Off-Junction Graphene/Water/Silicon Photodetector

• The paper introduces an off-junction graphene/water/silicon photodetector that produces transient pulse responses via lateral diffusion and drift of minority carriers.
• The device uses off-junction illumination with a light-shielded graphene/water interface, achieving ~2.12 μA photocurrent and 36.55 mA/W responsivity under 890 nm light.
• The study models carrier transport using exponential decay relations to extract precise minority-carrier lifetimes with errors as low as 1.2%, highlighting its metrological potential.

Searching arXiv for the cited papers and related topic to ground the article in the current literature. An off-junction graphene/water/silicon photodetector is a graphene/deionized-water/n-type silicon heterostructure in which the optically active region is intentionally displaced laterally from the junction region, so that illumination occurs on silicon away from a light-shielded graphene/water/silicon interface rather than directly on the interface itself. In the reported realization, water is encapsulated between graphene and silicon, two solid–liquid interfaces operate simultaneously, and the measured signal is a transient, pulse-like photocurrent generated by the diffusion and drift of minority carriers toward the water/silicon interface together with the dynamic polarization and depolarization of water molecules at the water/silicon and water/graphene interfaces (Wang et al., 26 Jul 2025). The device occupies a position at the intersection of graphene/semiconductor photodetectors, electrolyte-gated heterostructures, and off-junction carrier-transport metrology; it is also structurally analogous to the off-junction graphene–insulator–silicon configuration in which a parallel graphene–insulator–silicon path shapes both electrical response and photodetection (Pelella et al., 2021).

1. Device concept and structural configuration

The reported device is a graphene/deionized-water/n-type silicon stack in which water is encapsulated between graphene and silicon, with silver attached to graphene and gold attached to the backside of silicon (Wang et al., 26 Jul 2025). The junction region under graphene/water is covered by a top light-shield layer, and illumination is applied to bare silicon at a lateral distance from the covered junction. A commonly cited working distance is $0.2\ \mathrm{cm}$, at which the device shows a pulse-like current of $2.12\ \mu\mathrm{A}$ under $890\ \mathrm{nm}$ illumination at zero bias (Wang et al., 26 Jul 2025).

The operative geometry is explicitly off-junction. The light shield ensures that the graphene/water/silicon junction is not directly illuminated, and the photocurrent is unchanged compared to a fully light-isolated junction, confirming the shield’s effectiveness (Wang et al., 26 Jul 2025). This establishes that the observed signal originates from carriers generated at a remote spot and subsequently transported laterally toward the interface.

Graphene is transferred to a transparent PET substrate by wet transfer. Raman spectroscopy shows G and 2D peaks at $1590.71\ \mathrm{cm}^{-1}$ and $2693.67\ \mathrm{cm}^{-1}$; a weak D peak indicates few defects; and 2D intensity greater than G intensity is consistent with monolayer quality (Wang et al., 26 Jul 2025). On the silicon side, n-type wafers with resistivities of $50\ \Omega\!\cdot\!\mathrm{cm}$, $10\ \Omega\!\cdot\!\mathrm{cm}$, $5\ \Omega\!\cdot\!\mathrm{cm}$, and $1\ \Omega\!\cdot\!\mathrm{cm}$ were tested for lifetime extraction (Wang et al., 26 Jul 2025).

This architecture is best understood as an off-junction heterostructure whose readout is controlled by interfacial polarization rather than by direct photogeneration inside a conventional illuminated depletion region. A close analogue exists in graphene–silicon devices where graphene overlaps both a central Schottky window and a surrounding graphene–insulator–silicon region, producing a parallel “off-junction” path that modifies current transport and photoresponse (Pelella et al., 2021). In the water-based device, the off-junction path is not a solid-state MIS stack but a graphene/water/silicon heterointerface whose dynamics are governed by water polarization.

2. Interfacial electrostatics and pulse-generation mechanism

The operating mechanism begins with charge rearrangement at equilibrium. Upon contact, differences in Fermi levels or work functions cause holes to accumulate on graphene and electrons on n-type silicon surfaces, while water molecules align to screen these charges, with oxygen oriented toward n-Si and hydrogen toward graphene (Wang et al., 26 Jul 2025). The band-alignment data used in the reported analysis are a graphene work function of approximately $4.84\ \mathrm{eV}$, a water chemical potential of approximately $2.12\ \mu\mathrm{A}$0, and, for $2.12\ \mu\mathrm{A}$1 n-Si, a computed Fermi level $2.12\ \mu\mathrm{A}$2 (Wang et al., 26 Jul 2025).

Under off-junction illumination, minority holes are generated in silicon away from the junction, a local positive potential builds up in the illuminated region, and the junction region exhibits upward band bending with a negative potential there (Wang et al., 26 Jul 2025). Driven by both concentration and potential gradients, holes drift and diffuse laterally toward the water/silicon interface. Their accumulation modifies the interfacial charge and enhances water polarization, producing a positive pulse of current (Wang et al., 26 Jul 2025).

Under continuous illumination, transport and recombination reach dynamic equilibrium, and no steady-state photocurrent is observed (Wang et al., 26 Jul 2025). When illumination is removed, accumulated minority carriers recombine; water depolarizes because interface charge is reduced; and a negative transient current is generated as the device returns to its initial state (Wang et al., 26 Jul 2025). The device therefore exhibits transient, pulse-like responses only, rather than the persistent DC photocurrent characteristic of many conventional photodiodes.

This mechanism places the device within the broader class of polarized-liquid-triggered graphene photodetectors. A related graphene/NaCl(0.5 M)/n-GaN device was reported to operate under zero voltage bias through polarization and depolarization of water molecules driven by photogenerated carriers, yielding a repeatable photosensitive current (Lin et al., 2022). That broader result supports the interpretation that the water layer is not merely a passive dielectric spacer but an active polarization medium.

3. Transport framework and distance-dependent off-junction response

The transport description in n-type silicon is formulated through standard drift–diffusion relations for carrier current densities,

$2.12\ \mu\mathrm{A}$3

and

$2.12\ \mu\mathrm{A}$4

For excess minority carriers under off-junction illumination, the continuity or diffusion description is expressed as

$2.12\ \mu\mathrm{A}$5

and, in the reduced quasi-static form used for spatial decay,

$2.12\ \mu\mathrm{A}$6

The paper presents the same exponential form for the non-equilibrium minority concentration as Formula (7) (Wang et al., 26 Jul 2025).

The measured transient photocurrent is linked to minority-carrier concentration at the water/Si interface through

$2.12\ \mu\mathrm{A}$7

and the distance dependence is modeled as

$2.12\ \mu\mathrm{A}$8

where $2.12\ \mu\mathrm{A}$9 is an effective decay coefficient capturing losses during lateral drift–diffusion transport (Wang et al., 26 Jul 2025). The reported formulation is explicit that both drift and diffusion drive the carriers, and that $890\ \mathrm{nm}$0 therefore lumps both processes.

Distance-dependent fitting yielded the following expressions for n-Si wafers of different resistivity (Wang et al., 26 Jul 2025):

n-Si resistivity	Fitted current decay	Extracted $890\ \mathrm{nm}$1
$890\ \mathrm{nm}$2	$890\ \mathrm{nm}$3	$890\ \mathrm{nm}$4
$890\ \mathrm{nm}$5	$890\ \mathrm{nm}$6	$890\ \mathrm{nm}$7
$890\ \mathrm{nm}$8	$890\ \mathrm{nm}$9	$1590.71\ \mathrm{cm}^{-1}$0
$1590.71\ \mathrm{cm}^{-1}$1	$1590.71\ \mathrm{cm}^{-1}$2	$1590.71\ \mathrm{cm}^{-1}$3

Currents are in $1590.71\ \mathrm{cm}^{-1}$4 and distance $1590.71\ \mathrm{cm}^{-1}$5 is in $1590.71\ \mathrm{cm}^{-1}$6, as stated for the figure axes (Wang et al., 26 Jul 2025). The monotonic increase of $1590.71\ \mathrm{cm}^{-1}$7 with decreasing resistivity indicates faster decay of the off-junction transient with lateral distance in more heavily doped material. This suggests a shorter effective propagation length for the minority-carrier signal under the conditions used.

The transport picture is conceptually related to the off-junction MIS-assisted behavior in graphene–silicon devices, where carriers generated under a surrounding graphene–Si$1590.71\ \mathrm{cm}^{-1}$8N$1590.71\ \mathrm{cm}^{-1}$9–Si region can diffuse laterally toward a central graphene–silicon junction and modify the observed photocurrent or photovoltage even when the junction itself is not illuminated (Pelella et al., 2021). In both cases, lateral transport from a non-junction region is central to the measured response.

4. Photoresponse metrics, spectral behavior, and temporal characteristics

At $2693.67\ \mathrm{cm}^{-1}$0 off-junction distance under $2693.67\ \mathrm{cm}^{-1}$1 illumination and zero bias, the device produces a transient current peak of $2693.67\ \mathrm{cm}^{-1}$2 (Wang et al., 26 Jul 2025). Responsivity is computed as

$2693.67\ \mathrm{cm}^{-1}$3

or, as used in the paper,

$2693.67\ \mathrm{cm}^{-1}$4

where $2693.67\ \mathrm{cm}^{-1}$5 is the illuminated spot area (Wang et al., 26 Jul 2025). Detectivity is reported using the shot-noise-limited expression

$2693.67\ \mathrm{cm}^{-1}$6

with other noise terms such as Johnson noise and $2693.67\ \mathrm{cm}^{-1}$7 noise not included in the reported value (Wang et al., 26 Jul 2025).

Under $2693.67\ \mathrm{cm}^{-1}$8 illumination at $2693.67\ \mathrm{cm}^{-1}$9 from the junction, the device exhibits a typical responsivity of $50\ \Omega\!\cdot\!\mathrm{cm}$0 and a detectivity of $50\ \Omega\!\cdot\!\mathrm{cm}$1 (Wang et al., 26 Jul 2025). With increasing optical power density, both $50\ \Omega\!\cdot\!\mathrm{cm}$2 and $50\ \Omega\!\cdot\!\mathrm{cm}$3 decrease due to photocurrent saturation and enhanced trapping, meaning that a smaller fraction of photocarriers contributes to the transient current (Wang et al., 26 Jul 2025).

The temporal response is likewise transient. Measured rise and fall times are approximately $50\ \Omega\!\cdot\!\mathrm{cm}$4 and $50\ \Omega\!\cdot\!\mathrm{cm}$5, respectively, for $50\ \Omega\!\cdot\!\mathrm{cm}$6 illumination (Wang et al., 26 Jul 2025). These millisecond-scale dynamics are attributed to interfacial polarization and depolarization coupled to minority-carrier transport, rather than to purely electronic transit in a solid-state depletion region.

The spectral boundary is governed by silicon absorption. Using $50\ \Omega\!\cdot\!\mathrm{cm}$7, the absorption edge is $50\ \Omega\!\cdot\!\mathrm{cm}$8 from $50\ \Omega\!\cdot\!\mathrm{cm}$9, and the responsivity drops rapidly above approximately $10\ \Omega\!\cdot\!\mathrm{cm}$0, consistent with intrinsic absorption (Wang et al., 26 Jul 2025). This is broadly consistent with graphene–silicon Schottky photodetectors, in which above-bandgap operation in silicon yields substantially larger responsivity than sub-bandgap operation driven by graphene absorption and internal photoemission (Pelella et al., 2021).

A comparison with the graphene/NaCl(0.5 M)/n-GaN polarized-liquid photodetector clarifies the role of the absorber: under zero bias, that device reached $10\ \Omega\!\cdot\!\mathrm{cm}$1 responsivity and $10\ \Omega\!\cdot\!\mathrm{cm}$2 specific detectivity under $10\ \Omega\!\cdot\!\mathrm{cm}$3 illumination (Lin et al., 2022). The supplied account states that the response wavelength can be finely tuned through the free choice of semiconductor, because no lattice match between graphene and the semiconductor is required (Lin et al., 2022). In that sense, the silicon implementation extends the polarized-liquid principle into the visible and near-infrared range appropriate to silicon.

5. Minority-carrier lifetime metrology

A defining feature of the off-junction graphene/water/silicon photodetector is that the same transient signal used for photodetection can be used to infer silicon minority-carrier lifetime (Wang et al., 26 Jul 2025). The method measures photocurrent as a function of lateral distance from the junction and fits the decay with the exponential form $10\ \Omega\!\cdot\!\mathrm{cm}$4, after which minority-carrier lifetime $10\ \Omega\!\cdot\!\mathrm{cm}$5 is obtained by further decomposition of $10\ \Omega\!\cdot\!\mathrm{cm}$6 and substitution of known quantities using the Einstein relation and the diffusion-length concept (Wang et al., 26 Jul 2025).

The paper reports the following theoretical and fitted lifetimes (Wang et al., 26 Jul 2025):

n-Si resistivity	$10\ \Omega\!\cdot\!\mathrm{cm}$7	$10\ \Omega\!\cdot\!\mathrm{cm}$8
$10\ \Omega\!\cdot\!\mathrm{cm}$9	$5\ \Omega\!\cdot\!\mathrm{cm}$0	$5\ \Omega\!\cdot\!\mathrm{cm}$1
$5\ \Omega\!\cdot\!\mathrm{cm}$2	$5\ \Omega\!\cdot\!\mathrm{cm}$3	$5\ \Omega\!\cdot\!\mathrm{cm}$4
$5\ \Omega\!\cdot\!\mathrm{cm}$5	$5\ \Omega\!\cdot\!\mathrm{cm}$6	$5\ \Omega\!\cdot\!\mathrm{cm}$7
$5\ \Omega\!\cdot\!\mathrm{cm}$8	$5\ \Omega\!\cdot\!\mathrm{cm}$9	$1\ \Omega\!\cdot\!\mathrm{cm}$0

Reported errors are no more than $1\ \Omega\!\cdot\!\mathrm{cm}$1, and the best case is $1\ \Omega\!\cdot\!\mathrm{cm}$2, corresponding to a maximum accuracy rate of $1\ \Omega\!\cdot\!\mathrm{cm}$3 (Wang et al., 26 Jul 2025). The paper states that the method provides a straightforward route toward nondestructive tests in the semiconductor industry and does not require passivation or complex optical transient photoconductance setups (Wang et al., 26 Jul 2025).

The significance of this metrological use is that the off-junction geometry transforms lateral carrier transport into an experimentally accessible current–distance decay. Rather than measuring lifetime through a direct optical transient or contactless conductivity method, the device infers it from how efficiently photogenerated minority carriers can reach a remote interfacial polarization transducer. This suggests a hybrid readout in which the water layer functions as the key element for deducing the carrier lifetime, not just as a gate dielectric or electrolyte.

At the same time, the supplied description is explicit about the model’s scope: $1\ \Omega\!\cdot\!\mathrm{cm}$4 aggregates drift and diffusion contributions, and extracting $1\ \Omega\!\cdot\!\mathrm{cm}$5 assumes a relation between $1\ \Omega\!\cdot\!\mathrm{cm}$6 and $1\ \Omega\!\cdot\!\mathrm{cm}$7; in strong-field or nonuniform cases, $1\ \Omega\!\cdot\!\mathrm{cm}$8 may not map uniquely to $1\ \Omega\!\cdot\!\mathrm{cm}$9 without more detailed modeling (Wang et al., 26 Jul 2025). Validation is benchmarked against theoretical lifetimes for the given resistivities, whereas comparison to established metrology such as QSSPC, $4.84\ \mathrm{eV}$0-PCD, or DLTS is not provided (Wang et al., 26 Jul 2025).

6. Relation to off-junction MIS photodetectors, advantages, and limitations

The water-based off-junction device is closely related in concept to the graphene–silicon device in which graphene spans both a central graphene–silicon Schottky junction and a surrounding graphene–Si$4.84\ \mathrm{eV}$1N$4.84\ \mathrm{eV}$2–Si region in parallel (Pelella et al., 2021). In that solid-state case, the surrounding MIS path behaves as a capacitor in forward bias and, in reverse bias, can accumulate holes at the Si/Si$4.84\ \mathrm{eV}$3N$4.84\ \mathrm{eV}$4 interface, invert the silicon surface, and eventually conduct through Fowler–Nordheim tunneling. Under illumination, photocarriers generated under the nitride can be collected by the MIS path and can feed the Schottky junction region via diffusion, so photocurrent and photovoltage can be observed even when the central junction is not directly illuminated (Pelella et al., 2021).

The off-junction graphene/water/silicon photodetector replaces that solid dielectric environment with deionized water. The reported water-based behavior is therefore analogous in spatial logic but different in microscopic mechanism. In the silicon nitride case, the parallel path is described as a graphene–Si$4.84\ \mathrm{eV}$5N$4.84\ \mathrm{eV}$6–Si MIS branch with electrostatic coupling, inversion, and tunneling (Pelella et al., 2021). In the water case, the dominant effect is the dynamic polarization process of water molecules at the water/silicon and water/graphene interfaces, and the signal is pulse-like rather than a steady reverse-bias photocurrent (Wang et al., 26 Jul 2025).

The reported advantages are clear. The method enables nondestructive testing of minority-carrier lifetime, uses simple off-junction illumination and current-distance fitting, has spatial mapping capability through scanning of the illumination spot, and also functions as a photodetector with $4.84\ \mathrm{eV}$7 and $4.84\ \mathrm{eV}$8 under $4.84\ \mathrm{eV}$9 illumination at $2.12\ \mu\mathrm{A}$00 (Wang et al., 26 Jul 2025). The light-shielded geometry allows the junction itself to remain optically isolated while still producing a measurable electrical signal.

The limitations are equally important. The device operates through transient pulses only, with no steady-state photocurrent under continuous illumination (Wang et al., 26 Jul 2025). Water-layer thickness, purity, and ionic content are not quantified in the report and could affect the conversion coefficient $2.12\ \mu\mathrm{A}$01 and the decay coefficient $2.12\ \mu\mathrm{A}$02 (Wang et al., 26 Jul 2025). Interface traps and hysteresis at water/Si or water/graphene interfaces could alter polarization dynamics, especially at high power densities (Wang et al., 26 Jul 2025). Temperature and humidity are not specified, and environmental sensitivity is therefore uncharacterized (Wang et al., 26 Jul 2025).

A common misconception would be to treat the water layer as a passive medium comparable to a conventional solid insulator. The reported mechanism does not support that simplification. The water layer is the active polar liquid whose polarization and depolarization are essential to signal generation (Wang et al., 26 Jul 2025). Another misconception would be to interpret the measured current decay as a pure diffusion-length measurement. The paper explicitly states that both drift and diffusion contribute, and that $2.12\ \mu\mathrm{A}$03 is an effective decay coefficient rather than a uniquely diffusion-defined quantity (Wang et al., 26 Jul 2025).