Electron beam shaping enables control over the electron wavefront—its phase and amplitude—to generate structured beams and enhance imaging, spectroscopy, and aberration correction in transmission electron microscopy (TEM). This is achieved using phase plates, which modulate the phase of electrons at specific planes of the microscope. Among various implementations (thin-film, magnetic, photonic, and holographic), electrostatic phase plates are especially promising because they offer voltage-tunable, low-loss phase control with minimal scattering material in the beam path. Their operation relies on the Aharonov–Bohm effect: an electron acquires a phase proportional to the electrostatic potential it experiences, allowing a designed potential distribution to sculpt the beam wavefront. Modern electrostatic phase plates are MEMS-based devices fabricated on silicon chips and mounted on biasing TEM holders. However, their design faces two main challenges: (i) limited numbers of biasing channels in commercial holders restrict the achievable potential profiles, and (ii) complex electrode geometries are difficult to fabricate near the active region. To overcome these limits, this work introduces Current-Controlled Electrostatic Potential Phase Plates, a novel concept that achieves continuous potential gradients using simple, low-channel architectures. In this design, boundary electrodes are fabricated from resistive doped silicon and connected through high-conductivity gold layers. By driving a controlled current along these electrodes, a linear potential drop is established, reproducing complex boundary gradients with minimal independent biases. This approach enables compact, tunable, and reconfigurable devices. Using gradient-based optimization, we designed current-controlled vortex beam generators, orbital angular momentum (OAM) sorters, and electrostatic multipoles (quadrupole and hexapole) for aberration compensation. The hexapole phase plate demonstrates bias-driven phase rotation, allowing alignment to any A3 orientation with a single device—an advantage over conventional multipole correctors that require complex mechanical realignment. Finite-element electrostatic simulations and wave-optical propagation confirm the desired phase profiles, and prototype MEMS chips have been successfully fabricated and characterized. To complement hardware development, we implemented an AI-assisted optimization and control pipeline. A deep neural network (DNN) analyzes a single focused STEM probe image to estimate the orientation of twofold (A2) and threefold (A3) astigmatism within milliseconds. Integrated into a Python-based graphical interface, the DNN drives a feedback loop that adjusts the phase-plate biases via a custom PCB controller, enabling automatic aberration correction without switching to conventional Ronchigram or Zemlin procedures. The project’s main contributions include: • Definition and demonstration of Current-Controlled Electrostatic Potential Phase Plates, reducing required bias channels while increasing design flexibility. • Gradient-based optimization linking target phase maps to boundary potentials. • MEMS fabrication and characterization of vortex, OAM-sorter, and hexapole phase plates. • A DNN-based probe-shape method for real-time A2/A3 orientation estimation and automated feedback control. In summary, this work establishes current-controlled electrostatic phase plates as a compact, low-loss platform for programmable wavefront control in TEM. Combining electrostatic design, AI-assisted optimization, and machine learning–based feedback, it opens a path toward adaptive beam shaping and real-time aberration correction for next-generation electron microscopy.
La strutturazione del fascio elettronico consente di controllare il fronte d’onda — in fase e ampiezza — per generare fasci strutturati e migliorare imaging, spettroscopia e correzione delle aberrazioni nella microscopia elettronica a trasmissione (TEM). Ciò è ottenuto tramite piatti di fase, che modulano la fase degli elettroni in piani specifici del microscopio. Tra le varie implementazioni (a film sottile, magnetiche, fotoniche e olografiche), i piatti di fase elettrostatici risultano particolarmente promettenti, poiché offrono un controllo di fase regolabile in tensione, a bassa perdita e con minimo materiale diffusivo nel cammino del fascio. Il loro funzionamento si basa sull’effetto Aharonov–Bohm: un elettrone acquisisce una fase proporzionale al potenziale elettrostatico attraversato, consentendo di modellare il fronte d’onda attraverso una distribuzione di potenziale progettata. Le moderne piastre di fase elettrostatiche sono dispositivi MEMS su chip di silicio montati su portacampioni con contatti elettrici per TEM. Tuttavia, la progettazione è limitata da (i) pochi canali di biasing nei portacampioni commerciali, che riducono i profili di potenziale realizzabili, e (ii) difficoltà di fabbricazione di elettrodi complessi vicino alla regione attiva. Per superare questi limiti, questo lavoro introduce i Piatti di Fase Elettrostatici a Potenziale Controllato da Corrente, un concetto innovativo che genera gradienti di potenziale continui con architetture semplici e pochi canali. Gli elettrodi di bordo, in silicio drogato resistivo e collegati tramite strati d’oro ad alta conducibilità, creano una caduta di potenziale lineare applicando una corrente controllata, riproducendo gradienti complessi con un numero minimo di potenziali variabili indipendenti. Questo approccio consente dispositivi compatti, sintonizzabili e riconfigurabili. Attraverso ottimizzazione basata su gradiente, sono stati progettati generatori di fasci a vortice, sorter di momento angolare orbitale (OAM) e multipoli elettrostatici (quadrupolo ed esapolo) per la compensazione delle aberrazioni. I piatti di fase esapolari mostrano rotazione di fase dipendente dal potenziale, consentendo l’allineamento di A₃ a qualunque orientazione con un singolo dispositivo, un vantaggio rispetto ai correttori multipolari convenzionali che richiedono riallineamenti meccanici complessi. Le simulazioni elettrostatiche FEM e la propagazione ottica d’onda confermano i profili di fase desiderati, e il primi prototipi di chip MEMS sono stati fabbricati e caratterizzati con successo. Parallelamente, è stata sviluppata una pipeline di ottimizzazione e controllo assistita da IA. Una rete neurale profonda (DNN) analizza un’unica immagine STEM focalizzata per stimare in millisecondi l’orientamento degli astigmatismi al secondo (A₂) e terzo (A₃). Integrata in un’interfaccia Python, la rete guida un ciclo di retroazione che regola i potenziali del piatto di fase tramite un controller PCB dedicato, consentendo la correzione automatica delle aberrazioni senza ricorrere a procedure Ronchigram o Zemlin. In sintesi, questo lavoro dimostra che i piatti di fase elettrostatici controllati in corrente costituiscono una piattaforma compatta e a bassa perdita per il controllo programmabile del fronte d’onda in TEM. Combinando progettazione elettrostatica, ottimizzazione basata su IA e controllo mediante apprendimento automatico, apre la strada alla strutturazione del fascio elettronico e alla correzione in tempo reale delle aberrazioni nella microscopia elettronica di nuova generazione.
Strutturazione del fascio elettronico con un piatto di fase regolabile
HABIBZADEH KAVKANI, PAYAM
2026
Abstract
Electron beam shaping enables control over the electron wavefront—its phase and amplitude—to generate structured beams and enhance imaging, spectroscopy, and aberration correction in transmission electron microscopy (TEM). This is achieved using phase plates, which modulate the phase of electrons at specific planes of the microscope. Among various implementations (thin-film, magnetic, photonic, and holographic), electrostatic phase plates are especially promising because they offer voltage-tunable, low-loss phase control with minimal scattering material in the beam path. Their operation relies on the Aharonov–Bohm effect: an electron acquires a phase proportional to the electrostatic potential it experiences, allowing a designed potential distribution to sculpt the beam wavefront. Modern electrostatic phase plates are MEMS-based devices fabricated on silicon chips and mounted on biasing TEM holders. However, their design faces two main challenges: (i) limited numbers of biasing channels in commercial holders restrict the achievable potential profiles, and (ii) complex electrode geometries are difficult to fabricate near the active region. To overcome these limits, this work introduces Current-Controlled Electrostatic Potential Phase Plates, a novel concept that achieves continuous potential gradients using simple, low-channel architectures. In this design, boundary electrodes are fabricated from resistive doped silicon and connected through high-conductivity gold layers. By driving a controlled current along these electrodes, a linear potential drop is established, reproducing complex boundary gradients with minimal independent biases. This approach enables compact, tunable, and reconfigurable devices. Using gradient-based optimization, we designed current-controlled vortex beam generators, orbital angular momentum (OAM) sorters, and electrostatic multipoles (quadrupole and hexapole) for aberration compensation. The hexapole phase plate demonstrates bias-driven phase rotation, allowing alignment to any A3 orientation with a single device—an advantage over conventional multipole correctors that require complex mechanical realignment. Finite-element electrostatic simulations and wave-optical propagation confirm the desired phase profiles, and prototype MEMS chips have been successfully fabricated and characterized. To complement hardware development, we implemented an AI-assisted optimization and control pipeline. A deep neural network (DNN) analyzes a single focused STEM probe image to estimate the orientation of twofold (A2) and threefold (A3) astigmatism within milliseconds. Integrated into a Python-based graphical interface, the DNN drives a feedback loop that adjusts the phase-plate biases via a custom PCB controller, enabling automatic aberration correction without switching to conventional Ronchigram or Zemlin procedures. The project’s main contributions include: • Definition and demonstration of Current-Controlled Electrostatic Potential Phase Plates, reducing required bias channels while increasing design flexibility. • Gradient-based optimization linking target phase maps to boundary potentials. • MEMS fabrication and characterization of vortex, OAM-sorter, and hexapole phase plates. • A DNN-based probe-shape method for real-time A2/A3 orientation estimation and automated feedback control. In summary, this work establishes current-controlled electrostatic phase plates as a compact, low-loss platform for programmable wavefront control in TEM. Combining electrostatic design, AI-assisted optimization, and machine learning–based feedback, it opens a path toward adaptive beam shaping and real-time aberration correction for next-generation electron microscopy.| File | Dimensione | Formato | |
|---|---|---|---|
|
HabibzadehKavkani.pdf
accesso aperto
Licenza:
Tutti i diritti riservati
Dimensione
262.27 MB
Formato
Adobe PDF
|
262.27 MB | Adobe PDF | Visualizza/Apri |
I documenti in UNITESI sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/20.500.14242/358386
URN:NBN:IT:UNIMORE-358386