TEM/STEM Training

4D-STEM

Nov 10, 2025

Check ceta not inserted – TEM Balnker Monitor
Press the “insert” button for the Arena detector handpad
Unblank the beam from Velox
Zoom in a bit 20,000 x
Convergenace angle (30 mrad)
Turn on “Descan”
Make sure “interna” button is pressed
In TEM User Interface  Beam Settings  Probe  Normal  Check (MF/Y Convergence angle), probe convergence angle larger using multi-function angle X and Y (Higher)
Choose the aperture 0 10 Bul for C2
Firefox – initialize (if the green button is already on)
TVIPS EMpilitied – scan tap  Filenaem, folder name insert
Dectris – Survey tab
Make sure you have the EDX SCAN turned on
Use 50 micro s, is good.

STEM Lens System Overview

Understanding the STEM column and crossovers

What is a crossover? A crossover is where the electron beam focuses to a point (focal point), then diverges again. Smaller crossover = More concentrated beam = Higher current density.

COMPLETE STEM COLUMN WITH CROSSOVERS
====================================

ELECTRON GUN (≈200 kV)
     ║
    ║║║   ← Emission from filament or field emitter tip
    \│/
     •    ← 1st CROSSOVER (gun crossover)
    /|\
   ↓ ↓ ↓
┌────────────────┐
│   C1 LENS      │   ← Demagnifies gun crossover
└────────────────┘
    \│/
     •    ← 2nd CROSSOVER (after C1; smaller & brighter)
    /|\
   ↓ ↓ ↓
══════════════════════  C2 APERTURE (50–150 μm)
     ↑               Limits beam current / outer rays
┌────────────────┐
│   C2 LENS      │   ← Controls convergence semi-angle (α)
└────────────────┘
       ↓
      \│/       ← converging rays
       ↓
══════════════════════  SAMPLE (thin, ~50–100 nm)
       ↑ beam converges *to* this point (focus; 3rd crossover)
       ↓ transmitted & scattered electrons *diverge* afterward
      /│\       ← diverging rays
       ↓
┌────────────────┐
│  OBJECTIVE     │
│     LENS       │   ← Collects divergent electrons
└────────────────┘
       ↓
       ●   ← BACK FOCAL PLANE (Diffraction pattern forms here)
       ↓
══════════════════════  OBJECTIVE APERTURE
       ↓               (selects diffraction angles/spots)
┌────────────────┐
│ PROJECTOR      │
│   LENSES       │   ← Magnify:
│                │       - Diffraction pattern (diffraction mode)
│                │       - Real-space image (image mode)
└────────────────┘
       ↓
       ●   ← IMAGE PLANE (real-space image forms here)
       ↓
DETECTOR PLANE (HAADF, BF, DF, etc.)

Controlling the STEM Beam

How to control beam intensity, convergence angle, and focal length

STEM BEAM CONTROL PARAMETERS
=============================

1. BEAM INTENSITY (Current at Sample)
──────────────────────────────────────

Control via:
• Extraction voltage (gun)       → Higher voltage = more electrons emitted
• Spot size setting              → Larger spot = more current
• C2 aperture size               → Larger aperture = more electrons pass

Effect on sample:
┌─────────────┐              ┌─────────────┐
│ LOW CURRENT │              │ HIGH CURRENT│
│   (nA)      │              │   (μA)      │
└─────────────┘              └─────────────┘
     ↓                            ↓
  Less beam                   More beam
  damage                      damage
     ↓                            ↓
  Sample survives              Better signal
  longer                       but faster damage


2. CONVERGENCE ANGLE (α)
──────────────────────────────────────

Control via:
• C2 aperture size              → Larger aperture = larger α
• C2 lens strength              → Stronger lens = larger α

SMALL α (5 mrad):                LARGE α (30 mrad):

C2 Aperture (50 μm)              C2 Aperture (150 μm)
══════════════════               ══════════════════
     \│/                              ╲ │ ╱
      •   ← small probe                 •   ← tiny probe
     /│\    large depth                ╱│╲    small depth
    / │ \   of focus                  ╱ │ ╲   of focus
 ════════════                      ════════════
   Sample                            Sample

α = 5 mrad:                      α = 30 mrad:
• Larger probe (~1-2 nm)         • Smaller probe (~0.1 Å)
• Larger depth of focus          • Atomic resolution
• Good for thick samples         • Needs thin samples
• Less sensitive to focus        • Very sensitive to focus


What is "depth of focus"?
─────────────────────────

Depth of focus = the range of sample heights where the probe
stays reasonably focused (sharp).

SMALL α (large depth of focus):

     \│/
      │  ← probe focused here
     /│\
    / │ \
   /  │  \    ← Still reasonably focused over
  /   │   \      large z-range (~100-200 nm)
 /    │    \
──────────────── Top of sample
      │
      │         ← Probe stays relatively sharp
      │            throughout thick sample
──────────────── Bottom of sample

Result: Sample height variations don't blur image much.
        Good for thick/bent samples.


LARGE α (small depth of focus):

      ╲│╱
       •   ← probe ONLY focused at this exact height
      ╱│╲
     ╱ │ ╲   ← Rapidly defocuses (~10-20 nm range)
    ╱  │  ╲
   ╱   │   ╲
──────────────── Top of sample (blurry if not at focus)
       │
       •   ← Only sharp at exact focal plane
       │
──────────────── Bottom of sample (blurry if not at focus)

Result: Must focus precisely at sample surface.
        Sample must be very flat and thin.
        Any height variation = blur!


Mathematical relation:
Depth of focus ∝ probe size / α

Example:
• α = 5 mrad, probe = 2 nm   → depth ≈ 200 nm (forgiving)
• α = 30 mrad, probe = 0.1 nm → depth ≈ 10 nm (critical!)


3. FOCAL LENGTH (Working Distance)
──────────────────────────────────────

Control via:
• C2 lens current                → Higher current = shorter focal length

Why does higher current = shorter focal length?

Electromagnetic lenses work like optical lenses:
• Higher current → Stronger magnetic field
• Stronger field → Bends electron rays more sharply
• Sharper bending → Focus happens sooner (closer to lens)

Analogy: Like going from a weak eyeglass lens to a strong one.

WEAK C2 (low current):           STRONG C2 (high current):

┌────────────────┐               ┌────────────────┐
│   C2 LENS      │               │   C2 LENS      │
│ Low current    │               │ High current   │
│ Weak B-field   │               │ Strong B-field │
└────────────────┘               └────────────────┘
      \│/                              \│/
       \│/  ← gentle bending            ╲│╱  ← sharp bending
        \│/                              •   ← focus NEAR lens
         \│/                            /|\     (short focal length)
          •   ← focus FAR from lens    / │ \
         /|\     (long focal length)  /  │  \
        / │ \                        /   │   \
════════════════                  ════════════════
   Sample (out of focus)            Sample (too close,
   Need to move sample UP           also out of focus)
   or increase current


PROPER FOCUS:

┌────────────────┐
│   C2 LENS      │
│ Correct current│  ← Adjusted so focal point
└────────────────┘     lands exactly at sample
      \│/
       \│/
        ↓
        •   ← focus ON sample surface
       /|\
      / │ \
════════════════
   Sample (sharp!)


Practical tips:
• Sample too blurry? → Increase C2 current (stronger lens)
• Over-focused? → Decrease C2 current (weaker lens)
• Focal length formula: f ≈ 1/(μ₀ n I) where I = current


4. PROBE SIZE vs CURRENT vs α (The Trade-off)
──────────────────────────────────────────────

You cannot have ALL at once:
• Small probe (high resolution)
• High current (good signal)
• Small α (large depth of focus)

Choose your compromise:

┌──────────────────┬──────────┬──────────┬────────┐
│ Application      │ Probe    │ Current  │   α    │
├──────────────────┼──────────┼──────────┼────────┤
│ Atomic STEM      │ 0.1 Å    │ Low nA   │ 30 mrad│
│ imaging          │          │          │        │
├──────────────────┼──────────┼──────────┼────────┤
│ 4D-STEM          │ 1-2 Å    │ Med nA   │ 10 mrad│
│ (ptychography)   │          │          │        │
├──────────────────┼──────────┼──────────┼────────┤
│ EDS mapping      │ 1-5 nm   │ High nA  │ 5 mrad │
│                  │          │          │        │
├──────────────────┼──────────┼──────────┼────────┤
│ EELS             │ 0.5-1 Å  │ Low nA   │ 20 mrad│
│ (core-loss)      │          │          │        │
└──────────────────┴──────────┴──────────┴────────┘


5. PRACTICAL CONTROLS ON THE MICROSCOPE
────────────────────────────────────────

In TEM User Interface:

Beam Settings → Probe:
• Spot size (1-11)           → Controls extraction voltage & lens
• α (MF/Y knobs)             → Multifunction X/Y adjusts C2 strength
• C2 aperture position       → Insert/select aperture size

In practice:
• Start with spot size 3-5 (balance)
• Choose C2 aperture (50, 70, 100, 150 μm)
• Adjust C2 lens (α) using MF knobs
• Focus using Z-height or objective lens fine control

EDS and EELS Analysis

EDS: Energy Dispersive X-ray Spectroscopy

Introduction

Energy Dispersive X-ray Spectroscopy (EDS) is a powerful analytical technique for determining the elemental composition of materials in the transmission electron microscope (TEM). It provides quantitative and qualitative analysis of elements present in your sample.

Setup and Hardware

EDS uses 4 detectors positioned around the sample (shown under TEM User Interface → EDX User). These detectors:

Maximize collection efficiency by capturing X-rays from multiple angles
Enable better statistics and faster acquisition
Reduce shadowing effects from sample holders
Allow analysis even if one detector is shadowed

Working Principles

EDS identifies elements and their relative abundances by detecting characteristic X-rays emitted when an electron beam bombards the sample. The energy and intensity of these X-rays reveal the elemental composition.

When the electron beam ejects an inner-shell electron, a higher-energy electron fills the vacancy, releasing an X-ray photon. The photon energy equals the difference between the two shells. Since each element has unique energy levels, the emitted X-rays have characteristic energies for element identification.

Sample Considerations

Beam damage

Is inner-shell electron ejection considered beam damage?: Yes, but it’s less severe than sputtering (atom removal) or heating (thermal degradation). Inner-shell ejection only affects the electronic structure of individual atoms without causing physical changes to the sample. The electronic configuration quickly restabilizes as higher-energy electrons fill the vacancies.

Sample preparation

What sample preparation is needed for EDS?

Sample must be thin enough for electron transmission (typically < 100 nm)
Clean surface to avoid contamination
Conductive coating may be needed for insulating samples
Carbon support films are common but contribute background signal

Common artifacts

What are common artifacts to watch for?

Peak overlap between elements
Absorption effects in thick samples
Fluorescence from neighboring areas
Contamination build-up during analysis
Detector dead time at high count rates

Detection limits

What are the detection limits?

Typically 0.1-1 weight% for most elements
Better detection limits for heavier elements
Light elements (Z < 11) have poorer detection limits
Minimum detectable concentration depends on:
- Acquisition time
- Beam current
- Sample thickness
- Element atomic number

Data Analysis

Peak Identification

Identify characteristic X-ray peaks

Let’s use gold (Au) as an example:

Gold Characteristic X-ray Lines
Line	Energy (keV)	Origin	Relative Intensity
Mα	2.123	M → L transition	100
Lα	9.713	L₃ → M₅ transition	80
Lβ₁	11.442	L₂ → M₄ transition	50
Lγ₁	13.381	L₂ → N₄ transition	15

The peaks form characteristic families (K, L, M series) based on the electron shell where the initial vacancy occurs.

How do you deal with peak overlap?

Example: Au Mα (2.123 keV) vs P Kα (2.014 keV):

Use higher energy peaks if available (Au L-series)
Employ peak deconvolution through iterative optimization:

Step 1: Model the spectrum The measured spectrum is modeled as:

\[I(E) = \sum_i A_i \cdot G(E - E_i, \sigma_i) + B(E)\]

where:

\(I(E)\) is measured intensity
\(A_i\) is peak area
\(G\) is Gaussian function
\(E_i\) is peak position
\(\sigma_i\) is peak width
\(B(E)\) is background function

Step 2: Define loss function Mean squared error between model and data:

\[L = \frac{1}{N}\sum_j [I_{measured}(E_j) - I_{model}(E_j)]^2\]

Step 3: Gradient descent optimization Update parameters iteratively:

\[\begin{split}A_i^{(n+1)} &= A_i^{(n)} - \eta \frac{\partial L}{\partial A_i} \\ E_i^{(n+1)} &= E_i^{(n)} - \eta \frac{\partial L}{\partial E_i} \\ \sigma_i^{(n+1)} &= \sigma_i^{(n)} - \eta \frac{\partial L}{\partial \sigma_i}\end{split}\]

where:

\(\eta\) is learning rate
\(n\) is iteration number
Gradients are computed analytically

Step 4: Advanced Optimization Methods (FURTHER REVIEW NEEDED)

Modern software packages use more sophisticated optimization methods:

Levenberg-Marquardt Algorithm:
- Combines gradient descent and Gauss-Newton methods
- Uses damped least squares:
\[\theta_{n+1} = \theta_n - (H + \lambda I)^{-1}\nabla L\]
where:
\(H\) is Hessian matrix

\(\lambda\) is damping parameter

\(I\) is identity matrix
- Adapts \(\lambda\) during optimization:
  
  Large \(\lambda\): behaves like gradient descent
  
  Small \(\lambda\): behaves like Gauss-Newton
  
  Automatically adjusted based on improvement in loss
Trust-Region Methods:
- Define region where model is trusted
- Solve subproblem within radius \(\Delta\):
\[\min_p \{L(\theta_n + p) : \|p\| \leq \Delta\}\]
- Update trust region based on model accuracy:
  
  Good prediction: increase \(\Delta\)
  
  Poor prediction: decrease \(\Delta\)

Convergence Criteria Stop when either:

Change in loss \(\Delta L < \epsilon\)

Maximum iterations reached

Parameter changes \(\Delta \theta < \delta\)

Trust region or damping parameter stabilizes

Quantification

How are elements quantified?

The Cliff-Lorimer method relates peak intensities to element concentration ratios.

Step 1: Peak Integration Integrate the characteristic X-ray peak after background subtraction:

\[I_A = \int_{E_1}^{E_2} [I_{total}(E) - I_{background}(E)] dE\]

Step 2: Cliff-Lorimer Equation For elements A and B:

\[\frac{C_A}{C_B} = k_{AB} \frac{I_A}{I_B}\]

where:

\(C_A, C_B\) are concentrations
\(I_A, I_B\) are integrated intensities
\(k_{AB}\) is the Cliff-Lorimer factor

Step 3: Multiple Elements For a system with n elements:

\[\sum_{i=1}^n C_i = 100\%\]

Step 4: Error Analysis The relative error in concentration ratio is:

\[\frac{\Delta(C_A/C_B)}{C_A/C_B} = \sqrt{\left(\frac{\Delta I_A}{I_A}\right)^2 + \left(\frac{\Delta I_B}{I_B}\right)^2 + \left(\frac{\Delta k_{AB}}{k_{AB}}\right)^2}\]

Background correction

What is the drift area?: The drift area is a region outside the ROI used to measure background X-ray signal. It accounts for gradual changes in background signal over time due to instrument conditions or sample environment.
How is background correction performed?: The background signal from the drift area is subtracted from the total ROI signal to obtain the net sample signal.
Why use a special holder?: Special holders minimize background noise and interference from the holder material, ensuring detected X-rays originate primarily from the sample.

Recommended Settings

Dwell time: 50 µs per pixel
Imaging mode: STEM with HAADF detector (BF mode also acceptable)
Optical conditions:
- Defocus: 2.98 µm
- Convergence angle: 10 mrad
- Spot size: 3-5 (balance between spatial resolution and count rate)

EELS: Electron Energy Loss Spectroscopy

Comparison with EDS

EELS vs EDS (quick comparison)
Feature	EELS	EDS
Signal	Energy lost by transmitted electrons	Characteristic X-rays emitted from atoms
Spatial resolution	Can reach sub-ångström (atomic-scale) in modern STEM-EELS	Usually a few nanometres (limited by X-ray delocalization and interaction volume)
Detectable elements	All elements (including light elements such as B, C, N, O)	Inefficient for very light elements (typically Z < 5–6)
Information type	Chemical bonding, oxidation state, fine structure (ELNES/EXELFS), electronic structure	Elemental composition, characteristic peak energies
Quantification sensitivity	High sensitivity for thin samples (signal scales with transmitted electrons and specimen thickness)	Often better for thicker samples where more X-rays are generated (but depends on geometry and detector)
Spectral features	Fine structure near ionization edges (ELNES, EXELFS) that reports bonding and coordination	Sharp elemental peaks (no direct bonding information)

Introduction

What is the main goal of EELS?: EELS analyzes energy losses of transmitted electrons to reveal sample composition, electronic structure, and bonding by measuring how much energy electrons lose when interacting with the sample.
What problems does EELS solve compared to EDS?: EELS provides higher energy resolution and sensitivity to light elements (e.g., C, N, O) that are difficult to detect with EDS. It also reveals electronic structure and bonding information not accessible via EDS.
What can you do with the EELS data?: We can identify elements, determine their concentrations, analyze chemical bonding, and study electronic structure and band gaps.
How can bonding and electronic structure be revealed through EELS?: EELS reveals bonding and electronic structure through features such as Energy-Loss Near-Edge Structure (ELNES) and Extended Energy-Loss Fine Structure (EXELFS). ELNES provides information about the unoccupied density of states and local chemical environment, while EXELFS gives insights into interatomic distances and coordination numbers. By analyzing these fine structures, we can infer details about the chemical bonding and electronic properties of the material.

Alignment and Setup

Zero loss peak alignment

What are the main goals of alignment?

There are three main goals:

Center the Zero-Loss Peak (ZLP) at 0 eV
Ensure the spectrum is not tilted
Optimize resolution and signal by properly focusing the spectrometer

To achieve these goals:

In DigitalMicrograph: Use EELS → Align → Zero Loss Peak
In Velox: Use EELS Alignment → Energy Shift or Zero-loss Centering

With the ZLP visible, use the spectrometer focus (sometimes called EELS focus or energy filter focus) to make the peak as sharp and narrow as possible. If the spectrum appears tilted (dispersion direction not horizontal), adjust the rotation alignment or EELS dispersion axis.

Understanding the ZLP

What am I seeing in the peak?: The y-axis represents the number of electrons detected at each energy loss value (x-axis). The peak indicates the most probable energy loss, typically corresponding to the Zero Loss Peak (ZLP) where electrons pass through without energy loss.
What is ZLP?: The Zero Loss Peak (ZLP) represents electrons that pass through the sample without energy loss. It serves as an energy reference point and indicates the energy resolution of the system.
Why is it important to align ZLP to the center position?: Aligning the ZLP to the center position ensures accurate energy calibration and optimal resolution for detecting energy loss features. Centering the ZLP reduces asymmetries in the energy distribution, which can cause peak broadening and shifts. This leads to clearer, more accurate spectra.

EELS Physics

Energy ranges and what they tell us

What are the main energy ranges in EELS?

EELS spectra are typically divided into three regions:

Zero-loss region (0-5 eV):
- Contains elastically scattered electrons (no energy loss)
- Reveals sample thickness via \(I_0/I_t\) ratio
- Width indicates energy resolution
Low-loss region (5-50 eV):
- Plasmon excitations (collective electron oscillations)
- Interband transitions
- Band gap measurements
- Example: Free-electron plasmon in Al at ~15 eV
Core-loss region (>50 eV):
- Inner-shell ionizations
- Chemical bonding information (ELNES)
- Extended fine structure (EXELFS)
- Element-specific edges (e.g., C-K at 284 eV, Fe-L at 708 eV)

Plasmon physics

How do plasmons work physically?

Plasmons are collective oscillations of valence electrons. When fast electrons pass through a material:

\[E_p = \hbar\sqrt{\frac{ne^2}{m\epsilon_0}}\]

where:

\(E_p\) is plasmon energy
\(n\) is valence electron density
\(e\) is electron charge
\(m\) is electron mass
\(\epsilon_0\) is vacuum permittivity

This explains why metals (high \(n\)) have higher plasmon energies than semiconductors.

Mean free path and inelastic scattering

What is mean free path in EELS?

The inelastic mean free path (λ) is the average distance an electron travels between inelastic scattering events. It’s fundamental because:

Defines interaction volume:
- For 200 keV electrons in silicon: λ ≈ 100 nm
- In metals: λ typically 50-150 nm
- In light elements (carbon): λ can be >200 nm

2. Determines signal strength: The probability of single scattering is:

\[P_1 = \frac{t}{\lambda}e^{-t/\lambda}\]

This is maximum when t = λ, but we usually work at t ≈ 0.3λ to minimize plural scattering.

Why is mean free path energy-dependent?

The inelastic mean free path varies with:

Incident electron energy \(E_0\):

\[\lambda \propto \frac{E_0}{E_m \ln(2\beta E_0/E_m)}\]

where:

\(E_m\) is mean energy loss
\(\beta\) is collection semi-angle

Atomic number Z:

\[\lambda \propto \frac{1}{n\sigma} \propto \frac{A}{Z\rho}\]

where:

\(n\) is atomic density
\(\sigma\) is scattering cross-section
\(A\) is atomic weight
\(\rho\) is density

How does mean free path affect data collection?

Optimal thickness selection:
- Too thin (t << λ): Weak signal
- Too thick (t >> λ): Multiple scattering
- Optimal (t ≈ 0.3-0.5λ): Best signal/noise ratio
Collection angle effects:
- Larger β increases effective λ
- But reduces energy resolution
- Typical compromise: β ≈ 10-20 mrad
Energy window selection:
- Core-loss: Need thinner samples (more λ-limited)
- Low-loss: Can use thicker samples
- Zero-loss: Used for t/λ measurement

Sample thickness effects

How does thickness affect EELS spectra?

The probability of multiple scattering increases with thickness. For a sample of thickness \(t\):

\[P(n) = \frac{1}{n!}\left(\frac{t}{\lambda}\right)^n e^{-t/\lambda}\]

where:

\(P(n)\) is probability of \(n\) scattering events
\(\lambda\) is mean free path
\(t\) is sample thickness

This Poisson distribution explains why:

Thin samples (t/λ < 0.3): Mostly single scattering
Medium samples (0.3 < t/λ < 1): Multiple plasmon peaks
Thick samples (t/λ > 2): Complex plural scattering

Thickness measurement

How do you measure absolute thickness?

There are several methods:

Log-ratio method (relative thickness):

\[\frac{t}{\lambda} = \ln\left(\frac{I_t}{I_0}\right)\]

where:

\(I_t\) is total integrated intensity
\(I_0\) is zero-loss peak intensity

Kramers-Kronig method (absolute thickness):

\[t = -\lambda_e \ln\left(\frac{I_{ZLP}}{I_t}\right) \cdot F(\beta, E_0)\]

where:

\(\lambda_e\) is electron mean free path
\(F(\beta, E_0)\) is angular correction
\(\beta\) is collection semi-angle
\(E_0\) is incident electron energy

3. Convergent beam method: Using thickness fringes in CBED patterns:

\[t = \frac{2\xi_g}{s_i}\]

where:

\(\xi_g\) is extinction distance
\(s_i\) is deviation parameter

What thickness is ideal for EELS?

Optimal t/λ ≈ 0.3-0.5 (balances signal and plural scattering)
For 200 keV electrons in silicon: λ ≈ 100 nm
Therefore, ideal thickness ≈ 30-50 nm

EELS Imaging

How does energy filtering work?

Energy-filtered TEM (EFTEM) uses a slit to select specific energy losses:

Zero-loss filtering:
- Select only unscattered electrons (0-10 eV)
- Improves resolution by removing inelastic background
- Essential for thick samples
Elemental mapping:
- Select characteristic edge (e.g., Fe-L at 708 eV)
- Subtract pre-edge background
- Generate elemental distribution maps

\[I_{net} = I_{post} - A\cdot I_{pre}\exp(-r(E-E_{pre}))\]

where:

\(I_{post}\) is post-edge intensity
\(I_{pre}\) is pre-edge intensity
\(r\) is background slope
\(A\) is scaling factor

How do you get spatial information?

Two main approaches:

EFTEM Imaging:
- Use energy slit to form images
- Parallel acquisition but poor energy resolution
- Good for elemental mapping
STEM-EELS:
- Acquire spectrum at each probe position
- Better energy resolution
- Can map both elemental and electronic structure
- Spatial resolution limited by probe size (< 1Å possible)

Spectrum imaging combines both:

Full EELS spectrum at each pixel
3D dataset (x, y, energy-loss)
Allows post-acquisition analysis

Advanced Analysis

Plural scattering removal

What is plural scattering?

When electrons undergo multiple energy losses:

Double plasmon peak: Energy = 2×E₁
Core-loss + plasmon: Edge energy + plasmon energy
Multiple events: Convolution of single-loss probabilities

How do you remove plural scattering?

Using Fourier-log deconvolution:

\[s_1(\omega) = \mathcal{F}^{-1}\{\ln[\mathcal{F}\{s(\omega)\}]\}\]

where:

\(s(\omega)\) is recorded spectrum
\(s_1(\omega)\) is single-scattering spectrum
\(\mathcal{F}\) is Fourier transform

Fine structure analysis

What physics gives us ELNES?

Energy Loss Near Edge Structure (ELNES) reveals bonding through:

Matrix element effects:
\[\sigma(\Delta E) \propto |\langle\psi_f|e^{i\mathbf{q}\cdot\mathbf{r}}|\psi_i\rangle|^2\rho(E)\]
Density of states modulation:
- Unfilled states above Fermi level
- Modified by bonding environment

Common examples:

sp² vs sp³ carbon (graphite vs diamond)
Ti oxidation states (Ti⁴⁺ vs Ti³⁺)
Oxygen K-edge pre-peak in transition metal oxides

Energy Dispersive X-ray Spectroscopy (EDS) is a powerful analytical technique for determining the elemental composition of materials in the transmission electron microscope (TEM). It provides quantitative and qualitative analysis of elements present in your sample.

Setup

EDS uses 4 detectors positioned around the sample (shown under TEM User Interface → EDX User). These detectors:

Maximize collection efficiency by capturing X-rays from multiple angles
Enable better statistics and faster acquisition
Reduce shadowing effects from sample holders
Allow analysis even if one detector is shadowed

Goal

EDS identifies elements and their relative abundances by detecting characteristic X-rays emitted when an electron beam bombards the sample. The energy and intensity of these X-rays reveal the elemental composition.

Why do elements show peaks at specific energies?

When the electron beam ejects an inner-shell electron, a higher-energy electron fills the vacancy, releasing an X-ray photon. The photon energy equals the difference between the two shells. Since each element has unique energy levels, the emitted X-rays have characteristic energies for element identification.

Beam damage

Is inner-shell electron ejection considered beam damage?: Yes, but it’s less severe than sputtering (atom removal) or heating (thermal degradation). Inner-shell ejection only affects the electronic structure of individual atoms without causing physical changes to the sample. The electronic configuration quickly restabilizes as higher-energy electrons fill the vacancies.

Quantification

How are elements quantified?

The Cliff-Lorimer method relates peak intensities to element concentration ratios:

Integrate characteristic X-ray peaks to obtain peak areas
Apply the Cliff-Lorimer equation with sensitivity factors (k-factors) to convert intensity ratios into concentration ratios
Normalize results to determine relative abundances

Background correction

What is the drift area?: The drift area is a region outside the ROI used to measure background X-ray signal. It accounts for gradual changes in background signal over time due to instrument conditions or sample environment.
How is background correction performed?: The background signal from the drift area is subtracted from the total ROI signal to obtain the net sample signal.
Why use a special holder?: Special holders minimize background noise and interference from the holder material, ensuring detected X-rays originate primarily from the sample.

Sample preparation and artifacts

What sample preparation is needed for EDS?

Sample must be thin enough for electron transmission (typically < 100 nm)
Clean surface to avoid contamination
Conductive coating may be needed for insulating samples
Carbon support films are common but contribute background signal

What are common artifacts to watch for?

Peak overlap between elements
Absorption effects in thick samples
Fluorescence from neighboring areas
Contamination build-up during analysis
Detector dead time at high count rates

What are the detection limits?

Typically 0.1-1 weight% for most elements
Better detection limits for heavier elements
Light elements (Z < 11) have poorer detection limits
Minimum detectable concentration depends on:
- Acquisition time
- Beam current
- Sample thickness
- Element atomic number

Recommended settings

Dwell time: 50 µs per pixel
Imaging mode: STEM with HAADF detector (BF mode also acceptable)
Optical conditions:
- Defocus: 2.98 µm
- Convergence angle: 10 mrad
- Spot size: 3-5 (balance between spatial resolution and count rate)

EELS

Goal

What is the main goal of EELS?: EELS analyzes energy losses of transmitted electrons to reveal sample composition, electronic structure, and bonding by measuring how much energy electrons lose when interacting with the sample.
What problems does EELS solve compared to EDS?: EELS provides higher energy resolution and sensitivity to light elements (e.g., C, N, O) that are difficult to detect with EDS. It also reveals electronic structure and bonding information not accessible via EDS.
What can you do with the EELS data?: We can identify elements, determine their concentrations, analyze chemical bonding, and study electronic structure and band gaps.

Zero loss peak alignment

What are the main goals of alignment?

There are three main goals:

Center the Zero-Loss Peak (ZLP) at 0 eV
Ensure the spectrum is not tilted
Optimize resolution and signal by properly focusing the spectrometer

To achieve these goals:

In DigitalMicrograph: Use EELS → Align → Zero Loss Peak
In Velox: Use EELS Alignment → Energy Shift or Zero-loss Centering

With the ZLP visible, use the spectrometer focus (sometimes called EELS focus or energy filter focus) to make the peak as sharp and narrow as possible. If the spectrum appears tilted (dispersion direction not horizontal), adjust the rotation alignment or EELS dispersion axis.

What am I seeing the peak?

The y-axis represents the number of electrons detected at each energy loss value (x-axis). The peak indicates the most probable energy loss, typically corresponding to the Zero Loss Peak (ZLP) where electrons pass through without energy loss. The sharp peak at the low-energy-loss end is the zero-loss peak (ZLP), corresponding to electrons that passed through the sample without losing energy.

What is the Aligning spectrum dialog box?

The Aligning Spectrum dialog box allows you to adjust the spectrometer settings to center the Zero Loss Peak (ZLP) in the energy loss spectrum for optimal resolution and accuracy.

What’s the final state of alinged ZLP?

The energy loss is centered around zero, with the ZLP peak aligned to the center of the spectrum.

What is the noise grey background area?

The grey background area represents likely the real-space TEM image or the diffraction pattern, depending on your camera mode, used for positioning before spectrum acquisition.

What is ZLP?

The Zero Loss Peak (ZLP) represents electrons that pass through the sample without energy loss.

Why is it important to align ZLP to the center position?

Aligning the ZLP to the center position ensures accurate energy calibration and optimal resolution for detecting energy loss features. Centering the ZLP reduces asymmetries in the energy distribution, which can cause peak broadening and shifts. This leads to clearer, more accurate spectra.

Why does it take so long to align ZLP?

Aligning the ZLP can be time-consuming because it requires fine-tuning multiple spectrometer parameters (e.g., energy dispersion, focus) to achieve optimal centering. Small adjustments can have significant effects, necessitating iterative testing and measurement to find the best settings.

Is the aligning ZLP automated?

Some modern EELS systems offer automated ZLP alignment features, but manual adjustments are often needed for precise optimization, especially in challenging samples or conditions.

What do you need to align ZLP?

To align the Zero Loss Peak (ZLP) in EELS, you typically need to adjust the spectrometer settings, including the energy dispersion and focus, to ensure the ZLP is centered.