Papers
Topics
Authors
Recent
Search
2000 character limit reached

Photoinduced Modulation Doping

Updated 5 July 2026
  • Photoinduced modulation doping is a light-activated process where a remote charge reservoir modulates carrier density in 2D materials without chemical substitution.
  • It employs various platforms, such as MoS2/ITO and graphene/BN, to achieve rewritable, nonlocal, or persistent doping via optical excitation.
  • This technique offers versatile tuning of electronic properties, impacting 2D electronics, optoelectronics, and nanoscale device engineering.

Photoinduced modulation doping is a light-activated analogue of conventional modulation doping in which the states that donate, accept, or trap charge are spatially separated from the transport channel, so optical excitation changes carrier density without conventional chemical substitution of the channel lattice. In reported implementations, the remote charge reservoir may be a defect-rich dielectric, a nanocrystal layer, a gate electrode, or an adjacent van der Waals acceptor; the channel may be graphene, monolayer MoS2_2, a graphene nanoelectromechanical membrane, or a twisted WSe2_2 bilayer. The resulting doping can be contactless, rewritable, quasi-permanent, strongly local, or nonlocal over tens of micrometers, depending on the heterostructure and the mechanism that stabilizes the transferred charge (Kriegel et al., 2018, Ju et al., 2014, Tiberj et al., 2013, Miller et al., 2019, Gadelha et al., 2020, Li et al., 23 Oct 2025).

1. Conceptual foundation and relation to conventional modulation doping

In conventional modulation doping, dopant charges are placed in a nearby layer rather than inside the semiconductor channel, so carriers are transferred electrostatically while impurity scattering is reduced. Recent work on 2D electronics reformulates this in terms of workfunction-mediated charge transfer across van der Waals interfaces, especially under type-III (“broken-gap”) alignment, with vacuum-referenced band-edge descriptors

EIP=EvacEVBM,EEA=EvacECBM.E_\mathrm{IP} = E_\mathrm{vac} - E_\mathrm{VBM}, \qquad E_\mathrm{EA} = E_\mathrm{vac} - E_\mathrm{CBM}.

Within this framework, high-electron-affinity materials such as α\alpha-RuCl3_3, MoO3_3, and V2_2O5_5 are identified as strong p-type dopants for 2D channels, because they pull electrons from the semiconductor without introducing defects into the channel itself (Arora et al., 2024).

A closely related non-photonic implementation relocates the dopant function into a dielectric: aluminium in SiO2_2 forms acceptor states that can capture electrons from adjacent Si through tunnelling, leaving holes in silicon while keeping the Si active volume essentially intrinsic (König et al., 2016). In a correlated oxide variant, modulation-doped VO2_2-based heterostructures achieve carrier densities greater than 2_20 without measurable structural changes, explicitly separating filling control from direct lattice modification (Mondal et al., 2023).

Photoinduced modulation doping preserves the same remote-doping logic but adds an optical trigger. Light either excites defect transitions, charges nanocapacitors, photoionizes traps, or generates hot carriers in remote electrodes; the key outcome is that a nearby layer or interface becomes the charge reservoir that electrostatically dopes the channel. This suggests that photoinduced modulation doping is best understood not as a single microscopic process but as a family of optically activated remote charge-transfer and charge-storage phenomena.

2. Mechanistic classes

Several distinct microscopic realizations have been demonstrated. In monolayer MoS2_21/ITO nanocrystal hybrids, UV illumination at 2_22 nm excites the nanocrystals above their bandgap; the photo-generated valence-band holes in the nanocrystals are filled by MoS2_23 electrons, leaving stored electrons in the nanocrystals and effectively p-type photodoping the MoS2_24. In graphene/BN heterostructures, visible light excites defect states in BN; under gate bias, the photoexcited carriers move and the resulting fixed charge in BN screens the gate field and shifts the graphene doping. In monolayer MoS2_25 transistors on SiO2_26/Si, laser illumination plus back-gate bias creates persistent electron accumulation tied to the gate-insulator interface. In graphene NEMS, a focused laser and applied bias photoionize defect states in the heterostructure and oxide, and the trapped charge then shifts the mechanical charge-neutrality point 2_27. In dual-gated twisted WSe2_28 bilayers, optical pumping heats carriers in graphite gates, and holes are then thermionically injected across thick hBN into the moiré bilayer (Kriegel et al., 2018, Ju et al., 2014, Gadelha et al., 2020, Miller et al., 2019, Li et al., 23 Oct 2025).

Platform Photoactive reservoir Reported effect
Monolayer MoS2_29/ITO nanocrystals UV-excited ITO nanocrystals EIP=EvacEVBM,EEA=EvacECBM.E_\mathrm{IP} = E_\mathrm{vac} - E_\mathrm{VBM}, \qquad E_\mathrm{EA} = E_\mathrm{vac} - E_\mathrm{CBM}.0; EIP=EvacEVBM,EEA=EvacECBM.E_\mathrm{IP} = E_\mathrm{vac} - E_\mathrm{VBM}, \qquad E_\mathrm{EA} = E_\mathrm{vac} - E_\mathrm{CBM}.1 electrons per nanocrystal
Graphene/BN BN defect states under visible light and gate bias CNP pinning; writing and erasing of doping; EIP=EvacEVBM,EEA=EvacECBM.E_\mathrm{IP} = E_\mathrm{vac} - E_\mathrm{VBM}, \qquad E_\mathrm{EA} = E_\mathrm{vac} - E_\mathrm{CBM}.2-EIP=EvacEVBM,EEA=EvacECBM.E_\mathrm{IP} = E_\mathrm{vac} - E_\mathrm{VBM}, \qquad E_\mathrm{EA} = E_\mathrm{vac} - E_\mathrm{CBM}.3 junction formation
Monolayer MoSEIP=EvacEVBM,EEA=EvacECBM.E_\mathrm{IP} = E_\mathrm{vac} - E_\mathrm{VBM}, \qquad E_\mathrm{EA} = E_\mathrm{vac} - E_\mathrm{CBM}.4 on SiOEIP=EvacEVBM,EEA=EvacECBM.E_\mathrm{IP} = E_\mathrm{vac} - E_\mathrm{VBM}, \qquad E_\mathrm{EA} = E_\mathrm{vac} - E_\mathrm{CBM}.5/Si Gate-bias-assisted photodoping tied to gate-insulator interface EIP=EvacEVBM,EEA=EvacECBM.E_\mathrm{IP} = E_\mathrm{vac} - E_\mathrm{VBM}, \qquad E_\mathrm{EA} = E_\mathrm{vac} - E_\mathrm{CBM}.6 shifts from EIP=EvacEVBM,EEA=EvacECBM.E_\mathrm{IP} = E_\mathrm{vac} - E_\mathrm{VBM}, \qquad E_\mathrm{EA} = E_\mathrm{vac} - E_\mathrm{CBM}.7 V to EIP=EvacEVBM,EEA=EvacECBM.E_\mathrm{IP} = E_\mathrm{vac} - E_\mathrm{VBM}, \qquad E_\mathrm{EA} = E_\mathrm{vac} - E_\mathrm{CBM}.8 V
Graphene NEMS Trapped charge written by focused laser plus bias Persistent resonance-frequency tuning
Twisted WSeEIP=EvacEVBM,EEA=EvacECBM.E_\mathrm{IP} = E_\mathrm{vac} - E_\mathrm{VBM}, \qquad E_\mathrm{EA} = E_\mathrm{vac} - E_\mathrm{CBM}.9 bilayer Photo-thermionic hole injection from graphite gates Gate-voltage shifts of correlated-insulator signatures

These mechanisms separate naturally into four categories. One is defect-mediated photodoping, exemplified by graphene/BN and graphene on hydrophilic SiOα\alpha0/Si, where optically active states in a neighboring dielectric or interfacial environment trap charge and thereby dope graphene (Ju et al., 2014, Tiberj et al., 2013). A second is interfacial photoinduced charge transfer, as in MoSα\alpha1/ITO hybrids, where the illuminated remote layer stores charge capacitively and extracts carriers from the channel (Kriegel et al., 2018). A third is bias-assisted local trap writing, as in MoSα\alpha2 transistors and graphene NEMS, where illumination and an applied field together create long-lived trapped charge distributions (Gadelha et al., 2020, Miller et al., 2019). A fourth is photo-thermionic remote injection, where the photonic absorber and the doped quantum material are different layers, as in graphite-gated twisted WSeα\alpha3 (Li et al., 23 Oct 2025).

A distinct, more chemically mediated variant appears in exfoliated graphene under intense α\alpha4 nm, α\alpha5 ps laser pulses. There, the stable local increase in hole doping is accompanied by a reduction in compressive strain, and the proposed mechanisms are photo-induced oxygenation together with buckling or slippage of the graphene; the effect is local, stable for months, and reversible by solvent soaking (Alexeev et al., 2013).

3. Experimental signatures and quantitative observables

The observables used to identify photoinduced modulation doping depend on the channel. In n-type monolayer MoSα\alpha6, reduced electron density is diagnosed optically by a decrease in trion photoluminescence, an increase in neutral-exciton photoluminescence, a PL blue-shift, and a reduction of the exciton–trion energy separation toward the intrinsic value. Using a mass-action / Boltzmann equilibrium model for the exciton–trion–electron system, the MoSα\alpha7/ITO study extracted a carrier-density reduction of approximately α\alpha8 and a final carrier density of about α\alpha9, corresponding to roughly 3_30 electrons stored per nanocrystal and an attofarad-range capacitor-like response (Kriegel et al., 2018).

In monolayer MoS3_31 transistors, threshold-voltage shifts provide the carrier-density measure:

3_32

The reported threshold voltage moved from 3_33 V to 3_34 V, corresponding to 3_35. The decay was weak enough that after 3_36 hours the photodoping had barely decreased, and an exponential fit predicted that 3_37 of the initial photodoping remains long-term (Gadelha et al., 2020).

In graphene/BN devices, the fundamental transport signature is the anomalous 3_38 response under illumination: with light on and negative 3_39, resistance rises to the charge-neutrality point and becomes pinned, indicating that the BN charges screen the field. After controlled writing, the entire 3_30 curve shifts in the dark so that the CNP lands at the selected write voltage. The photo-induced charge density in BN acts like an effective built-in gate, consistent with the usual field-effect relation

3_31

A further signature of remote-dopant behavior is that the extracted electron mobility stays nearly constant over the full photo-doping range, while the charge-density fluctuations near the CNP change only weakly (Ju et al., 2014).

Raman spectroscopy is the main diagnostic in graphene systems exposed directly to light. On hydrophilic SiO3_32/Si, visible illumination at 3_33 nm reversibly tuned graphene from 3_34 at 3_35 mW to 3_36 at 3_37 mW, with the neutral point marked by maximum 3_38, maximum 3_39, and minimum 2_20; no D band appeared (Tiberj et al., 2013). Under intense picosecond irradiation, correlated Raman shifts of the G and 2D peaks were decomposed using the characteristic slopes 2_21 for uniaxial strain and 2_22 for hole doping above 2_23, leading to the conclusion that photoexcitation causes both increased p-doping and reduced compressive strain (Alexeev et al., 2013).

4. Spatial extent, locality, and persistence

Spatial behavior is strongly platform dependent. In the MoS2_24/ITO hybrid, the UV excitation was localized to about 2_25, yet photoluminescence changes proliferated tens of micrometers away; the largest reported extent was up to 2_26 from the excitation spot, and in some maps spectral-median shifts of up to 2_27 meV appeared over areas around 2_28. The authors attributed this to carrier redistribution in MoS2_29 driven by local band-structure variations, particularly in regions with lower initial PL energy, higher initial trion content, edges and grain boundaries, and strain or defect variations (Kriegel et al., 2018).

By contrast, the monolayer MoS5_50 transistor study emphasized locality. Local writing used a 5_51 nm laser with a spot size of about 5_52, and after illumination only the selected region showed photocurrent suppression in later SPCM images; the rest of the device remained photoactive. In that system, the photodoping was persistent but micrometer-scale and tied to the gate-insulator interface rather than to long-range redistribution across the flake (Gadelha et al., 2020).

Graphene/BN photodoping combined long persistence with spatial programmability. Charges in BN remained trapped for long times, so the induced doping persisted for days in the dark, while erasure by illumination near 5_53 usually required much higher light dose. Patterned illumination under negative gate bias wrote inhomogeneous doping profiles and produced a graphene 5_54-5_55 junction whose dark 5_56 trace exhibited two resistance peaks of similar height separated by about 5_57 V (Ju et al., 2014). In graphene on hydrophilic SiO5_58/Si, the response time was preliminarily evaluated to be less than 5_59 s, whereas suspended graphene remained neutral over the full power sweep, showing that the interface rather than the graphene alone sets the dynamical scale (Tiberj et al., 2013).

Graphene NEMS and twisted WSe2_20 bilayers introduce two additional time regimes. In NEMS, 2_21 saturated in about 2_22 ms, persistence extended for many days with a measured slow drift of about 2_23/hour after an initial small transient, and the spatial resolution was about 2_24, roughly the laser spot size (Miller et al., 2019). In graphite-gated twisted WSe2_25, injection occurs on sub-picosecond to picosecond timescales, the transient reflectance signal persists at a delay of 2_26s, pre-time-zero features indicate pulse-to-pulse accumulation, and the estimated electrostatic discharge time is 2_27 ms (Li et al., 23 Oct 2025).

5. Functional consequences and device implementations

The principal functional consequence of photoinduced modulation doping is that carrier density becomes an optically writable state variable. In monolayer MoS2_28/ITO, this enabled all-optical carrier-density control, contactless and quasi-permanent p-type photodoping, remote tuning over distances up to 2_29, and an energy-storage analogy in which the ITO nanocrystals act as optically charged nanocapacitors (Kriegel et al., 2018).

In graphene/BN heterostructures, the preserved high mobility and rewritable doping profiles allowed light-written and light-erased 2_20-2_21 junctions, suggesting photoresist-free photolithography of carrier-density landscapes (Ju et al., 2014). In graphene on hydrophilic SiO2_22/Si, the reversible ambipolar tuning under moderate laser power showed that visible light can function as an optical gate, but it also established that Raman spectroscopy is not always non-invasive when the electronic state is the quantity of interest (Tiberj et al., 2013).

In graphene NEMS, the modulation-doped state directly controlled mechanics through the effective voltage

2_23

The trapped charge produced a localized electrostatic strain and therefore a persistent frequency shift. Reported tuning ranged from about 2_24 MHz to 2_25 MHz in one device and from 2_26 MHz to 2_27 MHz in another, corresponding to nearly 2_28 tuning range. The write–erase cycle was repeated for 2_29 cycles with 2_200 repeatability (Miller et al., 2019).

In graphite-gated twisted WSe2_201, the functional effect was filling-factor control of correlated states. Photo-induced hole injection shifted the gate voltages at which optical signatures of correlated insulators appeared, generated persistent microsecond-scale signals consistent with charge accumulation, and produced photoinduced absorption interpreted as a transient move toward the 2_202 correlated-insulating regime (Li et al., 23 Oct 2025).

A recent graphene implementation in hBN/graphene/hBN/SiO2_203 heterostructures extended the device scope from carrier-density programming to in situ disorder control. Standard white light and a back-gate write voltage tuned the CNP over a 2_204 V range, corresponding to induced electron densities up to about 2_205. Drude and Landauer analyses showed mobility tunability between about 2_206 and 2_207 and a mean scattering length at 2_208 that could be tuned from about 2_209 to 2_210, enabling reversible switching between diffusive and quasi-ballistic transport regimes. The same procedure made quantum Hall isospin ferromagnetic states observable in a device whose initial quality would otherwise leave such states unobservable (Sanborn et al., 21 May 2026).

6. Misconceptions, limitations, and broader context

A recurring misconception is that photodoping is simply direct photoexcitation of the channel. The detailed literature shows otherwise. In MoS2_211/ITO, UV illumination is essential because it excites the ITO nanocrystals above their bandgap; without ITO nanocrystals, UV illumination does not produce the same MoS2_212 PL evolution (Kriegel et al., 2018). In monolayer MoS2_213 transistors, comparison between MoS2_214 on SiO2_215/Si and MoS2_216 on SiO2_217 glass showed that persistent photocurrent is not an intrinsic property of MoS2_218 alone but depends strongly on the gate-insulator interface (Gadelha et al., 2020). In graphene/BN, the BN-thickness dependence demonstrated that the relevant optical excitations occur in the BN bulk rather than merely in interfacial traps (Ju et al., 2014).

A second misconception is that all photoinduced modulation doping is either strictly local or necessarily long-range. The data show both limits. Micrometer-scale writing appears in SPCM studies of MoS2_219 and in graphene NEMS, whereas nonlocal redistribution over up to 2_220 occurs in MoS2_221/ITO hybrids (Gadelha et al., 2020, Miller et al., 2019, Kriegel et al., 2018). The distinction arises from the electronic landscape and the nature of the remote reservoir rather than from the optical trigger alone.

A third limitation is mechanistic specificity. In graphene on SiO2_222/Si, the effect requires hydrophilic substrates, is significantly affected by the substrate cleaning method, vanishes in suspended graphene, and was interpreted as a substrate-assisted electrochemical or photo-redox process whose detailed microscopic pathway remained unresolved (Tiberj et al., 2013). In picosecond-laser-modified graphene, the stable p-type change is likely tied to photo-induced oxygenation and buckling rather than to a purely electronic reservoir model (Alexeev et al., 2013). These cases indicate that “photoinduced modulation doping” covers both purely electrostatic remote charging and optically induced changes in the local chemical environment, provided the dopant function remains spatially separated from the main transport layer.

The broader non-photonic modulation-doping literature provides the equilibrium framework within which many photoinduced schemes can be interpreted. A single-layer 2_223-RuCl2_224 acceptor, for example, produces short-ranged lateral doping 2_225 nm, high homogeneity, and hole densities of 2_226 in monolayer graphene and 2_227 in bilayer graphene, while charge transfer through about 2_228 nm of hBN is reduced but not eliminated (Wang et al., 2020). First-principles screening then generalizes the design rule: type-III band alignment and large electron affinity are the central criteria for strong remote p-type doping in 2D electronics (Arora et al., 2024). A plausible implication is that future photoinduced modulation-doping platforms will increasingly combine these equilibrium band-engineering principles with optically addressable reservoirs, so that light controls not only nonequilibrium excitation but also the quasi-static carrier density landscape of low-dimensional devices.

Topic to Video (Beta)

No one has generated a video about this topic yet.

Whiteboard

No one has generated a whiteboard explanation for this topic yet.

Follow Topic

Get notified by email when new papers are published related to Photoinduced Modulation Doping.