E36 · Electron excitations

Vibrational EELS: Resolution & Aloof Probes

How far can you narrow the electron source's energy band to sharpen a vibrational spectrum? Until shot noise wins. Narrowing the source slit sharpens the energy resolution but throws electrons away. Past some point too few remain, and shot noise buries the weak vibrational peaks. Subtracting a zero-loss peak recorded separately in vacuum can expose the signal, but it cannot take away the noise.

Choose the energy band of the source, record a low-loss spectrum and a separate vacuum reference, and look at the difference in counts. Then move the probe outside the specimen edge and compare the vibrational signal with how many electrons hit the material directly.

Net recorded window counts $\text{—}$

Counting signal-to-noise ratio $\text{—}$

Gaussian-equivalent response width $\text{—}$

Monochromator transmitted fraction $\text{—}$

More measurements and target

Transmitted specimen-plane current $\text{—}$

Recorded sample counts in channel $\text{—}$

Independent reference counts in channel $\text{—}$

Direct material incident electrons $\text{—}$

Sample plus reference acquisition time $\text{—}$

Experiment target $\text{—}$

Simulated examples; each plot identifies its data source.

More examples

Selected source energy width

w

40\,\mathrm{meV}

Exposure per sample or reference

t

1\,\mathrm{s}

Geometry: 0 through foil, 1 aloof

g

1

Aloof clearance outside the edge

b

20\,\mathrm{nm}

Advanced settings

Current before monochromator

I_0

100\,\mathrm{pA}

Collection semi-angle

\beta

10\,\mathrm{mrad}

Linear energy drift within exposure

\dot E

0.1\,\mathrm{\frac{meV}{s}}

Specified mode temperature

T

300\,\mathrm{K}

Inspect recorded energy channel

\Delta E

146\,\mathrm{meV}

Learning goal and model limits

Make a prediction, then test it

How far can you narrow the electron source's energy band to sharpen a vibrational spectrum?

Target: resolve the specified pair of modes. The source and instrument together must have an equivalent width below 15 meV. Use the aloof geometry, with the probe 10–60 nm outside the specimen edge. Collect at least 300 net counts in the 105–160 meV window and reach a counting signal-to-noise ratio of at least 8.

Model assumptions

This teaching spectrum is synthetic, with a fixed 60 keV beam and two vibrational modes. It does not identify a material or calculate damage. A rectangular slit cuts out part of a Gaussian source with a 300 meV FWHM. The recorded response combines that selected distribution with a 5 meV Gaussian spectrometer response, a 4 meV intrinsic mode width and a uniform shift from linear drift. The displayed resolution is a Gaussian-equivalent RMS width, not the FWHM of the actual non-Gaussian line. The assumed modes at 120 and 145 meV follow Bose factors for loss and gain. Their absolute cross sections and the collection response are illustrative. Aloof coupling uses a planar exponential envelope with a single wave vector, not a full dielectric, Bessel or impact-scattering calculation. The sample and the vacuum reference get equal, independent exposures, so their difference keeps the counting variance of both and can be negative. Seeded counts use a normal approximation above 40 expected counts. Each exposure is centered in energy, so drift mismatch between exposures is not modeled. The finite recorded energy window can cut off the tails. An ideal aloof pencil beam puts no electrons directly into the material, but that does not guarantee safety or zero damage. Plural scattering, atomic resolution, diffraction, detector saturation, tail mismatch and calibrated temperature fitting are left out. The apparatus drawing enlarges the dispersion and the clearance from the specimen.

Physics tutorial

Vibrational EELS: Resolution & Aloof Probes

BackgroundYou choose the energy band of the source, record a low-loss spectrum and a separate vacuum reference, and look at the difference in counts. Then you move the probe outside the specimen edge and compare the vibrational signal with how many electrons hit the material directly.

Why it mattersThe goal is to resolve the specified pair of modes with a combined source and instrument width below 15 meV. Keep the probe outside the specimen at a clearance of 10–60 nm, and collect at least 300 net counts in the 105–160 meV window with a counting signal-to-noise ratio of at least 8.

Start with the essentials

Focus question: How far can you narrow the electron source's energy band to sharpen a vibrational spectrum?
One-sentence intuition: A narrower source slit throws away electrons. Subtracting a separately recorded zero-loss reference can bring out a small signal, but it cannot remove shot noise or restore detail that the response has blurred.

Core mathematical model

Fraction of the source passed by the slit

P=\operatorname{erf}\!\left(\frac{w}{2\sqrt{2}\sigma_0}\right)

The slit leaves a truncated Gaussian energy distribution. The slit width is not the line width you measure.

Sample minus vacuum reference

S=N_s-N_v,\qquad \operatorname{Var}(S)=\mu_s+\mu_v

The two exposures are equal and independent, so the difference carries the shot noise of both the sample and the vacuum reference.

Detailed balance between energy gain and loss

\frac{I_{gain}}{I_{loss}}=\exp\!\left(-\frac{\hbar\omega}{k_BT}\right)

The intrinsic weights of energy gain and loss obey detailed balance. Instrument broadening and the choice of energy window can mix the two.

How the aloof signal falls off with distance

C(b)=\exp\!\left(-\frac{2\omega b}{v}\right)

One chosen planar evanescent component shows how the coupling depends on distance. It is not the full aloof scattering probability.

Common difficulties

Brighter does not mean more local

Typical misconceptionA brighter or cleaner spectrum means the measurement is more local.

Better mental modelA narrower source slit throws away electrons. Subtracting a separately recorded zero-loss reference can bring out a small signal, but it cannot remove shot noise or restore detail that the response has blurred.

Run the experiment

01
Predict the cost in counts
Compare Broad source hides the pair, Resolve the pair outside the sample and Narrow but starved.
What to observe: A narrower energy band separates the modes better but lets less current through.
02
Inspect the two recorded spectra
Move the energy cursor through the zero-loss peak and the two modes.
What to observe: The sample, reference and difference all use the stored counts. The difference can be negative.
03
Tell electron exposure apart from signal
Move the probe outside the edge, vary the clearance and reach the target. Then compare Long exposure with energy drift.
What to observe: An ideal aloof beam misses the material but still couples to it through the near field. A longer exposure alone cannot overcome drift.

Vibrational EELS: Resolution & Aloof Probes

Vibrational EELS: Resolution & Aloof Probes

Core mathematical model

Fraction of the source passed by the slit

Sample minus vacuum reference

Detailed balance between energy gain and loss

How the aloof signal falls off with distance

Common difficulties

Brighter does not mean more local

Run the experiment

Predict the cost in counts

Inspect the two recorded spectra

Tell electron exposure apart from signal

Related experiments and articles

Molecules, forces and ultrafast dynamics