Mid-infrared feed-forward dual-comb spectroscopy

Zaijun Chena, Theodor W. Hänscha,b,1, and Nathalie Picquéa,1

a
Max-Planck-Institut für Quantenoptik, 85748 Garching, Germany; and bFakultät für Physik, Ludwig-Maximilians-Universität München, 80799 München,
Germany

Contributed by Theodor W. Hänsch, January 7, 2019 (sent for review November 6, 2018; reviewed by Federico Capasso and Y. Ron Shen)

Mid-infrared high-resolution spectroscopy has proven an invalu- and interband cascade (11) lasers, microresonators (12, 13),
able tool for the study of the structure and dynamics of molecules novel solid-state (14) or doped-fiber (15) gain media, improved
in the gas phase. The advent of frequency combs advances the approaches to nonlinear frequency conversion (16) benefiting
frontiers of precise molecular spectroscopy. Here we demonstrate, optical parametric oscillation, difference frequency generation,
in the important 3-μm spectral region of the fundamental CH and spectral broadening in nonlinear waveguides (17, 18).
stretch in molecules, dual-comb spectroscopy with experimental Furthermore, the technical requirements for a dual-comb spec-
coherence times between the combs that exceed half an hour. trometer are more challenging than those for a metrological fre-
Mid-infrared Fourier transform spectroscopy using two frequency quency comb. A dual-comb interferometer requires that the optical
combs with self-calibration of the frequency scale, negligible con- delay between the pairs of interfering pulses is controlled, from
tribution of the instrumental line shape to the spectral profiles, pulse pair to pulse pair, within interferometric accuracy. Timing
high signal-to-noise ratio, and broad spectral bandwidth opens up fluctuations in the attosecond range can be detrimental. In the
opportunities for precision spectroscopy of small molecules. Highly near-infrared domain, coherence between the two combs at most
multiplexed metrology of line shapes may be envisioned. reached the order of 1 s in different experimental realizations (19–
21). The measurement time can be extended to several minutes
frequency comb | laser spectroscopy | mid-infrared | molecular or hours by more or less complex techniques of phase correction,
spectroscopy that are implemented through analog (22), digital (23), or a pos-
teriori computer (24) treatment. Maintaining coherence in the mid-
infrared domain is even more difficult than in the near-infrared
M id-infrared frequency combs with unprecedented mutual
coherence times open the important spectral region
around 3 μm to broad-spectral-bandwidth gas-phase molecular
region, because, e.g., of the more complex laser systems and of
the difficulty to design continuous-wave mid-infrared lasers with a
spectroscopy with high precision and negligible instrumental hertz linewidth, that could serve as optical references for the combs.
linewidth. A frequency comb (1) is a spectrum that spans a broad Finally, experimenting in the mid-infrared region is generally
bandwidth and that is made up of phase-coherent evenly spaced more demanding than in the visible and near-infrared range,
narrow laser lines, the frequency of which may be known within because the optics technology is not as advanced.
the accuracy of an atomic clock. Frequency combs find numerous Consequently many reports of mid-infrared dual-comb spec-
new applications. In spectroscopy over broad spectral bandwidths, troscopy, although pointing to an intriguing potential, have
remained at a stage of promising proofs of principle. Many laser
they are used to directly interrogate the transitions of atomic and
systems developed lately have led to a variety of implementations
molecular species. One of the techniques of broadband frequency
(14, 25–30). For instance, the first demonstrations with fre-
comb spectroscopy, dual-comb spectroscopy (2), currently attracts
quency combs of large line spacing, such as microresonators (31)
significant interest. Dual-comb spectroscopy is a technique of
and quantum cascade lasers (32), highlight novel opportunities in
linear (3–6) and nonlinear (7, 8) Fourier transform spectroscopy
without moving parts. In most implementations of linear spec-
troscopy (Fig. 1A), a frequency comb interrogates an absorbing Significance
sample and is heterodyned against a second comb of slightly dif-
ferent line spacing, at a single fast photodetector. The measured Each molecule leaves characteristic absorptions, which enable
time-domain interference leads to a spectrum through the har- one to unambiguously identify and quantify it, and to un-
monic analysis provided by Fourier transformation. Dual-comb derstand its structure and dynamics. Such fingerprints, strong
interferometers hold promise for efficient spectrometers that in the mid-infrared region of the electromagnetic spectrum,
would offer a unique combination of distinguishing features: (i) can be accessed by measuring the intensity of light transmitted
absence of geometric limitations, unlike Michelson-based or through a medium as a function of its frequency. Here, we
describe a spectrometer for mid-infrared spectroscopy based
grating spectrometers, theoretically leading to unlimited spans
on two frequency comb generators, light sources emitting
and resolutions; (ii) absence of moving parts; (iii) multiplexed
narrow evenly spaced laser lines over a broad spectral band-
recording (in other words: all of the spectral elements are simul-
width. Our instrument without moving parts simulta-
taneously measured on a single photodetector), like other in-
neously measures broad high-resolution spectra with
terferential spectrometers, granting excellent overall consistency
calibration of the frequency scale directly to an atomic clock
of the spectral measurements and applicability in any spectral
and precise spectral line shapes. These features are obtained
regions; (iv) interrogation of the sample by laser lines of narrow from the experimental data without any need for corrective
width providing a negligible contribution of the instrumental line processing.
shape; (v) calibration of the frequency scale within the accuracy of
an atomic clock; (vi) short measurement times; (vii) use of laser Author contributions: Z.C., T.W.H., and N.P. designed research; Z.C. and N.P. performed
beams of high brightness. research; Z.C., T.W.H., and N.P. analyzed data; and Z.C., T.W.H., and N.P. wrote the paper.
The development of dual-comb spectroscopy in the regions of Reviewers: F.C., Harvard University; and Y.R.S., University of California, Berkeley.
interest to molecular spectroscopy, such as the mid-infrared The authors declare no conflict of interest.
range where most molecules have strong vibrational transitions, This open access article is distributed under Creative Commons Attribution-NonCommercial-
turns out to be rather challenging for several reasons. NoDerivatives License 4.0 (CC BY-NC-ND).
Novel laser technology is still emerging and many different 1
To whom correspondence may be addressed. Email: t.w.haensch@mpq.mpg.de or
approaches to mid-infrared frequency comb generation (9) have nathalie.picque@mpq.mpg.de.
been and are being explored. They include quantum cascade (10) Published online February 12, 2019.

3454–3459 | PNAS | February 26, 2019 | vol. 116 | no. 9 www.pnas.org/cgi/doi/10.1073/pnas.1819082116

A frep frep+ frep
called slave comb, has a slightly different repetition frequency frep +
master
comb
δfrep (with δfrep << frep) and a different CEO frequency fCEO +
δfCEO. Here, we set jδfrepj = 130 Hz, which leads to an optical free
slave spectral range of 38 THz. For interferometry measurements, the
optical frequency (THz) comb
frep optical retardation between pairs of interfering pulses must be ac-
radio curately controlled during the duration of the measurement. Our
frequency
technique is based on feed-forward control of the relative CEO
B radio frequency (MHz) comb
frequency: an acousto-optic frequency shifter, at the output of the
Master comb
Er fiber slave laser system, keeps δfCEO constant over arbitrary times by
Er fiber PPLN frequency translating, with a fast response time (about 550 ns), all of
amplifier
self-referenced
the comb lines of the slave comb. A slow feedback loop (<1-kHz

gas cell
frep+ frep Pump bandwidth) adjusts the length of the cavity of the slave comb to
laser
fceo+ fceo maintain a fixed difference in repetition frequencies δfrep. The sig-
nals that enable one to monitor the relative fluctuations between the
Slave comb Er fiber PPLN digitizer
combs are two beat notes between two pairs of optical comb lines,
AOFS
Er fiber amplifier
one from each comb. Two continuous-wave lasers, emitting at 189
and 195 THz, respectively, serve as intermediate oscillators to pro-
Fig. 1. (A) Frequency-domain principle of dual-comb spectroscopy. Beat duce the two beat notes. For the spectroscopy setup, the output of
notes between pairs of lines of two optical combs, with a spacing of frep and
the master comb and that of the first-order diffracted beam of the
frep+δfrep, respectively, generate a radio-frequency comb of line spacing
δfrep. The radio-frequency comb can be digitally processed. (B) Experimental
acousto-optic frequency shifter of the slave comb have an available
setup for mid-infrared broadband dual-comb spectroscopy using feed- average power of about 250 mW each.
forward control of the relative CEO frequency. AOFS, acousto-optic fre- Each near-infrared signal comb is converted to a mid-infrared
quency shifter. idler comb by difference-frequency generation in a periodically
poled lithium-niobate (PPLN) crystal of a length of 3 mm and with
seven poling periods of about 30 μm. The pump beam is provided

PHYSICS
time-resolved spectroscopy for physical chemistry in the con- by a continuous-wave laser that emits at fpump ∼ 281.8 THz
densed phase. Nevertheless, quantum cascade (33) and inter- (1,063.8 nm) and that is phase-locked to a line of the master
band (34) cascade lasers, difference-frequency generation using comb. Its free-running linewidth at 100 μs is specified to be 50
erbium-doped fiber lasers (3), and electro-optic modulators (6) kHz. For frequency calibration of the spectra, we count the ab-
have produced, over narrow spectral bandwidths, high-quality solute frequency of the pump laser fpump, that of the 189-THz
spectra with resolved comb lines. Recently, the control of de- continuous-wave intermediate-oscillator laser, and the radio-
generate optical parametric oscillators has also been remarkably frequency parameters of the two signal combs during the mea-
advanced (35), while a combination of difference-frequency surement of the interferograms. The pump is split into two beams.
comb generators stabilized in the near-infrared and of real- For each, an average power of 2 W is combined with 250 mW of
time digital correction proved successful in producing (4) spec- signal beam by means of a dichroic mirror and each superimposed
tra with resolved comb lines over a broad spectral bandwidth. beam is focused onto a mixing PPLN crystal. We generate an idler
Not long ago, we devised (5) a scheme of dual-comb in- comb spanning up to 8.2 THz––a span limited by the phase-
terferometry that reports coherence times more than three or- matching bandwidth of the PPLN crystal––with an average
ders of magnitude longer than the state of the art. The initial power up to 90 μW. Such a technique of difference-frequency
demonstration has been performed in the near-infrared region comb generation has already been harnessed in dual-comb spec-
with erbium-doped fiber oscillators spectrally broadened in troscopy with erbium fiber lasers (3) and with electro-optic
nonlinear fibers. The near-infrared interferometer, based on modulators (6). Mixing a continuous-wave laser and an ultrashort
feed-forward stabilization of the relative carrier-envelope offset pulse is a fairly inefficient process; the simplicity of its imple-
frequency, does not require any type of phase correction for mentation and the fact that, due to detector nonlinearities, we do
continuous averaging of interferograms up to half an hour, not require high mid-infrared powers, justifies though that we
resulting in greater experimental simplicity and reduction of retained this scheme. The efficiency could be much increased with
possible artifacts. Here we explore the extension of our scheme a build-up cavity for the pump beam and by introducing a pulse-
to broadband highly accurate dual-comb spectroscopy in the stretching chirp for the pulses of the near-infrared signal combs.
mid-infrared 3-μm range, where the fundamental CH, NH, OH The repetition frequency of the idler comb is the same as that of
stretches in molecules are found. the signal comb. The CEO frequency of the idler comb is shifted
by an amount fpump [modulo frep]. As the same pump laser is used
Experimental Setup for the generation of the two idler combs, the two CEO fre-
Our 3-μm dual-comb laser system (Fig. 1B) is based on difference- quencies are shifted by the same amount and their difference
frequency generation of near-infrared femtosecond erbium-doped remains δfCEO. The center frequency of the idler comb is adjust-
fiber lasers. able between 82 and 100 THz by changing the temperature and
The near-infrared dual-comb system has already been described the poling period of the PPLN crystal. After each PPLN crystal, an
in ref. 5. We summarize here its main features for the clarity of the optical long-wavelength-pass filter filters out the residual pump
description. Two amplified mode-locked erbium-doped fiber laser and signal beams. The master idler comb beam passes through a
systems emit trains of pulses at a repetition frequency of about 100 single-pass absorption cell of a length of 70 cm. The master and
MHz. Their spectrum is centered at an optical frequency of about slave idler beams are combined on a pellicle beamsplitter. One of
190 THz and it spans more than 20 THz. The repetition frequency the outputs of the beamsplitter is focused onto a fast thermo-
frep and the carrier-envelope offset (CEO) frequency fCEO of the electrically cooled HgCdTe photodetector. To avoid detector
first fiber laser system (called master comb) are stabilized against the nonlinearities that generate distortions and systematic effects in
radio-frequency clock signal of a hydrogen maser, and all electronic the spectra, the total average power at the detector is kept weaker
instruments in our setup are synchronized to this clock signal. In the than 40 μW. The electric signal is amplified and filtered. It is
experiments described below, we choose frep = 100.00 MHz and sampled synchronously to the master comb repetition frequency,
fCEO = 20.0 MHz and both radio frequencies are counted during the frep = 100 MHz, by a fast digitizer. To keep the individual inter-
recording of the interferograms. The second fiber laser system, ferograms in phase and enable their averaging in the time domain,

Chen et al. PNAS | February 26, 2019 | vol. 116 | no. 9 | 3455
we set δfCEO = 0 [modulo δfrep]. The time-domain interference to the finite measurement time, which convolves the comb beat
signal is Fourier transformed to reveal the spectrum. Integration notes of negligible width. The full width at half maximum of the
times of several minutes are required to get good signal-to-noise observed comb lines is 6.8 Hz in the radio-frequency domain,
ratios over a broad spectral bandwidth. We average the raw inter- which corresponds to the Fourier transform limit. The mea-
ferograms in the time domain by simply adding them up. We do not surement time is similar to that in our recent near-infrared
perform any numerical corrections (like phase correction) to the spectra (5) that use the same feed-forward technique and we
interferograms or to the spectra. The displayed transmission spectra observe no degradations in the retrieved instrumental line shapes
are the amplitude of the complex Fourier transform of the averaged due to the operation in the mid-infrared region.
interferograms while the dispersion spectra are their phase. The spectrum sampled at exactly the comb-line positions re-
veals (Fig. 3A) the absorption by ethylene. Ethylene C2H4 is a
Results near-prolate planar asymmetric top molecule. Our span covers
Fig. 2A displays a transmission spectrum with 82,000 resolved the region of the ν9,ν11-stretching dyad of 12C2H4, which includes
comb lines, resulting from an interferogram averaged during the ν2 + ν12, 2ν10 + ν12, and ν9 + ν10 cold bands. Fig. 3 B and C
29 min (1,742 s). For illustrating the resolved comb lines, in- illustrates details of the transmission and dispersion spectra. The
terferometric sequences of 0.146 s including 19 individual in- signal-to-noise ratio at 92.3 THz peaks at 1,275 for a measure-
terferograms are time averaged and the resulting interferogram ment time of 1,742 s, corresponding to 30.5 s−1/2. The average
is Fourier transformed. In Fig. 2A, the spectrum is shown over signal-to-noise ratio is 570 (or 13.6 s−1/2) across the full span of
the entire useful spectral span of 8.2 THz, from 88 to 96.2 THz, 8.2 THz, and the corresponding figure of merit, which is given by
with triangular apodization of the interferogram. Fig. 2B shows the product of the average signal-to-noise ratio per unit square
an expanded view of the unapodized spectrum, illustrating that root of measurement time and the number of spectral elements
the rovibrational transitions of the absorbing sample, here eth- 82,000, is 1.1 × 106 s−1/2. Such a figure of merit is comparable to
ylene C2H4 in natural abundance, with a Doppler full width at those reported in the same spectral region for other experiments
half maximum of 215 MHz at 296 K are satisfactorily sampled by of narrowband (3, 6) or digitally corrected broadband (4) dual-
the comb lines of 100-MHz spacing. The ethylene pressure is 147 comb spectroscopy. When the interferograms are averaged in the
Pa and the absorption path length is 70 cm. The individual comb time domain, the average signal-to-noise increases (Fig. 4) as
lines (Fig. 2C) have the expected line shape, a cardinal sine due the square root of the measurement time: the experimentally

88 90 92 94 96
Frequency (THz)
B

92.3655 92.3660 92.3665 92.3670

Frequency (THz)

Fig. 2. Experimental dual-comb spectrum with re-

C solved comb lines, recorded within 1,742 s. (A) Entire
apodized spectrum: 82,000 individual comb lines
spanning 8.2 THz are resolved. (B) Detail of the
unapodized representation of A showing the (J′ =
13, Ka′ = 1, Kc′ = 13)–(J″ = 14, Ka″ = 0, Kc″ = 14) line in
a P branch of the ν9 band of 12C2H4 sampled by the
comb lines of 100-MHz spacing. (C ) Magnified
unapodized representation of A showing eight in-
dividual comb lines with their cardinal sine instru-
mental line shape. The frequency scale is converted
92.4780 92.4782 92.4784 92.4786 to the optical scale, thus the transform-limited width
of 6.8 Hz of the radio-frequency comb lines appears
Frequency (THz) as 5.3 MHz in the optical domain.

3456 | www.pnas.org/cgi/doi/10.1073/pnas.1819082116 Chen et al.

A

88 90 92 94 96
Frequency (THz)

B J'=15 K'a =4 J'=24 K'a =3 17

11
{
Transmittance (%)

100

75

50
J'=10 K'a =5 7
0.5
Phase (rad)

0.0

PHYSICS
Fig. 3. Experimental dual-comb spectrum of ethyl-
ene sampled at exactly the comb-line spacing of 100
MHz in the region of 92 THz. (A) The amplitude of
-0.5 the FT provides the transmission spectrum. (B) Mag-
91.15 91.20 91.25 91.30 91.35 nified representation of the transmittance (Top) and
Frequency (THz) dispersion (Bottom) spectra around 91.25 THz. Note
that the transmittance y axis does not go to zero.
C J'=19 K'a =7 7 Some rotational assignments (39), in the P branch of
the ν9 band, are given as an illustration. The strong
Transmittance (%)

100
lines are unresolved doublets of the (K′a = 5, ΔKa =
K′a − K″a = −1, ΔKc = K′c − K″c = ±1) series, while
50 the ΔKc = ±1 doublets are resolved for (K′a = 4,
ΔKa = −1). Ka (respectively, Kc) is the quantum
number for the projection of the rotational angular
0 momentum (excluding electron and nuclear spin) onto
1.0 the inertial axis of smallest moment (respectively,
Phase (rad)

largest moment). Prime and double primes refer to

upper and lower states, respectively. (C) Magnified
0.0 representation of the transmittance (Top) and dis-
persion (Bottom) spectra around 94.56 THz. The in-
tensity of the spectral envelope of our dual-comb
-1.0 system is significantly weaker here, in the region of
94.54 94.55 94.56 94.57 94.58 94.59 Q branch of the (K′a = 7, K″a = 6, ΔKc = −1) lines of
Frequency (THz) the ν9 band.

achieved interferometric coherence reaches half an hour. Currently, the weaker ones belong to the ν3 + ν41 − ν41, ν3 + ν51 − ν51, ν2 +
the maximum averaging time is technically limited by our data (2ν4 + ν5)1II − ν41, and ν2 + (ν4 + 2ν5)1II − ν51 hot bands (their
acquisition system. Nevertheless, Fig. 4 indicates that coherence rotational assignment may be found in ref. 36, following nota-
over longer averaging times could likely be achieved. Our tions proposed by ref. 37) and to the ν3 band of 13C2H2. The
achievement of a coherence time of half an hour, obtained by pressure of acetylene in natural abundance is 10.7 Pa and the
time-domain averaging of the interferograms without any data absorption path length is 70 cm. The spectrum results from an
treatment, is already significantly longer than that of state-of- interferogram averaged during 2,050.2 s (34.2 min) in the time
the-art dual-comb systems, where an experimental mutual co- domain. The signal-to-noise ratio at around 98.4 THz is 1,360,
herence time of 1 s used to be an excellent figure. This provides corresponding to 30 s−1/2. The average signal-to-noise ratio, across
convincing evidence of the power of feed-forward dual-comb the span of 8 THz, is about 500 (or 11 s−1/2). The signal-to-noise
interferometry in the molecular fingerprint region and it renders ratio increases as the square root of the measurement time.
superfluous any types of phase corrections or other numerical As a preliminary analysis of our molecular spectra, we have
manipulations of the experimental data. fitted the profiles and determined the positions of a limited
Adjusting the temperature and poling period of the PPLN number of lines among the hundreds present in each crowded
crystals slightly tunes the idler spectrum and makes it possible to spectrum. Fig. 6 exemplifies the results of the nonlinear least-
optimize the spectrum to the transitions of different molecules, squares fit of Doppler profiles to three experimental lines in the
as illustrated in Fig. 5, which displays the region of the P branch spectrum of ethylene shown in Fig. 3. The standard deviation of
of the ν3 band of acetylene. The strongest lines in Fig. 5B are the “observed-fitted” residuals is 0.12% and no systematic re-
assigned to the ν3 and ν2+(ν4 + ν5)+0 cold bands of 12C2H2, while sidual signatures are distinguishable at the noise level. We

Chen et al. PNAS | February 26, 2019 | vol. 116 | no. 9 | 3457
 1000 comb that interrogate the sample. Their width is about 100 kHz
Average signal-to-noise ratio
Measured
at an integration time of 1 min, thus three orders of magnitude
Linear fit
narrower than the Doppler width. Therefore, the intrinsic mo-
lecular line shape can directly fit the line profile. Even for
spectroscopy of Doppler-broadened profiles at room tempera-
100
ture, improved line parameters––not limited to positions and
shifts––may be determined. To date, most of the precise spec-
troscopic individual line parameters retrieved from high-
resolution infrared spectra have derived from Michelson-based
10 Fourier transform spectroscopy (FTS) with an incoherent light
1 10 100 1000 source, where the wavenumber-scale calibration relies on the
Measurement time (s) presence of molecular lines, accurately measured by other
Fig. 4. Evolution of the average signal-to-noise ratio with the measurement means, which can serve as wavenumber standards in the spectra.
time in an ethylene spectrum. The fit of a line to the experimental data The instrumental line shape in Michelson-based FTS has a width
shows a slope of 0.506 (5) indicating that the signal-to-noise ratio shows generally of the same order of magnitude as the intrinsic width of
dependence with the square root of the measurement time. The spectrum the observed lines and has to be included in the fits. The use of
corresponding to a measurement time of 1,742 s is shown in Figs. 2 and 3. an incoherent light source in Michelson-based FTS enables re-
cordings over extended spectral bandwidths that are still out of
reach with frequency-comb spectroscopy. Over limited band-
compare the positions of about 250 selected isolated lines of 12 widths, which are commonly used in Michelson-based FTS for
C2H4 to those available in the literature. The retrieved positions signal-to-noise ratio improvement, coherent light sources of high
in our line selection present an average statistical uncertainty of brightness represent (41, 42) a significant increase in signal-to-
0.8 MHz with an standard deviation of 0.4 MHz. An extensive noise ratio (or decrease in measurement time). Our system
linelist is available in the 2016 HITRAN database (38), which benefits from this high brightness and adds the advantages of the
reproduces the report (39). In the report (39), the precision of accuracy of the frequency scale, of the precision of the line-
the line-position measurements is given to ±20 MHz and the
profile determination, and of the absence of moving parts.
accuracy is said to be “slightly worse.” The average value of the
A limitation of the present dual-comb system is the low average
difference between the positions in ref. 39 and ours is 0.3 MHz
power (90 μW) of each mid-infrared comb. For spectroscopy in a
and the standard deviation is 1 MHz. About 60 of our selected
single-pass cell or in a multipass cell with a small number of re-
lines, in the ν11 band, were also measured in ref. 40 with an
flections, the achieved average power is appropriate, as the total
accuracy of ±18 MHz. The average value of the difference be-
power on the detector needs to be kept lower than 30–40 μW to
tween the positions in ref. 40 and ours is 1.0 MHz and the
avoid detector nonlinearities. For laboratory spectroscopy over
standard deviation is 3.2 MHz. This indicates that precise posi-
long paths in a multipass cell or in a high-finesse cavity (43), the
tion measurements of Doppler-broadened transitions may be
achieved with our setup. The accuracy of the frequency scale is
granted by the direct calibration to the hydrogen maser. Special
care was also taken to operate the photodetector within its lin- A
earity range to avoid systematic shifts due to detector nonline-
arities. However, biases at the sample, such as pressure shifts,
have not been evaluated yet and will be the object of future
investigations.
Discussion
Here, we demonstrate self-referenced broadband mid-infrared
dual-comb spectroscopy, with resolved comb lines and negligible 94 96 98
Frequency (THz)
100 102

contribution of the instrumental line shape, which does not re-

quire any type of phase correction. We experimentally achieve B Pf/e(10) 3
1
4– 1
4 band Pf/e(3)
100
mutual coherence times that far exceed the state of the art.
Transmittance (%)

Meanwhile the instrumental setup for coherence control is rel-

atively simple. The implementation of real-time phase-correction
50
algorithms, that reconstruct the coherence, requires complex
computer hardware and programming. Compared with these,
our technique is straightforward. In contrast to locking the 0
Pe (14) band
3
Pe(7)
combs to optical references, our technique does not require 1.5 Pe(8) Pe(2)
+ 4+ band
reference cavities (nor other means of achieving a continuous- 2 5
Phase (rad)

wave laser with a linewidth of a hertz or below). As already

discussed in ref. 5, phase correction is not always implementable 0.0
and it may otherwise induce subtle artifacts detrimental to pre-
cision measurements. Thus, an instrument that does not require
any data corrections will be useful to broadband precision -1.5
spectroscopy and it will help ascertaining the precision frontiers 97.8 97.9 98.0 98.1 98.2 98.3
and identifying tiny systematic instrumental effects. Frequency (THz)
Our spectrometer provides self-calibration of the frequency
Fig. 5. Experimental spectrum of acetylene in the region of the ν3 band
scale directly within the accuracy of an atomic clock, as well as a with a resolution of 100 MHz. (A) The entire spectrum is measured within
negligible contribution of the instrumental line shape to the ex- 2,050.2 s. (B) Transmittance and dispersion spectra in the region of the P
perimental profile. As the achieved mutual coherence is excel- branch of the ν3 band of 12C2H2. The assignments for a selection of strong
lent, the strongest contribution to the instrumental profiles and weak lines are provided (36). The lines with a broader pedestal are due
comes from the optical comb lines of the self-referenced master to atmospheric water vapor, outside the cell, in the beam path.

3458 | www.pnas.org/cgi/doi/10.1073/pnas.1819082116 Chen et al.

 100 For experiments requiring significantly higher power, other comb
sources, such as optical parametric oscillators (35) or systems
based on difference-frequency generation between ultrashort pulses
Transmittance (%)

75
(4, 26, 44, 45), would have to be harnessed. The feed-forward rel-
ative control could be implemented on the mid-infrared beams.
Observed This still has to be experimentally demonstrated, but is technically
Fitted achievable. With our setup, the span currently covers 81–106 THz
50 Observed-fitted
(2.8–3.7 μm). Broader tunability over the entire 60–100-THz (3–
0.25 5-μm) atmospheric window could readily be achieved with the same
0.0
0.25 setup, by spectrally broadening the output of the erbium oscillators
92.681 92.682 92.683 92.684 92.685 92.686 in nonlinear waveguides. Lower frequency ranges, and particularly
Frequency (THz) the atmospheric window spanning 21–37 THz (8–14 μm), would
require a tunable continuous-wave laser as the pump in the
Fig. 6. Portion of the transmittance spectrum of 12C2H4 (black open dots, difference-frequency generation or, more interestingly, intrapulse
labeled observed), magnifying three rovibrational lines. The transmittance y
difference-frequency generation in a semiconductor (9), such as
scale in the spectrum stops at 50%.The experimental profile is satisfactorily
fitted by a Doppler line shape (red line, labeled fitted). The standard de- orientation-patterned gallium phosphide (16, 27).
viation of the residuals (“observed–fitted”) is at the noise level. The y scale The 3-μm region is rich in strong molecular transitions, be-
of the residuals is magnified compared with that of the top. longing, e.g., to fundamental bands of many small and large
organic, nitrogen-containing, and oxygen-containing molecules,
of relevance to fundamental and applied spectroscopy. Future
power would likely be too low. In our configuration though, work will exploit our feed-forward dual-comb spectrometer for
the pump photons from the continuous-wave laser are entirely precise spectroscopy of molecular line shapes.
depleted. By chirping the signal pulses to several picoseconds, we
increase the conversion efficiency and the average power by one ACKNOWLEDGMENTS. Support by the Munich Center for Advanced Photonics

PHYSICS
order of magnitude, which would be sufficient for multipass cells. and by the Carl-Friedrich-von-Siemens Foundation is gratefully acknowledged.

1. Hänsch TW (2006) Nobel lecture: Passion for precision. Rev Mod Phys 78:1297–1309. 26. Zhu F, et al. (2015) Mid-infrared dual frequency comb spectroscopy based on fiber
2. Picqué N, Hänsch TW, Frequency comb spectroscopy. Nat Photonics, in press. lasers for the detection of methane in ambient air. Laser Phys Lett 12:095701.
3. Baumann E, et al. (2011) Spectroscopy of the methane ν3 band with an accurate 27. Timmers H, et al. (2018) Molecular fingerprinting with bright, broadband infrared
midinfrared coherent dual-comb spectrometer. Phys Rev A 84:062513. frequency combs. Optica 5:727–732.
4. Ycas G, et al. (2018) High-coherence mid-infrared dual-comb spectroscopy spanning 28. Jin Y, Cristescu SM, Harren FJM, Mandon J (2015) Femtosecond optical parametric
2.6 to 5.2 μm. Nat Photonics 12:202–208. oscillators toward real-time dual-comb spectroscopy. Appl Phys B 119:65–74.
5. Chen Z, Yan M, Hänsch TW, Picqué N (2018) A phase-stable dual-comb interferometer. 29. Kara O, Maidment L, Gardiner T, Schunemann PG, Reid DT (2017) Dual-comb spec-
Nat Commun 9:3035. troscopy in the spectral fingerprint region using OPGaP optical parametric oscillators.
6. Yan M, et al. (2017) Mid-infrared dual-comb spectroscopy with electro-optic modu- Opt Express 25:32713–32721.
lators. Light Sci Appl 6:e17076. 30. Schliesser A, Brehm M, Keilmann F, van der Weide D (2005) Frequency-comb infrared
7. Meek SA, Hipke A, Guelachvili G, Hänsch TW, Picqué N (2018) Doppler-free Fourier spectrometer for rapid, remote chemical sensing. Opt Express 13:9029–9038.
transform spectroscopy. Opt Lett 43:162–165. 31. Yu M, et al. (2018) Silicon-chip-based mid-infrared dual-comb spectroscopy. Nat
8. Ideguchi T, Bernhardt B, Guelachvili G, Hänsch TW, Picqué N (2012) Raman-induced Commun 9:1869.
Kerr-effect dual-comb spectroscopy. Opt Lett 37:4498–4500. 32. Klocke JL, et al. (2018) Single-shot sub-microsecond mid-infrared spectroscopy on
9. Schliesser A, Picqué N, Hänsch TW (2012) Mid-infrared frequency combs. Nat protein reactions with quantum cascade laser frequency combs. Anal Chem 90:
Photonics 6:440–449. 10494–10500.
10. Hugi A, Villares G, Blaser S, Liu HC, Faist J (2012) Mid-infrared frequency comb based 33. Hangauer A, Westberg J, Zhang E, Wysocki G (2016) Wavelength modulated multi-
on a quantum cascade laser. Nature 492:229–233. heterodyne spectroscopy using Fabry-Pérot quantum cascade lasers. Opt Express 24:
11. Bagheri M, et al. (2018) Passively mode-locked interband cascade optical frequency 25298–25307.
combs. Sci Rep 8:3322. 34. Sterczewski LA, et al. (2018) Multiheterodyne spectroscopy using interband cascade
12. Wang CY, et al. (2013) Mid-infrared optical frequency combs at 2.5 μm based on lasers. Opt Eng 57:011014.
crystalline microresonators. Nat Commun 4:1345. 35. Muraviev AV, Smolski VO, Loparo ZE, Vodopyanov KL (2018) Massively parallel
13. Griffith AG, et al. (2015) Silicon-chip mid-infrared frequency comb generation. Nat sensing of trace molecules and their isotopologues with broadband subharmonic
Commun 6:6299. mid-infrared frequency combs. Nat Photonics 12:209–214.
14. Bernhardt B, et al. (2010) Mid-infrared dual-comb spectroscopy with 2.4 μm Cr2+:ZnSe 36. Mandin JY, Jacquemart D, Dana V, Régalia-Jarlot L, Barbe A (2005) Line intensities of
femtosecond lasers. Appl Phys B 100:3–8. acetylene at 3μm. J Quant Spectrosc Radiat Transf 92:239–260.
15. Duval S, et al. (2016) Watt-level fiber-based femtosecond laser source tunable from 37. Plíva J (1972) Spectrum of acetylene in the 5-micron region. J Mol Spectrosc 44:
2.8 to 3.6 μm. Opt Lett 41:5294–5297. 145–164.
16. Schunemann PG, Zawilski KT, Pomeranz LA, Creeden DJ, Budni PA (2016) Advances in 38. Gordon IE, et al. (2017) The HITRAN2016 molecular spectroscopic database. J Quant
nonlinear optical crystals for mid-infrared coherent sources. J Opt Soc Am B 33: Spectrosc Radiat Transf 203:3–69.
D36–D43. 39. Pine AS (1980) Final report. Tunable laser survey of molecular air pollutants: Doppler-
17. Lau RKW, et al. (2014) Octave-spanning mid-infrared supercontinuum generation in limited spectra of the C-H stretching bands of formaldehyde, ethylene, ethane and
silicon nanowaveguides. Opt Lett 39:4518–4521. allene (Lincoln Laboratory, Massachusetts Institute of Technology, Cambridge, MA),
18. Kuyken B, et al. (2015) An octave-spanning mid-infrared frequency comb generated Report prepared for the National Science Foundation under Contract NSF/ASRA/DAR
in a silicon nanophotonic wire waveguide. Nat Commun 6:6310. 78-24562.
19. Zolot AM, et al. (2012) Direct-comb molecular spectroscopy with accurate, resolved 40. Lebron GB, Tan TL (2013) High-resolution Fourier transform infrared spectrum of the
comb teeth over 43 THz. Opt Lett 37:638–640. ν11 band of ethylene (12C2H4). J Mol Spectrosc 288:11–13.
20. Okubo S, et al. (2015) Ultra-broadband dual-comb spectroscopy across 1.0–1.9 μm. 41. Mandon J, Guelachvili G, Picqué N, Druon F, Georges P (2007) Femtosecond laser
Appl Phys Express 8:082402. Fourier transform absorption spectroscopy. Opt Lett 32:1677–1679.
21. Millot G, et al. (2016) Frequency-agile dual-comb spectroscopy. Nat Photonics 10: 42. Sorokin E, Sorokina IT, Mandon J, Guelachvili G, Picqué N (2007) Sensitive multiplex
27–30. spectroscopy in the molecular fingerprint 2.4 μm region with a Cr2+:ZnSe femtosec-
22. Ideguchi T, Poisson A, Guelachvili G, Picqué N, Hänsch TW (2014) Adaptive real-time ond laser. Opt Express 15:16540–16545.
dual-comb spectroscopy. Nat Commun 5:3375. 43. Bernhardt B, et al. (2010) Cavity-enhanced dual-comb spectroscopy. Nat Photonics 4:
23. Roy J, Deschênes J-D, Potvin S, Genest J (2012) Continuous real-time correction and 55–57.
averaging for frequency comb interferometry. Opt Express 20:21932–21939. 44. Meek S, Poisson A, Guelachvili G, Hänsch TW, Picqué N (2014) Fourier transform
24. Burghoff D, Yang Y, Hu Q (2016) Computational multiheterodyne spectroscopy. Sci spectroscopy around 3 μm with a broad difference frequency comb. Appl Phys B 114:
Adv 2:e1601227. 573–578.
25. Villares G, Hugi A, Blaser S, Faist J (2014) Dual-comb spectroscopy based on quantum- 45. Seidel M, et al. (2018) Multi-watt, multi-octave, mid-infrared femtosecond source. Sci
cascade-laser frequency combs. Nat Commun 5:5192. Adv 4:eaaq1526.

Chen et al. PNAS | February 26, 2019 | vol. 116 | no. 9 | 3459