Large fluctuations at the lasing threshold of solid- and liquid-state dye lasers

Abstract

Intensity fluctuations in lasers are commonly studied above threshold in some special configurations (especially when emission is fed back into the cavity or when two lasers are coupled) and related with their chaotic behaviour. Similar fluctuating instabilities are usually observed in random lasers, which are open systems with plenty of quasi-modes whose non orthogonality enables them to exchange energy and provides the sort of loss mechanism whose interplay with pumping leads to replica symmetry breaking. The latter however, had never been observed in plain cavity lasers where disorder is absent or not intentionally added. Here we show a fluctuating lasing behaviour at the lasing threshold both in solid and liquid dye lasers. Above and below a narrow range around the threshold the spectral line-shape is well correlated with the pump energy. At the threshold such correlation disappears, and the system enters a regime where emitted laser fluctuates between narrow, intense and broad, weak peaks. The immense number of modes and the reduced resonator quality favour the coupling of modes and prepares the system so that replica symmetry breaking occurs without added disorder.

Full text

1
Scientific RepoRts | 6:32134 | DOI: 10.1038/srep32134
www.nature.com/scientificreports
Large uctuations at the lasing
threshold of solid- and liquid-state
dye lasers
Supratim Basak, Alvaro Blanco & Cefe López
Intensity uctuations in lasers are commonly studied above threshold in some special congurations
(especially when emission is fed back into the cavity or when two lasers are coupled) and related with
their chaotic behaviour. Similar uctuating instabilities are usually observed in random lasers, which are
open systems with plenty of quasi-modes whose non orthogonality enables them to exchange energy
and provides the sort of loss mechanism whose interplay with pumping leads to replica symmetry
breaking. The latter however, had never been observed in plain cavity lasers where disorder is absent
or not intentionally added. Here we show a uctuating lasing behaviour at the lasing threshold both
in solid and liquid dye lasers. Above and below a narrow range around the threshold the spectral line-
shape is well correlated with the pump energy. At the threshold such correlation disappears, and the
system enters a regime where emitted laser uctuates between narrow, intense and broad, weak
peaks. The immense number of modes and the reduced resonator quality favour the coupling of modes
and prepares the system so that replica symmetry breaking occurs without added disorder.
Lasers made with organic dyes in liquid solutions or embedded in solid matrices are appreciated for their high
eciency1. us random lasers (RL)2, a notable example combining disorder and gain media, but lacking a cavity
(which hinders feedback) were demonstrated3 and are most oen made with organic dyes. In this case the feed-
back is obtained from scattering o the disordered medium so no external cavity is needed. Due to their nature it
is reasonable to expect some uctuating behaviour in their emission4. Apart from the case where chaotic behav-
iour is purposefully provoked for technological applications5, laser uctuations were mostly observed to occur
(and fought against) above threshold, i.e. during laser action. For instance, uctuations in the emitted spectra of
ZnO in an organic solid matrix observed when excited with pulses longer that the chromophore lifetime were
described as a lasing instability due to interplay between pulse length and excited state lifetime6. Unlike these
intrinsic uctuations, a similar system, albeit liquid, showed uctuation attributed to the dynamic disorder within
the colloid realized for each pump pulse7. Mujumdar et al. introduced the idea of mode coupling between the long
lived extended modes for the chaotic behaviour of emission spectra8.
In any case, the lasing threshold of lasers, random or conventional, is perhaps the regime that garnered the
least attention9,10. Intensity uctuation between the emission from lasing and non-lasing modes at the threshold
in cw GaAs laser were modelled using coupled van der Pol oscillators11. A temperature dependent study of the
correlation between the uctuations of dierent modes for a semiconductor laser has also been carried out12. e
latter examples deal with very few modes and depend on direct energy exchange between one lasing mode and
few neighbouring non-lasing ones. ese modes are relatively far apart and respond dierently to changes in gain
so their noise eect is of an individual rather than collective character. However lasers involving many modes
require statistical approaches that treat them as o-equilibrium systems whose stationary regime is brought about
by a constant pumping that causes the system to behave as if in equilibrium with a thermal bath: the pumping
rate. is allows to liken modes to a liquid and permits to draw phase diagrams of the lasing function13. Typically
these systems require a mechanism, like disorder in RLs, that establishes the loss channel.
When a large number of modes are involved in the system emission and an interaction between them is con-
sidered, it is advantageous to treat the system as a spin glass14 in the sense that any actual state is composed of a
large number of interacting spins that can uctuate adopting random values to nd the equilibrium state. is
problem was solved through the use of the replica trick for the calculation of free energy15 providing an order
Instituto de Ciencia de Materiales de Madrid (ICMM) Consejo Superior de Investigaciones Cientícas (CSIC), Calle Sor
Juana Inés de la Cruz 3, Cantoblanco, 28049 Madrid, Spain. Correspondence and requests for materials should be
addressed to C.L. (email: c.lopez@csic.es)
received: 15 February 2016
accepted: 02 August 2016
Published: 25 August 2016
OPEN

www.nature.com/scientificreports/
2
Scientific RepoRts | 6:32134 | DOI: 10.1038/srep32134
parameter that was later on proven to bear a physical meaning16 and the phenomenon has ever since been referred
to as the replica symmetry breaking. In the search to minimize energy some of the possible congurations are
blocked because the spins involved cannot comply with the random distribution of couplings in the system
(frustration). is theory has been used to model the functioning of ordered and disordered lasers permitting
to draw a phase diagram13 and was found to account for the modal behaviour of random lasers17. Random lasers
base the mode interaction in the fact that a proper cavity is lacking and spatial overlap from unfullled orthog-
onality allows an ecient energy exchange. Many modes can be excited by the pumping pulse some of whose
interactions are frustrated so that the system is led to choosing between dierent but equivalent congurations.
e set of the activated mode congurations changes from pulse to pulse. Each of these congurations is a ther-
modynamic state characterized by the set of modes activated and their interactions. In our case, as in the case of
RLs, under the exact same conditions the systems ends up in dierent states but, because the distribution of Parisi
overlaps between states (of mode congurations) is the same as between replicas18, it is possible to identify the
RSB from the statistical analysis of overlaps among states. In this framework the result of successive instances of
pumping a lasing system can be viewed as equivalent states (conceived as modes congurations) that may present
correlations that depend on the states and with non-trivial statistical distributions. While the states are equivalent
their correlations may not be. In fact, replica symmetry breaking was so far observed only in RLs because they
provide a collection of light modes whose emissions are equivalent (degenerate) and susceptible to frustration.
On the contrary, ordinary lasers usually accumulate too few modes under the gain curve to lead to liquid-like
behaviour and well dened orthogonality pre-empting frustration. Although quenched disorder is oen the fun-
damental reason for frustration and RSB, some systems with complicated, though deterministic19, interactions
that can establish self-induced frustration20 were shown within the replica theory to display RSB.
In this work we demonstrate a uctuating behaviour at the threshold region of lasers made from pure liquid
dye solution in a cuvette and dye doped DNA lms without adding any scatterers. e uctuating behaviour is
evident from the direct intensity and full width at half maximum (FWHM) observations. We have performed
measurements to synchronously collect the energy of the pump pulses and the corresponding emission intensity.
Our performed measurements demonstrate that the uctuations are not due to changes in the pulse energy from
shot to shot or thermal eects from the sample. Further, the uctuations in the threshold region are not only pres-
ent in liquid state lasers but also in solid state although showing comparatively less marked uctuating behaviour.
We have also tested our laser uctuation for varying pulse duration (τ
p) and assessed the impact on uctuations
of spatial (through cuvette thickness) and temporal (through the pumping pulse duration) control. Finally, we
carried out statistical analyses to prove that mode coupling/frustration is responsible for the uctuations in the
threshold region. e system is the rst to show replica symmetry breaking with no intentional disorder because a
lousy cavity ultimately induces frustration. is comes about when the immense number of leaky modes involved
experience couplings that are frustrated: coherent oscillation of one mode simultaneously with two other coher-
ently oscillating modes can be impossible. is opens the way to equivalent states with dierent sets of activated
modes in each shot.
Two types of systems were tested: liquid-state dye solution laser and solid state dye lm lasers. For the former a
dye solution was placed in a cuvette and pumped with ns or ps pulses from Q-switched lasers. e latter consisted
of dye in DNA lms subjected to the same pumping. Figure1a shows the main features of lasing as a function of
pump energy for the typical liquid state conguration. On pumping the cuvette lled with 1 mg DCM in 1 mL
THF at low energy density broad photoluminescence from the sample was observed. e FWHM for the photo-
luminescence spectra is ∼ 60 nm. However, above a certain energy density, a spectral narrowing was observed,
accompanied by a signicant increase in intensity. For this sample a clear transition from regular broadband
(∼ 60 nm) to narrow band (∼ 20 nm) laser like emission was observed on increasing the pump energy which is
the accepted as a signature of the lasing regime21. In the threshold, however, series of spectra at constant pump
energy showed that some of the emission peaks have very high intensity with a FWHM ∼ 20 nm (lasing) and some
have low intensity and FWHM ∼ 60 nm (Fig.1b). e two events of lasing and non lasing can be captured on a
screen and are shown in the image Supplementary Fig. S1(a,b) respectively. Only rarely the emission has widths
and intensities in between (see Fig.1c) showing a departure from a monotonous behaviour. Figure1c shows the
intensity maxima for spectra collected over 6000 successive laser shots. e lasing or uorescence behaviour of
the successive spectra was clearly observed from the FWHM versus shot number plot (Fig.1d). Notice that, unlike
FWHM, recorded intensity is limited by the detector dynamic range and all shots of high intensity simply register
as the detector maximum making it look like there are many fewer.
To rule out the occurrence of the blinking due to uctuations of the pump energy that, owing to the rapid
change in slope of the emission intensity near threshold, might cause a similar uctuation in emission we have
measured the shot energy synchronously for each spectrum. e normalized correlation coecient between the
intensity and shot energy was calculated according to the expression:
=
∑−−
∑−∑−
r
IIPP
II PP
()()
()()
(1)
IP ii
ii
22
where Ii are the peak maxima and
Pi
the laser shot energies for the corresponding series of spectra and
I
and
P
their averages. From Fig.2a it is evident that the intensity from the sample is well correlated with the shot energy
below and above the threshold. But, at or near above the threshold the correlation factor reduces to a value of 0.5
indicating the spectral intensity is decoupled from the shot energy and presents strong uctuations. Figure2b
shows FWHM vs. shot energy for 3000 consecutive shots where we can see that at the threshold (∼ 0.32 mJ/pulse)
the peak widths have two most frequent values (∼ 60 nm and ∼ 20 nm) while these values collapse in one or the
other away from the threshold region (60 nm below and 20 nm above threshold). is is further detailed in Fig.2c

www.nature.com/scientificreports/
3
Scientific RepoRts | 6:32134 | DOI: 10.1038/srep32134
where a contour plot of the statistical distribution of peak width P(FWHM) is represented against the pumping
energy near the threshold. Shots were grouped in 30 energy segments and the distributions normalized for each
segment in the following way. A one hundred-bin histogram of each segment gives Pi(FWHM) (i = 1… 30) for
thirty energies in the range highlighted Fig.2a. Each Pi is normalised so that the total probability for each energy
adds up to one. ese thirty vectors are the columns in the contour plot shown. One can see clearly that at lower
energies most of the peaks have widths around 60 nm while for the higher pulse energies this population has
diminished and most peaks are narrow with widths of around 20 nm. In the region in between there is a bimodal
distribution.
e low intensity-power correlation obtained near threshold is comparable to the degree of time correlation
=
−
cIP IP/
liil ii
2
1
22
1
2 between successive shots viz. lasing intensity, Ii, and shot energy, Pi−l, as a function of lag
(l). Notice that shots are separated by one h of a second while emitted pulses are in the nanosecond range. Such
measurements, as expected, show no correlation between the two except when the lag imposes only two data
points in the calculation (l = ± 3000) making cross-correlation identically equal to 1 merely for mathematical
reasons. Away from the threshold region correlation simply attest to the pump lasers stability. (see Supplementary
Fig. S2.) is conrms that rIP values in the threshold represent fully uncorrelated behaviour. Time delayed inten-
sity–power correlation away from threshold (above or below) shows zero-delay correlation values signalling the
fact that in this regime the intensity for each shot is linked to the power causing it.
In order to ensure the ndings are not dye dependent other laser dyes like Rhodamine B and also dierent
solvent system (e.g. EtOH, H2O, DMSO, acetonitrile, ethylene glycol etc.) were tried. In every case we nd the
same behaviour at the threshold region. e uctuation behaviour is independent of the solvent-solute inter-
action, viscosity and boiling point of the solvents. To further ascertain that the uctuations of the intensity at
the threshold are not due to the thermal or other eects from the solution we have made several samples with
dierent concentration of DCM dye in THF so as to place the threshold at dierent absolute pulse energies. As
we increase the concentration the lasing threshold increases drastically as can be seen from the Supplementary
Fig. S3a. e required threshold energy increases one order of magnitude from 261 μ J to 2.4 mJ, as we increase
the concentration from 0.05 mg/mL to 3.2 mg/mL. is is believed to be due to the inner lter eect22 as well as
internal quenching. It is also interesting to note that at the highest concentration the samples show uctuation
not only at the threshold but also above threshold due to pump power limitations. Gain is low and the region
of ordinary lasing above threshold is pushed too far towards high energy and beyond the limits of the available
laser and probably the dye tolerance. However, the intensity uctuation at the threshold was observed for all the
Figure 1. Amplied spontaneous emission and lasing. (a) Emission intensity maxima versus excitation
energy (532 nm, 9 ns. 10 Hz) red open circles (le axis), and FWHM (blue open diamonds) versus pump energy
for a dye concentration of 0.04 mg/mL. (b) Series of emission spectra obtained from DCM (dye) 1 mg/1 mL
THF in successive excitation shots near threshold (pump energy per pulse 1 mJ). (c) Emission intensity maxima
collected for 6000 shots(pump energy per pulse 1 mJ). (d), Corresponding FWHM for each spectrum collected
in successive 6000 excitation shots. (pump energy per pulse 1 mJ).

www.nature.com/scientificreports/
4
Scientific RepoRts | 6:32134 | DOI: 10.1038/srep32134
samples. When the threshold is low there is ample margin to pump much above threshold before any sign of sat-
uration shows (see Fig.1a). If the threshold is pushed higher, when stimulated emission sets in the system is near
saturation and the uctuations cannot be damped.
In order to establish it further, beyond photophysical properties, we have prepared samples of various solu-
tions having a range of boiling point viz. THF → EtOH → DMSO → Diethylene Glycol. e solvents have an
increasing boiling point starting from 66 °C up to 244 °C. e concentration of DCM in all the cases was kept con-
stant. e intensity uctuations were present for all the samples. ese observations prove that the uctuations
in the intensity of the samples are not due to the thermal eect of the samples and clearly support the hypothesis
that the uctuations are not due to environmental factors but are intrinsic to the system. Next we intended to
measure solid state samples to check whether the uctuating behaviour is also present in that case. We prepare a
DNA-CTMA complex doped with DCM dye as previously developed in our laboratory23. e sample, in the form
of a thin lm of several hundred micrometres thickness, was pumped by a stripe formed by the cylindrical lens
and the emitted light was collected from the edge. For this particular sample the intensity uctuations were not
as evident from the successive spectra collected at dierent pump energies. e plot in Supplementary Fig. S4(b)
shows thirty consecutive spectra at the highest uctuation point.
To obtain more insight into the uctuation from dierent samples (liquid and solid), it is instructive to analyse
the uctuation coecient (f) dened as the standard deviation (σ
I) of the intensity of light emitted by the sample
divided by the mean of the intensities
=σ
If():I
I
so that we can describe the strength of the uctuations by exam-
ining their statistical distribution. At threshold when uctuation is strongest the probability distribution is not
Gaussian. Instead they follow U-quadratic distribution (see Supplementary Fig. S5.) Fig.3a shows the statistical
analysis of f as a function of pump energy for the liquid sample for the solid lm. It is evident from the plot that f
has maximum at the threshold and minima at the regions below and above threshold. is is clearly at odds with
related results in RLs where the variance of the emitted intensity scales as the intensity itself17. Figure3b shows the
statistical distribution of FWHM below (green), at (orange) and above (red) the threshold. In all three cases and
despite the dierence in ranges data was processed into one hundred bins which makes the histogram bars thin-
ner away from the threshold. Both below and above the distributions clearly resemble a normal distribution. At
threshold however the distribution (orange histogram) greatly departs from normal as can be seen in Fig.3b
where f takes its maximum value. It is interesting to note that f reaches values as large as 152% to be compared
with values for the laser that never go above a few. In fact the uctuations at the threshold are so large that, owing
to its limited dynamic range, the detector saturates which impedes to record a good distribution, something that
the FWHM permits very clearly. e fact that some of these pulses are so much more intense than the average
makes the sporadic lasing events directly observable by the naked eye as a red blinking point on a constant green
background (that corresponds to the pump laser). Figure3c shows a similar analysis for the intensity: here the
range of the random variable is so wide that separate logarithmic scale plots are needed. Again, while below and
above the distributions are ostensibly Gaussian, in the threshold a clear bimodal distribution appears. e thresh-
old region of the solid state laser also shows uctuations that reach a maximum value of ∼ 27%, well above the
pumping laser pulses.
Mode coupling
e dynamics of the single mode laser eld as it interacts with the lasing medium has been examined in the con-
text of non-equilibrium statistical mechanics so that a laser near threshold can nd a close parallel in the order/
disorder phase transition of a pure uid vapour-liquid second order phase transitions. e interaction of a mole-
cule’s emission with that of all other molecules is entirely similar to that of a magnetic dipole with its environment
in a ferromagnet. is invites to identify the laser electric eld as the variable corresponding to the ferromagnetic
Figure 2. e threshold signatures. (a) Normalized emission-pumping correlation coecient calculated
over 3000 single shorts vs. the average shot energy; (b) the corresponding FWHM vs. shot energy at the lowest
correlation coecient and (c), probability density of FWHM (normalized for each energy) as a function of pulse
energy for a dye concentration of 0.04 mg/mL.

www.nature.com/scientificreports/
5
Scientific RepoRts | 6:32134 | DOI: 10.1038/srep32134
order parameter and the population inversion as the temperature24. Our systems, despite their apparent similarity
to ordinary rather than RLs, contain many modes and, owing to low quality of the cavity, we believe that mode
coupling is responsible for the intensity uctuations. Other possible causes like random losses; local change of
molecules concentration which might rapidly change the threshold or the optical feedback like chaotic seeding
can probably be ruled out by the independence on physical and chemical environments as tested here.
In the case of the liquid sample the cuvette cavity acts as a resonator. e longitudinal mode spacing for a
regular Fabry-Perot is given by λ
∆=
λ
nL2
2
. Here L is the cavity length and n is the refractive index of the medium.
e calculated mode spacing, Δ λ = 0.068 nm and 0.0135 nm for 2 mm and 10 mm path length cuvette respec-
tively. ese values are well below the resolution limit of the spectrometer (0.4 nm). e number of modes per
unit volume supported by a cavity can be expressed as =
π
λ
N8
3
3; which gives 3.8 × 1010 modes per cubic millime-
tre, for light of 604 nm wavelength. For the excitation of an area of 25 μ m2 and considering the path length of the
cuvette 10 mm ~9.53 × 107 modes are activated along the cylindrical length (π R2L). is immense number of
modes sets this system apart from early studies in few modes semiconductor lasers just as the fact that the cavity
is regular rather than disordered gives it a novel character at variance with RLs. It is therefore an unlikely environ-
ment for ordinary lasers threshold instability and for replica symmetry breaking but the latter is proved by the
analysis of shot to shot correlations. To assess the establishment of a regime of replica symmetry breaking induced
by coupling between the modes as a function of pump energy we followed a statistical mechanics approach. Just
because the distribution of overlaps between mode states is the same as between mode replicas, it is possible to
detect the replica symmetry breaking from the statistical analysis of the former18, it is possible to detect the replica
symmetry breaking from the statistical analysis of the former. For each shot, α, we evaluate the intensity uctua-
tion of the modes (labelled by wavelength, k)
∆= −
αα
kIkIk() () ()
where Iα(k) is the intensity of mode k for
shot α and
Ik()
is the average over all shots at mode k. e overlap of the spectral uctuation from shot to shot can
be calculated by the correlation between intensity uctuation of any two shots α and β:
=∑
∆∆
∑∆∑∆
αβ
=αβ
=α=β
q
kk
kk
() ()
() ()
(2)
k
N
k
Nk
N
1
1
2
1
2
where the sums run over the number of modes (N). is analysis is based on the intensity alone, as the phase
is hard to obtain, but it has shown its power in revealing the dierent stages RL presents in terms of replica
symmetry breaking25. Figure4 shows the plot of statistical distribution of P(q) versus q calculated for α, β = 1,
… 500 shots that provide a total 500 × (500 − 1)/2 values of q. e plots are in log scale to try to show as much
detail as possible in the cases where the distribution presents a large range of probability densities. At very low
pulse energies (0.25 μ J in Fig.4 upper row) the distribution presents a peak centred at q = 0 on top of a much
Figure 3. Emission uctuations at the threshold. (a) Fluctuation coecient (f) for DCM dye solution, DCM
dye doped DNA-CTMA lm and pump. (b) Statistical distribution of FWHM for dye solution. (c) Statistical
distribution of intensity. Both distributions expressed as relative frequency so as to insure the integral equals unity.

www.nature.com/scientificreports/
6
Scientific RepoRts | 6:32134 | DOI: 10.1038/srep32134
weaker background. is same behaviour is found at very high pulse energy (326 μ J in Fig.4 upper row) and can
most possibly be a sign that a regime where full RSB is attained. At pulse energies very near above and below the
threshold the distribution presents the signature of one-step plus full RSB with a peak around q = 0 and two wings
reaching q = ± 1. In the threshold energy range (101 μ J in Fig.4 upper row) the distribution is totally dierent
with two strong maxima at q = ± 1 and a largely depleted region around q = 0 pointing to a one-step RSB. In these
circumstances where emission consists of broad weak peaks and narrow, intense lasing pulses if the statistical
analysis is performed aer separating both kinds of spectra, two diering statistics are found: weak peaks obey
a replica symmetric zero-centred distribution while intense laser bursts follow a one-step RSB distribution with
high density at q = 0 (refer to SI Fig. S6).
It is very clear from (Fig.4) that in the threshold regime (P = 101 μ J) the range of small |q| is depleted indicat-
ing that any two shots have very strongly correlated intensity uctuations over wavelength. Either upswings in
some modes coincide with upswings in others (q ~ 1) or upswings coincide with downswings (q ~ − 1). In other
words, in this regime the electromagnetic modes strongly interact and become coupled. On the other hand, if we
prepare the system in the region below or above threshold the modes are mostly uncorrelated as revealed by zero
peaked distribution.
If the uctuations in the intensity at threshold are mediated by coupling of the modes and frustration, which
are activated during the pump pulses, they should respond to the time dependence of the excitation. e coupling
is more eective if the pulse duration (τ
p) is comparable to or much longer than the mode lifetime. e mode
lifetime is in the order of hundreds of picoseconds26,27. e experimental set up (Supplementary Fig. S1) allows
nanosecond laser pulses (τ
p = 9 ns) and picosecond laser pulses (τ
p = 30 ps) to be sent to the sample without dis-
turbing any other part of the set up. Unlike in the case of nano pumping, during the pico pumping as we sweep the
pulse energy through the threshold region, no uctuations are observed in the intensity emitted. Moreover the
eective spectral window over which the lasing peaks are found for ns pulses (14 nm) is narrower as compared
to the range generated by picosecond pulses (32 nm) as can be seen in Fig.5. e time required to establish the
coupling between modes makes that in case of nano pumping a larger number of modes is activated whereas in
the case of pico pumping only strongly interacting modes can establish an eective coupling in the duration of
the exciting pulse6,28.
It is also possible to act on the intensity uctuation through a spatial control. By making the sample thinner
and thinner, eventually we decrease the number of modes to a point where there is not sucient direct (mode to
mode) and mediated (through a third mode) coupling between modes so that they can’t couple eectively and
subsequently be liable to frustration, to produce a large intensity uctuation. ey can only show a relatively small
uctuating behaviour as was seen from our solid state sample.
In conclusion, we have studied the uctuating behaviour of the emission spectra of fabricated solid state and
solution state lasers. Because of the high gain of the liquid dye solution the coupling between the modes is so
eective that replica symmetry breaking becomes a readily observable phenomenon. Although our system has no
deliberate disorder, its modes are allowed to exchange energy and interact but some of these interactions are frus-
trated obliging the system to choose and leading to equivalent but distinct states. e uctuation can be explained
in terms of mode coupling and frustration. is behaviour is completely intrinsic and can be observed for many
Figure 4. Overlap distribution as a function of pumping. Plot of P(q) versus q for dierent pumping rates
from below threshold to above threshold and at the threshold region for liquid state sample (upper row) and
below and on threshold for solid state sample (lower row). Each panel carries the pump energy as a label and
can be identied in Fig. 1a and Fig. 3a. by the larger symbols. All distributions in relative frequency to insure
integral equals unity.

www.nature.com/scientificreports/
7
Scientific RepoRts | 6:32134 | DOI: 10.1038/srep32134
dyes. We also show how to manage the uctuation behaviour by temporal and spatial control. It is also interesting
to study this behaviour for large uorescence lifetime samples (e.g. lanthanides) and in presence of scattering
media as well as diusion time. e ease of fabrication and the simplicity of sample preparation may bring a large
variety of application such as use in security marker and photonic displays or even in random number generation.
Methods
Preparation of liquid dye solution. 1 mg DCM dye was dissolved in 1 mL of THF (tetrahydrofuran). A
glass cuvette (internal path length 10 mm) was lled with the THF solution of DCM and mounted on the sample
stage for studies.
Preparation of Solid sample. In a typical procedure at rst 10 mg DCM dye was dissolved in 20 mL EtOH
by sonication. A previously prepared DNA-CTMA complex (100 mg) was dissolved in 5 mL EtOH. e above
two solutions were mixed and stirred for thirty minutes. Films of DCM doped DNA-CTMA on a quartz substrate
were prepared by dip coating, keeping them overnight in a calibrated oven at temperature 40 °C.
Optical set up. e 532 nm laser line was cleaned by using appropriate lters. e beam was split into two
parts by a pellicle beam splitter, one part was sent to the energy meter and the other part was focused on the sam-
ple using a plano-convex lens (f = 10 cm) on a DCM dye solution containing cuvette. e ASE/uorescence was
collected using another biconvex lens (f = 10 mm), aer removing the pump (532 nm) using a 532/1064 nm notch
lter and LP 532 nm, the collimated ASE/uorescence light was focused on an optical ber (600 μ m core) cou-
pled to the Ocean Optics USB2000+ /USB4000 spectrometer (Supplementary Fig. S1). e sample was pumped
either by a ND:YAG laser (Litron model Nano-T 250-10) providing 20 mJ, 9 ns, 532 nm pulses operating at 10 Hz
or 5 Hz repetitions rate or by a ND:YAG laser (EKSPLA model PL2250) providing 30 ps, 20 mJ, 532 nm pulses
operating at 10 Hz repetition rate. To measure correlation between pulse and emission the 532 nm laser beam was
split using a 45:55 (R:T) pellicle beam splitter. e energy of the each shot was measured using an energy meter
(Gentec-Solo-2) coupled with a detector. To ensure a good synchronization the laser frequency was kept at 5 Hz.
3000 spectra were collected for 10 minutes at several shot energy ranging from 0.0-0.9 mJ.
References
1. Dye Laser Principles, (eds. Duarte, F. J. & Hillman, L. W.) (Academic, New Yor, 1990).
2. Letohov, V. S. Light generation by a scattering medium with a negative resonant absorption. Sov. Phys. -JETP 16, 835, (1968).
3. Lawandy, N. M., Balachandran, . M., Gomes, A. S. L. & Sauvain, E. Laser action in strongly scattering media. Nature (London) 368,
436 (1994).
4. Leonetti, M., Conti, C. & Lopez, C. e mode-locing transition of random lasers. Nature Photon. 5, 615–617 (2011).
5. Sciamanna, M. & Shore, . a. Physics and applications of laser diode chaos. Nat. Photon. 9, 151–162 (2015).
6. Anglos, D. et al. andom laser action in organic–inorganic nanocomposites. J. Opt. Soc. Am. B 21, 208 (2004).
7. van der Molen, . L., Mos, A. P. & Lagendij, Ad. Intrinsic intensity uctuations in random lasers. Phys. ev. A 74, 053808 (2006).
8. Mujumdar, S., Türc, V., Torre, . & Wiersma, D. S. Chaotic behavior of a random laser with static disorder. Phys. ev. A 76, 033807
(2007).
9. Chua, S.-L. et al. Modeling of threshold and dynamics behavior of organic nanostructured lasers. J. Mater. Chem. C 2, 1463 (2014).
Figure 5. Pumping pulse length. e spectra obtained for a same sample keeping the exact set up conditions.
(a), for pico pumping and (b), for nano pumping condition. Shaded areas represent the spectral range for two
pumping conditions showing that sustained pumping enables mode interaction and laser emission narrowing.

www.nature.com/scientificreports/
8
Scientific RepoRts | 6:32134 | DOI: 10.1038/srep32134
10. Zhu, G., Gu, L. & Noginov, M. Experimental study of instability in a random laser with immobile scatterers. Phys. ev. A 85, 043801
(2012).
11. Armstrong, J. A. & Smith, A. . Intensity uctuations in GaAs laser emission. Phys. ev A 155, 140 (1965).
12. Aabane, H., Tsuchiya, T., awaubo, T. & Abe, H. Correlation between mode uctuations in a semiconductor laser. Jpn. J. Appl.
Phys. 24, L501 (1985).
13. F. Antenucci, C. Conti & A. Crisanti General Phase Diagram of Multimodal Ordered and Disordered Lasers in Closed and Open
Cavities, L. Leuzzi, Phys. ev. Lett. 114, 43901 (2015).
14. Edwards, S. F. & Anderson, P. W. eory of spin glasses. J. Phys. F. 5, 965–974 (1975).
15. Parisi, G. Innite number of order parameters for spin-glasses. Physical eview Letters, 43, 1754–1756 (1979).
16. Parisi, G. Order Parameter for Spin-Glasses. Phys. ev. Lett. 50, 1946–1948 (1983).
17. Ghofraniha, N. et al. Experimental evidence of replica symmetry breaing in random lasers. Nat. Commun. 6, 6058 (2015).
18. Mezard, M., Parisi, G. & Virasoro, M. A. Spin Glass eory and Beyond (World Scientic, 1987).
19. Marinari, E., Parisi, G. & itort, F. eplica eld theory for deterministic models: I. Binary sequences with low autocorrelation.
Journal of Physics A: Mathematical and General, 27, 7615 (1994).
20. E. Marinari, G. P. & itort, F. eplica eld theory for deterministic models:II. A non-random spin glass with glassy behavior. J. Phys.
A. 27, 7647 (1994).
21. Ganiel, U., Hardy, A., Neumann, G. & Treves, D. Amplied spontaneous emission and signal amplication in dye-laser systems.
IEEE J. Quantum Electron. 11, 881–892 (1975).
22. Terada, M. & Muto, J. Concentration dependent laser threshold of rhodamine 6g in methanol solution. Opt. Commun. 59, 199–202
(1986).
23. Leonetti, M., Sapienza, ., Ibisate, M., Conti, C. & Lopez, C. Optical gain in DNA-DCM for lasing in photonic materials. Opt. Lett.
34, 3764–3766 (2009).
24. DeGiorgio, V. & Scully, M. O. Analogy between the Laser reshold egion and a Second-Order Phase transition. Phys. ev. A 2,
1170 (1970).
25. Antenucci, F., Crisanti, A. & Leuzzi, L. e glassy random laser: replica symmetry breaing in the intensity uctuations of emission
spectra. Sci. ep. 5, 16792 (2015).
26. Sououlis, C. M., Jiang, X., Xu, J. Y. & Cao, H. Dynamic response and relaxation oscillations in random lasers. Phys. ev. B 65,
041103 (2002).
27. Siddique, M., Alfano, . ., Berger, G. A., empe, M. & Genac, A. Z. Time-resolved studies of stimulated emission from colloidal
dye solutions. Opt. Lett. 21, 450–452 (1996).
28. Leonetti, M., Conti, C. & Lopez, C. Dynamics of phase-locing random lasers. Phys. ev. A 88, 043834 (2013).
Acknowledgements
We wish to thank André Espinha for his help with the experimental set up. is work was partially funded by the
Spanish MINECO MAT2015-68075-R (SIFE) and the Comunidad de Madrid S2013/MIT-2740 (PHAMA_2.0)
projects.
Author Contributions
S.B. carried out the experiments, S.B., A.B. and C.L. contributed to writing the manuscript, S.B., A.B. and C.L.
contributed to interpreting the results, C.L. supervised the research.
Additional Information
Supplementary information accompanies this paper at http://www.nature.com/srep
Competing nancial interests: e authors declare no competing nancial interests.
How to cite this article: Basak, S. et al. Large uctuations at the lasing threshold of solid- and liquid-state dye
lasers. Sci. Rep. 6, 32134; doi: 10.1038/srep32134 (2016).
is work is licensed under a Creative Commons Attribution 4.0 International License. e images
or other third party material in this article are included in the article’s Creative Commons license,
unless indicated otherwise in the credit line; if the material is not included under the Creative Commons license,
users will need to obtain permission from the license holder to reproduce the material. To view a copy of this
license, visit http://creativecommons.org/licenses/by/4.0/
© e Author(s) 2016