1. No category

New light sources for time correlated single photon counting in
commercially available spectrometers
Roger Fenske*, Dirk U. Näther, Mark Goossens, S. Desmond Smith
Edinburgh Instruments, 2 Bain Square, Kirkton Campus, Livingston, UK, EH54 7DQ;
ABSTRACT
The method of Time Correlated Single Photon Counting requires high repetitive light sources (>100kHz) with pulse
widths of ideally less than approximately 20ps. While these light sources have been available for some time now in the
form of Ti:Sapphire lasers, picosecond pulsed diode lasers (<90ps) and light emitting diodes (<700ps), they all have the
drawback of either having no spectral tunability, or tunability over a very narrow spectral range (10nm-100nm).
While this is often sufficient for specific laboratory setups for measurements of fluorescence lifetimes, commercial
Fluorescence Lifetime Spectrometers have suffered for a long time from the lack of the availability of simple, compact
and relatively inexpensive broad spectral band light sources that can be employed for Time Correlated Single Photon
Counting.
A new light source as an integral part of a commercial fluorescence lifetime spectrometer will be discussed that allows
tunability over a wide spectral band of more than 500nm.
Keywords: Time Correlated Single Photon Counting, supercontinuum light source
1
1.1
INTRODUCTION
Characteristics of light sources for TCSPC
Time Correlated Single Photon Counting (TCSPC) is an established technique for the measurement of fluorescence
lifetimes in the picosecond and nanosecond time scale. The technique makes use of the fact that repeated detection of
single photons following (relatively low peak power) pulsed sample excitation results in a probability distribution that is
equal to the fluorescence decay that would follow a single powerful excitation pulse. The TCSPC technique requires a
high repetition rate light source to accumulate a sufficient number of photon events for statistical data precision in the
probability histogram of arrival times. In addition, it is preferable for the light source to have temporal pulse widths that
either match or supersede the temporal resolution that is set by the fastest available detectors. It is also highly desirable
to have spectral tunability of the light source so that it can be tuned to the absorption maximum of the fluorophore being
measured.
Although a high repetition rate and narrow pulses are prerequisites for a light source suitable for TCSPC other light
source parameters are not as important for the performance of a system based on TCSPC due to nature of the technique.
Only a single emission photon is recorded for each excitation pulse, therefore a high intensity source is not required.
TCSPC is by nature a digital technique, it only measures whether a photon has been detected, and thus the pulse to pulse
peak power stability of the light source is not a concern. In addition, only the temporal position of single photons with
respect to each excitation pulse is measured, therefore as long as the same pulse that is used for excitation is used to
trigger the start signal– a pulse period jitter does not affect the result.
*[email protected]; phone +44 1506 425 300; fax +44 1506 425 320; edinst.com
1.2
Current light sources for TCSPC
Before the advent of picosecond visible lasers most TCSPC fluorescence measurement were made using a nanosecond
flashlamp. The nanosecond flashlamp works by discharging a high voltage between two electrodes; the discharge is
gated using a thyratron. Both electrodes are charged to a high voltage and when the thyratron is gated on, it rapidly
discharges the top electrode to ground potential. When this occurs there is a spark discharge across the electrodes. The
spectral profile of this discharge is dependent on the filler gas in the electrode chamber and the temporal profile is
dependent on the size of the gap; pressure and purity of the gas and cleanliness of the electrodes.
The nanosecond flashlamp provides excitation pulses with less power than a laser however they can produce a spectral
continuum from the vacuum-UV to infra red (<1000nm) depending on the filler gas chosen. The most popular filler gas
is hydrogen as it has a continuous spectrum with maximum emission in the UV and the narrowest temporal profile. The
repetition rate of the nanosecond flashlamp is typically 40kHz.
The nanosecond flashlamp is an established, versatile light source and there are many positive aspects to its use in timeresolved spectrofluorometers, however there are some drawbacks. The low repetition rate means a maximum detectable
emission rate of 2000 photon counts per second, no matter how fluorescent the sample, in order to ensure no more than
one photon per light flash is detected; this is because multiphoton events affect the statistics and lead to erroneous results
(pulse pile-up) [1,2]. In addition the nanosecond flashlamp requires a regular maintenance regime in order to maintain its
performance (its electrodes must be cleaned and realigned after every 50 hours operation). The pulse width of typically
1ns is often limiting fluorescence decay studies. Also the temporal profile of the lamp can vary with wavelength and can
alter with time.
The Bremsstrahlung of synchrotron radiation has been used for TCSPC measurements, too [1]. The benefit is the broad
spectrum and the potential tunability. However, the broad pulse width (>1ns) and the limited availability today make this
light source practically obsolete.
Titanium-Sapphire (Ti:Sapphire) lasers have the benefits of short pulse width, tunability and high repetition rate
(typically 80MHz). Depending on the version pulse width can vary from around 20 femtoseconds to a number of
picoseconds with a tunability from 700-980nm [3]. Ti:Sapphire lasers used to require fine-tuning and realignment to
maintain performance however computer controlled self-adjusting Ti:Sapphire lasers are now available. The high power
and short pulse width mean that two photon excitation of the sample is possible using these lasers (i.e. samples with
absorption at 350nm may be excited using the fundamental output). The spectral range of the Ti:Sapphire laser can be
extended to lower wavelengths with the use of Second Harmonic Generation (SHG), Third Harmonic Generation (THG)
and by pumping Optical Parametric Oscillators (OPO’s) with SHG crystals on there outputs.
The high repetition rate of Ti:Sapprire lasers, which is fixed by the resonator length, can be a disadvantage for
fluorescence measurements. Although it avoids problems with pulse pile-up it can cause problems if fluorescence
lifetimes are longer than 3ns as the fluorescence does not decay completely within the pulse period. The repetition rate of
the laser may be reduced by employing a pulse picker, however most pulse pickers generate RF noise synchronous with
the pulses which can make TCSPC measurements difficult. Also, the pulse picker generally does not fully suppress the
unwanted pulses, around 1% of the intermediate pulses still get through (leakage), this is a large amount in TCSPC
lifetime measurements as they are made in a semi-log plot of photon counts against time.
However, the main draw back to Titanium Sapphire lasers is cost. Ti:Sapphire lasers are an order of magnitude too
expensive to be used in a economically priced spectrofluorometer and the second harmonic generators and so on that are
required for visible or UV output adds considerably to the cost.
Diode lasers are extremely reliable and do not require any maintenance and alignment during normal operation. They are
monochromatic sources (typically 2nm linewidth) but the repetition rate is variable from 0 to 100MHz, depending on the
manufacturer. Diode lasers are commercially available at 375nm 405nm, 440nm, 475nm and 635nm and at many
wavelengths above 635nm. The lasers typically have sub-100 picosecond pulse widths (FWHM).
Diode lasers have a small cavity (a few μm) and operate in the single-transversal mode. This implies that the radiation
can, in theory, be focused into a diffraction-limited spot. However, the small cavity size means light is emitted over a
wide angle and additionally the cavity geometry means that the laser has an elliptical beam profile. Collimating this
source is difficult, an elliptical collimated beam can be produced using an aspherical lens, though the elliptical nature of
the beam is not important for bulk fluorescence measurements. Picosecond pulsed diode lasers are reliable, cheap
sources that are very good for use in TCSPC based lifetime fluorimeters, however they are inherently monochromatic,
therefore it can become expensive depending on how many excitation wavelengths are of interest to the researcher.
Additionally there are regions of the spectrum where pulsed diode lasers are not readily available (for instance between
475-635nm).
Light Emitting Diodes (LEDs) like laser diodes are extremely reliable, robust, compact and require no maintenance.
They have a typical bandwidth of 20nm and they can be pulsed at up to MHz frequencies. Interference filters should be
used on the output of the diodes in order to shape the spectral profile of the source. LED sources are difficult to collimate
and focus and their pulse width is in the order of 600ps. They are available in spectral ranges where picosecond pulsed
diode lasers are not available, i.e. between 500 and 600nm and below 375nm.
1.3
Supercontinuum generation from photonic crystal fibres
Supercontinuum generation is the spectral broadening of the narrow bandwidth of a laser pulse into a broad continuum
spanning more than an optical octave (from a wavelength to twice that wavelength) through nonlinear processes.
Supercontinuum generation has been performed for many years [4]; early experiments involved focussing femtosecond
pulses of low repetition rate or single shot amplified lasers into sapphire, glass or water. The expensive pump laser had to
be able to deliver pulses with megawatt peak power, and the low repetition rates was incompatible with TCSPC.
However at the Conference on Lasers and Electro-Optics (CLEO) 2000 a photonic crystal fibre (PCF) was demonstrated
that allowed the use of much lower laser pump powers to generate the supercontinua [5, 6]. Photonic crystal fibres are a
type of glass optical fibre in which the cladding contains a regular array of microscopic air holes along the entire length
of the fibre. The size and position of the holes determine the optical behaviour of the fibre. In order to generate a
supercontinuum in an optical fibre, the zero dispersion wavelength of the fibre must be close to that of the pump
wavelength. The zero dispersion wavelength of bulk silica is 1270nm, therefore conventional optical fibres have a zero
dispersion wavelength greater than this (typically 1300-1600nm), however most ultrafast lasers operate at wavelengths
below this. By careful design of the PCF it is possible to change its zero dispersion wavelength to anywhere in the region
of 560-1600nm, therefore the PCF can be designed to match the pump laser.
Photonic crystal fibres produce white light when pumped with kilowatt peak powers, and therefore require much smaller,
comparatively inexpensive pump lasers that operate in the megahertz pulse repetition regime. A supercontinuum light
source that contains a compact chip or fibre laser as pump source and produces a broad visible spectrum with picosecond
pulses and megahertz repetition rates has the possibility of becoming a versatile and popular light source for fluorescence
lifetime spectrometers based on time correlated single photon counting.
There are two different regimes for supercontinuum generation in photonic crystal fibres; in both regimes a single
wavelength is converted into a broad spectrum through a combination of nonlinear effects stemming from a small third
order nonlinearity in photonic crystal fibres. When a suitable photonic crystal fibre is pumped with ultrashort
(femtosecond and less than tens of picosecond) pulses; the high intensity of the pulses within the confines of the fibre
core leads to self-phase modulation and soliton formation (solitons are isolated optical pulses that propagate down the
fibre without experiencing any dispersion) due to a nonlinear refractive index intensity dependence. The solitons undergo
fission as they propagate down the fibre causing a shift in wavelength toward the red and the corresponding generation
of blue wavelength shifted non solitonic radiation [7]. The nature of the supercontinuum radiation produced is dependent
on the pulse dispersion properties of the fibre [8], the mode field diameter [9], the input pulse polarisation [10,11] and
the input pulse chirp [12]. Further spectral components of the supercontinuum are produced due to the interaction
between (the linear) dispersive waves and (the nonlinear) solitons in the photonic crystal fibre [13]. The length of the
fibre (propagation distance), is a crucial parameter in determining the shape and extent of the spectra produced, however
it is not the case that the longer the fibre the broader the spectral range. It has recently been shown [14] that even when
the fibre is too short (5.7mm) for multiple soliton fission, the self phase modulation alone may produce a broad
continuum (350nm to 3μm). However when longer pump pulses are used (beyond some picoseconds to continuous wave
radiation) the effects at the temporal edges of the pulse are insignificant, therefore the pump pulses may be considered to
be quasi-CW. In this case the major nonlinear process is phasematched four wave mixing, which generates sidebands at
equal frequencies from the pump when a specifically designed photonic crystal fibre is used [8,11]. As the radiation
propagates down the fibre, the sidebands increase in intensity and act as pumps for further four-wave mixing processes
[15, 16]. This can become a cascade process generating a broad, reasonably flat continuum spanning the visible and near
infrared part of the spectrum. The gain is again provided by a refractive index intensity dependence.
In this work we evaluate a new supercontinuum light source (SC450, Fianium) for use as a universal excitation source in
a fluorescence lifetime spectrometer. We show advantages and limitations, and device suitable options for wavelength
separation with the particular focus on high quality, high temporal resolution TCSPC measurements.
2
EXPERIMENTAL
To study the spectral and temporal characteristics of the supercontiuum light source with TCSPC detection, the laser was
coupled to a fluorescence lifetime spectrometer as shown in figure 1.
The supercontinuum light source has a specified spectral range from 450-2000nm, a repetition rate of 20MHz and an
output power of 2W over the full spectral range (although other repetition rates and output powers are available). The
supercontinuum light source consists of an Ytterbium (Yb) based fibre laser master pump source and a specially
designed photonic crystal fibre. The pump laser is specified with a fundamental pulse width of <8ps and with a timing
jitter of <1ps.
The TCSPC based fluorescence lifetime spectrometer (FLS920, Edinburgh Instruments) was used as standard, but with
the following modifications: The side port of the sample chamber, opposite the standard excitation arm, was used to
direct the laser radiation to the sample. A beam dump behind the sample is used to absorb the remainder of the beam not
absorbed by the sample. This greatly reduces reflections that can severely affect the quality of TCSPC measurements, as
they are typically shown in semi-logarithmic scale and therefore disclose even very small artefacts. In addition, the
standard PMT was replaced by the optional cooled MCP-PMT for measurements with ultimate time resolution. The
instrumental response function for a 17ps diode laser is 45ps.
For spectral resolution of the supercontinuum light source a 1800grooves/mm grating was used, which was illuminated
by the collimated laser beam. A slit was used for wavelength selection. Two amplified silicon pin diodes (OT900,
Edinburgh Instruments) were used to pick up radiation at different parts of the setup in order to provide trigger pulses to
the TCSPC data acquisition electronics of the FLS920 spectrometer. It should be noted that the optical system was setup
to minimize any temporal effects that would originate from varying beam propagations by varying the wavelengths. By
comparing measurements that were made at zero order and at first order (with additional wavelength selection filter) we
could confirm that temporal pulse displacements were less than 20ps and pulse broadening less than 5ps per 500nm.
Monochromators themselves can introduce temporal dispersion and shift effects. In order to separate temporal dispersion
effects caused by the emission monochromator from those of the laser source under investigation, all measurements
reported here were made with the emission monochromator in Zero order.
A
G
D
F
Sp
L
S
B
BD
D
Fig. 1: Experimental setup: Sp- FLSP920 spectrometer with emission monochromator and cooled MCP-PMT detector; Lsupercontinuum laser; F- interference filter; D- fast trigger diode; G- grating on rotational mount; BD- Beam dump;
S- slit; A- setup for triggering at 700nm, B- setup for triggering on the excitation wavelength.
3
3.1
RESULTS
Optical power measurements
The optical power of the supercontinuum source was measured at the sample position (see figure 1) for wavelength
between 450nm and 800nm (figure 2). Note that the efficiency of the optical elements, in particular the efficiency of the
grating has not been corrected for. The grating efficiency is responsible for the drop of optical power beyond 700nm.
The more than two-fold increase in optical power in the region of 480nm and the dramatic cut-off at around 460nm is a
feature of the laser. Although the laser is specified to have optical output only down to 450nm, it is possible to perform
lifetime measurements down to excitation wavelength of 440nm, due to the high sensitivity of the TCSPC based
instrument.
Average Power (uW)
300
250
200
150
100
50
0
450
550
650
750
Wavelength (nm)
Fig. 2.: Optical power measured at the sample position, using a power meter (LM2, Coherent).
3.2
Pulse width measurements
The width and the shape of the instrumental response function (IRF) are crucial indicators for the temporal resolution
that can be achieved with the fluorescence lifetime spectrometer. In the absence of a suitable trigger output at the laser
(the TTL output that is available does not provide sufficient timing accuracy), a beam splitter with trigger diode
(attachment A in Figure 1) was used to provide synchronization pulses for the TCSPC electronics. The smallest widths
of the IRFs were only obtained with an interference filter placed in the reference beam. A suitable transmission
wavelength for this interference filter was found to be 700nm or higher. 700nm was used for the experiments described
here.
Figure 3 shows two example measurements of IRFs at different wavelength. IRFs measured at shorter wavelength than
the reference wavelength are temporally delayed and have a larger full width at half maximum (FWHM) than those
measured at the reference wavelength of 700nm.
Figure 4 shows the compiled results of pulse shift and the pulse width with respect to the laser wavelength. The shape of
the graph “pulse position versus wavelength” has the familiar look of a fiber dispersion curve. For the laser that is
evaluated here the shift is 300ps when the wavelength is changed by 200nm (from 500nm to 700nm). The temporal shift
is not really surprising; the shift is simply a consequence of the fact that blue light travels through the fiber with reduced
speed in comparison to red or infrared light. For practical single TCSPC measurements this shift is not really a problem,
as long as the IRF is measured at the wavelength of excitation. The temporal shift becomes more problematic when time
resolved emission spectra (TRES) are to be generated from measurement series with fixed emission wavelength and
incremental increase of the excitation wavelength (Excitation TRES). In this case, triggering on the excitation
wavelength would be desirable as this would eliminate the temporal shift of the excitation pulses.
Figure 4 also shows the variation in pulse width with respect to wavelength; with the maximum IRF pulse width of
170ps at around 520nm. The pulse width increase is a bigger problem, as it means reduced temporal resolution in this
wavelength region. For the experimental setup of figure 1, with the trigger pickup attachment A, a pulse width increase
for shorter wavelengths could result from either: temporal jitter of the short wavelength pulses in respect to the pulses of
the reference wavelength 700nm, or broader optical laser pulses in the shorter wavelength region, compared to the
reference wavelength.
9.0
8.0
Counts/10
3
7.0
6.0
5.0
4.0
3.0
2.0
1.0
0.0
0.0
0.5
1.0
1.5
2.0
2.5
3.0
Time/ns
Fig. 3.: IRF measurements at 500nm (right) and at 700nm (left), using setup A.
FWHM(500nm)=150ps, FWHM (700nm) = 100ps
600
pulse position, width / ps
500
pulse position
400
300
200
100
0
400
pulse width
500
600
700
800
900
wavelength / nm
Fig. 4: Pulse shift and pulse width with respect to the wavelength for experimental setup A
The use of the experimental setup of figure 1 with the trigger pickup attachment B, i.e. triggering on the pulses of the
colour that are used to excite the sample, should not only eliminate the pulse shift caused by the dispersion of the
photonic crystal fiber, but should also provide an answer to whether jitter or broadening is the cause for the pulse width
dependence on wavelength. The results are shown in figure 5.
With the pulse triggering on the colour of the excitation pulses the dramatic shift in the peak position is eliminated. A
much smaller, yet significant, sinusoidal shaped pulse shift with respect to wavelength remains; most likely caused by
the optical system used for these experiments. However, shape and general tendency of the change of the pulse width
with respect to wavelength remains the same. This experiment leads to the conclusion that laser pulse broadening, not
pulse jitter, is the more likely reason for the pulse width increase at 520nm.
It should be noted that the pulse width reduces again at wavelengths below 520nm. The wavelength with the broadest
pulsewidth is therefore different than the wavelength with maximum average power (480nm).
600
pulse position, width / ps
500
pulse position
400
300
200
pulse width
100
0
400
500
600
700
800
900
wavelength / nm
Fig. 5: pulse shift and pulse width in dependence of the wavelength for experimental setup B
12
750
60
Counts / 1000
10
525
475 nm
8
6
4
2
0
0
100
200
300
400
500
600
700
800
900
1000
800
850
1100
time / ns
100
jitter pulse width / ps
80
60
40
20
0
400
450
500
550
600
650
700
750
900
wavelength / nm
Fig. 6: a) (top) examples of jitter measurements at four different wavelength
b) (bottom) measured (‘) and corrected (z) width of the pulse jitter for different wavelength
3.3
Pulse jitter measurements
By attaching both trigger pulses (from the setup attachments A and B) to the TCSPC data acquisition electronics,
independent pulse jitter measurements can be made. The peaks measured this way are not affected by the intrinsic pulse
height distribution of the MCP-PMT detector, but are solely caused by jitter of the timing electronics and the pulse jitter
of the light source. The results of the measurement are summarized in figure 6. For this particular experiment, the jitter
of the timing electronics, verified by measuring the signal from one trigger diode and using a signal splitter to connect it
to both TCSPC timing channels, was 40ps. Using the approximation that the square of the electronics jitter plus the
square of the jitter of the optical pulses results in the square of the width of the measured pulse, the jitter of the optical
pulses can be calculated. The results are shown in figure 6b.
3.4
Pulse height distribution measurements
It is apparent that the pulse width and pulse jitter measurements appear to give different results. One important
characteristic, the pulse height distribution of the laser pulses, has so far not been considered.
The pulse height distribution was measured using the trigger diode of the setup attachment B. (compared to PMT and
MCP-PMT detectors with high amplification noise, silicon photodiodes have negligible intrinsic pulse height
distribution). The lower threshold of the detection electronics was systematically increased in steps of 5mV and the count
rate of signal pulses was recorded. Subsequent differentiation of those curves resulted in the shape of the pulse height
distribution of the laser pulses with respect to the laser wavelength (Figure 7). The position of the curves of Figure 7 was
independently determined by monitoring the threshold at which the count rate starts to drop below the laser repetition
rate of 20MHz. This is reasonably accurate for narrow distributions (800nm), but has increased uncertainty for the
broader distributions.
By moving from longer (800nm) to shorter (475nm) wavelengths, the distribution of the laser pulses become
systematically broader, and the average pulse height decreases. While TCSPC, as a digital technique, is known to be
insensitive towards pulse height fluctuations, light source fluctuations this significant are likely to cause additional
problems. The constant fraction discriminator, which usually overcomes trigger problems from pulse height fluctuations
of photomultiplier detectors, could cause additional broadening to the resultant IRF when pulses with extremely different
heights and shape have to be processed.
13
12
11
No. of Pulses / MHz
10
9
8
7
6
5
4
3
2
1
0
10
30
50
70
90
110
130
150
pulse height [mV]
Fig. 7: Pulse height distribution at different wavelength: U at 475nm,  at 525nm, ο at 600nm, at 800nm
4
CONCLUSIONS
For a supercontinuum laser source of the type that was evaluated, best performance is achieved when the synchronisation
(trigger) pulses are derived at the wavelength that is used for the TCSPC experiments. This eliminates the temporal shift
for pulses at different wavelength caused by the dispersion of the photonics crystal fibre.
Even with the light source synchronised this way to the TCSPC electronics, an increase of the width of the instrumental
response function in the wavelength region of 520nm cannot be avoided. Optical pulses from the fibre in this wavelength
region have a significantly broadened pulse height distribution. Compared to the pulse height distribution of high gain
photomultipliers, where the pulse shape remains largely unaffected, the pulse height distribution of the pulses from the
fibre appear to be accompanied by changes in the temporal width of the pulses.
The supercontinuum laser has a maximum of optical power in the region of 480nm. At this spectral range a noticeable
reduction of the pulse width is observed. This phenomenon remains unexplained at present.
Further optimisation of the optical setup shown in Figure 1 is required for a robust, universal light source that is
spectrally tuneable and for use in a spectrometer based on TCSPC.
For wavelength selection the grating approach appears to be the safest route as all wavelength of the collimated laser
beam are evenly spectrally dispersed and wavelength selection is easily achieved by a slit. Higher order effects do not
impose problems, as the wavelengths of most interest are below 1000nm.
A variable interference filter is an alternative approach. This would come with a fixed bandwidth, which we do not see to
be a major limitation. However, interference filters often have side bands of transmission that could affect the overall
performance, both spectrally and temporally and would have to be considered. A major advantage of variable
interference filter would be the absence of temporal shifts and dispersion of this optical element.
The supercontinuum light source is very likely to become a common light source in fluorescence lifetime spectrometers.
It is already a light source that has optical pulse widths of less than 90ps throughout the spectrum; a pulse width that is
typical for picosecond pulsed lasers at a defined wavelength. The supercontinuum light source does not have the
afterpulse problem of picosecond diode lasers, and most importantly, it is spectrally tuneable, currently from 450nm to
wavelengths longer than the upper limit for single photon counting using PMTs of 1650nm, given by the R5509 infrared
sensitive photomultipliers. Today a tuneable supercontinuum laser is cost comparable to 4-5 picosecond diode lasers.
The currently biggest limitation of the supercontinuum light source is the lower wavelength cut-off of 450nm. There are
already reports that wavelength below 400nm can be generated [14,17,18,] by tapered or cascaded photonic crystal
fibres. We remain optimistic that wavelengths in this region can also be reliably generated using 10MHz, economically
priced, fibre lasers as pump sources.
5
ACKNOWLEDGEMENTS
We would like to thank Fianium for supplying us with the supercontinuum source used to perform the measurements.
We also wish to thank Dr Sandra Majno for her help in analysing the data produced during the course of the experiments.
REFERENCES
1.
2.
3.
4.
D.V. O’Connor, D. Phillips, Time Correlated Single Photon Counting, Academic Press, London, 1984.
W. Becker, Advanced Time Correlated Single Photon Counting Techniques, Springer, Heidelberg, 2005.
W. Koechner, Solid State Laser Engineering, Fifth edition, Springer, Berlin, 1999.
R. R. Alfano and S. L. Sharpiro, “Emission in the region 4000 to 7000 Å via four-photon coupling in glass,” Phys.
Rev. Lett. 24, 584-587, 1970
5. J. Ranka et al, (postdeadline paper) CLEO, 2000
6. J. K. Ranka, R. S. Windeler, and A. J. Stentz, "Visible continuum generation in air silica microstructure optical
fibers with anomalous dispersion at 800nm ," Opt. Lett. 25, 25-27, 2000.
7.
A.V. Husakou and J. Herrmann, “Supercontinuum generation, four-wave mixing, and fission of higher-order
solitons in photonic crystal fibres” J. Opt. Soc. Am. B, 19, 2171-2182, 2002.
8. W. H. Reeves, D.V. Skryabin, F. Biancalana, J.C. Knight, P. St.J. Russell, F. Ominetto, A. Efimov and A.J. Taylor,
“Transformation and control of ultra-short pulses in dispersion-engineered photonic crystal fibres,” Nature, 424, 511515 (2003)
9. B. Kibler, J. M. Dudley, and S. Coen, “Supercontinuum generation and nonlinear pulse propagation in photonic
crystal fibre: influence of the frequency-dependent effective mode area,” Appl. Phys. B, 81, 337-342, 2005
10. L. Tartara, I. Cristiani, V. Degiorgio, F. Carbone, D. Faccio, M. Romagnoli, and W. Belardi, “Phase matched
nonlinear interactions in a holey fiber induced by infrared supercontinuum generation,” Opt. Commun, 215, 191-197,
2003
11. S. Coen, A.H.L. Chau, R. Leonhardt, J.D. Harvey, J.C. Knight, W.J. Wadsworth and P. St. J. Russell,
“Supercontinuum generation by stimulated Raman scattering and parametric four-wave mixing in photonic crystal
fibers”, J. Opt. Soc. Am. B 19, 753-764, 2002.
12. G. Sansone, G. Steinmeyer, C. Vozzi, S. Stagira, M. Nisoli, S. De Silvestri, K. Starke, D. Ristau, B. Schenkel, J.
Biegert, A. Gosteva and U. Keller, “Mirror dispersion control of a holey fiber supercontinuum,” Appl. Phys. B, 78,
551-555, 2004.
13. D. V. Skyabin, F. Luan, J. C. Knight and P. S. Russell, “Soliton self-frequency shift cancellation in photonic crystal
fibers,” Science, 301, 1705-1708, 2003
14. F. G. Omenetto, N. A. Wolchover, M. R. Wehner, M. Ross, A. Efimov, A. J. Taylor, V. V. R. K. Kumar, A. K.
George, J. C. Knight, N. Y. Joly, and P. S. J. Russell, "Spectrally smooth supercontinuum from 350 nm to 3 μm in
sub-centimeter lengths of soft-glass photonic crystal fibers," Opt. Express 14, 4928-4934, 2006
15. 16. C. Xiong, A. Witkowska, S. G. Leon-Saval, T. A. Birks, and W. J. Wadsworth, "Enhanced visible continuum
generation from a microchip 1064nm laser ," Opt. Express 14, 6188-6193, 2006
16. W. Wadsworth, N. Joly, J. Knight, T. Birks, F. Biancalana, and P. Russell, "Supercontinuum and four-wave mixing
with Q-switched pulses in endlessly single-mode photonic crystal fibres," Opt. Express 12, 299-309, 2004
17. A. Kudlinski, A. K. George, J. C. Knight, J. C. Travers, A. B. Rulkov, S. V. Popov, and J. R. Taylor, "Zerodispersion wavelength decreasing photonic crystal fibers for ultraviolet-extended supercontinuum generation," Opt.
Express 14, 5715-5722, 2006
18. J. C. Travers, S. V. Popov, and J. R. Taylor, "Extended blue supercontinuum generation in cascaded holey fibers,"
Opt. Lett. 30, 3132-3134, 2005.