REU Student: Si (Athena) Pan Connecticut College Mentor: Dimitre Ouzounov Graduate Student Mentor: Heng Li Summer 008
Ultrashort pulses, its measurement and motivation of my project Two-photon absorption measurement in UV diodes Auto-correlation measurement error quantification
Electromagnetic Pulses whose time duration is in the femtosecond (fs = 10-15 s) to picosecond (ps = 10-1 s) range. Electronics devices (diodes, oscilloscopes, etc) are not fast enough to allow direct measurement of picosecond and femtosecond pulses. Electric Field Intensity Cross-Correlation: A ( τ ) = I ( t) I ( t τ ) dt c Unknown pulse s r Time Slow detector Reference pulse E r (t τ) Lens V det (τ ) C(τ ) Delay
Michelson Interferometer Mirror E(t) Beamsplitter Rick Trebino Mirror Input pulse E(t τ) Delay Lens SHG crystal E(t ) + E(t τ ) Filter Slow detector [E(t) + E(t τ)] G( τ ) = [ E( t) + E( t τ )] dt Interferometric Autocorrelation: Split the pulse in two with a Michelson Interferometer. Overlap them as they recombine. Shaker arm creates the delay time.
1.3 Oscillator GHz oscillator 0 ps 14 mw Compressor ps 7 mw SMF pre-amplifier 50 mw DC-LMA amplifier Shaping Crystals 50 mw SHG 5 watts
ERL requires Flat-top pulses. Polarization Sensitive Unable to measure the whole pulse in real time. We stack -ps pulses through a sequence of 3 birefringent crystals to produce pulse with nearly flattop. Problem: SHG crystal is polarization sensitive. Intensity (a. u.) 0.35 0.30 0.5 0.0 0.15 0.10-0 0 0 40 Time (ps)
ERL requires Flat-top pulses. Solution: Replace SHG & PMT with a single photodiode. We stack -ps pulses through a sequence of 3 birefringent crystals to produce pulse with nearly flattop. Problem: SHG crystal is polarization sensitive. Intensity (a. u.) 0.35 0.30 0.5 0.0 0.15 0.10-0 0 0 40 Time (ps)
Photo-detector that absorbs two photons of ω each, but not one at ω. hν CB hν < E g < hν E g hν hν (One photon absorption) Two-photon induced photocurrent: signal is a quadratic function of power ~ I VB Problem: Impurity linear absorption signal obscure quadratic signal Solution? ~ Find one that works!
Measurement: photo-current as a function of the laser beam power. Setup λ/-plate Polarizer Beamsplitter Cube Pre- Amp Volt Meter Trial Experiment: Diode = G1116, λ = 1µm, I peak ~ 10 7 W/cm, w ~ 15 µm Result: nice quadratic response (as expected) Photocurrent (µα) 10 4 1 4 0.1 4 G1116 3 4 5 6 7 y =.08x + 0.09 10 Power (mw) 3 4 5 6 7 100
Diode = UVTOP60 f lens = 1 inch = 5.4 mm w ~ 8 µm Photocurrent (na) I peak ~ 1.8 10 7 W/cm 1 8 6 y = x + 1.40 4 Reduce spot size, w, of the focused beam, hence, increase I peak Increase the twophoton absorption rate Diode = UVTOP60 f lens = 14 mm w ~ 4.6 µm Photocurrent (na) 10 8 6 4 10 UVTOP60 f = 5.4 mm 3 4 5 6 7 8 9 100 Power (mw) 6 y = 1.08x - 1.35 4 I peak ~ 6 10 8 W/cm 5 6 3 4 5 6 3 10 1 6 4 UVTOP60 f = 14 mm 10 100 Power (mw)
Photcurrent (na) Photocurrent (na) 100 8 6 8 I peak ~ 9.8 10 8 W/cm y = 1.04x - 0.51 6 10 8 10 4 UVTOP60 f = 11 mm Diode = UVTOP60 6 f lens = 11 mm 4 w ~ 4 µm I peak ~ 9.8 10 8 W/cm y = 1.03x - 1.1 Diode = UVTOP300 f lens = 11 mm w ~ 4 µm 3 4 5 6 7 8 9 100 3 UVTOP300 f = 11 mm Power (mw) 10 3 4 5 6 7 8 100 3 Power (mw)
Two major sources of error associated with our measurement: - Linear absorption signal distortion - Misalignment while scanning the delay
Linear absorption signal can distort the shape of the pulse. Simulation: assume pulse FWHM = ps, two-photon signal = linear signal + = The background to peak ratio is distorted. The FWHM measured is slightly greater than the correct one.
Two-photon + Linear Signal Two-photon Absorption Signal Relative Difference in BG to Peak Ratio Intensity Autocorrelation: The background to peak ratio is also distorted. Linear signal = c two-photon signal (0 c 1) Background to peak ratio distortion, as c. Difference in FWHM also. Max discrepancy (c = 1) ~ 3% Relative Difference in FWHM
Misalignment Error - Align at zero delay - Due to shaker wobbling and alignment difficulty Reference Beam misalignment occurs except at zero delay. - Partially overlapping in time & space Delayed Beam Perfectly Aligned Misaligned Zero delay Misaligned Negative delay Misaligned Positive delay
Assume: D beam = 0 µm, pulse FWHM = ps Misalignment: 50 µm Perfectly Aligned Misaligned by 50 µm Perfectly Aligned Misaligned by 50 µm Simulation shows: Artificially shortened pulses are measured due to misalignment.
Under normal conditions error within < 10%. Relative Difference in FWHM Between Perfectly Aligned and Misaligned Pulses 5 µm ~ 1% 10 µm ~ 5% 15 µm ~ 10%
Misalignment ~ 5µm, linear signal = two-photon signal Perfect Measurement Misaligned + Linear Distortion Perfect Measurement Misaligned + Linear Distortion * These plots are intended to show what happens. Under normal conditions, the difference may not be so obvious.
Setup Reference Arm Michelson Interferometer Mirror E(t) Beamsplitter Input pulse E(t τ) Lens C C D Mirror Delay Standard Deviation, σ, shows how stable the beam is. Reference Arm: σ ~ 0.6 µm
Shaker Arm σ ~ 5 µm Further Misaligned Shaker Arm σ ~ 1 µm Wobbling effect increases substantially Intensity [a. u.] 3 1 5 micron misalignm ent FWHM =.9 ps 0-0 -10 0 10 0 Time [ps] Replace CCD with photo-diode and measure the traces. Intensity [a. u.] 3 1 0 1 micron misalignment FW HM =.83 ps -0-10 0 10 Time [ps]
FWHM FWHM 5µ m 1µ m Experimental Results: 3% FWHM 5µ m Theoretical Results: FWHM 5µ m FWHM FWHM 5µ m 1µ m 1.6% D beam = 30 µm Discrepancy arises due to the non-negligible wobbling effect of the shaking mirror.
1. None of the photo-diode exhibits quadratic response signal at the current power level. We will continue to search for more suitable diodes.. However, it is likely that these diodes we tested will work at higher beam power, which will happen once the second stage amplifier is in operation. 3. Computer simulation shows that linear absorption signal can - distort the background to peak ratio of both the interferometric and intensity autocorrelation measurement. - artificially lengthen the pulse. (< 5%) 4. Computer simulation also shows misalignment between the reference and shaker arm will yield artificially shortened pulses. There is about a factor of discrepancy between the experimental and theoretical results. One of the main reason for such discrepancy is due to the increasing wobbling effect of the shaker mirror when it is misaligned purposely.
I would like to thank my mentor Dimitre Ouzounov, grad student, Heng Li, and professor Frank Wise for their guidance and encouragement. I would also like to thank the LEPP REU program for providing me with the opportunity to work and study here at Cornell University this summer.