Designing electrochemical microfluidic multiplexed biosensors for on-site applications

Clinical assessment based on a single biomarker is in many circumstances not sufficient for adequate diagnosis of a disease or for monitoring its therapy. Multiplexing, the measurement of multiple analytes from one sample and/or of the same target from different samples simultaneously, could enhance the accuracy of the diagnosis of diseases and their therapy success. Thus, there is a great and urgent demand for multiplexed biosensors allowing a low-cost, easy-to-use, and rapid on-site testing. In this work, we present a simple, flexible, and highly scalable strategy for implementing microfluidic multiplexed electrochemical biosensors (BiosensorX). Our technology is able to detect 4, 6, or 8 (different) analytes or samples simultaneously using a sequential design concept: multiple immobilization areas, where the assay components are adsorbed, followed by their individual electrochemical cells, where the amperometric signal readout takes place, within a single microfluidic channel. Here, first we compare vertical and horizontal designs of BiosensorX chips using a model assay. Owing to its easier handling and superior fluidic behavior, the vertical format is chosen as the final multiplexed chip design. Consequently, the feasibility of the BiosensorX for multiplexed on-site testing is successfully demonstrated by measuring meropenem antibiotics via an antibody-free β-lactam assay. The multiplexed biosensor platform introduced can be further extended for the simultaneous detection of other anti-infective agents and/or biomarkers (such as renal or inflammation biomarkers) as well as different (invasive and non-invasive) sample types, which would be a major step towards sepsis management and beyond. Graphical Abstract Supplementary Information The online version contains supplementary material available at 10.1007/s00216-022-04210-4.


Wafer-level fabrication process of biosensors
The fabrication procedure of the miLab and Multilab biosensors was already described in [1] and [2] in detail and obeys a strict protocol consisting of ten steps ( Figure S1). In the following, these steps will be shortly explained in the order they are executed during fabrication. After each step, the wafers are washed with DI-water and dried with pressurized air.

Cutting of polyimide and copper etching
First, the 6" wafers are cut from the Pyralux AP8545 (DuPont, USA) substrate using a pair of scissors. The polyimide film is covered with copper on both sides which is removed in a bubble etch tank (PA104, MEGA Electronics, UK) filled with 20% sodium persulfate at a temperature of 45 °C for 1 hour.

Spin coating and development of MA-N 1420
At this step, a special resist allowing a lift-off process is used in order to form the metallization area for creating the contact pads and electrodes. Herein, 2-3 µl of the resist MA-N1420 (Micro resist technology GmbH, Germany) is spin-coated on the wafer and then it is soft baked for 2 minutes at 100 °C. This is followed by a 2-minute UV exposure on an exposure unit (Hellas, Bungard Elektronik, Germany) using a foil mask (for metallization) (MKD Kramer, Germany) and then development in two baths of ma-D 533s (Micro resist technology GmbH, Germany).

Platinum deposition and lift-off step
To realize the metallization, 200 nm of platinum (Pt) is deposited onto the wafer by a physical vapor deposition (PVD) process in the clean room at IMTEK, University of Freiburg. For removing the excess Pt, the wafers are dipped in a remover bath, containing ma-R 404s (Micro resist technology GmbH, Germany), on a shaker for 40 minutes.

Formation of insulation layer with SU-8
At this step, the epoxy resist SU-8 (3005, Microchem, USA) is applied on the wafer to define the electrode areas / contact pads, forming the wells for Teflon stopping barrier and isolating the not-active metal parts.
First, the wafers are soft-baked in an oven at 120 °C for 10 minutes prior to spin coating 6 ml of the resist onto the wafer. This step is followed by two baking steps on different hotplates, for 2 minutes at 65°C and 3 minutes at 90°C, respectively. After letting the wafers dry overnight in closed carriers, the UV-exposure for 1:20 minutes using a foil mask (for isolation) is performed, followed by development in three baths of 1-methoxy-2-propyl acetate solutions (Merck KGaA, Germany) and one bath of isopropanol is conducted.
After the washing step, the wafers are hard-baked for 3 hours at 160°C starting from / ending at the room temperature in an oven (Binder, Germany).

Plasma cleaning step
To remove the SU-8 residues, the wafers are placed into a standard plasma unit (Tetra30-LF-PC, Diener, Germany). The Pt electrodes are "cleaned" by using 300 W low-frequency power with 100% oxygen flow, at a flow rate-controlled pressure of 0.4 mbar, for a total plasma reaction time of 3 minutes.

Silver deposition and chlorination of reference electrodes
To create on-chip Ag/AgCl reference electrodes on wafer level, an electrodeposition step is employed. To protect the contact pads of the chips during this step, they are covered with a UV-sensitive tape. The silver deposition takes place in an alkaline silver cyanide solution (Arguna S, Umicore, Germany), immersed in an ultrasonic bath (Sonorex Super 10 P, Bandelin, Germany). The bulk contact pad of the wafer is the cathode of the process, while as anode a bare silver wire, immersed to the Arguna S solution is used. At the next step, the wafer is placed in an 0.1 % potassium chloride (KCl) solution for partial chlorination of the silver deposited to create on-chip reference electrodes. Finally, the UV-tape is removed after a short UV-illumination.

Dispensing of Teflon for formation of hydrophobic stopping barriers
To prevent the bioreagents from getting into the area of the electrochemical cell and thus, from poisoning the electrodes, hydrophobic stopping barriers (single for miLab and multiple for BiosensorX) are formed.
Therefore, a small drop of a 3% Teflon (AF 1600, DuPont, USA) in a FC-75 solution is dispensed into the wells (formed by the SU-8), using a hand dispenser (1500 XL, Nordson EFD).

Production of DFR layers
For the realization of the microfluidic channels, the dry-film photoresist (DFR) Pyralux PC1025® (DuPont, USA) is used. The DFR is cut in appropriate pieces, and then exposed in the standard vacuum exposure unit using the respective foil masks. For one wafer, four layers are needed: a channel, a cover and two backside layers. Last two ones prevent the chips from bending during the final backing step.
Upon illumination, the layers are developed in two dishes filled with 1% sodium carbonate (Na2CO3), placed in an ultrasonic bath. To stop the development, prevent an over development and release stress from the layers, they are shaked for one minute in a 1% hydrochloric acid (HCl) bath. After that, the layers are carefully but thoroughly rinsed with DI-water and dried pressurized air. Before continuing with lamination, the layers should be kept for some hours (better for couple days) at room temperature to reduce their stickiness while covered with a lid to prevent any dust formation.

Lamination of DFR layers
Herein, the wafer is sticked on an overhead foil and the appropriate DFR layers are aligned under a microscope, starting from the channel layer, then the two backsides and finally the cover. The layers are laminated one after the other with a standard hot roll laminator (HRL 350, Ozatec). It is important to remove the shiny foil of each Pyralux PC layer before continuing with the lamination of the next layer.

Chip cutting and hard bake
After all layers are laminated, the wafer can be finalized by pulling the last two protective foils from top and bottom and cutting the wafer in stripes, using a pair of scissors. Finally, the chips are baked in an oven for 3 hours at 160 °C starting from / ending at the room temperature. After that step, the chips are ready to be used.

Biosensor designs
Additionally to the 4-plex vertical Biosensor, discussed in the main text, several chip designs have been estimated and measured, namely 4-plex horizontal (Figure S2a and b), 6-plex vertical (Figure S2c and d) and horizontal, 8-plex vertical and horizontal and milab chip (Figure S2e and f).  and assembled chips, respectively. a) and b) 4-plex horizontal, c) and d) 6-plex vertical and e) and f) miLab chip.

Cost estimation of biosensors designed and fabricated in this study
To estimate the fabrication costs, a batch of four wafers produced under research laboratory conditions is considered. Table S1 shows the cost calculation including all fabrication steps, and total costs per measurement channel for all biosensor designs: miLab chip (two biosensors per chip) and the different designs of BiosensorX (18.5 mm and 22 mm). The costs per incubation area range from 3.400 € (8-plex version) to 2.948 € (4-plex version).

Chip connection and integration
For the assay incubation and electrochemical signal readout, different components are necessary to ensure an easy handling and accurate results. These components, therefore, must perfectly fit together and be suitable for the designed multiplexed biosensors.
During the assay incubation, a washing step is needed to remove unbound biomolecules from the channel after each consecutive biomolecule incubation. Herein, a vacuum adapter (Figure S3) is      Figure S7. Initial characterization plots obtained with different multiplexed chip designs: a) 4-plex vertical, b) 4plex horizontal, c) 6-plex vertical and d) 6-plex horizontal. Obtained current peaks did not show a rectangular shaped signal. This indicates that the channels were not completely saturated with the signaling biomolecule employed: StrGOx. Therefore, we repeated these tests (Figure 4) with a higher concentration of StrGOx (200 µg ml -1 ), to ensure the deviations were solely due to the design differences and not due to limited biomolecule concentration.

Limit of detection
The limit of detection (LOD) of an assay describes the lowest measurable analyte concentration and is also referred to as "sensitivity". It is defined through a certainty level of 3.3 (corresponds to 95%), where is the standard deviation of the blank value. For most assays, the correlation between the analyte concentration and the signal measured is non-linear and can be fitted with a 4-or 5-parameter logistic fit. Here, 4-parametric logistic model was employed to determine the LOD by using the limit of blank (LOB) where 8 is the inflection point of the curve, 0%: the response at low concentration and the slope factor of the fitting curve [4,5].