Ultrafast laser technology; characterization of ultrafast ...
Development and Implementation of Near- Infrared Ultrafast ... Final... · Infrared Ultrafast Laser...
Transcript of Development and Implementation of Near- Infrared Ultrafast ... Final... · Infrared Ultrafast Laser...
Development and Implementation of Near-
Infrared Ultrafast Laser Sources Generated by
Nonlinear Fiber Propagation
Final Exam for Scott R Domingue, PhD candidateDepartment of Electrical and Computer Engineering
Colorado State University
Welcome to my committee members:
Prof. Amber Krummel
Prof. Diego Krapf
Prof. Mario Marconi
and my adviser
Prof. Randy Bartels
Development and Implementation of Near-
Infrared Ultrafast Laser Sources Generated by
Nonlinear Fiber Propagation
Final Exam Overview:Examining a couple systems utilizing our 1060 nm
ultrafast technology for an end-goal application
Imaging ModalityNonlinear Optical
ElementUltrafast Sources
ANDi fiber Laser
3-Photon
Excitation
Fluorescence
Hyperspectral
Imaging via
Excitation Labeled
Fluorescence
ANDi seeded NPA
Dual-Band SC
ANDi: all normal dispersion laser
NPA: nonlinear power amplifier
SC: supercontinuum generation
Part I
Part III
Part IINonlinear Pulse
Compression w/
Normal Dispersion SC
Second-Harmonic
Generation
• (I) Chapter 6
• (I) Chapter 7
• (II) Chapter 10
• (III) Chapter 14
Multi-Photon Microscopy based on Labels,
Endogenous Fluorophores, and Harmonic-
Generation Imaging
Laser-scanning, multi-photon microscopy:
• High Resolution (submicron)
• Optical Sectioning
• Imaging Through Scattering media
(scattering dominated attenuation, 𝜆−2 −𝜆−4)
Recent Advances: Quantitative diagnosis of
disease based on label-free tissue interrogation
• 2nd-harmonic generation from collagen:
quantification of fibrosis
• 2-photon excitation fluorescence from
intrinsic fluorophores: metabolic response
from NADH concentration
• 3rd-harmonic generation from lipids:
functional imaging of retinas (𝜏−2 signal
sensitivity)
Denk, W.et al., Science 248, (1990)
Campagnola, P., Anal. Chem. 83, (2011)
Xu, C. et al., Proc. Natl. Acad. Sci. 93, (1996)
Masihzadeh, O. et al., Mol. Vis. 21, (2015)
Part IIntro.
Experimental Super-Resolved Multiphoton
Microscopy with Spatial Frequency-Modulated
Imaging
Field, JJ, et al., ARXIV, (2015)
Second-harmonic
generation from CdTe
Solar Cell grain
boundaries
Two-photon auto-
fluorescence (or
photoluminescence)
Increased Resolution
Line Focus => average
power limitations
Part IIntro.
Scaling Relationships between
Excitation Pulses and nonlinear signal
level
power in Nth-order nonlinear
signal, from pulses with intensity
𝐼 𝑡 and repetition rate 𝑓𝑊 ∝ 𝐼𝑁 = 𝑓∫ 𝐼 𝑡
𝑁𝑑𝑡
𝑊 ∝𝐼 𝑁
𝑓𝑁−1 𝜏𝑁−1
Scaling the power in Nth-order nonlinear signal with excitation pulse
parameters (for clean, Gaussian pulses)
𝐼 : Average Power
𝜏 : Pulse Duration
𝑓 : Pulse Repetition Rate
𝐼 𝑡 =𝐼
𝑓𝜏 𝜋𝑒−
𝑡𝜏
2
Part IIntro.
The Color Spectrum of Multi-Photon
Imaging ModalitiesHow might we access the extensive color palette of ultrafast microscopy starting with
1060 nm pulses? (Answer: Harnessing nonlinear propagation)
500 750 1000 1250 1500 1750 2000
Ti:SapphireYb-dopedEr-dopedCr:Forsterite
Wavelength [nm]
Imaging
Modalities
2nd and 3rd Harmonic GenerationSHG/ THG
2-PEF 3/4-PEF
CRS
TAS Transient Absorption Spectroscopy
Coherent Raman Spectroscopy
2-, 3-, and 4-Photon Excitation Fluorescence
Ultrafast
Sources
Part IIntro.
Nonlinear Bulk
and Fiber based
Extensions
A-SC
DB-SCDB-SC
A/N-SC
SSFS
Dual-Band Supercontinuum
Supercontinuum (Normal vs. Anomalous Dispersion)
Soliton Self-Frequency Shifting
OPO Optical Parametric Oscillator
A-SC
How might we access the extensive color palette of ultrafast microscopy starting with
1060 nm pulses? (Answer: Harnessing nonlinear fiber propagation)
Imaging
Modalities
2nd and 3rd Harmonic GenerationSHG/ THG
2-PEF 3-PEF
CRS
TAS Transient Absorption Spectroscopy
Coherent Raman Spectroscopy
2-, 3-Photon Excitation Fluorescence
500 750 1000 1250 1500 1750 2000
Wavelength [nm]
Yb-doped fiber lasers
Supporting ~150 fs pulses
Nonlinear Fiber
Based Spectral
Solutions for 1060
nm pulses
DB-SCDB-SC
N-SC
Dual-Band Supercontinuum
Normal Dispersion Supercontinuum
Leveraging Nonlinear Fiber Propagation to
Extend the Application Space of Yb-doped laser
Part IIntro.
Our Goals for 1060 nm Ultrafast Source Development via Nonlinear Pulse Compression
Part IIntro.
1060 nm sources
beyond the Yb
gain bandwidth
750 1000 1250
Yb-doped fiber lasers
Supporting ~150 fs pulses
N-SC
Wavelength [nm]
Normal Dispersion
Supercontinuum
1. Bandwidths/Durations comparable to Ti:Sapphire Oscillators: <30 fs2. Average Powers/Pulse Energies sufficient for creative microscope
development: >0.5 W / >10 nJ
3. High quality pulse intensity profiles: >90% Normalized Compression
4. Stability and Reliability exceeding that of Ti:Sapphire Oscillators: up-time
of Months to Years
5. Simplest (most elegant) possible optical systems: price point low enough
for system replication
6. High Quality Power Spectra: power spectral smoothness comparable to
Gaussian Pulses
Our Starting Ultrafast Source at 1060 nm
ANDi
`
NPA
`
`
MC
Master Oscillator Nonlinear and Power Amplifier, using Yb-doped
optical fiber
• 27 nJ or 1.7 W at 63
MHz
• 136 fs FWHM, 180 kW
peak power
• 97% of TL peak powerMaster Oscillator Nonlinear and Power Amplifier
ANDi: all normal dispersion laser
NPA: nonlinear power amplifier
MC: Martinez compressor
1045 1065 10850
0.2
0.4
0.6
0.8
1
Pow
er S
pec
trum
[a.
u.]
Wavelength [nm]-200 0 200
0
0.2
0.4
0.6
0.8
1
Time [fs]
Inte
nsi
ty [
a.u
.]
Part IInitialSource
Domingue, S. R. et al., Opt. Lett. 39, (2014)
850 950 1050 1150 1250-200
-150
-100
-50
0
50
Wavelength [nm]D
isper
sion C
oef
fici
ent
[ps/
nm
-km
]
Bulk FSLMA-PM-10, PCFLMA-PM-5, PCF
SC-4-975, PCFNL-1050-NEG1, PCF
UHNA3
Spectral Broadening in Optical Fiber
ANDi
`
NPA
`
`
MC
SC
Generating shorter pulses from the Yb-doped ultrafast source, using
nonlinear propagation in optical fiber
ANDi: all normal dispersion laser
NPA: nonlinear power amplifier
MC: Martinez compressor
SC: supercontinuum generation
solitons
𝛽2 = −𝐷10−6𝜆2
2𝜋𝑐Conversion to Group
Velocity Dispersion
Part INonlinearPulseCompression
`
MC
1060 nm
Our Early Lessons in Nonlinear Pulse Compression
ΦRSN
RMS-N
RSN
S.R. Domingue et al., Optics Express 21 (2013)
What we thought we knew• PM nonlinear fiber => eliminates polarization
instability AND average power limitations
Yb:KYWPulse Shaper +
Grating
Compressor
SC in Normal
Dispersion, “Fairly”
nonlinear fiber
• Pulse shaper + prism compressor => a general
solution to nonlinear pulse compression
Non-PM PM
Part INonlinearPulseCompression
350 mW
Transmissive Pulse Shaping in the Naïve limit
Transform-Limit supported by the SC bandwidth dropped
from 18(dashed) to 36 fs (solid)
None of which are the Dominant Issue
Initially attributed to sub-optimal pulse shaper: visible SLM, 4λ SLM
Wavefront Error, Spherical Aberration, and Spatial Chirp
Wav
elen
gth
[n
m]
Time [fs]
Mea
sure
d
Rec
on
stru
cted
-100 0 100
490
500
510
520
530950 1000 1050 1100
0
0.2
0.4
0.6
0.8
1
1.2
Pow
er S
pec
tru
m [
a.u
.]
Wavelength[nm]
SeedSC
-1000 0 10000
0.5
1
Inte
nsi
ty [
a.u
.]
Time [fs]
SeedSC
-1
0
1
2
Sp
ectr
al P
has
e [r
ad]
Part INonlinearPulseCompression
SLM: spatial light modulator
930nm990nm1060nm
1150nm
1230nm
4-F Martinez Pulse Shapers with an Achromatic Doublet
Extensive Petzval Field Curvature, roughly quadratic in Field Angle (i.e.
Diffraction Angle / Wavelength)
-10 -5 0 5 10
-14
-12
-10
-8
-6
-4
-2
0
Field Angle, Δθ [degrees]
Fie
ld C
urv
ature
[m
m]
170nm
302nm15fs
127nm
224nm20fs
102nm
179nm
25fs
72nm
127nm
35fs
51nm
89nm
50fs
Rayleigh
Lengths
S.R. Domingue et al., Optics Letters, 40 (2015)
Out of Plane Focusing =>
Spectral Wings are NOT 4-F =>
Spectral Apodization
sin 𝛾 − sin 𝜃 𝜆 = 𝑁𝜆
Δ𝜃 = 𝜃 𝜆 − 𝜃(𝜆0)
950 1000 1050 1100Pow
er S
pec
trum
[a.
u.]
Wavelength [nm]
Trans. Grating
Part INonlinearPulseCompression
4-F Martinez Pulse Shapers with a Plӧssl Lens
Significant reduction in Petzval Field Curvature
S.R. Domingue et al., Optics Letters, 40 (2015)
A. Negrean et al., Biomed. Opt. Express. 5, (2014)
= 𝑧𝑡𝑎𝑛 (𝜆)
𝑧𝑅𝑎𝑦𝑙𝑒𝑖𝑔ℎ 𝜆∗ 100 %
Tangential
Relative
Curvature0 2 4 6 8 10 12
-10
-8
-6
-4
-2
0
Field Angle [degrees]
Tan
gen
tial
Rel
ativ
e
Cu
rvat
ure
[100%
]
Wavelength [nm]
60mm Plӧssl100mm Plӧssl
200mm Ach.
150mm Ach.
100mm Ach. same as before
75mm Ach.
1060 1089 1117 1145 1171 1197 1222Flat-Field Pulse Shaper
930nm990nm1060nm
1150nm
1230nm
Trans. Grating
Kernel for Idea:
Swap grating for Scan Mirror (galvo)
and it’s a Laser Scanning system,
Plӧssl lenses are nearly telecentric.
Part INonlinearPulseCompression
ANDi
`
NPA
`
`
MC
SC
Armed with the Flat-Field Pule shaper, we try to use the Master Oscillator Nonlinear
and Power fiber Amplifier to generate broader bandwidths
Pulses injected into SC fiber
• 27 nJ or 1.6 W at 60 MHz
• 136 fs FWHM
• 97% of TL peak power
Notice: ~1 W coupled
power vs. 0.35 W before
ANDi: all normal dispersion laser
NPA: nonlinear power amplifier
MC: Martinez compressor
SC: supercontinuum generation
FFPS: flat-field pulse shaper
So what happens in
“fairly” nonlinear PM
fiber (more nonlinear than
telecom fiber) at these
higher pulse energies?
`
FFPS
Part INonlinearPulseCompression
Nonlinear Pulse Compression beyond the Naïve limit (almost)
The PM, “fairly” nonlinear fiber fails
The polarization
maintenance fails after
~30 min exposure of
500 mW average
power
ANDi
`
NPA
`
`
MC
SC
ANDi: all normal dispersion laser
NPA: nonlinear power amplifier
MC: Martinez compressor
SC: supercontinuum generation
FFPS: flat-field pulse shaper
`
FFPS
Similar failure reports
in PM photonic
crystal fiber.
Part INonlinearPulseCompression
Nonlinear Pulse Compression beyond the Naïve limit (almost)
-150 -100 -50 0 50 100 1500
0.2
0.4
0.6
0.8
1
Time [fs]
Inte
nsi
ty [
a.u
.]
0
10
20
30
40
Sp
ec.
Ph
ase
[rad
]
980 1030 1080 1130
Wavelength [nm]
State-of-the-Art Nonlinear Pulse Compression
`
ANDi
`
NPA
`
MC
SC
Master Oscillator Nonlinear and Power fiber Amplifier + Telecom-like fiber (low
nonlinearity) + Flat-Field Pulse Shaper:
ANDi: all normal dispersion laserNPA: nonlinear power amp.MC: Martinez compressorSC: supercontinuum generationFFPS: flat-field pulse shaperTL: transform-limited duration
Move away from “nonlinear” fiber and
towards telecom and amplifier fibers: 𝛾𝑁𝐿
∼3
`
FFPS
FFPS
TL
27 fs
FWHM
State-of-the-Art
at 10 nJ
Shaper-off
Fiber
Shaper-on
S.R. Domingue et al., Optics Letters, 40 (2015)
Part INonlinearPulseCompression
0 2.5 5 7.5 10 12.50
5
10
15
20
25
SLM Pixel [k-pix.]
Vo
ltag
e B
its
[k-b
its]
0
10
20
30
40
50
0
0.2
0.4
0.6
0.8
1
What Sources hath Nonlinear Pulse Compression wrought?
Generating state-of-the-art pulses beyond the typical Yb-doped gain bandwidth
using our series of pulse source and pulse shaping technologies
S.R. Domingue et al., Optics Letters, 40 (2015)TL: transform-limited duration
Part INonlinearPulseCompression
950 1000 1050 1100 1150 1200
0
10
20
30
40
50
Wavelength [nm]
Spec
tral
Phas
e [r
ad]
-80 -60 -40 -20 0 20 40 60 800
0.2
0.4
0.6
0.8
1
Time [fs]
Inte
nsi
ty [
a.u
.]
95 % TL
27 fs
FWHM
FFPS
TL
Shaper-off
Fiber
Shaper-on
10
0 m
m P
lӧss
l
92 % TL
19 fs
FWHM
60
mm
Plӧ
ssl 10 nJ
600 mW
Revisiting Our Goals for 1060 nm Ultrafast Source Development via Nonlinear Pulse
Compression1060 nm sources
beyond the Yb
gain bandwidth
750 1000 1250
Yb-doped fiber lasersSupporting ~150 fs pulses
N-SC
Wavelength [nm]
Normal Dispersion Supercontinuum
1. Bandwidths/Durations comparable to Ti:Sapphire Oscillators: <30 fs2. Average Powers/Pulse Energies sufficient for creative microscope
development: >0.5 W / >10 nJ
3. High Quality Pulse Intensity profiles: >90% Normalized Compression
4. Stability and Reliability exceeding that of Ti:Sapphire Oscillators: up-time
of Months to Years
5. Simplest (most elegant) possible optical systems: price point low enough
for system replication
6. High Quality Power Spectra: power spectral smoothness comparable to
Gaussian Pulses
Part IFinale
Revisiting Our Goals for 1060 nm
ultrafast source development1060 nm sources beyond
the Yb gain bandwidth
750 1000 1250
Yb-doped fiber lasers
Supporting ~150 fs pulses
N-SC
Wavelength [nm]
Normal Dispersion Supercontinuum
5. Simplest (most elegant) possible optical systems: price point low enough
for system replication
6. High Quality Power Spectra: power spectral smoothness comparable to
Gaussian Pulses
Part IFinale
`
AN
Di `
MC
2 W
ANDi
Spectral Clipping
Enclosed Short
(<50 mm)
Photonic Crystal
Fiber
(Hyper) Prism
Compressor
Without a Pulse
Shaper
> 10 nJ
< 20 fs
1 MW Peak Power
• Chapter 2.3
• Chapter 4.3
• Chapter 6.3
• Chapter 10
Moving from Source Development to Applications
Some possible applications utilizing our ultrafast 1060 nm pulse source, in
particular the two that we describe in remainder of the presentation.
`
ANDi
`
NPA
`
MC
SC
ANDi: all normal dispersion laser
NPA: nonlinear power amp.
MC: Martinez compressor
SC: supercontinuum generation
FFPS: flat-field pulse shaper
`
FFPS
• ~25 fs
• >350 kW
• 10 nJ @ 60 MHz
• 600 mW
`
Laser Scanning
Multi-photon
Microscopy
`Nonlinear Laser
Machining
`
3-Photon
Excitation
Fluorescence
Microscopy
`
Hyperspectral
Imaging via
Excitation Labeled
Fluorescence
`
Nonlinear 1250
nm pulse
generation`
Second
Harmonic
Generation
Part IFinale
Part IPart III Part II
Motivation for Part II: Source Development in the 1300 nm Biological Imaging Window
The attenuation of ballistic photons with propagation in turbid
(biological) media fundamentally limits the depth of image penetration
for multi-photon microscopy.
800 1000 1200 1400 1600 18000
250
500
750
1000
1250
1500
1750
Att
enu
atio
n L
eng
th [
μm
]
Wavelength [nm]
Mie Scattering
Water Absorption
Combined (Mouse
Cortex Model)
𝑙𝑒 =1
𝑙𝑠𝑐𝑎𝑡𝑡𝑒𝑟𝑖𝑛𝑔+
1
𝑙𝑎𝑏𝑠𝑜𝑟𝑝𝑡𝑖𝑜𝑛
−1
Effective multi-photon fluorescence
attenuation length (depth where signal
decays by 𝑒−2)
D. Kobat et al., J. Biomed. Opt. 16, (2011)
N.G. Horton et al., Nat. Photonics 7, (2013)
Fundamental image penetration
depth limit (in-focus fluorescent
signal to background ratio >1):
• 2-photon excitation ~5-6× 𝑙𝑒• 3-photon excitation ~8-9× 𝑙𝑒
Part IIIntro.
Motivation for Source Development in the 1300 nm Biological Imaging Window
To reach the fundamental depth limit, you do still need enough average
power (and/or pulse energy) to get ballistic photons down to 5𝑙𝑒
800 1000 1200 1400 1600 18000
250
500
750
1000
1250
1500
1750
Att
enu
atio
n L
eng
th [
μm
]
Wavelength [nm]
Mie Scattering
Water Absorption
Combined (Mouse
Cortex Model)
Canonical Ultrafast Lasers
Early Work• Ti:sapphire, 830 nm
• 𝑙𝑒 = 155 𝜇m
• Maximum Imaging Depth =
650 𝜇m (~4𝑙𝑒 ≠ 5𝑙𝑒)
D. Kleinfeld et al., Proc. Natl. Acad. Sci. 95, (1998)
J. Mertz, Introduction to Optical Microscopy, (2009)
• Reducing 𝜏 is limited to ~5x
• Instead reduce f by ~500x
𝑊 ∝𝐼 𝑁
𝑓𝑁−1 𝜏𝑁−1
Part IIIntro.
To reach the fundamental depth limit, you do still need enough average
power (and/or pulse energy) to get ballistic photons down to 5𝑙𝑒
Canonical Ultrafast Lasers
Later Work• Ti:sapphire, 925 nm
• 𝑙𝑒 = 200 𝜇m
• Regenerative amplifier: 200 kHz
repetition rate (150 fs, 3 𝜇J
pulses)
• Imaging GFP-labeled neurons
1 mm below brain surface (5𝑙𝑒)
P. Theer et al., OL 28, (2003)
This hits the limit of penetration
depth. To get deeper have to
scale something else… 𝜆800 1000 1200 1400 1600 18000
250
500
750
1000
1250
1500
1750
Att
enu
atio
n L
eng
th [
μm
]
Wavelength [nm]
Mie Scattering
Water Absorption
Combined (Mouse
Cortex Model)
Motivation for Source Development in the 1300 nm Biological Imaging Window
Part IIIntro.
Longer wavelength excitation affects not only the attenuation length, but also
can enable the possibility of 3-photon excitation fluorescence: win, win
800 1000 1200 1400 1600 18000
250
500
750
1000
1250
1500
1750
Att
enu
atio
n L
eng
th [
μm
]
Wavelength [nm]
Mie Scattering
Water Absorption
Combined (Mouse
Cortex Model)
Increasing the Excitation Wavelength
𝑙𝑒 800 nm = 150 𝜇m
𝑙𝑒 1060 nm = 250 𝜇m
𝑙𝑒 1300 nm = 350 𝜇m
Fundamental Imaging Depth Limits
for 2PM (5𝑙𝑒) / 3PM (8𝑙𝑒)
• 750 / 1200 𝜇m
• 1250 / 2000 𝜇m
• 1750 / 2800 𝜇m
N.G. Horton et al., Nat. Photonics 7, (2013)2PM: 2-photon microscopy
3PM: 3-photon microscopy
No “Simple” ultrafast sources
at 1300 nm!!
Cr:forsterite lasers (reliability) and Ti:sapphire
pumped optical parametric oscillators ($$$)
Motivation for Source Development in the 1300 nm Biological Imaging Window
Part IIIntro.
Generating pulses at 1250 nm in a photonic crystal fiber for 3-photon excitation
fluorescence microscopy.
`
ANDi
`
NPA
`
MC
SC
ANDi: all normal dispersion laser
NPA: nonlinear power amp.
MC: Martinez compressor
SC: supercontinuum generation
FFPS: flat-field pulse shaper
PCF: photonic crystal fiber
`
FFPSNonlinear 1250 nm pulse
generation in Dual Zero-
Dispersion Wavelength
Photonic Crystal Fiber
(PCF)PCF
Goal: • >1 nJ pulses at ~1300 nm
• minimum pulse duration (as
close to 28 fs as possible)
D. Kobat et al., J. Biomed. Opt., 16 (2011)
Part IIIntro.
Nonlinear Frequency Conversion from 1060 nm out to the 1300 nm Biological Imaging
Window
Nonlinear Frequency Conversion from 1060 nm out to the 1300 nm via Photonic Crystal
Fiber Generating pulses at 1250 nm in a photonic crystal fiber with two, closely
spaced zero-dispersion wavelengths (NL-1050-ZERO2)
850 950 1050 1150 1250-50
-40
-30
-20
-10
0
10
Wavelength [nm]
Dis
per
sion
Co
effi
cien
t
[ps/
nm
-km
]
Bulk FSLMA-PM-5
NL-1050-NEG1NL-1050-ZERO2
800 1000 1200 14000
1
Inte
nsi
ty 30 fs, 0.5 nJ
Wavelength [nm]
Fib
er L
ength
[m
m]
800 1000 1200 14000
5
10
15
20
25
30
Part IIIntro.
Numerical Model
Constructing a fiber cocktail composed of telecom and photonic crystal fiber
(PCF) with 2 zero-dispersion wavelengths (NL-1050-Zero2)
Part IINonlinearFiber Construction
Nonlinear Fiber Cocktails for Broadband Pulse Coupling with Minimal Instability
HI-1060
8 mm
PCF-Zero-2
30 mm
Optical
Fiber:
Off-Axis
Parabolic
Collimator
Free-Space
Aspheric Lens
Angle-Polished
ConnectorsLaser In• 0.67 dB Splice Loss
• 59% Transmission
• Fiber-Fiber splice
hermetically Seals PCF
PCF: photonic crystal fiber
1250 nm Pulse Characteristics after Nonlinear Frequency Conversion from 1060 nm
1250 nm generation at low powers: 0.6 nJ, 37 mW @ 61 MHz
Time [fs]
Wav
elen
gth
[nm
]
Mea
sure
d
Rec
on
stru
cted
-100 0 100
580
600
620
640
Efficiency:• 5% Depolarization
• 35% Conversion
efficiency including
fiber coupling (!)
Durations:• TL: 23 fs FWHM
• TL: 28 fs Square
• Reconstructed: 44 fs
Square
• 64% compressed
S.R. Domingue et al., Opt. Express 22, (2014)
TL: transform-limited duration
FROG: frequency resolved optical gating
Part II1250 nmPulseGeneration
Wavelength [nm]
Sp
ectr
al P
has
e [r
ad]
1150 1250 1350
-10
12
3
Pow
er S
pec
trum
[a.
u.]
-100 0 1000
0.2
0.4
0.6
0.8
1
Time [fs]
Inte
nsi
ty [
a.u
.]TL
Recon.
𝜏𝑠𝑞𝑢𝑎𝑟𝑒 =𝐸𝑛𝑒𝑟𝑔𝑦
𝑃𝑒𝑎𝑘 𝑃𝑜𝑤𝑒𝑟
Phantom-FROG
1250 nm generation at higher powers: 2 nJ, 122 mW @ 61 MHz
Efficiency:• 10% Depolarization
• Still 35% Conversion
efficiency
• Record pulse energy
at 1250 nm via fiber
nonlinear frequency
conversion
Durations:• TL: 21 fs FWHM
• TL: 29 fs Square
• FROG: 49 fs Square
• 59% compression
Time [fs]
Wav
elen
gth
[nm
]
Mea
sure
d
Rec
on
stru
cted
-100 0 100
600
620
640
660
S.R. Domingue et al., Opt. Express 22, (2014)
Part II1250 nmPulseGeneration
1250 nm Pulse Characteristics after Nonlinear Frequency Conversion from 1060 nm
TL: transform-limited duration
FROG: frequency resolved optical gating
Wavelength [nm]
Spec
tral
Phas
e [r
ad]
1150 1250 1350
-10
12
3
Pow
er S
pec
trum
[a.
u.]
-100 0 1000
0.2
0.4
0.6
0.8
1
Time [fs]
Inte
nsi
ty [
a.u
.]
TL
Recon.
Phantom-FROG
At 2 nJ (200 mW coupled) there is a problematic instability. We suspect back reflections
from the fiber exit face feeding back into the fiber amplifier
1250 nm pulses have
nearly binary changes of
state.• Small changes in spectral
phase
• Larger changes in power
spectrum
• Huge change in intensity
profile
Output of Photonic
Crystal Fiber
S.R. Domingue et al., Opt. Express 22, (2014)
Part II1250 nmPulseGeneration
Average Power Limitations in Nonlinear Frequency Conversion to 1250 nm Pulses
-1
0
1
2
3
4
5
Wavelength [nm]
Spec
tral
Phas
e [r
ad]
1150 1250 1350
-1
0
1
2
3
4
5
Pow
er S
pec
trum
[a.
u.]
-100 0 1000
0.2
0.4
0.6
0.8
1
Time [fs]
Inte
nsi
ty [
a.u
.]
Three-Photon Excitation Fluorescence Microscopy in the 1250 nm Biological Imaging Window
Preliminary results indicating feasibility and needs for future system
`
ANDi
`
NPA
`
MC
SC
ANDi: all normal dispersion laser
NPA: nonlinear power amp.
MC: Martinez compressor
SC: supercontinuum generation
FFPS: flat-field pulse shaper
PCF: photonics crystal fiber
`
FFPS
Stage-Scanning 3-Photon
Excitation Fluorescence
Microscopy (0.5 NA)
PCF-100 0 100
0
0.2
0.4
0.6
0.8
1
Time [fs]
Inte
nsi
ty [
a.u
.] 0.6 nJ
44 fs
xY
Part II1250 nmMicroscopy
S.R. Domingue et al., Opt. Express 22, (2014)
100 𝝁m
Three-Photon Excitation Fluorescence Microscopy in the Second Biological Imaging
Window
Preliminary results indicating feasibility and needs for future system
GFP labeled mouse pancreas
Fluorescein dyed lens tissue
S.R. Domingue et al., Opt. Express 22, (2014)
30 𝝁m
apart
Part II1250 nmMicroscopy
Goals and Future Work Generating 1250 nm pulses
100 𝝁m
Fluorescein dyed lens tissue
Future Work:• Reduce Parasitic Back-Reflections from
Photonic Crystal Fiber systems• Increase 1250 nm pulse energy
• Improve Spatial Mode (Degrading exit fiber
face => reduces spatial mode quality)
Part IIFinale
Goal• >1.5 nJ incident on
surface
HI-1060
1-2 mm
PCF-Zero-2
30 mm
Sealed Chamber WindowsFerrules /
Potted Fiber
Seeding Dual-Zero Dispersion Photonic Crystal Fiber with 1060 nm pulses to directly generate
1250 nm pulses for Nonlinear Microscopy
• Limited to < 3nJ (~200 mW) by
Polarization Instability in Fiber
Future Work Generating ~1300 nm pulses for Nonlinear Microscopy
• Photonic Crystal Fiber with single
Zero-Dispersion Wavelength at < 1060
nm
• Soliton Self-Frequency Shifting • 0.3 nJ, 50 fs seed pulse at 1340 nm
• Amplify in Nd-Doped ZBLAN fibers• >5 nJ
• <100 fs pulse
Part IIFinale
`
ANDi
`
NPA
`
MC
ANDi: all normal dispersion laser
NPA: nonlinear power amp.
MC: Martinez compressor
SSFS: soliton self-frequency shifting
SS
FS
Pump
Module #1
4 copies
2 copies
+ GDD
Chirped-Pulse
Amplifier
Divided-
Pulse
Amplifier
- GDD
8 copies>5 nJ 1340 nm
Amplified pulse
output
1050 nm 795 nm
LD’s
PBS
DM1
DM2
Nd-ZBLAN #1
Pump
Module #3
FR+M
Nd-ZBLAN #2Pump
Module #2
Pump
Module #4
Move to Nd-doped Fiber Amplifier for surplus pulse energy in the 1300
nm biological imaging window
Chapter 9
Chapter 11
Part III: Hyperspectral Imaging
Multiplexing spectral measurements in time using unique frequency signatures
`
ANDi
`
NPA
`
MC
SC
ANDi: all normal dispersion laser
NPA: nonlinear power amp.
MC: Martinez compressor
SC: supercontinuum generation
FFPS: flat-field pulse shaper
`
FFPS
• ~25 fs
• >350 kW
• 10 nJ @ 60 MHz
• 600 mW
`
Laser Scanning
Multi-photon
Microscopy
`Nonlinear Laser
Machining
`
3-Photon
Excitation
Fluorescence
Microscopy
`
Hyperspectral
Imaging via
Excitation Labeled
Fluorescence
`
Nonlinear 1250
nm pulse
generation`
Second
Harmonic
Generation
Part IIIIntro.
Hyperspectral Imaging and Spectral Multiplexing more Generally
The desire on both fronts is to eliminate the requirement of a
1- or 2-dimensional photo-detector array
Futia, G. et al, Opt. Express 19, (2011)
Diebold, E. D. et al., Nat. Photonics 7, (2013)
Spatial Frequency Modulated
Imaging
Encodes Spatial Information
into the time-domain
Decouples imaging speed from
camera frame rate:• Single-Element Photodetector
bandwidths virtually unlimited
• Demonstrated 4 kHz confocal
fluorescence frame rates*
To add a hyperspectral modality,
we desire something other than
spatially disperse and capture with
an array detector!
Part IIIIntro.
The desire on both fronts is to eliminate the requirement of a
1- or 2-dimensional photo-detector array
Weber, J. R. et al., J. Biomed. Opt. 16, (2011)
Hyperspectral information
enables:
Encodes Spatial Information
into the time-domain
• Scattering, absorption, and
fluorescence (tissue optical
properties => disease indicators)
• Imagination limited
applications: • electronic to vibrational
interactions
• photo-thermal imaging
• excitation labeled fluorescence
Part IIIIntro.
Hyperspectral Imaging and Spectral Multiplexing more Generally
Spatial Frequency Modulated
Imaging
Light Labeling for Hyperspectral Imaging and General Spectroscopy: a New Method
Down shifting the angular carrier frequencies of an optical bandwidth into a directly
measureable frequency range: 500 THz @ 2000 THz => 100’s Hz @ 100-10 kHz
Related to Fourier Transform Spectroscopy,
except we encode a wavelength dependent
intensity modulation onto power spectrum
Moves the “Spectrometer” from
the back-end of the experiment
to the front-end
Transfers wavelength/spectroscopic
information into the time-domain
Fluorescent Intensity
Fluorescent Intensity and
Fluorophore Identification
Alexa
Fluor 514
Alexa
Fluor 546
Part IIIIntro.
S.R. Domingue et al., Optica, Publication Pending
Modulating the spectral line-focus within a folded 4-F Martinez with a
frequency modulated reticle
Light Labeling via Spectral Intensity Modulation
The color to space mapping:
𝑓 Ω = 𝑟0 + 𝛼Ω
Diffraction
Grating
Side ViewF
F
Motor
Mirror
Reticle Side view of reticle and
spectral line focus
M(r,t) = 1
2+
1
2𝑠𝑖𝑔𝑛(cos 2𝜋Δ𝑘 𝑟𝜈𝑅𝑡 )
Δ𝑘 is in cycles / unit radius:
Δ𝑘 = 5 𝑚𝑚−1 → 𝑀(𝑟 = 1 𝑚𝑚) has 5
cycles (or periods) per revolution of the
reticle and
𝑥(𝜔) = 𝛼Ω
𝑥(𝜔0) = 0
The lateral displacement of
wavelets after the lens:
𝑥
Ω = 𝜔 − 𝜔0.
sign() enforces binary modulations
𝜈𝑟: rotation rate𝛼 ≈
2𝜋𝑐𝑁𝐹
𝜔02 cos 𝜃
Part IIILightLabeling
𝜈𝑅
Modulating the spectral line-focus within a folded 4-F Martinez with a
frequency modulated reticle
The Physical Means of Encoding the Spectral
Modulation for Light Labeling
𝜈𝑅
The color to space mapping:
𝑓 Ω = 𝑟0 + 𝛼Ω
Combined with the spatial mask fcn:
M r, t =1
2+1
2𝑠𝑖𝑔𝑛(cos 2𝜋Δ𝑘 𝑟𝜈𝑅𝑡 )
Affects a Power Spectral Modulation:
M r, t → M(Ω, 𝑡)
𝐼 Ω, 𝑡 = M Ω, 𝑡 𝐴 Ω
𝐴(Ω): power spectral envelope
Ω = 𝜔 − 𝜔0.
𝜈𝑟: rotation rate
sign() enforces binary modulations
Diffraction
Grating
F
F
Motor
Mirror
Reticle
𝑥(𝜔) = 𝛼Ω
𝑥(𝜔0) = 0
The lateral displacement of
wavelets after the lens:
𝑥
Ω = 𝜔 − 𝜔0.
Part IIILightLabeling
-90 -60 -30 0 30 60 90
0
0.25
0.5
0.75
Transmitted Power Spectrum as a function of Reticle Angle
The Temporal Intensity imparted by the Frequency Modulated Reticle
Photodetector
Wavelength-Independent, Single-Element Detection:
s 𝑡 = ∫ 𝐼 Ω, 𝑡 𝑑ΩWavelength
Pow
er S
pec
tru
m
𝜃 = 0∘
Excitation
Spectrum
Transmitted
Spectrum,
𝐼(Ω, 𝑡)
𝜃 = 3.5∘𝜃 = 7∘𝜃 = 10.5∘𝜃 = 17.5∘𝜃 = 70∘
Reticle Angle [deg] / 2𝜋 𝜈𝑟𝑡
s(t)
Part IIILightLabeling
Isolating the first sideband in the Fourier transform of 𝑠(𝑡), 𝑆(𝜈) = 𝐹𝐹𝑇{𝑠(𝑡)}, returns the excitation power spectrum
𝐹𝐹𝑇{𝑠(𝑡)}
The electric field carrier frequencies (𝜔) of the optical
bandwidth down shifted to a frequency range Hz-MHz
LiL
aP
ow
er
Spec
trum
, 𝑆(𝜈)
Modulation Frequency →← Wavelength
The Temporal Intensity imparted by the Frequency Modulated Reticle
LiL
a S
ignal
, 𝑠(𝑡)
Reticle Angle
/ 2𝜋 𝜈𝑟𝑡
Part IIILightLabeling
Measuring Power Spectra with the Light Labeling (LiLa) system
Spectral Resolution of Single-Element Detector Spectroscopy with Light Labeling
Wavelength [nm]
Po
wer
Sp
ectr
a [a
.u.]
510 530 550
LiLa
Ocean Optics
with:
Spectral resolution depends on:
• reticle Δ𝑘• monochromatic wavelet focal spots sizes
incident on the reticle, 𝑤𝐹𝑊𝐻𝑀
1
Δ𝑘𝑤𝐹𝑊𝐻𝑀
For the case 𝑤𝐹𝑊𝐻𝑀 ≪1
Δ𝑘, the Spectral Resolution:
𝛼𝛿Ω =1
Δ𝑘→ 𝛿𝜆 =
cos 𝜃
Δ𝑘 𝑁 𝐹(= 0.8 nm)
Numerical Values for our System
𝑤𝐹𝑊𝐻𝑀 = 20 𝜇m and 1
Δ𝑘= 200 𝜇m
sin 𝜃 = 𝑁𝜆0 − sin(17.5∘).𝑁=1200 L/mm: grating ruling density
𝐹=200 mm : focal length in Martinez S.R. Domingue et al., Optica, Publication Pending
Part IIILightLabeling
90 spectral points (limited by
available bandwidth)
Time / Reticle AngleL
iLa
Sig
nal
, 𝑠(𝑡)
FFT{𝑠(𝑡)}
Modulation
Frequency
LiL
aP
ow
er
Spec
trum
, 𝑆(𝜈)
𝜈′
Calibrated
Spectrometer
Razor Blade
Wav
elen
gth
Modula
tion
Fre
quen
cy
𝜈′
Razor Position
𝜆′
𝑥′
Co-Locating EdgesCaptured by
spectrometer
Captured by Light Labeling
With RazorNo Razor
Wavelength
Pow
er
Spec
trum
, C
ounts
𝜆′
𝜆[𝑛𝑚] = −0.14𝜈[𝐻𝑧] + 620
Measured mapping between 𝜈 and 𝜆
Locating known edges in power spectrum: for example, a razor blade in
spectral line focus
Calibrating the Light Labeling system
Wav
elen
gth
[nm
]
Modulation Frequency [Hz]500 650 800
510
530
550Data
Part IIILightLabeling
Background-Free Absorption Imaging via Light Labeling and Excitation Labeled
FluorescenceTransferring the Light Labeling modulations from the excitation power
spectrum to a fluorophore via absorption
Wavelength
No
rmal
ized
Sp
ectr
a Excitation
Fluorophore A
Fluorophore B
Absorbed
spectrum
(filled)
Fluorophore A
Fluorophore B
Wavelength
Em
issi
on
Sp
ectr
a
The broadband fluorescent emission retains
the temporal label of absorbed wavelengths
LiLa Excitation
Pulse
TimeE
mis
sio
n I
nte
nsi
ty
Higher
Frequencies
Time-
Domain
Wavelength-Domain
507090 Hz
Modulation
Frequency
S.R. Domingue et al., Optica, Publication Pending
Part IIIExcitationLabeledFluorescence
Background-Free Absorption Imaging via Light Labeling: Excitation Labeled Fluorescence
The fluorescent emission intensity contains the temporal Light Labeling
modulations, from which the absorbed power spectrum is recovered
Wavelength
Em
issi
on
Sp
ectr
a
Time
Em
issi
on
In
tensi
ty
Excitation
Fluorophore A
Fluorophore B
FFT{𝑠(𝑡)}
Exci
tati
on
Time
Wavelength →L
iLa
Rec
ov
ered
Ab
sorb
ed S
pec
tra
← Modulation Frequency
Absorption Spectra
Absorbed Spectra
Part IIIExcitationLabeledFluorescence
For weakly separated absorbed spectra the un-absorbed spectra can
magnify the contrast for algorithmic species identification.
FrequencyU
n-a
bso
rbed
Spec
trum
Part IIIHI-ELFMicroscope
𝑆un,546 𝜈 .
𝑆un,514 𝜈 .
𝑆un 𝑥𝑖 , 𝑦𝑗 , 𝜈 .
Absorbed Spectra Characteristics:
Alexa Fluor 514
• Δ𝜆 = 16 nm FWHM
• Δ𝜆0 = 524 nm
Alexa Fluor 546
• Δ𝜆 = 22 nm FWHM
• Δ𝜆0 = 527 nm
500 510 520 530 540 5500
0.2
0.4
0.6
0.8
1
Wavelength [nm]
Ab
sorp
tio
n S
pec
tra
[a.u
.]
0
0.2
0.4
0.6
0.8
1𝑆excitation𝑆546𝑆514.
𝑆546.
𝑆514.𝑆(𝑥𝑖 , 𝑦𝑗)
Using Excitation Labeled Fluorescence for Fluorescent Species Identification
𝑆𝑢𝑛−𝐴𝑏𝑠 𝜆 = 𝑆𝐸𝑥𝑐 𝜆 − 𝑆𝐴𝑏𝑠𝑜𝑟𝑏𝑒𝑑(𝜆)
y
x
t
𝑠(𝑥, 𝑦; 𝑡)
The frame rate of a high speed camera and Light Labeling converts the time-domain
trace from each camera pixel into a spectral measurement
Hyperspectral Imaging via Excitation Labeled Fluorescence (HI-ELF)
Sequential HI-
ELF Micrographs
of Fluorescent
Intensity
reference spectrum
Camera
Dichroic
Mirror
Long Pass
Filter
HI-ELF Microscope
Part IIIHI-ELFMicroscope
10x /
0.2 NA
200 250 300 350
552 535 518 504
Frequency [Hz]
Spec
trum
[a.
u.]
Wavelength [nm]
FFT{s(t)}𝑆𝐸𝑥𝑐 𝜆 −
𝑆′(𝜆)
0 0.25 0.5-6
-4
-2
0
2
4
Time [s]
Sig
nal
, s(
t) [
V]
200 250 300 350-0.5
0
0.5
1
1.5
2552 535 518 504
Frequency [Hz]U
n-A
bso
rbed
Spec
trum
[a.
u.]
Wavelength [nm]
Processing the Light Labeled signal at each pixel
y
x
t
𝑠(𝑥, 𝑦; 𝑡)
The frame rate of a high speed camera and Light Labeling converts the time-domain
trace from each camera pixel into a spectral measurement
Hyperspectral Imaging via Excitation Labeled Fluorescence (HI-ELF)
Sequential HI-
ELF Micrographs
of Fluorescent
Intensity
Species weighting at each pixel by least mean-squared-error search
𝑆 𝑥𝑖 , 𝑦𝑗 , 𝜈 = 𝑎546 𝑥𝑖 , 𝑦𝑗 𝑆un,546 𝜈 + 𝑎514 𝑥𝑖 , 𝑦𝑗 𝑆un,514(𝜈)
Species weighted micrographs:
𝐼514,546 𝑥𝑖 , 𝑦𝑗 = ∫ 𝑆 𝑥𝑖 , 𝑦𝑗 , 𝜈𝑎514,546(𝑥𝑖,𝑦𝑗)
𝑎514 𝑥𝑖,𝑦𝑗 +𝑎546(𝑥𝑖,𝑦𝑗)𝑑𝜈
reference spectrum
Camera
Dichroic
Mirror
Long Pass
Filter
HI-ELF Microscope
Frequency
Un-a
bso
rbed
Sp
ectr
um
𝑎546 𝑥𝑖 , 𝑦𝑗 𝑆un,546 𝜈 .
𝑎514 𝑥𝑖 , 𝑦𝑗 𝑆un,514 𝜈 .
𝑆 𝑥𝑖 , 𝑦𝑗 , 𝜈 .
These fits are distinguishing between Absorbed
Spectra with centroids separated by 3 nm
Part IIIHI-ELFMicroscope
10x /
0.2 NA
y
x
t
𝑠(𝑥, 𝑦; 𝑡)
Micrographs of differentiated fluorescent species based on absorbed spectra, recovered
from the temporal dynamics of the broadband fluorescent emission intensity
Hyperspectral Imaging via Excitation Labeled Fluorescence (HI-ELF)
Sequential HI-
ELF Micrographs
of Fluorescent
Intensity
Images of total emission intensity (Left) and emission intensity weighted
by species coefficients, 𝐼514,546 𝑥𝑖 , 𝑦𝑗 , (Right)
Alexa Fluor 514 and 546 in blue and red, respectively
Camera
Dichroic
Mirror
Long Pass
Filter
HI-ELF Microscope
reference spectrum
Part IIIHI-ELFMicroscope
S.R. Domingue et al., Optica, Publication Pending
10x /
0.2 NA
Future Work for Hyperspectral Imaging via Light Labeling
Part IIIFinale
Future Work:• A Generalized Theory of Light
Labelingy
x
t
Future Work for Hyperspectral Imaging via Light Labeling
Part IIIFinale
Future Work:• A Generalized Theory of Light
Labeling
• Add Spatial Frequency Modulated
Imaging to current green Light
Labeling system
Single-
Element
Detector Cylindrical
Lens
Dichroic
Mirror
Future Work for Hyperspectral Imaging via Light Labeling
Part IIIFinale
Future Work:• A Generalized Theory of Light
Labeling
• Add Spatial Frequency Modulated
Imaging to current green Light
Labeling system
• Calibrate a Light Labeling
spectrometer at 2 𝜇m to capture
Second-Harmonic Generation
Frequency Resolved Optical
Gating (SHG-FROG) traces
Single-
Element
Detector
4 𝜇m
SHG-
FROG
Special Thanks
The Bartels Ultrafast Lab
Philip Schlup
Omid Masihzadeh
Jesse Wilson
Jeff Field
David Winters
David Kupka
David Smith
Keith Wernsing
Patrick Stockton
Committee Members
Amber Krummel
Diego Krapf
Mario Marconi
Funding Sources
Department of Energy
Office of Naval Research
National Institute of Health
The Keck Foundation
Colorado State University
Special Thanks
My ANDis
Blandi (CSU)
Negato (CSU)
Blazer (CSU)
Kandi (CSU)
Boomer/Athena (CSU)
Mandi (CU-Denver)
Jandi (CSM)
Nandi (NREL)
Extra-Special Thanks
My Adviser
Randy Bartels
My Fiancée
Stephanie Krueger
Q.E.D.
Nonlinear Pulse Compression a la Warhol