Pocket-effect: Localized Impact in Diverse Fields

• Pocket-effect is a term describing localized substructures that alter overall system behavior in a context-dependent manner across various disciplines.
• In device physics, engineered pockets in TFETs sharpen tunneling junctions, thereby enhancing on-current performance and reducing subthreshold swing.
• Across fields like optics, protein docking, and cosmology, pocket-effect manifests as localized confinement, resonance, or selective activation that drives observable outcomes.

Pocket-effect is a context-dependent technical term rather than a single standardized concept. Across the literature, it denotes the system-level consequences of a localized “pocket,” but the physical meaning of that pocket varies sharply by field: a source-side doped region in a tunnel field-effect transistor, an optical-potential well behind a barrier, a small field-induced thermodynamic phase region, a protein binding cavity, a $^{13}$C-rich layer in an AGB star, or a pocket universe in eternal inflation. The most explicit device-physics usage appears in source-pocket TFETs, where the pocket reshapes electrostatics at the tunneling junction and enhances band-to-band tunneling (Fariha et al., 10 Jun 2025). In other domains, the same word marks localized confinement, activation, or selection effects with different observables, including oscillatory cross sections, ARPES contour topology, docking success, $s$-process yields, and measure-weighted cosmological histories (Lee et al., 2021).

1. Terminological scope

Across the cited literature, “pocket” refers to a spatially or functionally localized substructure whose presence alters global behavior. A plausible cross-disciplinary implication is that the term marks regions with disproportionate dynamical leverage, but the operative mechanism remains field-specific.

Domain	Pocket definition	Principal consequence
TFETs (Fariha et al., 10 Jun 2025, Liu et al., 2013, Bishnoi et al., 2014)	Source-side doped region near tunneling junction	Higher $I_{ON}$, altered subthreshold swing, ambipolar suppression or resonance
Low-energy scattering (Lee et al., 2021)	Optical-potential pocket behind a barrier	Narrow oscillatory annihilation and elastic resonances
Cuprate ARPES (Cao, 2010)	Small nodal closed zero-energy contour	Fermi pocket rather than Fermi arc
Field-induced order (Kittaka et al., 2024)	Small thermodynamic phase region in the $B$–$T$ diagram	Double anomalies in specific heat and entropy signatures
Protein science and ML (Balytskyi et al., 4 Feb 2025, Gao et al., 2022, Zhang et al., 2023, Lin et al., 27 Jan 2026, Wei et al., 2 Jun 2026, Mendez-Otalvaro et al., 25 Aug 2025)	Ligand-binding or annotated functional site	Altered docking, affinity prediction, allostery, or generative design
Stellar and cosmological theory (Bisterzo et al., 2010, Bisterzo et al., 2017, Lehners, 2012)	$^{13}$C-rich layer or pocket universe	Modified $s$-process yields or measure-biased cosmological histories

This range is important because identical vocabulary can conceal non-equivalent mechanisms. In semiconductor transport, the pocket is deliberately engineered. In scattering and condensed matter, it may be an emergent region of the potential or phase diagram. In protein modeling, it may be an annotated prior, a binding cavity, or a design target. In astrophysics and cosmology, it is part of the underlying physical environment rather than a fabricated perturbation.

2. Source-pocket engineering in tunnel transistors

In TFET research, pocket-effect most directly denotes source-side doping engineering that modifies the tunneling barrier. The silicon-based homojunction DMDG-SP TFET uses the sequence $p^+$ source, $n^+$ pocket, $p$ channel, $s$0 drain, with the pocket inserted between source and channel so that tunneling proceeds effectively as source $s$1 pocket $s$2 channel rather than source $s$3 channel directly. In the optimized structure, the pocket length is $s$4 nm, the pocket doping is $s$5, the source is doped at $s$6, the homogeneous dielectric is HfO$s$7, and the device is analyzed in Silvaco Atlas 2-D TCAD using nonlocal BTBT, bandgap narrowing, field- and impurity-dependent mobility, and SRH and Auger recombination. The pocket produces a local minimum in $s$8 and a stronger lateral electric field. Relative to a pocket-less DMDG TFET, it yields a $s$9 higher ON current and a $I_{ON}$0 lower subthreshold swing. In pocket-based devices, dual-material gates further increase ON current by $I_{ON}$1 and ON/OFF ratio by $I_{ON}$2. After optimization, the reported metrics are $I_{ON}$3, $I_{ON}$4, $I_{ON}$5, $I_{ON}$6 mV/decade, a tunneling barrier width of $I_{ON}$7 nm, and a peak electric field of $I_{ON}$8 MV/cm (Fariha et al., 10 Jun 2025).

The graphene nanoribbon TFET literature uses the same structural logic in a quantum-transport setting. There, an additional heavily doped $I_{ON}$9-type pocket is placed on the source side of an armchair GNR TFET, converting a $B$0–$B$1–$B$2 device into a $B$3–$B$4–$B$5–$B$6 device. The pocket doping density is $B$7 dopant/atom, and pocket length $B$8 is varied from $B$9 to $T$0 nm. The ballistic 3D real-space NEGF simulation shows that the pocket thickens the effective barrier in the off-state but creates a thinner local tunneling barrier in the on-state. At $T$1 V, the on-current increases by about one order of magnitude, the subthreshold swing improves from $T$2 to $T$3 mV/dec, and ambipolar behavior is suppressed as pocket length increases. For longer pockets, resonance in the pocket-induced quantum well produces negative transconductance at negative gate voltage (Liu et al., 2013).

A heterojunction realization appears in the $T$4-Al$T$5Ga$T$6Sb pocket-implanted L-shaped GaSb/InAs vertical $T$7-TFET. Here the pocket is inserted in the source injector region, and the L-shaped geometry aligns the gate electric field with the tunnel-junction internal field. Using a 3D full-band atomistic $T$8 spin-orbit-coupled tight-binding simulator, the study reports an order-of-magnitude increase in $T$9 from the pocket implant. For the optimized pocket-implanted structure at $^{13}$0 V and $^{13}$1 V, the reported values are $^{13}$2, $^{13}$3, and $^{13}$4 mV/dec. The effect depends strongly on undercut geometry, oxide thickness, dielectric constant, and drain thickness (Bishnoi et al., 2014).

Taken together, these TFET studies support a common device-level reading of pocket-effect: the pocket intensifies the local field at the desired tunneling junction, narrows the barrier, and makes the BTBT rate more gate-sensitive. In the silicon study, this is summarized qualitatively as higher field $^{13}$5 narrower tunneling barrier $^{13}$6 larger BTBT rate, with the standard subthreshold-swing definition written as $^{13}$7 (Fariha et al., 10 Jun 2025).

3. Resonant, spectral, and phase-pocket phenomena

Outside device engineering, pocket often refers to a confining region of a potential or phase diagram. In very low-energy antineutron–nucleus reactions, the optical-model analysis identifies “pocket resonances” arising from quasi-bound states inside an optical-potential pocket. For heavy nuclei such as Pb and Ag, annihilation and elastic cross sections in the $^{13}$8–$^{13}$9 MeV region exhibit small-amplitude oscillatory structures with narrow periodicity in $s$0. The semiclassical decomposition $s$1 separates a barrier-reflected wave from an internal pocket wave, and the oscillations originate in the interference term $s$2. Resonance maxima satisfy $s$3, equivalently $s$4, which is the quasi-bound quantization condition inside the pocket. The effect depends sensitively on the absorptive width $s$5; for $s$6Pb, an onset estimate $s$7 MeV is compared with a fitted $s$8 MeV, so the oscillations are weakened but remain visible (Lee et al., 2021).

A thermodynamic usage appears in PrIr$s$9Zn$p^+$0, where a magnetic-field-induced “phase pocket” denotes a small A-phase region near the antiferro-quadrupole boundary. For $p^+$1, specific-heat measurements show coexistence of two anomalies in roughly the $p^+$2–$p^+$3 T range, with the higher-temperature anomaly most visible near $p^+$4 T. Rotational magnetocaloric measurements at $p^+$5 T and $p^+$6 K identify a plateau-like entropy feature for $p^+$7, consistent with the same pocket. Entering the A phase from the NFL state reduces entropy by about $p^+$8, whereas the transition from A phase to AFQ reduces entropy by about $p^+$9. The pocket is strongly anisotropic and is absent for $n^+$0 up to $n^+$1 T (Kittaka et al., 2024).

A related but conceptually distinct pocket construct occurs in cuprate ARPES. There the “Fermi pocket” is interpreted not as a conventional reconstructed Fermi surface alone, but as a zero-excitation contour influenced by free-electron-like final states and by pairing away from the Fermi level. The analysis uses an excitation condition $n^+$2. In the zero-order argument, if the work function satisfies $n^+$3, two nodal zero-crossings can occur and close into a Fermi pocket; if $n^+$4, only a Fermi arc remains. The contrast between Bi2201, with valence bandwidth about $n^+$5 eV and work function about $n^+$6 eV, and Bi2212, with about $n^+$7 eV and $n^+$8 eV, is used as the main illustration (Cao, 2010).

These examples show that pocket-effect need not imply intentional insertion of a material region. It can instead denote the observable consequences of confinement, phase-space restriction, or contour closure.

4. Protein pockets in docking, affinity prediction, and generative design

In computational structural biology, pocket-effect often means that pocket identification or pocket conditioning materially changes downstream task performance. RAPID-Net formulates the central issue as binding-site-agnostic docking: pocket predictors should not be judged only by cavity geometry, but by whether they guide docking successfully. With AutoDock Vina on PoseBusters, RAPID-Net-guided docking achieves $n^+$9 RMSD-correct top-1 predictions and $p$0 PB-success, compared with $p$1 and $p$2 for DiffBindFR. When evaluated as “at least one correct pose in the ensemble,” RAPID-Net-guided Vina reaches $p$3. On Coach420 and BU48, the RAPID ensemble reports pocket-ligand intersection values of $p$4 and $p$5, respectively, surpassing Kalasanty and PUResNet. In this usage, the pocket-effect is the practical impact of pocket prediction quality on docking accuracy and search efficiency (Balytskyi et al., 4 Feb 2025).

HonestAffinity uses the phrase in a more formal statistical sense. Its pocket prior is a binary residue-level pocket-position marker $p$6 injected directly into the residue embedding stream through $p$7. The central result is a split-conditioned reversal. HonestAffinity-Pocket is best on familiar or canonical settings, with Pearson $p$8 on validation, $p$9 on CASF-2016, and $s$00 on the CASF non-train subset. Under strict LP-PDBBind no-leak evaluation, however, HonestAffinity-Pocket-NoESM is best on all three tiers, with Pearson $s$01, $s$02, and $s$03 on test_cl1, test_cl2, and test_cl3. The paper interprets this as a distribution-specific prior: pocket annotation helps when test targets resemble training pockets, but can hurt when similarity leakage is removed (Wei et al., 2 Jun 2026).

CoSP addresses pocket-ligand interaction at the representation-learning level. It jointly pretrains pocket and ligand encoders using a gated geometric message passing layer that includes chemical features, 3D coordinates, and orientation vectors, and it injects binding knowledge through contrastive learning with chemical-similarity-enhanced negative sampling. The reported pocket-matching results on ProSPECCTs include D1 $s$04, D1.2 $s$05, D2 $s$06, D3 $s$07, D4 $s$08, and D7 $s$09. In virtual screening on DUD-E, CoSP reports AUC-ROC $s$10, with $s$11, $s$12, $s$13, and $s$14 enrichment scores of $s$15, $s$16, $s$17, and $s$18, respectively (Gao et al., 2022).

Generative design work turns the pocket from annotation into design target. FAIR defines the pocket as residues with atoms within $s$19 Å of any ligand atom and co-designs pocket residue types, backbone coordinates, side-chain atoms, and ligand structure through coarse-to-fine full-shot iterative refinement. On CrossDocked, it reports AAR $s$20, RMSD $s$21, and Vina $s$22; on Binding MOAD, AAR $s$23, RMSD $s$24, and Vina $s$25. The paper states average improvements of $s$26 in AAR and $s$27 in RMSD and a runtime of $s$28 s versus $s$29 s for DEPACT (Zhang et al., 2023). EnzyPGM makes pocket-substrate interaction explicit in enzyme design through Residue Function Fusion and Residue-atom Bi-scale Attention. On EnzyPock, it reports AAR $s$30, Vina $s$31, pLDDT $s$32, scRMSD $s$33, and scTM $s$34, while reducing average binding energy by $s$35 kcal/mol relative to EnzyGen (Lin et al., 27 Jan 2026).

5. Pocket-mediated activation and nucleosynthetic pockets

In ion-channel biophysics, the relevant construct is the modulator pocket of K2P channels. This cryptic ligand-binding site was first identified in TREK1 and lies adjacent to SF1 at the P1–M4 interface, exposed to both membrane and aqueous environment. The review describes a largely amphipathic architecture with a hydrophobic lower region and a more polar upper region. Ligands such as ML335 and ML402 bind the TREK1 pocket in an L-shaped conformation and stabilize the conductive selectivity filter, reducing C-type inactivation and suppressing the S3 carbonyl-flipping inactive state. Gain-of-function mutations in the same region, including G137I and W275S in TREK1 and G124I and W262S in TRAAK, support the same pocket-mediated gating logic (Mendez-Otalvaro et al., 25 Aug 2025).

In stellar astrophysics, the $s$36C-pocket is a thin layer formed after third dredge-up when a few protons from the H-rich envelope partially mix into the He-rich intershell and react through

$s$37

At interpulse temperatures of about $s$38 K, radiative burning through

$s$39

provides the main neutron source for the $s$40-process. In the $s$41, $s$42 model, the hydrogen profile inside the pocket controls the $s$43C/$s$44N balance and hence the neutron economy. Standard three-zone pockets span $s$45 to $s$46, and an added zone 4 of mass $s$47 is used to mimic an extended tail. For standard pockets, $s$48 ranges from $s$49 to $s$50, $s$51 from $s$52 to $s$53, and $s$54 from $s$55 to $s$56. Adding a tail with $s$57 lowers the maximum $s$58 to about $s$59 dex and the maximum $s$60 to about $s$61 dex (Bisterzo et al., 2010).

Galactic chemical evolution work treats the same uncertainty at population scale. A standard three-zone $s$62C pocket has total mass $s$63, while other cases test larger pockets, additional $s$64N-rich zones, or a smoother internal $s$65C tail. With an appropriate weighted average of pocket strengths, the model reproduces solar $s$66Sm within about $s$67 and the $s$68-only isotopes with $s$69 within $s$70. Isotopes in the interval $s$71, however, remain underproduced by about $s$72. Fast-rotating massive stars add about $s$73 to solar Sr, $s$74 to solar Y, and $s$75 to solar Zr, but do not eliminate the heavier-element discrepancy (Bisterzo et al., 2017).

In these biophysical and astrophysical settings, the pocket is not primarily a tunneling junction or a docking grid. It is a localized structural region that controls activation pathways, neutron production, or abundance redistribution.

6. Pocket universes and cross-disciplinary interpretation

Cosmology uses the term in yet another sense. In eternal inflation, pocket universes are bubble regions produced by vacuum transitions, and the relevant effect is measure-induced selection among different pocket histories. The light-cone time cut-off and causal diamond measure favor pockets that begin with a small initial Hubble rate rather than high-scale inflationary pockets. For a non-cyclic emergent model, the paper writes

$s$76

so the relative weighting depends on the tunneling rate from the master vacuum into a small-$s$77 pocket compared with that into an inflationary pocket. Cyclic pockets receive additional enhancement because they can be habitable repeatedly, and later cycles dominate over the first when $s$78 (Lehners, 2012).

This cosmological usage underscores the semantic breadth of pocket-effect. In some fields, the pocket is engineered and its consequences are evaluated in terms of current, slope, or docking accuracy. In others, the pocket is an emergent feature of a potential, a phase diagram, a stellar shell, or a multiverse landscape. This suggests a family resemblance rather than a strict definition: the pocket is a localized region that concentrates fields, states, interactions, neutrons, entropy changes, or probability currents strongly enough to alter macroscopic observables. The term nevertheless remains irreducibly context-bound, and precise interpretation requires the surrounding model, geometry, and observable set.