Spectroscopy with Free Electron Lasers

David BernsteinSASS Talk

January 28th, 2009

Overview

• Who am I?!• What is FLASH?!• The promise of Free Electron Lasers

(FELs)• The Trouble with Spectroscopy• Sample Fabrication• The Actual Experiment• Data Analysis

Who am I?!

Who am I?!

Who am I?!

Who am I?!

What is FLASH?!Free election LASer at Hamburg

BOGUS ACRONYM

The promise of FELs

• Intense pulses containing upto 1012

photons• Short temporal pulse widths, on the order

of 50-100 fs• Represents a 108 improvement in peak

intensity over 3rd generation synchrotron sources

The promise of FELs

H. Chapman, et.al., Nature Physics, 2, 839 (2006)

The trouble with Spectroscopy

• Near Edge X-Ray Absorption Fine Structure (NEXAFS) spectroscopy:– Commonly used at synchrotrons– X-rays excite a core electron into a valence

state– Element Specific– Sensitive to bond angle and length– Can probe subsystems such as spin

(magnetic) and charge (electric) subsystems independently

The trouble with Spectroscopy

http://unicorn.mcmaster.ca/research/stxm-intro/STXM_poly3.JPG http://ssrl.slac.stanford.edu/stohr/xmcd.htm

The trouble with Spectroscopy• So whats the problem?

– FELs are difficult to tune (sort of…)

– FEL radiation is produced by Self-Amplified Spontaneous Emission. This process is inherently stochastic.

– FEL radiation is therefore characterized by humungous fluctuations in relevant beam parameters such as energy, position, intensity and mode profile.

http://hasylab.desy.de/facilities/flash/machine/how_it_works/sase_self_amplified_spontaneous_emission/index_eng.html

The trouble with Spectroscopy

To deal with these problems… we need to capture as much information as possible on a shot-by-shot basis.

The experiment!

Grating disperses the beam by 0.3-0.4eV/mm in the vertical direction

Different parts of the sample get hit with different energy photons

We record an entire spectrum in each shot

The experiment!

Sample Fab

• In order to do this, we designed a sample with large silicon nitride windows in a silicon wafer.

• Half of the window is covered by the metal, the other half is blank nitride. This records the “Io” information

METAL Si3N4 membrane

Sample Fab

• Samples consist of 100nm thick silicon nitride windows sitting on top of silicon wafers

• 100nm of metal (LaMnO or Aluminum) deposited via shadow mask

Analytical Challenges• We need to linearize

the detector• We need to normalize

the transmitted intensity by the incident intensity based on the information we have from a little sliver of window next to the sample.

Linearizing the DetectorWe recorded a run with no sample in the beamline. A gold mesh upstream of the sample chamber recorded total pulse intensity:

We also have the intensity on each pixel recorded by the (very nonlinear) detection scheme consisting of a Ce:YAG crystal imaged by an intensified CCD camera.

What we want to know is the true intensity incident on a given pixel,

Linearizing the DetectorTo do this, we use the ansatz:

We then construct an error function summed over all images,

…and iteratively adjust our parameters, ai, until the error function reaches an acceptable value.

Linearizing the Detector

Normalizing the DataWe calculate the mode intensity in the x-direction by summing over y for each value of x.

Now the spectra are flat in the x-direction. We then normalize by the average value at each y point to get the full normalized spectrum.

Normalizing the DataFor this to work, we can only accept images with a single mode in the x-direction. In other words, we have good shots and bad shots.

GOOD SHOT BAD SHOT

Finally… the spectra!

Finally… more spectra!

Acknowledgements

• Stanford/SSRL/Pulse– Yves Acremann– Andreas Scherz– Mark Burkhard– Jo Stohr

• Stanford Nanofabrication Laboratory– Mahnaz Mansourpour

• DESY/UH– Bill Schlotter– Martin Beye, – Torben Beeck– F. Sorgenfrei– Annette Pietzch– Wilfred Wurth– A Foehlisch