Volume 7 | Number 9 | September 2008 | Pages 993–1100
An international journal
PERSPECTIVE www.rsc.org/pps | Photochemical & Photobiological Sciences
Time-resolved methods in biophysics. 7. Photon counting vs. analog
time-resolved singlet oxygen phosphorescence detection†
Ana Jim´ enez-Banzo,aXavier Rag` as,aPeter Kapustaband Santi Nonell*a
Received 13th March 2008, Accepted 9th May 2008
First published as an Advance Article on the web 25th June 2008
Two recent advances in optoelectronics, namely novel near-IR sensitive photomultipliers and
inexpensive yet powerful diode-pumped solid-state lasers working at kHz repetition rate, enable the
time-resolved detection of singlet oxygen (O2(a1Dg)) phosphorescence in photon counting mode,
thereby boosting the time-resolution, sensitivity, and dynamic range of this well-established detection
technique. Principles underlying this novel approach and selected examples of applications are provided
in this perspective, which illustrate the advantages over the conventional analog detection mode.
The lowest electronic excited state of molecular oxygen, singlet
oxygen or O2(a1Dg), is a non-radical, highly-reactive oxidising
species (ROS) that is produced very often in chemical,1enzyme2
is indeed a very general process that results from the interaction
of a light-activated molecule, referred to as the photosensitiser,
with ambient molecular oxygen. It is also a most convenient
method to create a controlled population of O2(a1Dg) molecules
in a system under study. Thus, the electronically-excited states of
the photosensitiser, particularly the longer-lived triplet state, are
the two species resulting in the production of O2(a1Dg):
aGrup d’Enginyeria Molecular, Institut Qu´ ımic de Sarri` a, Universitat
Ramon Llull, Via Augusta 390, 08017, Barcelona, Spain. E-mail:
firstname.lastname@example.org; Fax: +34 932 05 62 66; Tel: +34 932 67 20 00
bPicoQuant GmbH, Rudower Chaussee 29, D-12489, Berlin, Germany
†Edited by T. Gensch and C. Viappiani. This paper is derived from the
spectroscopic methods in biophysics (organized by the Italian Society of
Pure and Applied Biophysics), held in Venice in January 2006.
Ana Jim´ enez-Banzo graduated
in Chemistry in 2003 from
the Institut Qu´ ımic de Sarri` a
(Barcelona), where she also
earned her PhD in 2008 under
the supervision of Santi Nonell.
Her research focused on the pro-
duction and diffusion of singlet
oxygen in biological media. She
is currently working as a tech-
nology transfer consultant.
istry in 2005 from the Institut
Qu´ ımic de Sarri` a (Barcelona),
where he is currently pursuing
a PhD. His research involves
the use of time-resolved singlet
oxygen detection techniques for
the study of photodynamic pro-
Peter Kapusta graduated in 1993
(Faculty of Science, Charles
University, Prague) and ob-
tained his PhD in 1998 (Czech
Technical University, Prague),
both under supervision of Prof.
Vlastimil Fidler. Since 2001 he
has worked as a senior sci-
entist at PicoQuant GmbH in
Berlin, where he develops instru-
ments and methods based on flu-
orescence and advanced photon
ical Chemistry at the Institut
Qu´ ımic de Sarri` a in Barcelona.
He obtained his PhD in 1988
from the Max-Planck-Institut
f¨ ur Strahlenchemie where he was
awarded the Otto-Hahn Medal
of the Max-Planck-Society for
his work with Silvia Braslavsky.
He has authored over 70 papers
in peer-reviewed journals on the
chemical aspects of photobiolog-
ical processes, particularly pho-
O2(a1Dg) is a strong electrophile and thus reacts readily with
electron-rich substrates such as unsaturated hydrocarbons, phe-
nols, amines or sulfides.4This is relevant for biological systems
because such reactive substrates are ubiquitous in membrane
lipids, proteins, and DNA.5–8Not surprisingly, O2(a1Dg) plays a
major role in several biomedical and biological processes ranging
from photodynamic therapy (PDT)9to plant defense,10where it
is being increasingly regarded as a signaling species.11–13O2(a1Dg)
can also be deactivated by physical interactions with surrounding
molecules. In chemically-unreactive systems, such interactions
are the major deactivation mode for O2(a1Dg) and ultimately
determine its lifetime, which ranges from 3.3 ls in neat water
to hundreds of milliseconds in the gas phase.14A minor fraction
of O2(a1Dg) molecules decays by emitting a photon at ca. 1270 nm
(Fig. 1). The quantum yield of such emission is extremely low,
ranging from 10−5to 10−7depending on its environment.15Never-
theless, the time-resolved detection of O2(a1Dg) phosphorescence
(TRPD) is regarded as the most direct and specific means for
This journal is © The Royal Society of Chemistry and Owner Societies 2008 Photochem. Photobiol. Sci., 2008, 7, 1003–1010 | 1003
Near-infrared phosphorescence spectrum of O2(a1Dg).
monitoring this species, with the added benefit of providing
information about the kinetics of its production and decay.
detector to a conventional laser flash photolysis system. With
minor variations, the set-up uses a pulsed laser for excitation,
working at 1–25 Hz repetition rate and delivering millijoules
pulses of 1–20 ns pulse duration. The emission arising from the
sample is monitored at 90◦through either a monochromator or a
combination of long-pass and interference filters, which block the
excitation wavelengths and isolate the O2(a1Dg) phosphorescence.
germanium or indium-gallium arsenide photodiode and recorded
using a digital oscilloscope.16This approach, firmly established
since the middle of the 1980s, has provided most of the O2(a1Dg)
data underpinning the current understanding of this intermediate
species. Despite a long list of successes, analog optical techniques
suffer from a number of problems (vide infra), which have limited
the detection of O2(a1Dg) in biological samples in a time-resolved
manner. The remarkable evolution of optolectronics over the
last few years makes now possible to overcome such difficul-
ties. Specifically, the availability of near-infrared photomultiplier
tubes (PMTs), in conjunction with inexpensive diode-pumped
high-repetition pulsed lasers, allows the detection of O2(a1Dg)
phosphorescence in photon counting mode, improving the time
resolution and sensitivity of O2(a1Dg) detection. This concept was
pioneered in the early nineties by Egorov et al. using a custom-
made PMT.17With the recent advent of commercially-available
photon counting near-IR PMTs, a number of laboratories are
rapidly adopting it.18–24In this perspective, the two approaches
are compared and examples that illustrate the advantages of the
photon counting technique over the analog mode for O2(a1Dg)
detection are discussed.
as well as platinum(II)-2,3,7,8,12,13,17,18-octaethyl-21H,23H-
porphine (PtOEP) from Frontier Scientific, were used as received.
Toluene of spectroscopic grade was purchased from Solvents
Documentation Syntheses (SDS).
Singlet oxygen measurements
The singlet oxygen spectrophotometer used is based on the
PicoQuant Fluotime 200 fluorescence lifetime system with the fol-
lowing modifications: (1) A diode-pumped solid-state Q-switched
Nd:YAG laser (CryLas, FTSS355-Q) is used for excitation. This
laser pulses at either 355 nm (5 mW, 0.5 lJ per pulse) or 532 nm
(12 mW, 1.2 lJ per pulse). The repetition rate can be decreased if
necessary to ensure that the singlet oxygen emission has dropped
sufficiently when the subsequent laser pulse arrives (4–5 lifetimes).
(2) Instead of the standard single-grating monochromator a dual-
grating model has been selected, which allows extending its
dispersion range from 200 to 2000 nm. A flip mirror is used to
direct the dispersed light beam either to the visible or to the
near-IR detector ports. (3) A TE-cooled Hamamatsu near-IR
is used to detect the weak O2(a1Dg) phosphorescence. (4) The
output of the PMT is sent to a multichannel scaler (Becker and
Hickl, model MSA300 or PicoQuant’s Nanoharp 250). In order
to block NIR background radiation from the excitation source, a
at the exit port of the laser. Likewise, a cold mirror (Edmund
Scientific, Barrington, USA) is placed at the entry port of the
form reaching the detector. Data are processed using PicoQuant’s
FluoFit software. The reasons underlying the choice of this set-up
are discussed in the following sections.
A. Overview of photon counting techniques
Photon counting detection arises naturally from the response
of photomultiplier tubes to detected photons, as the intrinsic
response of a PMT to a photon striking its surface is a pulse
of electrical current. There are basically three main photon
counting techniques for time-resolved measurements: gated pho-
ton counting (GPC), multichannel scaling (MCS), and time-
summary is presented here.
counted per laser shot. This approach provides the most accurate
timing of the photon among all photon counting techniques,
down to a few ps per channel. Thus, TCSPC is the technique of
choice when time resolution is the prime need, at the expense of a
much longer acquisition time in order to build the complete signal
In GPC, all pulses in a pre-set time window above a minimum
threshold are counted. Repeating the measurements at different
ing is used to select only a portion of the total emitted photons. A
typical application is to discriminate between prompt and delayed
emissions, e.g., fluorescence from phosphorescence, or to remove
light scattering after excitation. The main disadvantage of GPC is,
again, that it does not count all detected photons but only those
within the gate. The minimum gate width is currently ∼500 ps.
Reconstruction of the signal’s time profile requires repeating the
acquisition at several delays between excitation and the counting
window, which results in much longer acquisition times.
MCS finally can be viewed as a multi-gated photon counting
in which all detected photons are counted and sorted out in the
1004 | Photochem. Photobiol. Sci., 2008, 7, 1003–1010This journal is © The Royal Society of Chemistry and Owner Societies 2008
different positions of the board memory. The time distribution of
all detected photons is thus obtained at once. The time resolution
of the technique is given by the memory speed and is currently
ca. 1 ns per channel, enough for O2(a1Dg) applications (vide infra).
Fig. 2 gives a pictorial representation of the three techniques.
counting (b) gated photon counting (GPC), and (c) multichannel scaling
(MCS). Each row represents the outcome of a single experiment, i.e., a
single laser shot. The full shape of the signal could be recovered in a
single experiment using the MCS. In TCSPC, (only) one count is obtained
generally every 100 experiments. The waveform intensity (i.e. histogram
height) may be orders of magnitude smaller than that obtained using
MCS with an equivalent number of shots. This is the slowest technique for
building up the count histogram, but the time resolution can be as high
as 1 ps, i.e., 1000-fold better than with MCS. In GPC, the position of the
gate must be varied to reconstruct the time profile. The resulting waveform
amplitude is, in comparison to MCS, 10- to 1000-fold lower.
A pictorial representation of (a) time-correlated single photon
Given the considerations above, what is then the best option
for photon counting time-resolved O2(a1Dg) phosphorescence
a tedious task, as the time gate must be scanned. The smaller the
gate width, the larger the number of acquisition cycles necessary
but one, at most, in TCSPC makes data acquisition unnecessarily
long, as there is hardly a need for picosecond time resolution in
O2(a1Dg) experiments. With emission quantum yields in the 10−5–
10−7range, NIR PMTs 20-fold noisier than visible PMTs, and
kinetics in the ns–ls range, the best option for O2(a1Dg) detection
is currently MCS.
B. Emission optics
Two fundamentally different approaches can be used for condi-
tioning the emission signal before sensing it with the detector.
The use of a monochromator allows to spectrally resolve the
luminescence arising from the sample at the expense of signal
pass and interference filters provides the highest throughput at the
expense of spectral resolution. There are situations which may
favour either approach, hence it is not possible to make a single
C. Choosing the detector
Currently, Hamamatsu is the only manufacturer marketing pho-
ton counting NIR PMTs. For the purpose of O2(a1Dg) detection,
only the series R5509 and H10330 need to be considered, the
high speed of the R3809U NIR microchannel plate providing no
distinct advantage for the measurement of lifetimes in the nano-
and microsecond time range.
The R5509 family offers two models, the R5509-43 and the
ranging from 300 nm to 1.4 lm or to 1.7 lm, respectively.
Alternatively, the H10330 series is a family of compact NIR-
PMT modules with a convenient internal thermoelectric cooler
that eliminates the need for liquid nitrogen and cooling water of
the R5509 family. Due to the substrate material (InP) of their
transmission-type photocathode, the spectral sensitivity range
starts at 950 nm. The longest detectable wavelength is determined
by the cathode material itself as follows 1.2 lm (H10330-25),
1.4 lm (H10330-45) and 1.7 lm (H10330-75).
PMTs sensitive up to 1.2 lm are inadequate for O2(a1Dg)
phosphorescence detection as it peaks at 1270 nm. On the
other hand, extending the spectral range to 1.7 lm increases
the dark count rate by one order of magnitude in both PMT
series, deteriorating the sensitivity of the photon counting system.
Therefore, the PMTs R5509-43 and H10330-45 are currently
the most suitable ones for O2(a1Dg) phosphorescence detection
In comparing both PMTs, the dark count levels are basically
the same, therefore other parameters must be taken into account
for making a purchase decision. The most remarkable features of
the H10330-45 are its faster time response and the convenience
of thermoelectric cooling. The transit time spread (TTS), defined
as the uncertainty of the delay between the moment of photon
absorption at the photocathode and the output pulse from the
anode, is also an advantage: 300 ps for the H10330-45 vs. 1.5 ns
for the R5509-43. Likewise, the rise time is ∼3-fold faster for
the H10330-45 (900 ps) than for the R5509-43 (3 ns). Thus,
convenience and time resolution point to the H10330-45 PMT as
the detector of choice for the dedicated detection of O2(a1Dg). On
the other hand, the R5509-43 shows a wider spectral response,
which may be advantageous for detection of the sensitiser’s
phosphorescence at the expense of lower time resolution and a
less comfortable system to work with.
D. The laser excitation source
Photon counting requires that the individual photons remain
distinguishableinthedetector output.Itis thereforeconvenient to
use lasers delivering energy pulses in the microjoule range to avoid
derived from such low-energy laser pulses, high repetition rates
are required, the upper limit being given by the inherent O2(a1Dg)
systems, a time window up to 100 ls is the largest one needed (see
below), for which lasers working at 10 kHz repetition rate are best
A typical experimental set-up for time- and spectrally-resolved
phosphorescence detection in photon counting mode is depicted
in Fig. 3.
This journal is © The Royal Society of Chemistry and Owner Societies 2008Photochem. Photobiol. Sci., 2008, 7, 1003–1010 | 1005
ly-resolved O2(a1Dg) phosphorescence detection.
Experimental set-up for the photon counting time- and spectral-
Analog vs. digital detection: a comparison
Over the last three decades, much has been learnt about analog
time-resolved detection of O2(a1Dg) phosphorescence, and the
limitations and pitfalls of this approach are now well understood.
Three main problems have been identified: (i) lack of time
resolution, which precludes obtaining kinetic information in the
the accurate determination of singlet oxygen production quantum
yields below 0.01; and (iii) interference of spurious signals such
as sensitiser fluorescence or scattered laser light, which result
in “spikes” at the earlier part of the signal that mask O2(a1Dg)
phosphorescence, particularly when its lifetime is only a few
microseconds as in aqueous solvents, or even create malfunctions
of the detection system. The following sections address these
problems and show how the photon counting approach can be
successfully used to overcome them.
A. Time resolution: what is the minimum lifetime that can be
measured with analog and photon counting techniques?
The lifetime of O2(a1Dg) in a particular environment is limited by
the presence of physical and chemical quenchers.27To the best
of our knowledge, the shortest O2(a1Dg) lifetime reported to date
from analog phosphorescence data is 209 ns, which was obtained
by the group of Schmidt using a 3 mm germanium photodiode.28
As shown in Fig. 4, this value is comparable to the full width
at half maximum (FWHM) of the response function of a typical
germanium diode (ca. 350 ns); hence, even using deconvolution
techniques, O2(a1Dg) lifetimes below this value can hardly be
would improve the time resolution,29but the cost in sensitivity is
unacceptable for most applications. In contrast, the FWHM of
the new near-IR photomultipliers is ca. 2 ns working in analog
mode. When used in photon counting mode, the time resolution
is better described by the transit-time-spread (TTS), and is as low
as 300 ps for these PMTs. Combined with a 1 ns pulsewidth laser,
the FWHM of our instrument’s response function (IRF) is ca.
7 ns, still 2 orders of magnitude faster than the typical germanium
photodiodes (Fig. 4).
In a triplet-photosensitized experiment, the time profile of
O2(1aDg) phosphorescence, St, is given by eqn (1):16
matsu H9170-45 PMT (the predecessor of H10330-45 PMT) in photon
counting mode (blue) and with the NorthCoast EO-817P Ge diode in
analog mode (red). Inset: detail of the IRF of the PMT.
Instrument response function (IRF) recorded with the Hama-
St= A(e−t/s1− e−t/s2) (1)
where A is an empirical constant and s1and s2are the lifetimes
of the decay and rise components of the signal, respectively. Stis
triplet excited state of the sensitiser and molecular oxygen, kd
is the rate constant for the sensitiser’s triplet-state decay by
oxygen-independent pathways, kq
quenching of the sensitiser’s triplet state, sTis the actual lifetime
of the photosensitiser triplet-state (1/sT= kd
is the actual O2(a1Dg) lifetime. According to eqn (2), the [O2(a1Dg)]
is zero at time t = 0, then grows to a maximum, and finally decays
back to zero. The factor sD/(sT− sD) implies that the signal rises
with the fastest of the two lifetimes, either sTor sD. In most air-
equilibrated neat solvents sT< sDand the natural observation is
made that the O2(a1Dg) phosphorescence grows with the triplet
lifetime and decays with the O2(a1Dg) own lifetime.
Conversely, under conditions of very strong quenching or low
oxygen availability, the counterintuitive observation is made that
O2(a1Dg) grows with its own lifetime and decays with the lifetime
of its precursor, the sensitiser’s triplet state (i.e., s1= sDand s2=
sTin eqn (1)).30Thus, the unambiguous assignment of s1and s2
requires that the triplet lifetime sTbe determined independently,
e.g. by transient absorption or by phosphorescence spectroscopy.
Fig. 5 shows the luminescence signals recorded from an
toluene in the presence of 0.18 M DABCO, a well-known singlet
oxygen quencher (kq= 2.2 × 108M−1s−1).27Porphine derivatives
are well-known, highly efficient O2(1Dg) photosensitisers, whose
strong phosphorescence provides a convenient means for
monitoring their triplet state.31Reconvolution fitting of eqn
(1) to the signal at 1275 nm gives two lifetimes s1 = 173 ns
and s2= 24 ns. When the emission is observed at 645 nm, the
wavelength for PtOEP phosphorescence,31the signal is found to
decay monoexponentially with lifetime sT = 175 ns. Thus, the
lifetime of O2(a1Dg) is sD= 24 ns in our system, which is one order
O2is the rate constant of energy transfer between the
O2is the rate constant for oxygen
O2[3O2]), and sD
1006 | Photochem. Photobiol. Sci., 2008, 7, 1003–1010This journal is © The Royal Society of Chemistry and Owner Societies 2008
fitting (red) gives sT= 175 ns; kexc= 375 nm. B: O2(a1Dg) phosphorescence at 1275 nm (black) and instrument response function (blue). Deconvolution
fitting (red) gives s1= 173 ns and s2= 24 ns; kexc= 532 nm.
Phosphorescence of PtOEP solutions in air-saturated toluene containing 0.18 M DABCO, A: PtOEP phosphorescence at 645 nm (black). Tail
of magnitude below the shortest lifetime ever reported for this
reactive oxygen species.28
As a further proof of the validity of the lifetime determinations,
the function sD
expected to increase linearly with [DABCO] according to sD
1 + kq
the rate constant found in literature, kq= 2.2 × 108M−1s−1.27
0/sD is plotted vs. [DABCO] in Fig. 6. sD
DABCO[DABCO].27The slope of the straight line is therefore
DABCO= (2.3 ± 0.2) × 108M−1s−1, in excellent agreement with
liner least-squares fit is kq
DABCO= (2.3 ± 0.2) × 108M−1s−1.
Clearly, the use of fast photomultiplier tubes as detectors
provides the ability to measure O2(a1Dg) lifetimes reliably down
to a few nanoseconds. Such time resolution paves now the way for
unravelling the behaviour of O2(a1Dg) in biological systems.
B. Sensitivity: what is the minimum O2(a1Dg) quantum yield that
can be measured?
The ability of a given photosensitiser to produce O2(a1Dg) is
quantitatively described by the O2(a1Dg) production quantum
yield, UD.16The method for UD determination relies on the
comparison of signal intensities for solutions of a sample and
a reference photosensitiser measured under matched conditions.16
of the analog and of the photon counting approach, i.e., which
is the minimum signal that can be distinguished from the
background noise in either case. In an analog detection system,
the output of the detector is fed to a digital oscilloscope, where the
the signal-to-noise ratio SNR can be indefinitely increased by
averaging a large number of independent signals (N) according
, the 8-bit resolution of the ADC converter built
in most digital oscilloscopes implies that a signal buried in a
noise background 28-fold higher (i.e., SNR = 0.004) cannot be
distinguished from zero, no matter how many signals are averaged
(Fig. 7). In practice, the situation is even worse due to baseline
fluctuations and other sources of noise.
Signals smaller than the resolution of the ADC cannot be distinguished
Effect of analog-to-digital conversion (ADC) in oscilloscopes.
The intrinsic nature of photon counting, on the contrary,
overcomes such problems: as long as there is a signal, i.e.,
non-random PMT pulses correlated with the laser flash, the
of signals acquired, i.e., by the acquisition time. For practical
reasons, this should not exceed a few hours - and in these cases
thestabilityofthe systeminvestigated shouldbe tested.Therefore,
photon counting is endowed with the largest intrinsic sensitivity.
To test this conclusion, we assessed the ability to gen-
erate O2(a1Dg) of substances generally regarded as non-
photosensitising. Two such substances are the sunscreens 4-
methoxy-2-hydroxybenzophenone (benzophenone-3, BP3) and
2-ethyl-2-cyano-3,3-diphenylacrylate (octocrylene, OCR).32BP3
and OCR failed to produce any O2(a1Dg) signal in methanol.
This journal is © The Royal Society of Chemistry and Owner Societies 2008 Photochem. Photobiol. Sci., 2008, 7, 1003–1010 | 1007
However, when a solution containing either BP3 or OCR was
irradiated in air-saturated cyclohexane at 355 nm, clear O2(a1Dg)
disappeared when the emission was monitored at 1150 nm, where
O2(a1Dg) does not emit, and also when either the sunscreen or
oxygen were excluded from the system. Comparison of the signal
amplitudes to that recorded for 1H-phenalen-1-one (PN), used as
a reference sensitiser yielded, UDvalues (2.7 ± 0.3) × 10−3for BP3
and (5.0 ± 0.1) × 10−4for OCR, respectively, which suggests that
these sunscreens may be acting as photosensitisers if localised in
In conclusion, the photon counting detection technique pro-
vides us with the ability to reliably measure UD values below
0.001. It might be possible to push this frontier further by
observing the signal through long-pass and interference filters
rather than through a monochromator. Values below 10−4have
sensitivity for kinetic information.
C. Effect of sensitiser’s fluorescence and scattered laser light
One of the most severe problems in analog O2(a1Dg) detection is
the presence of background scattered laser light and sensitiser
fluorescence. These appear as large spikes in the early stages
of the signal that, at the very least, mask the O2(a1Dg) signal
and, in the worst cases, may even saturate the detection system
creating artefacts. The end result is that the O2(a1Dg) signal
is severely distorted and rendered useless. Spectral isolation of
the 1270 nm emission band through a monochromator or a
combination of long-pass and interference filters is the simplest
and most common approach to minimize this problem, although
it is often still insufficient. Photon counting techniques are less
sensitive to this problem: as long as each detector pulse can be
counted independently, there is no upper limit for the number of
counts in any given channel and the spikes do not lead to artefacts
even if they are orders of magnitude larger than the O2(a1Dg)
component (Fig. 9).
Now, because the IRF of the detection system is ca. 7 ns, the
spike vanishes in a few tens of nanoseconds and the growth of the
O2(a1Dg) signal appears well resolved in most cases (see Fig. 9). In
practice, though, highly-scattering samples such as suspensions
(e.g. of solids or cells) result in spikes lasting sometimes up
to a few hundred nanoseconds, most likely due to long-lived
autofluorescence of the optical materials in the detection pathway
solution in toluene, kexc = 532 nm, 20 ns per channel, 2 million laser
pulses per signal.
Time-resolved emission spectra of an air-saturated PdTPPo
(filters and lenses)35or to the saturation-recovery cycle of the
PMT and preamplifier system: above a certain photon density
impinging on the photocathode, single photon pulses are not
resolved anymore and a quasi-continuous current flows through
the PMT’s dynode system. Such current changes the voltage
gradients between the dynodes and also saturates the input
the decay of the photon burst. The preamplifier’s input circuitry
also may need some time to recover.
sensitized by OCR (grey) and PN (black) in cyclohexane, fitted functions (red). Fitted parameters: BP3, OCR and PN: sD= 23 ls.
O2(a1Dg)transient at 1275 nm sensitized by A: BP3 (grey) and PN (black) in cyclohexane, fitted functions (red); B: O2(a1Dg) transient at 1275 nm
1008 | Photochem. Photobiol. Sci., 2008, 7, 1003–1010 This journal is © The Royal Society of Chemistry and Owner Societies 2008
earlier part of the signal, showing an artefact due to saturation of the detection optoelectronics; inset: full apparent signal; B: analog mode, 20 lJ cm−2,
the detector is not saturated but then the O2(a1Dg) signal is below the system’s sensitivity; inset: detail of the spike of the signal; C and D: photon counting
mode, 6 lJ cm−2. Note the spike in the early channels, more than three orders of magnitude stronger than the O2(a1Dg) signal. kexc= 532 nm.
Detected signals at 1275 nm upon irradiation of a ∼200 lM solution of EGFP in deuterated PBS. A: Analog mode, 200 lJ cm−2, detail of the
Two examples may serve to illustrate the points above: the
first one demonstrates the benefits of the time- and spectral-
resolution of the photon counting system to discriminate between
the sensitiser’s fluorescence, the sensitiser’s phosphorescence, and
O2(a1Dg) phosphorescence; the second one compares the attempts
to detect a O2(a1Dg) phosphorescence transient from a solution of
the green fluorescent protein mutant EGFP using the analog and
the photon counting approaches.
Porphycenes are structural isomers of porphyrins that are
phenylporphycene (PdTPPo) has demonstrated interesting pho-
tosensitising potential against a variety of tumour cells.37,38This
compound is weakly fluorescent though it phosphoresces strongly
and exhibits a high O2(a1Dg) production quantum yield.39Fig. 9
shows the luminescence signals recorded from an air-saturated
solution of PdTPPo in toluene: the earliest part of the signals
is dominated by the fluorescence decay; at later times a slower,
∼200 ns decay component can also be observed, whose spectrum
matches that of PdTPPo phosphorescence:39concomitant to this
decay, the growth of the O2(a1Dg) band at 1270 nm is well
resolved. Thus, the time and spectral resolution of the photon
counting system is able to provide information on the kinetics of
singlet oxygen photosensitization even in the presence of strong
background luminescence signals.
The second example is provided by green fluorescent proteins,
which have become very popular in cell biology for imaging of
organelles and processes.40,41These proteins tend to bleach upon
prolonged irradiation and it has been suggested that O2(a1Dg)
photosensitized by the protein is the main species responsible for
this process.42,43When a solution of the enhanced-GFP mutant
a clear transient could be observed with the germanium photo-
diode operating in analog mode (Fig. 10A, inset). However, the
lifetime of this transient, ca. 200 ls, was not consistent with the
of the sample with argon had no effect on the signal, suggesting
that it was an artifact due to saturation of the detection system
rather than a O2(a1Dg) transient. Inspection of the early part of
the signal revealed the saturation of the detection electronics by
the huge EGFP fluorescence burst, a problem well-known to the
O2(a1Dg) community for highly fluorescent samples.44Reducing
the laser energy down to 20 lJ cm−2prevented this problem but
then no O2(a1Dg) transient could be observed even after averaging
103laser shots (Fig. 10B), likely because it was below the system’s
When the EGFP sample was studied with the photon counting
system, a large fluorescence spike of almost 106counts could still
be observed, but it did not preclude the detection of an additional
Clearly, the use of the photon counting approach solves, or
at least substantially alleviates, the long-standing problem of the
signal spikes precluding the acquisition of clean O2(a1Dg) signals
in highly scattering and/or highly fluorescent samples. Again,
this improvement is judged essential for studying the behavior
of O2(a1Dg) in biological media. The first results obtained in our
as well as in other laboratories are already providing a wealth of
novel and highly exciting data.22,24,46,47As a final recommendation,
it should be noted that lasers with pulsewidth as small as possible
should be used to minimize the spike problem.
Conclusions and outlook
The advantages of photon counting time-resolved O2(a1Dg)
phosphorescence detection over the analog technique have been
highlighted in the preceding sections. Time resolution in the
nanosecond domain, higher sensitivity, and better ability to cope
with scattered laser light and sensitiser fluorescence give photon
This journal is © The Royal Society of Chemistry and Owner Societies 2008Photochem. Photobiol. Sci., 2008, 7, 1003–1010 | 1009
counting detection a clear edge over the analog mode. It can be
safely anticipated that progress in optoelectronic technology will
eventually provide new NIR PMTs with lower dark-count rates
and new lasers with the right balance of energy, repetition rate,
and wavelength tunability. Combined with parallel developments
in space resolution46these advances will push further our ability
to temporally and spatially detect O2(a1Dg) in systems of ever
increasing complexity, thereby widening and deepening our un-
derstanding of the roles of O2(a1Dg) in such systems.
This work was supported by a grant of the Spanish Ministerio
de Educaci´ on y Ciencia (CTQ2007-67763-C03-01/BQU). A. J.-B.
and X. R. thank the Generalitat de Catalunya and Fons Social
Europeu for their predoctoral fellowships.
1 C. S. Foote, Acc. Chem. Res, 1968, 1, 104.
2 J. R. Kanofsky, J. Biol. Chem., 1983, 258, 5991.
3 I. E. Kochevar and R. W. Redmond, Photosensitized production
of singlet oxygen, in Singlet oxygen, UV-A, and Ozone. Methods in
Enzymology, ed. L. Packer and H. Sies, Academic Press, San Diego,
2000, vol. 319, pp. 20–28.
4 C. S. Foote, J. S. Valentine, A. Greenberg and J. F. E. Liebman, Active
Oxygen in Biochemistry, Blackie Academic and Professional, London,
5 A. Michaeli and J. Feitelson, Photochem. Photobiol., 1994, 59(3),
6 A. Michaeli and J. Feitelson, Photochem. Photobiol., 1995, 61(3), 255.
7 J. L. Ravanat, P. Di Mascio, G. R. Martinez, M. H. G. Medeiros and J.
Cadet, J. Biol. Chem., 2000, 275, 40601.
8 X. S. Zhang, B. S. Rosenstein, Y. Wang, M. Lebwohl and H. C. Wei,
Free Radical Biol. Med., 1997, 23, 980.
9 D. E. J. G. Dolmans, D. Fukumura and R. K. Jain, Nat. Rev. Cancer,
2003, 3, 380.
10 C. Flors and S. Nonell, Acc. Chem. Res., 2006, 39, 293.
11 C. Laloi, K. Apel and A. Danon, Curr. Opin. Plant Biol., 2004, 7, 323.
12 C. H. Foyer and G. Noctor, Physiol. Plant., 2003, 119, 355.
13 S. W. Ryter and R. M. Tyrrell, Free Radical Biol. Med., 1998, 24, 1520.
14 C. Schweitzer and R. Schmidt, Chem. Rev., 2003, 103, 1685.
15 R. D. Scurlock, S. Nonell, S. E. Braslavsky and P. R. Ogilby, J. Phys.
Chem., 1995, 99, 3521.
16 S. Nonell and S. E. Braslavsky, Time-resolved singlet oxygen detection,
in Singlet oxygen, UV-A, and Ozone. Methods in Enzymology, ed.
L. Packer and H. Sies, Academic Press, San Diego, 2000, vol. 319,
Toleutaev and S. V. Zinukov, Chem. Phys. Lett., 1989, 163, 421.
18 M. Niedre, M. S. Patterson and B. C. Wilson, Photochem. Photobiol.,
2002, 75, 382.
19 O. Shimizu, J. Watanabe and K. Imakubo, J. Phys. Soc. Jpn., 1997, 66,
20 J. W. Snyder, E. Skovsen, J. D. C. Lambert and P. R. Ogilby, J. Am.
Chem. Soc., 2005, 127, 14558.
21 R. Dedic, A. Molnar, M. Korinek, A. Svoboda, J. Psencik and J. Hala,
J. Lumin., 2004, 108, 117.
22 A. Baier, M. Maier, R. Engl, M. Landthaler and W. Baumler, J. Phys.
Chem. B, 2005, 109, 3041.
23 D. B. Tada, L. L. R. Vono, E. L. Duarte, R. Itri, P. K. Kiyohara, M. S.
Baptista and L. M. Rossi, Langmuir, 2007, 23, 8194.
24 F. Postigo, M. L. Sagrista, M. A. De Madariaga, S. Nonell and M.
Mora, Biochim. Biophys. Acta, 2006, 1758, 583.
25 W. Becker, Advanced Time-Correlated Single Photon Counting Tech-
niques, Springer, Germany, 2005.
26 J. R. Lakowicz, Principles of fluorescence spectroscopy, Kluwer Aca-
demic/Plenum Publishers, New York, 2006.
27 F. Wilkinson, W. P. Helman and A. B. Ross, J. Phys. Chem. Ref. Data,
1995, 24, 663.
28 R. Schmidt and C. Tanielian, J. Phys. Chem., 2000, 104, 3177.
29 K.-K. Iu, R. D. Scurlock and P. R. Ogilby, J. Photochem., 1987, 37, 19.
30 J. G. Parker and W. D. Stanbro, Dependence of photosensitized singlet
oxygen production on porphyrin structure and solvent, in Porphyrin
localization and treatment of tumors, ed. D. R. Doiron and C. J. Gomer,
Alan R. Liss, New York, 1984, pp. 259–284.
31 S. L. Pan and L. J. Rothberg, J. Am. Chem. Soc., 2005, 127, 6087.
32 J. M. Allen, C. J. Gossett and S. K. Allen, Chem. Res. Toxicol., 1996,
33 L. De Sola, A. Jimenez-Banzo and S. Nonell, Afinidad, 2008, 64, 251.
34 S. Yamaguchi and Y. Sasaki, J. Photochem. Photobiol., A, 2001, 142,
35 R. D. Scurlock, K.-K. Iu and P. R. Ogilby, J. Photochem., 1987, 37,
36 J. C. Stockert, M. Ca˜ nete, A. Juarranz, A. Villanueva, R. W. Horobin,
J. Borrell, J. Teixido and S. Nonell, Curr. Med. Chem., 2007, 14, 997.
37 M.Ca˜ nete,A.Ortiz,A.Juarranz,A.Villanueva,S.Nonell,J.I.Borrell,
J. Teixid´ o and J. C. Stockert, Anti-Cancer Drug Des., 2000, 15, 143.
38 M. Ca˜ nete, C. Ortega, A. Gavalda, J. Cristobal, A. Juarranz, S. Nonell,
J. Teixido, J. I. Borrell, A. Villanueva, S. Rello and J. C. Stockert,
Int. J. Oncol., 2004, 24, 1221.
39 N. Rubio, F. Prat, N. Bou, J. I. Borrell, J. Teixido, A. Villanueva, A.
Juarranz, M. Canete, J. C. Stockert and S. Nonell, New J. Chem., 2005,
40 R. Y. Tsien, Annu. Rev. Biochem., 1998, 67, 509.
41 B. N. G. Giepmans, S. R. Adams, M. H. Ellisman and R. Y. Tsien,
Science, 2006, 312, 217.
42 A. F. Bell, D. Stoner-Ma, R. M. Wachter and P. J. Tonge, J. Am. Chem.
Soc., 2003, 125, 6919.
43 L. Greenbaum, C. Rothmann, R. Lavie and Z. Malik, Biol. Chem.,
2000, 381, 1251.
44 A. Beeby, A. W. Parker and C. F. Stanley, J. Photochem. Photobiol., B,
1997, 37, 267.
45 A. Jimenez-Banzo, S. Nonell, J. Hofkens and C. Flors, Biophys. J.,
2008, 94, 168.
46 J. W. Snyder, E. Skovsen,J. D. C. Lambert, L. Poulsen and P. R. Ogilby,
Phys. Chem. Chem. Phys., 2006, 8, 4280.
Biol. Med., 2008, 44, 1926.
1010 | Photochem. Photobiol. Sci., 2008, 7, 1003–1010This journal is © The Royal Society of Chemistry and Owner Societies 2008