linear dynamic ranges necessitate significant sample dilutions (typically up to 1:10,000 for L-malic; up to 1:1000 for L-lactic), adding a sample-preparation step. - Biosensor responses exhibit saturation above 1 × 10⁻⁵ M (L-lactate) and 1 × 10⁻⁶ M (L-malate), restricting direct analysis of undiluted high-concentration samples. - Measurements were performed under stop-flow conditions with conditioned buffer (0.5 M KCl) and added reagents (ferrocyanide or NAD+), which may differ from fully in-line, reagentless scenarios. - The current work demonstrates accuracy and stability in controlled laboratory flow setups using a peristaltic pump; fully automated on-site integration (multi-valve manifolds, remote operation) is proposed but not experimentally demonstrated here. - Validation was conducted on three red wines from a single region/vintage; broader generalizability across diverse wine matrices and process conditions remains to be assessed.