A sandwich immunoassay (ELISA) works in two steps. One antibody grabs the target protein, and a second antibody, carrying an enzyme, sticks to the captured protein. The enzyme then turns many substrate molecules into colored product, which makes even trace amounts of protein visible. This is accurate but slow, and it needs a lab. Rapid tests are fast, but they lose sensitivity or get confused by sticky biological fluids such as mucus, saliva and serum. We propose a different way to read out the same sandwich. The detection label builds a tiny junction between two semiconductors only where a sandwich has formed. Electrons flow across that junction to balance the energy levels of the two materials, and this flips the direction of the light-driven current through the electrode. We would read the electrode voltage at which the current changes sign, rather than how large the current is. Non-specific gunk on the surface makes the current weaker in both directions, but it should not move the sign-change point.
The physical idea comes from oxide and two-dimensional heterostructure physics. There, electrons migrate across an interface to equalize the Fermi levels (the energy up to which electron states are filled), and this creates states that neither material has on its own (Calderón et al.; Di Yi et al.). We borrow only the principle. We do not borrow those materials, because their emergent phases survive only at 0.03–120 K (Bi2Ir2O7/Dy2Ti2O7; WS2/FePS3) and they degrade in aqueous electrolytes. The bet is that an interface created by the binding event itself gives a readout that is ratiometric, meaning it compares two signals rather than measuring one absolute size. It would also be insensitive to the sample matrix, and it could compete with the best rapid catalytic immunoassay in the file.
Core Claim
The proposed sensor has two parts. The base is an n-type photoelectrode, meaning one whose light-driven current under bias is anodic (electrons flow into the electrode). Each sandwich complex then deposits or carries a p-type label: either a catalytic label that generates a p-type deposit in place, or a p-type nanocrystal attached directly to the second antibody. Where the label sits, a local p–n junction forms, and that patch contributes a cathodic photocurrent (current in the opposite direction). The total photocurrent density at applied potential $E$ is then approximately
$$j(E) = a\,[\,(1-\theta)\,j_n(E) - \theta\,j_p(E)\,],$$
where $j_n$ and $j_p$ are the anodic and cathodic photocurrent densities of bare and junction-covered surface, $\theta$ is the fraction of surface covered by label (set by the analyte concentration), and $a \le 1$ is the attenuation from non-specific fouling and light-source drift. The reversal potential $E_0$ is defined by $j(E_0)=0$. It depends on $\theta$ but not on $a$, so long as fouling attenuates both current components by the same factor.
What would be new is threefold. The immunocomplex creates the charge-transfer interface rather than blocking one. The readout is a potential rather than a current. And the decisive test is whether $E_0$ really ignores the matrix. If fouling attenuates the two components unequally, $E_0$ will drift, and the experiments below are built to catch exactly that.
Relation to Prior Work
Two photoelectrochemical biosensors in the file motivate the design and also mark its limits. The CdSe–ZnO core-shell sensor uses a type-II heterojunction, an interface arranged so electrons and holes separate into different materials. It reads Norovirus RNA through photocurrent quenching: hybridized RNA insulates the surface, and current falls along $I = 0.03256 - 0.0033c$. That gives an LOD (limit of detection) of 0.50 nM. Quenching readouts cannot separate specific binding from fouling, since both only reduce current, and that is the failure mode we target. The MXene/Co-TiO2 magneto-photoelectrochemical platform shows that a room-temperature magnetic field lowers charge-transfer resistance and improves the LOD for kinase activity from 0.00027 to 0.00016 U/mL. However, it detects peptide phosphorylation, not antibody sandwiches, and it still reads current magnitude. We reuse its Co-TiO2 photoanode as one candidate base and its field arm as an optional ablation. We do not make magnetism central, because a 68.75% gain is modest next to the orders-of-magnitude gains that catalytic amplification gives.
The strongest competitor comes from the rapid-immunoassay side. The FeAu@AuIr nanozyme immunochromatographic assay replaces the enzyme horseradish peroxidase with a synthetic catalytic particle. It reaches 1.2 pg/mL for SARS-CoV-2 spike in 20 minutes, which is 58-fold better than a commercial ELISA and 238–416-fold better than colloidal gold strips, with 100%/100% agreement on 170 clinical samples. Its readout is optical, and its magnetic core serves only to pull particles together. We borrow its nanozyme chemistry as one way to generate the label. The standardized nasal IgA ELISA provides the benchtop reference (intermediate precision under 17%, agreement coefficient of 0.87 against an electrochemiluminescence assay) and the sampling protocol. A polyvinyl alcohol sponge detected nasal IgA in 95.5% of cases, against 68.8% for nasopharyngeal swabs.
One result from interface physics constrains how we interpret success. In WS2/FePS3, density functional theory suggested that sulfur-mediated bonding, more than charge transfer, drives the emergent order. Mismatched-chalcogen controls showed no effect (Ferromagnetism emerged from non-ferromagnetic atomic crystals). By analogy, a shift in $E_0$ could come from surface chemistry rather than band alignment, so we include a matched-mismatch control in the same spirit.
Research Program
The work proceeds in three stages. In the first, we build the base photoanodes and label pairs. For bases we use TiO2 and Co-TiO2 films on indium tin oxide glass, both room-temperature stable and compatible with antibody immobilization through silane or EDC/NHS linkers (a standard chemistry for joining amines to carboxyl groups). For labels we screen two routes. In the first route, the AuIr nanozyme on the second antibody catalyzes oxidation of a soluble precursor into a p-type deposit. In the second, a p-type oxide nanocrystal is conjugated directly to the antibody. We characterize band positions with Mott–Schottky plots (capacitance measurements that reveal semiconductor band edges) and photoluminescence, as the magneto-photoelectrochemical paper did. Crucially, we also make an energetically mismatched control label: one of similar size, conductivity and surface chemistry, but with band edges that cannot form a charge-separating junction with the base.
In the second stage, we calibrate the readout. Each electrode receives one short potential sweep under chopped light, taking seconds, from which $E_0$ is extracted. We map $E_0$ against analyte concentration and against deliberately applied fouling. The fouling agents are bovine serum albumin, mucin and whole diluted serum, applied with no analyte present, and we also vary light intensity.
In the third stage, we run the full assay on clinical matrices. One target is SARS-CoV-2 spike in nasal sponge eluates, which matches the nanozyme assay's analyte. The other is plasma TNFR1/TNFR2. Cao et al. found by ELISA that these two markers correlate strongly with albuminuria, whereas CRP does not, so they give a second, serum-heavy matrix with clinical meaning. The total time budget is 20 minutes, matching the strongest baseline.
Baselines
The naive reference point is a colloidal-gold lateral flow strip run on the same samples. The nanozyme paper reports it to be 238–416-fold less sensitive than catalytic readout, so it sets the floor that any amplified method should clear easily.
The benchtop reference is a conventional sandwich ELISA using the same capture and detection antibody pair, run as in the standardized nasal IgA protocol. It sets clinical truth for concordance, and we will report agreement coefficients against it. The bar to clear is the 0.87 that the IgA ELISA achieved against electrochemiluminescence.
The strongest existing method is the FeAu@AuIr nanozyme immunochromatographic assay. It reports an LOD of 1.2 pg/mL for SARS-CoV-2 spike, 20-minute runtime, CV of at most 7.5%, four orders of dynamic range, and 100% sensitivity and specificity on 170 samples (paper). We will reproduce it with the same antibody lot, sample aliquots and 20-minute budget. Beating it means one of two outcomes. The first is equal or lower LOD with at least four orders of range. The second, which is the more important claim, is a matrix-induced bias (difference in measured concentration between buffer and mucus or serum spikes) at least threefold smaller at equal LOD. For TNFR1/TNFR2 there is no reported rapid-assay result found on Lacuna, so ELISA is the reference there.
Two ablations isolate the key ingredient. The first reads the same sandwich on the same electrode in two other ways: as plain photocurrent magnitude at fixed bias, and as dark amperometry of the label's catalytic product. Both use the same data and the same calibration-fitting procedure. This separates the gain from the reversal-potential readout from the gain from catalysis alone. The second is the energetically mismatched label, which should leave $E_0$ unmoved if the effect really comes from interfacial charge transfer. An optional third arm adds the external magnetic field on Co-TiO2. Its only reported comparison is the 0.00027 to 0.00016 U/mL kinase-activity LOD improvement, in a different assay.
Why not just…?
Why not just use the nanozyme lateral flow strip, which already works? This is the real threat. At 1.2 pg/mL in 20 minutes with perfect clinical agreement, it may be good enough for respiratory antigens. Its weak point, which the authors themselves note, is that the boronic acid capture chemistry binds host glycoproteins in heavy mucus, and optical density cannot tell specific signal from background. If our matrix-bias comparison shows no advantage on mucus and serum, the proposal reduces to a costlier strip, and we would say so and stop.
Why not just read the enzyme product electrochemically, with no heterojunction? Dark amperometry of a catalytic label is simpler and needs no light source. But it is still a magnitude readout: fouling that slows electron transfer lowers it exactly as low analyte would. The amperometry ablation tests this directly. If amperometric matrix bias matches that of $E_0$, the heterojunction adds nothing and we would drop it.
Why not just use a quenching photoelectrode like CdSe–ZnO? That design is already demonstrated, but the binding event can only reduce current, and an insulating fouling layer produces the same signal. Its reported 0.50 nM nucleic-acid LOD comes without catalytic amplification. The photocurrent-magnitude ablation, run with an unlabeled sandwich, reproduces this mode on our electrodes.
Why not just use the correlated-oxide or two-dimensional magnetic interfaces themselves? Their phenomena exist at cryogenic temperatures and several tesla. The graphone/PbTiO3 magnetoelectric effect (DFT study) has not been fabricated. None has been shown to survive physiological fluids. We take the Fermi-level-equalization principle and leave the materials.
Proposed Experiments and Decisive Checks
The first check is whether $E_0$ is invariant to fouling. On analyte-free electrodes exposed to albumin, mucin and serum, and with light intensity varied, photocurrent magnitude should drop substantially while $E_0$ stays within the run-to-run spread seen in clean buffer. If $E_0$ moves by an amount comparable to a calibration step, the core claim fails.
The second check is specificity of mechanism. The mismatched control label should produce magnitude changes but no shift in $E_0$, mirroring the chalcogen-mismatch logic of the WS2/FePS3 work.
The third check is analytical performance. For spike in buffer and in nasal sponge eluate, we report LOD, dynamic range, CV and recovery against the nanozyme strip, the ELISA and both ablations, all on shared aliquots within 20 minutes.
The fourth check is clinical concordance. On banked nasal samples, and on plasma for TNFR1/TNFR2, we compute the agreement coefficient against the standardized ELISA and sensitivity and specificity against qPCR, alongside the nanozyme strip.
The proposal succeeds if checks one and two hold and the matrix-bias advantage over both the strip and the amperometric ablation appears at comparable LOD. It fails informatively if it reaches only LOD parity without that advantage.
References
- Magnetoelectric coupling at the interface of BiFe<mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" display="inline"><mml:mrow><mml:msub><mml:mi mathvariant="normal">O</mml:mi><mml:mrow><mml:mn>3</mml:mn></mml:mrow></mml:msub></mml:mrow></mml:math>/La<mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" display="inline"><mml:mrow><mml:msub><mml:mi/><mml:mrow><mml:mn>0</mml:mn><mml:mo>.</mml:mo><mml:mn>7</mml:mn></mml:mrow></mml:msub></mml:mrow></mml:math>Sr<mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" display="inline"><mml:mrow><mml:msub><mml:mi/><mml:mrow><mml:mn>0</mml:mn><mml:mo>.</mml:mo><mml:mn>3</mml:mn></mml:mrow></mml:msub></mml:mrow></mml:math>Mn<mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" display="inline"><mml:mrow><mml:msub><mml:mi mathvariant="normal">O</mml:mi><mml:mrow><mml:mn>3</mml:mn></mml:mrow></mml:msub></mml:mrow></mml:math>multilayers
- Tuning Perpendicular Magnetic Anisotropy by Oxygen Octahedral Rotations in <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" display="inline"><mml:mrow><mml:mo stretchy="false">(</mml:mo><mml:msub><mml:mrow><mml:mi>La</mml:mi></mml:mrow><mml:mrow><mml:mn>1</mml:mn><mml:mo>−</mml:mo><mml:mi>x</mml:mi></mml:mrow></mml:msub><mml:mrow><mml:msub><mml:mrow><mml:mi>Sr</mml:mi></mml:mrow><mml:mrow><mml:mi>x</mml:mi></mml:mrow></mml:msub></mml:mrow><mml:mrow><mml:msub><mml:mrow><mml:mi>MnO</mml:mi></mml:mrow><mml:mrow><mml:mn>3</mml:mn></mml:mrow></mml:msub></mml:mrow><mml:mo stretchy="false">)</mml:mo><mml:mo>/</mml:mo><mml:mo stretchy="false">(</mml:mo><mml:msub><mml:mrow><mml:mi>SrIrO</mml:mi></mml:mrow><mml:mrow><mml:mn>3</mml:mn></mml:mrow></mml:msub><mml:mo stretchy="false">)</mml:mo></mml:mrow></mml:math> Superlattices
- Anomalous magnetoresistance by breaking ice rule in Bi2Ir2O7/Dy2Ti2O7 heterostructure
- Ferromagnetism emerged from non-ferromagnetic atomic crystals
- Development of CdSe-ZnO Flower-Rod Core-Shell Structure Based Photoelectrochemical Biosensor for Detection of Norovirous RNA
- Magneto-photoelectrochemical 2D heterojunction platform for biosensing detection
- Mercaptophenylboronic Acid‐Mediated Nanozyme Immunochromatographic Assay for Simultaneous Detection of Respiratory Bacteria and Virus
- Exploring the standardized detection and sampling methods of human nasal SARS-CoV-2 RBD IgA
- Inflammation and Kidney Injury in Diabetic African American Men
- Giant magnetoelectric effect at the graphone/ferroelectric interface