1 of 37

1

All-Fiber Frequency-Agile Dual Comb Spectroscopy

with High Sensitivity

Debanuj Chatterjee,

PhLAM, University of Lille, France

Vanguard Seminar, Rabat, Morocco

2 of 37

2

All-Fiber Frequency-Agile Dual Comb Spectroscopy

with High Sensitivity

3 of 37

3

Motivation : Optical Frequency Comb

What is an optical frequency comb?

  • Equally spaced frequencies
  • Ruler of light
  • Pulse train in time

How can we use it?

  • Detect chemicals
  • Chemical composition of astronomical objects
  • Multiplexing of data for optical communication
  • Analog signal distribution (radar signals)

Is it a big deal?

  • Theodor Hӓnsch and John Hall received half of the Nobel Prize in Physics in 2005
  • Contributions to the development of laser-based precision spectroscopy, including the optical frequency comb technique

(source : nobelprize.org)

(source : Victor Torres-Company et al, Laser Photonics Rev. 2014)

4 of 37

4

All-Fiber Frequency-Agile Dual Comb Spectroscopy

with High Sensitivity

5 of 37

5

Spectroscopy : Principle

(Nathalie Picqué et al, Nature Photonics, 2019)

Dual comb spectroscopy

6 of 37

6

All-Fiber Frequency-Agile Dual Comb Spectroscopy

with High Sensitivity

7 of 37

7

Dual Comb Spectroscopy : Principle

Time domain

(Takuro Ideguchi, Optics & Photonics News, 2017)

Frequency domain

Vernier effect

f1

f2=f1+Δf

Δf

fAOM

1/f1

1/f2

Optical

RF

Optical

RF

Pay attention to scales!

  • fc ~ 192 THz (optical carrier)
  • f1, f2 ~ 500 MHz
  • Δf ~ 50 kHz
  • fAOM ~ 100 MHz

fc

fAOM

8 of 37

8

All-Fiber Frequency-Agile Dual Comb Spectroscopy

with High Sensitivity

9 of 37

9

All Fiber vs Free Space

  • Mirror alignment is time consuming
  • Susceptible to vibrations
  • Bulky

Free space

All fiber

  • Limited to near infrared
  • Compact
  • No alignment

10 of 37

10

All-Fiber Frequency-Agile Dual Comb Spectroscopy

with High Sensitivity

11 of 37

11

Electro-Optic Frequency Combs

Agile

Input

Output

frequency

frequency

Mach-Zehnder modulator

Electro-optic crystal (Pockel’s effect)

Agile EO comb!

Time

RF voltage

T

1/T

MZM transfer function

12 of 37

12

All-Fiber Frequency-Agile Dual Comb Spectroscopy

with High Sensitivity

  • Configuration (symmetric/asymmetric)
  • Broadband
  • Something else

13 of 37

13

Asymmetric vs Symmetric Dual Comb Spectroscopy

Asymmetric/Dispersive Configuration

Symmetric/Collinear Configuration

  • Only one comb passes through the sample
  • Both amplitude and phase is recovered
  • Amplitude sensitivity weaker by 3 dB
  • Both combs pass through the sample
  • Only amplitude is recovered
  • Amplitude sensitivity is stronger by 3 dB

We focus on symmetric/collinear configuration

14 of 37

14

All-Fiber Frequency-Agile Dual Comb Spectroscopy

with High Sensitivity

  • Configuration (symmetric/asymmetric)
  • Broadband
  • Something else

15 of 37

15

Spectral Broadening

  • Frequency spectrum broadened with a nonlinear medium
  • Thanks to four wave mixing

Sample’s absorption

Cascaded four wave mixing

Nonlinear medium

Nonlinear medium

16 of 37

16

Partial Summary

Basics of Dual Comb Spectroscopy

All-fiber dual comb source

Symmetric configuration (more sensitive amplitude detection)

EOM based comb with frequency agility

What can we do better?

Agile

Time

RF voltage

In

Out

17 of 37

17

All-Fiber Frequency-Agile Dual Comb Spectroscopy

with High Sensitivity

  • Configuration (symmetric/asymmetric)
  • Broadband
  • Signal to noise ratio (SNR)

18 of 37

18

Signal-to-Noise Ratio

SNR = Signal power/noise floor level

Frequency

Power (dBm)

Signal

Noise floor

SNR

How to increase the SNR?

Increase the number of ifg pulses in the time domain

19 of 37

19

SNR Improvement Strategy

Unused space

(Yu Zhang et al, Optics Letters, 2021)

Strategy : Increase Δf

(to pack more ifg pulses)

1/Δf

Time domain interferogram

20 of 37

20

SNR Improvement Strategy

Strategy : Increase Δf

1/Δf

Time domain interferogram

But, there is a problem…

21 of 37

21

AOM Shift

Frequency domain

f1

f2=f1+Δf

Δf

fAOM

Pay attention to scales!

  • fc ~ 192 THz (optical carrier)
  • f1, f2 ~ 500 MHz
  • fAOM ~ 100 MHz (~f1/4)
  • Δf ~ 50 kHz

zoom

zoom

fc

fAOM

 

fAOM

Optical

RF

22 of 37

22

Problem of RF Spectral Aliasing

f1

Strategy

f1

f

f

RF spectrum

f1/2

f1/2

f1/4

f1/4

Available spectral width

aliasing

fAOM

[Constant number of comb lines in both case]

LPF

0

0

23 of 37

23

SNR Improvement Strategy

Strategy : Spectral filtering

(to shape the comb and increase effective Δf)

Standard

Spectral filtering

(Akiko Nishiyama, Optics Express, 2017)

(Nazanin Hoghooghi, Applied Physics B, 2021)

  • Avoids aliasing
  • Filter induces losses
  • Complicated to implement

24 of 37

24

SNR Improvement Strategy

Numerical simulation of time multiplexing with EOM combs

Comb 1

Comb 2

(standard)

Comb 2

(multiplexed)

standard

multiplexed

Phase shift

Strategy : Time domain shaping of optical pulses (time multiplexing)

Δf=1 MHz

f1=500 MHz

f2=499 MHz

CW source

EOM

EOM comb

RF pulse

25 of 37

25

SNR Improvement Strategy

Numerical simulation of time multiplexing

SNR increase

  • SNR improved / Faster acquisition speed (for given SNR)
  • Aliasing avoided
  • Spectral resolution is reduced

26 of 37

26

Experimental Scheme

50%

50%

Arm1

Arm 2

100 MHz

f1 = 500 MHz

f2 = 500 MHz – 50 kHz

Laser

AOM

IM

IM

Common clock

AWG

AWG

PD

PD

DCF

50%

50%

Channel 1

Channel 2

Oscilloscope

AWG : arbitrary waveform generator, AOM : acousto optic modulator, EDFA : Erbium doped fiber amplifier, DCF : dual core fiber (nonlinear), IM : intensity modulator, PD : phorodetector, WS : waveshaper, VOA : variable optical attenuator

EDFA

EDFA

VOA

WS

Multiplexing

Lorentzian filter

(50 GHz FWHM)

Time domain ifg (1.4 multiplexed)

Spectral broadening

50%

50%

Reference

multiplexed

standard

20 μs

14.3 μs

80 ps pulsewidth

27 of 37

27

SNR improvement / Faster Acquisition

  • Multiplexed case provides a larger SNR (for same acquisition time)
  • Multiplexed case provides a faster acquisition (for same SNR)
  • Multiplexed case show extra spiking features due to a sinc envelope

Time domain ifg (mul=1.4)

Frequency domain ifg (mul=1.4)

standard

multiplexed

28 of 37

28

Theoretical Analysis

 

0

t

3T

6T

-3T

-6T

 

 

0

t

3T

6T

-3T

-6T

S(t)=P(t)

P(t)

G(t)

w

0

t

3T

6T

-3T

-6T

L(t)=P(t).G(t)

0

t

3T

-3T

S(t)

 

 

Standard

Multiplexed

29 of 37

29

Theoretical Analysis

  • For the simplest case of w=T, the multiplexed case gives the Fourier domain signal for the standard case
  • For the multiplexed case, we see a dense Dirac comb, with a frequency spacing of 1/w (smaller than 1/T), which is further shaped by a sinc like envelope

0

t

3T

-3T

S(t)

 

 

Multiplexed

 

For w=T,

 

 

30 of 37

30

11.5 dB

SNR improvement with Scan of Multiplexing

  • Multiplexed case provides a larger SNR
  • For multiplexing rate 3.3, about 11.5 dB increase in SNR
  • Experiment matches theory

Standard

Part of frequency domain interferogram spectrum

31 of 37

31

Proof of Concept for Time Multiplexed DCS

f1=500 MHz

Δf=50 kHz

Lorentzian filter

No filter

Frequency domain ifg

Comb 1

Comb 2

PD

PD

Filter

VOA

50%

50%

50%

50%

Reference

FT

FT

Attenuation

Frequency

32 of 37

32

Proof of Concept for Time Multiplexed DCS

  • f1=500 MHz, Δf=50 kHz
  • Filter shape reconstructed for standard and multiplexed case
  • Residue (Experiment-Truth) is lower for multiplexed case
  • Larger SNR / faster acquisition for multiplexed case

Lorentzian filter

No filter

Frequency domain ifg

Waveshaper filter reconstruction

Time domain ifg

multiplexed

standard

residue

Data

processing

33 of 37

33

Single Shot Measurements with Multiplexing

Multiplexed

Standard

Waveshaper reconstruction

Ifg freq (no absorption)

Ifg freq (absorption)

Ifg time (no absorption)

Ifg time (absorption)

Ifg peak missed for std case

  • Standard case, ifg peak (time domain) is missed, we cannot reconstruct the waveshaper function
  • Multiplexed case, we get the waveshaper function since there is always an ifg peak
  • Delta frep=2.5 kHz (ifg beating period for standard case=400 micro sec)
  • Multiplexing = 3.33 (ifg beating period for multiplexed case=121 micro sec)
  • Comparison : 1. standard case (blue) 2. multiplexed case (mul=3.3) (red)
  • Waveshaper : Lorentzian profile with 10 dB attenuation, 50 GHz FWHM
  • Oscilloscope : 1 G samples/s, 200 k pts, total recording time : 200 micro sec (less than standard beating period, but more than multiplexed beating period)

Lorentzian shape

detected

34 of 37

34

Conclusion

Conclusion

  • EOM-based, all-fiber, frequency-agile DCS
  • Temporal folding or time multiplexing for larger SNR
  • Demonstration of proof-of-concept of the time multiplexed DCS
  • Striking improvement for single shot measurement

All-Fiber Frequency-Agile Dual Comb Spectroscopy

with High Sensitivity

35 of 37

35

Perspectives

Multidimensional spectroscopy with multicore fibers

Compressive sensing : hybrid approach

Fast sensing in engines

(Eve-Line Bancel et al, Nature Communications, 2023)

(Anthony Draper et al, Optics Express, 2019)

(Akira Kawai et al, Scientific Reports, 2021)

36 of 37

36

Acknowledgements

Nonlinear optics group at PhLAM

Matteo Conforti

(faculty)

Arnaud Mussot

(group leader)

Francesco Tani

(faculty)

Siddharth Sivankutty

(faculty)

Thomas Bunel

(PhD student)

Eve-line Bancel

(PhD student)

Stefano Negrini

(PhD student)

Debanuj Chatterjee

(Postdoc)

37 of 37

37

Thank You!