Introduction to

1 Fluorescence
During the past 20 years there has been a remarkable light. Light travels 30 cm, or about one foot, in one
growth in the use of fluorescence in the biological sciences. nanosecond. Many fluorophores display subnanosecond
Fluorescence spectroscopy and time-resolved fluorescence lifetimes. Because of the short timescale of fluorescence,
are considered to be primarily research tools in biochem- measurement of the time-resolved emission requires
istry and biophysics. This emphasis has changed, and the sophisticated optics and electronics. In spite of the added
use of fluorescence has expanded. Fluorescence is now a complexity, time-resolved fluorescence is widely used
dominant methodology used extensively in biotechnology, because of the increased information available from the
flow cytometry, medical diagnostics, DNA sequencing, data, as compared with stationary or steady-state measure-
forensics, and genetic analysis, to name a few. Fluorescence ments. Additionally, advances in technology have made
detection is highly sensitive, and there is no longer the need time-resolved measurements easier, even when using mi-
for the expense and difficulties of handling radioactive trac- croscopes.
ers for most biochemical measurements. There has been Phosphorescence is emission of light from triplet excit-
dramatic growth in the use of fluorescence for cellular and ed states, in which the electron in the excited orbital has the
molecular imaging. Fluorescence imaging can reveal the same spin orientation as the ground-state electron. Transi-
localization and measurements of intracellular molecules, tions to the ground state are forbidden and the emission
sometimes at the level of single-molecule detection. rates are slow (103 to 100 s–1), so that phosphorescence life-
Fluorescence technology is used by scientists from times are typically milliseconds to seconds. Even longer
many disciplines. This volume describes the principles of lifetimes are possible, as is seen from "glow-in-the-dark"
fluorescence that underlie its uses in the biological and toys. Following exposure to light, the phosphorescence sub-
chemical sciences. Throughout the book we have included stances glow for several minutes while the excited phospho-
examples that illustrate how the principles are used in dif- rs slowly return to the ground state. Phosphorescence is
ferent applications. usually not seen in fluid solutions at room temperature. This
is because there exist many deactivation processes that
1.1. PHENOMENA OF FLUORESCENCE
compete with emission, such as non-radiative decay and
quenching processes. It should be noted that the distinction
Luminescence is the emission of light from any substance, between fluorescence and phosphorescence is not always
and occurs from electronically excited states. Lumines- clear. Transition metal–ligand complexes (MLCs), which
cence is formally divided into two categories—fluores- contain a metal and one or more organic ligands, display
cence and phosphorescence—depending on the nature of mixed singlet–triplet states. These MLCs display interme-
the excited state. In excited singlet states, the electron in the diate lifetimes of hundreds of nanoseconds to several
excited orbital is paired (by opposite spin) to the second microseconds. In this book we will concentrate mainly on
electron in the ground-state orbital. Consequently, return to the more rapid phenomenon of fluorescence.
the ground state is spin allowed and occurs rapidly by emis- Fluorescence typically occurs from aromatic mole-
sion of a photon. The emission rates of fluorescence are cules. Some typical fluorescent substances (fluorophores)
typically 108 s–1, so that a typical fluorescence lifetime is are shown in Figure 1.1. One widely encountered fluo-
near 10 ns (10 x 10–9 s). As will be described in Chapter 4, rophore is quinine, which is present in tonic water. If one
the lifetime (∏) of a fluorophore is the average time between observes a glass of tonic water that is exposed to sunlight, a
its excitation and return to the ground state. It is valuable to faint blue glow is frequently visible at the surface. This
consider a 1-ns lifetime within the context of the speed of glow is most apparent when the glass is observed at a right
1
2 INTRODUCTION TO FLUORESCENCE

Figure 1.2. Sir John Fredrich William Herschel, March 7, 1792 to

May 11, 1871. Reproduced courtesy of the Library and Information
Centre, Royal Society of Chemistry.
Figure 1.1. Structures of typical fluorescent substances.

light, or a white object, it yet exhibits in certain

angle relative to the direction of the sunlight, and when the aspects, and under certain incidences of the light, an
dielectric constant is decreased by adding less polar sol- extremely vivid and beautiful celestial blue colour,
vents like alcohols. The quinine in tonic water is excited by which, from the circumstances of its occurrence,
the ultraviolet light from the sun. Upon return to the ground would seem to originate in those strata which the light
state the quinine emits blue light with a wavelength near first penetrates in entering the liquid, and which, if not
strictly superficial, at least exert their peculiar power
450 nm. The first observation of fluorescence from a qui-
of analysing the incident rays and dispersing those
nine solution in sunlight was reported by Sir John Frederick which compose the tint in question, only through a
William Herschel (Figure 1.2) in 1845.1 The following is an very small depth within the medium.
excerpt from this early report: To see the colour in question to advantage, all
that is requisite is to dissolve the two ingredients
On a case of superficial colour presented by a homo- above mentioned in equal proportions, in about a
geneous liquid internally colourless. By Sir John hundred times their joint weight of water, and
Frederick William Herschel, Philosophical Translation having filtered the solution, pour it into a tall nar-
of the Royal Society of London (1845) 135:143–145. row cylindrical glass vessel or test tube, which is
Received January 28, 1845 — Read February 13, 1845. to be set upright on a dark coloured substance
before an open window exposed to strong day-
"The sulphate of quinine is well known to be of light or sunshine, but with no cross lights, or any
extremely sparing solubility in water. It is however strong reflected light from behind. If we look
easily and copiously soluble in tartaric acid. Equal down perpendicularly into the vessel so that the
weights of the sulphate and of crystallised tartaric visual ray shall graze the internal surface of the
acid, rubbed up together with addition of a very little glass through a great part of its depth, the whole
water, dissolve entirely and immediately. It is this of that surface of the liquid on which the light
solution, largely diluted, which exhibits the optical first strikes will appear of a lively blue, ...
phenomenon in question. Though perfectly transparent If the liquid be poured out into another vessel,
and colourless when held between the eye and the the descending stream gleams internally from all
PRINCIPLES OF FLUORESCENCE SPECTROSCOPY 3

its undulating inequalities, with the same lively ecules, atoms are generally nonfluorescent in condensed
yet delicate blue colour, ... thus clearly demon- phases. One notable exception is the group of elements
strating that contact with a denser medium has no commonly known as the lanthanides.4 The fluorescence
share in producing this singular phenomenon. from europium and terbium ions results from electronic
The thinnest film of the liquid seems quite as
transitions between f orbitals. These orbitals are shielded
effective in producing this superficial colour as a
considerable thickness. For instance, if in pour- from the solvent by higher filled orbitals. The lanthanides
ing it from one glass into another, ... the end of display long decay times because of this shielding and low
the funnel be made to touch the internal surface emission rates because of their small extinction coeffi-
of the vessel well moistened, so as to spread the cients.
descending stream over an extensive surface, the Fluorescence spectral data are generally presented as
intensity of the colour is such that it is almost emission spectra. A fluorescence emission spectrum is a
impossible to avoid supposing that we have a
plot of the fluorescence intensity versus wavelength
highly coloured liquid under our view."
(nanometers) or wavenumber (cm–1). Two typical fluores-
cence emission spectra are shown in Figure 1.3. Emission
It is evident from this early description that Sir Her-
spectra vary widely and are dependent upon the chemical
schel recognized the presence of an unusual phenomenon
structure of the fluorophore and the solvent in which it is
that could not be explained by the scientific knowledge of
dissolved. The spectra of some compounds, such as pery-
the time. To this day the fluorescence of quinine remains
lene, show significant structure due to the individual vibra-
one of the most used and most beautiful examples of fluo-
tional energy levels of the ground and excited states. Other
rescence. Herschel was from a distinguished family of sci-
compounds, such as quinine, show spectra devoid of vibra-
entists who lived in England but had their roots in Ger-
tional structure.
many.2 For most of his life Sir Herschel did research in
An important feature of fluorescence is high sensitivi-
astronomy, publishing only a few papers on fluorescence.
ty detection. The sensitivity of fluorescence was used in
It is interesting to notice that the first known fluo-
1877 to demonstrate that the rivers Danube and Rhine were
rophore, quinine, was responsible for stimulating the devel-
connected by underground streams.5 This connection was
opment of the first spectrofluorometers that appeared in the
demonstrated by placing fluorescein (Figure 1.1) into the
1950s. During World War II, the Department of War was
Danube. Some sixty hours later its characteristic green flu-
interested in monitoring antimalaria drugs, including qui-
orescence appeared in a small river that led to the Rhine.
nine. This early drug assay resulted in a subsequent pro-
Today fluorescein is still used as an emergency marker for
gram at the National Institutes of Health to develop the first
locating individuals at sea, as has been seen on the landing
practical spectrofluorometer.3
of space capsules in the Atlantic Ocean. Readers interested
Many other fluorophores are encountered in daily life.
in the history of fluorescence are referred to the excellent
The green or red-orange glow sometimes seen in antifreeze
summary by Berlman.5
is due to trace quantities of fluorescein or rhodamine,
respectively (Figure 1.1). Polynuclear aromatic hydrocar-
bons, such as anthracene and perylene, are also fluorescent, 1.2. JABLONSKI DIAGRAM
and the emission from such species is used for environmen-
tal monitoring of oil pollution. Some substituted organic The processes that occur between the absorption and emis-
compounds are also fluorescent. For example 1,4-bis(5- sion of light are usually illustrated by the Jablonski6 dia-
phenyloxazol-2-yl)benzene (POPOP) is used in scintilla- gram. Jablonski diagrams are often used as the starting
tion counting and acridine orange is often used as a DNA point for discussing light absorption and emission. Jablon-
stain. Pyridine 1 and rhodamine are frequently used in dye ski diagrams are used in a variety of forms, to illustrate var-
lasers. ious molecular processes that can occur in excited states.
Numerous additional examples of probes could be pre- These diagrams are named after Professor Alexander
sented. Instead of listing them here, examples will appear Jablonski (Figure 1.4), who is regarded as the father of flu-
throughout the book, with a description of the spectral orescence spectroscopy because of his many accomplish-
properties of the individual fluorophores. An overview of ments, including descriptions of concentration depolariza-
fluorophores used for research and fluorescence sensing is tion and defining the term "anisotropy" to describe the
presented in Chapter 3. In contrast to aromatic organic mol- polarized emission from solutions.7,8
4 INTRODUCTION TO FLUORESCENCE

Figure 1.4. Professor Alexander Jablonski (1898–1980), circa 1935.

Courtesy of his daughter, Professor Danuta Frackowiak.

Figure 1.3. Absorption and fluorescence emission spectra of perylene

and quinine. Emission spectra cannot be correctly presented on both
Throughout the 1920s and 30s the Depart-
the wavelength and wavenumber scales. The wavenumber presenta-
tion is correct in this instance. Wavelengths are shown for conven-
ment of Experimental Physics at Warsaw Univer-
ience. See Chapter 3. Revised from [5]. sity was an active center for studies on lumines-
cence under S. Pienkowski. During most of this
period Jablonski worked both theoretically and
Brief History of Alexander Jablonski
experimentally on the fundamental problems of
Professor Jablonski was born February 26, 1898 in photoluminescence of liquid solutions as well as
Voskresenovka, Ukraine. In 1916 he began his study on the pressure effects on atomic spectral lines in
of atomic physics at the University of Kharkov, which gases. A problem that intrigued Jablonski for
was interrupted by military service first in the Russian many years was the polarization of photolumi-
Army and later in the newly organized Polish Army nescence of solutions. To explain the experimen-
during World War I. At the end of 1918, when an inde- tal facts he distinguished the transition moments
pendent Poland was re-created after more than 120 in absorption and in emission and analyzed vari-
years of occupation by neighboring powers, Jablonski ous factors responsible for the depolarization of
left Kharkov and arrived in Warsaw, where he entered luminescence.
the University of Warsaw to continue his study of Jablonski's work was interrupted once again
physics. His study in Warsaw was again interrupted in by a world war. From 1939 to 1945 Jablonski
1920 by military service during the Polish-Bolshevik served in the Polish Army, and spent time as a
war. prisoner of first the German Army and then the
An enthusiastic musician, Jablonski played Soviet Army. In 1946 he returned to Poland to
first violin at the Warsaw Opera from 1921 chair a new Department of Physics in the new
to 1926 while studying at the university under Nicholas Copernicus University in Torun. This
Stefan Pienkowski. He received his doctorate in beginning occurred in the very difficult postwar
1930 for work "On the influence of the change of years in a country totally destroyed by World
wavelengths of excitation light on the fluores- War II. Despite all these difficulties, Jablonski
cence spectra." Although Jablonski left the opera with great energy organized the Department of
in 1926 and devoted himself entirely to scientific Physics at the university, which became a scien-
work, music remained his great passion until the tific center for studies in atomic and molecular
last days of his life. physics.
PRINCIPLES OF FLUORESCENCE SPECTROSCOPY 5

His work continued beyond his retirement in

1968. Professor Jablonski created a spectroscop-
ic school of thought that persists today through
his numerous students, who now occupy posi-
tions at universities in Poland and elsewhere.
Professor Jablonski died on September 9, 1980.
More complete accounts of his accomplishments
are given in [7] and [8].

A typical Jablonski diagram is shown in Figure 1.5. The

Figure 1.5. One form of a Jablonski diagram.
singlet ground, first, and second electronic states are depict-
ed by S0, S1, and S2, respectively. At each of these electron-
trum is typically a mirror image of the absorption spectrum
ic energy levels the fluorophores can exist in a number of
of the S0 6 S1 transition. This similarity occurs because
vibrational energy levels, depicted by 0, 1, 2, etc. In this
electronic excitation does not greatly alter the nuclear
Jablonski diagram we excluded a number of interactions that
geometry. Hence the spacing of the vibrational energy lev-
will be discussed in subsequent chapters, such as quenching,
els of the excited states is similar to that of the ground state.
energy transfer, and solvent interactions. The transitions
As a result, the vibrational structures seen in the absorption
between states are depicted as vertical lines to illustrate the
and the emission spectra are similar.
instantaneous nature of light absorption. Transitions occur in
Molecules in the S1 state can also undergo a spin con-
about 10–15 s, a time too short for significant displacement of
version to the first triplet state T1. Emission from T1 is
nuclei. This is the Franck-Condon principle.
termed phosphorescence, and is generally shifted to longer
The energy spacing between the various vibrational
wavelengths (lower energy) relative to the fluorescence.
energy levels is illustrated by the emission spectrum of
Conversion of S1 to T1 is called intersystem crossing. Tran-
perylene (Figure 1.3). The individual emission maxima
sition from T1 to the singlet ground state is forbidden, and
(and hence vibrational energy levels) are about 1500 cm–1
as a result the rate constants for triplet emission are several
apart. At room temperature thermal energy is not adequate
orders of magnitude smaller than those for fluorescence.
to significantly populate the excited vibrational states.
Molecules containing heavy atoms such as bromine and
Absorption and emission occur mostly from molecules with
iodine are frequently phosphorescent. The heavy atoms
the lowest vibrational energy. The larger energy difference
facilitate intersystem crossing and thus enhance phospho-
between the S0 and S1 excited states is too large for thermal
rescence quantum yields.
population of S1. For this reason we use light and not heat
to induce fluorescence.
Following light absorption, several processes usually 1.3. CHARACTERISTICS OF FLUORESCENCE
occur. A fluorophore is usually excited to some higher EMISSION
vibrational level of either S1 or S2. With a few rare excep-
tions, molecules in condensed phases rapidly relax to the The phenomenon of fluorescence displays a number of gen-
lowest vibrational level of S1. This process is called internal eral characteristics. Exceptions are known, but these are
conversion and generally occurs within 10–12 s or less. Since infrequent. Generally, if any of the characteristics described
fluorescence lifetimes are typically near 10–8 s, internal in the following sections are not displayed by a given fluo-
conversion is generally complete prior to emission. Hence, rophore, one may infer some special behavior for this com-
fluorescence emission generally results from a thermally pound.
equilibrated excited state, that is, the lowest energy vibra-
tional state of S1.
1.3.1. The Stokes Shift
Return to the ground state typically occurs to a higher
excited vibrational ground state level, which then quickly Examination of the Jablonski diagram (Figure 1.5) reveals
(10–12 s) reaches thermal equilibrium (Figure 1.5). Return to that the energy of the emission is typically less than that of
an excited vibrational state at the level of the S0 state is the absorption. Fluorescence typically occurs at lower energies
reason for the vibrational structure in the emission spectrum or longer wavelengths. This phenomenon was first observed
of perylene. An interesting consequence of emission to by Sir. G. G. Stokes in 1852 at the University of Cam-
higher vibrational ground states is that the emission spec- bridge.9 These early experiments used relatively simple
6 INTRODUCTION TO FLUORESCENCE

instrumentation (Figure 1.6). The source of ultraviolet exci- sidered as in some way or other qualitatively different
tation was provided by sunlight and a blue glass filter, from the original light."
which was part of a stained glass window. This filter selec-
tively transmitted light below 400 nm, which was absorbed Careful reading of this paragraph reveals several
by quinine (Figure 1.6). The incident light was prevented important characteristics of fluorescent solutions. The qui-
from reaching the detector (eye) by a yellow glass (of wine) nine solution is colorless because it absorbs in the ultravio-
filter. Quinine fluorescence occurs near 450 nm and is let, which we cannot see. The blue color comes only from a
therefore easily visible. region near the surface. This is because the quinine solution
It is interesting to read Sir George's description of the was relatively concentrated and absorbed all of the UV in
observation. The following is from his report published in the first several millimeters. Hence Stokes observed the
1852:9 inner filter effect. After passing through the solution the
light was "enfeebled" and no longer capable of causing the
On the Change of Refrangibility of Light. By G. G. blue glow. This occurred because the UV was removed and
Stokes, M.A., F.R.S., Fellow of Pembroke College, the "enfeebled" light could no longer excite quinine. How-
and Lucasian Professor of Mathematics in the ever, had Sir George used a second solution of fluorescein,
University of Cambridge. Phil. Trans. Royal rather than quinine, it would have still been excited because
Society of London (1852) pp. 463-562. of the longer absorption wavelength of fluorescein.
Received May 11, — Read May 27, 1852. Energy losses between excitation and emission are
observed universally for fluorescent molecules in solution.
"The following researches originated in a considera-
One common cause of the Stokes shift is the rapid decay to
tion of the very remarkable phenomenon discovered
by Sir John Herschel in a solution of sulphate of qui- the lowest vibrational level of S1. Furthermore, fluo-
nine, and described by him in two papers printed in the rophores generally decay to higher vibrational levels of S0
Philosophical Transactions for 1845, entitled "On a (Figure 1.5), resulting in further loss of excitation energy by
Case of Superficial Colour presented by a Homoge- thermalization of the excess vibrational energy. In addition
neous Liquid internally colourless," and "On the to these effects, fluorophores can display further Stokes
Epipolic Dispersion of Light." The solution of quinine, shifts due to solvent effects, excited-state reactions, com-
though it appears to be perfectly transparent and
plex formation, and/or energy transfer.
colourless, like water, when viewed by transmitted
light, exhibits nevertheless in certain aspects, and
under certain incidences of the light, a beautiful celes- Brief History of Sir. G. G. Stokes:
tial blue colour. It appears from the experiments of Sir
Professor Stokes was born in Ireland, August 3, 1819
John Herschel that the blue colour comes only from a
(Figure 1.7). He entered Pembroke College, Cam-
stratum of fluid of small but finite thickness adjacent
bridge, in 1837, and was elected as a fellow of Pem-
to the surface by which the light enters. After passing
broke immediately upon his graduation in 1841. In
through this stratum, the incident light, though not
1849 Stokes became Lucasian Professor at Cam-
sensibly enfeebled nor coloured, has lost the power of
bridge, a chair once held by Newton. Because of poor
producing the same effect, and therefore may be con-
endowment for the chair, he also worked in the Gov-
ernment School of Mines.
Stokes was involved with a wide range of sci-
entific problems, including hydrodynamics, elas-
ticity of solids, and diffraction of light. The wave
theory of light was already known when he
entered Cambridge. In his classic paper on qui-
nine, he understood that light of a higher "refran-
gibility" or frequency was responsible for the
blue glow of lower refrangibility or frequency.
Thus invisible ultraviolet rays were absorbed to
produce the blue light at the surface. Stokes later
suggested using optical properties such as
absorption, colored reflection, and fluorescence,
Figure 1.6. Experimental schematic for detection of the Stokes shift. to identify organic substances.
PRINCIPLES OF FLUORESCENCE SPECTROSCOPY 7

Later in life Stokes was universally honored

with degrees and medals. He was knighted in
1889 and became Master of Pembroke College in
1902. After the 1850s, Stokes became involved
in administrative matters, and his scientific pro-
ductivity decreased. Some things never change.
Professor Stokes died February 1, 1903.

1.3.2. Emission Spectra Are Typically Independent

of the Excitation Wavelength

Another general property of fluorescence is that the same

fluorescence emission spectrum is generally observed irre-
spective of the excitation wavelength. This is known as
Kasha's rule,10 although Vavilov reported in 1926 that quan-
tum yields were generally independent of excitation wave-
length.5 Upon excitation into higher electronic and vibra-
tional levels, the excess energy is quickly dissipated, leav-
ing the fluorophore in the lowest vibrational level of S1.
This relaxation occurs in about 10–12 s, and is presumably a
result of a strong overlap among numerous states of nearly
equal energy. Because of this rapid relaxation, emission
spectra are usually independent of the excitation wave- Figure 1.7. Sir George Gabriel Stokes, 1819–1903, Lucasian
length. Exceptions exist, such as fluorophores that exist in Professor at Cambridge. Reproduced courtesy of the Library and
two ionization states, each of which displays distinct Information Centre, Royal Society of Chemistry.
absorption and emission spectra. Also, some molecules are
known to emit from the S2 level, but such emission is rare 1.8. These peaks are due to transitions from the lowest
and generally not observed in biological molecules. vibrational level of the S0 state to higher vibrational levels
It is interesting to ask why perylene follows the mirror- of the S1 state. Upon return to the S0 state the fluorophore
image rule, but quinine emission lacks the two peaks seen can return to any of the ground state vibrational levels.
in its excitation spectrum at 315 and 340 nm (Figure 1.3). These vibrational energy levels have similar spacing to
In the case of quinine, the shorter wavelength absorption those in the S1 state. The emission spectrum shows the
peak is due to excitation to the second excited state (S2), same vibrational energy spacing as the absorption spec-
which relaxes rapidly to S1. Emission occurs predominant- trum. According to the Franck-Condon principle, all elec-
ly from the lowest singlet state (S1), so emission from S2 is tronic transitions are vertical, that is, they occur without
not observed. The emission spectrum of quinine is the mir- change in the position of the nuclei. As a result, if a partic-
ror image of the S0 6 S1 absorption of quinine, not of its ular transition probability (Franck-Condon factor) between
total absorption spectrum. This is true for most fluo- the 0th and 1st vibrational levels is largest in absorption,
rophores: the emission is the mirror image of S0 6 S1 the reciprocal transition is also most probable in emission
absorption, not of the total absorption spectrum. (Figure 1.8).
The generally symmetric nature of these spectra is a A rigorous test of the mirror-image rule requires that
result of the same transitions being involved in both absorp- the absorption and emission spectra be presented in appro-
tion and emission, and the similar vibrational energy levels priate units.12 The closest symmetry should exist between
of S0 and S1. In most fluorophores these energy levels are the modified spectra ⊗(Φ!)/Φ! and F(Φ!)/Φ!3 , where ⊗(Φ!) is the
not significantly altered by the different electronic distribu- extinction coefficient at wavenumber (Φ!), and F(Φ!) is the
tions of S0 and S1. Suppose the absorption spectrum of a relative photon flux over a wavenumber increment ΘΦ!.
fluorophore shows distinct peaks due to the vibrational Agreement between these spectra is generally found for
energy levels. Such peaks are seen for anthracene in Figure polynuclear aromatic hydrocarbons.
8 INTRODUCTION TO FLUORESCENCE

Figure 1.9. Absorption (pH 1, 7.64, and 13) and emission spectra (pH
7) of 1-hydroxypyrene-3,6,8-trisulfonate in water. From [11].

Figure 1.8. Mirror-image rule and Franck-Condon factors. The ab-

sorption and emission spectra are for anthracene. The numbers 0, 1, Excited-state reactions other than proton dissociation
and 2 refer to vibrational energy levels. From [11]. can also result in deviations from the mirror symmetry rule.
One example is shown in Figure 1.10, which shows the
1.3.3. Exceptions to the Mirror-Image Rule emission spectrum of anthracene in the presence of diethyl-
aniline.13 The structured emission at shorter wavelengths is
Although often true, many exceptions to the mirror-image a mirror image of the absorption spectrum of anthracene.
rule occur. This is illustrated for the pH-sensitive fluo- The unstructured emission at longer wavelengths is due to
rophore 1-hydroxypyrene-3,6,8-trisulfonate (HPTS) in Fig- formation of a charge-transfer complex between the excited
ure 1.9. At low pH the hydroxyl group is protonated. The state of anthracene and diethylaniline. The unstructured
absorption spectrum at low pH shows vibrational structure emission is from this complex. Many polynuclear aromatic
typical of an aromatic hydrocarbon. The emission spectrum hydrocarbons, such as pyrene and perylene, also form
shows a large Stokes shift and none of the vibrational struc- charge-transfer complexes with amines. These excited-state
ture seen in the absorption spectrum. The difference complexes are referred to as exciplexes.
between the absorption emission spectra is due to ionization Some fluorophores can also form complexes with
of the hydroxyl group. The dissociation constant (pKa) of themselves. The best known example is pyrene. At low con-
the hydroxyl group decreases in the excited state, and this centrations pyrene displays a highly structured emission
group becomes ionized. The emission occurs from a differ- (Figure 1.11). At higher concentrations the previously invis-
ent molecular species, and this ionized species displays a ible UV emission of pyrene becomes visible at 470 nm.
broad spectrum. This form of HPTS with an ionized This long-wavelength emission is due to excimer forma-
hydroxyl group can be formed at pH 13. The emission spec- tion. The term "excimer" is an abbreviation for an excited-
trum is a mirror image of the absorption of the high pH state dimer.
form of HPTS.
Changes in pKa in the excited state also occur for bio-
chemical fluorophores. For example, phenol and tyrosine 1.4. FLUORESCENCE LIFETIMES AND
each show two emissions, the long-wavelength emission QUANTUM YIELDS
being favored by a high concentration of proton acceptors. The fluorescence lifetime and quantum yield are perhaps
The pKa of the phenolic hydroxyl group decreases from 11 the most important characteristics of a fluorophore. Quan-
in the ground state to 4 in the excited state. Following exci- tum yield is the number of emitted photons relative to the
tation, the phenolic proton is lost to proton acceptors in the number of absorbed photons. Substances with the largest
solution. Depending upon the concentration of these accep- quantum yields, approaching unity, such as rhodamines,
tors, either the phenol or the phenolate emission may dom- display the brightest emissions. The lifetime is also impor-
inate the emission spectrum. tant, as it determines the time available for the fluorophore
PRINCIPLES OF FLUORESCENCE SPECTROSCOPY 9

individual relaxation processes leading to the relaxed S1

state. Instead, we focus on those processes responsible for
return to the ground state. In particular, we are interested in
the emissive rate of the fluorophore (⋅) and its rate of non-
radiative decay to S0 (knr).
The fluorescence quantum yield is the ratio of the num-
ber of photons emitted to the number absorbed. The rate
constants ⋅ and knr both depopulate the excited state. The
fraction of fluorophores that decay through emission, and
hence the quantum yield, is given by

⋅
Q! (1.1)
⋅ " knr
The quantum yield can be close to unity if the radia-
tionless decay rate is much smaller than the rate of radiative
decay, that is knr < ⋅. We note that the energy yield of fluo-
rescence is always less than unity because of Stokes losses.
For convenience we have grouped all possible non-radiative
decay processes with the single rate constant knr.
The lifetime of the excited state is defined by the aver-
Figure 1.10. Emission spectrum of anthracene in toluene containing
0.2 M diethylaniline. The dashed lines show the emission spectra of
age time the molecule spends in the excited state prior to
anthracene or its exciplex with diethylaniline. Figure revised from return to the ground state. Generally, fluorescence lifetimes
[13], photo from [14]. are near 10 ns. For the fluorophore illustrated in Figure 1.12
the lifetime is
1
∏! (1.2)
⋅ " knr
Fluorescence emission is a random process, and few
molecules emit their photons at precisely t = ∏. The lifetime
is an average value of the time spent in the excited state. For
a single exponential decay (eq. 1.13) 63% of the molecules
have decayed prior to t = ∏ and 37% decay at t > ∏.
An example of two similar molecules with different
lifetimes and quantum yields is shown in Figure 1.13. The
differences in lifetime and quantum yield for eosin and ery-
throsin B are due to differences in non-radiative decay rates.
Eosin and erythrosin B have essentially the same extinction
Figure 1.11. Emission spectra of pyrene and its excimer. The relative coefficient and the same radiative decay rate (see eq. 1.4).
intensity of the excimer peak (470 nm) decreases as the total concen- Heavy atoms such as iodine typically result in shorter life-
tration of pyrene is decreased from 6 x 10–3 M (top) to 0.9 x 10–4 M
times and lower quantum yields.
(bottom). Reproduced with permission from John Wiley and Sons Inc.
From [12]. The lifetime of the fluorophore in the absence of non-
radiative processes is called the intrinsic or natural lifetime,
and is given by
to interact with or diffuse in its environment, and hence the
1
information available from its emission. ∏n ! (1.3)
The meanings of quantum yield and lifetime are best ⋅
represented by a simplified Jablonski diagram (Figure In principle, the natural lifetime ∏n can be calculated
1.12). In this diagram we do not explicitly illustrate the from the absorption spectra, extinction coefficient, and
10 INTRODUCTION TO FLUORESCENCE

and emission wavelength, and assumes no change in excit-

ed-state geometry. A more complete form of eq. 1.4 (not
shown) includes a factor G = gl/gu on the right-hand side,
where gl and gu are the degeneracies of the lower and upper
states, respectively. For fluorescence transitions G = 1, for
phosphorescence transitions G = 1/3.
The natural lifetime can be calculated from the meas-
ured lifetime (∏) and quantum yield
Figure 1.12. A simplified Jablonski diagram to illustrate the meaning
of quantum yields and lifetimes.
∏n ! ∏ / Q (1.5)

emission spectra of the fluorophore. The radiative decay which can be derived from eqs. 1.2 and 1.3. Many biochem-
rate ⋅ can be calculated using15,16 ical fluorophores do not behave as predictably as unsubsti-
tuted aromatic compounds. Hence, there is often poor
$ F(Φ ) dΦ ⊗(Φ ) agreement between the value of ∏n calculated from eq. 1.5
⋅ # 2.88 # 109 n2
$ F(Φ )dΦ / Φ
" 3 Φ
dΦ
and that calculated from its absorption and emission spec-
tra (eq. 1.4). These discrepancies occur for a variety of
unknown and known reasons, such as a fraction of the flu-
⊗(Φ ) dΦ
! 2.88 # 109n2 <Φ $3 > $1 " Φ
(1.4) orophores located next to quenching groups, which some-
times occurs for tryptophan residues in proteins.
The quantum yield and lifetime can be modified by
where F(Φ!) is the emission spectrum plotted on the
factors that affect either of the rate constants (⋅ or knr). For
wavenumber (cm–1) scale, ⊗(Φ!) is the absorption spectrum,
example, a molecule may be nonfluorescent as a result of a
and n is the refractive index of the medium. The integrals
large rate of internal conversion or a slow rate of emission.
are calculated over the S0 : S1 absorption and emission
Scintillators are generally chosen for their high quantum
spectra. In many cases this expression works rather well,
yields. These high yields are a result of large ⋅ values.
particularly for solutions of polynuclear aromatic hydrocar-
Hence, the lifetimes are generally short: near 1 ns. The flu-
bons. For instance, the calculated value15 of ⋅ for perylene
orescence emission of aromatic substances containing
is 1.8 x 108 s–1, which yields a natural lifetime of 5.5 ns.
–NO2 groups are generally weak, primarily as a result of
This value is close to that observed for perylene, which dis-
large knr values. The quantum yields of phosphorescence are
plays a quantum yield near unity. However, there are
extremely small in fluid solutions at room temperature. The
numerous reasons why eq. 1.4 can fail. This expression
triplet-to-singlet transition is forbidden by symmetry, and
assumes no interaction with the solvent, does not consider
the rates of spontaneous emission are about 103 s–1 or small-
changes in the refractive index (n) between the absorption
er. Since knr values are near 109 s–1, the quantum yields of
phosphorescence are small at room temperature. From eq.
1.1 one can predict phosphorescence quantum yields of
10–6.
Comparison of the natural lifetime, measured lifetime,
and quantum yield can be informative. For example, in the
case of the widely used membrane probe 1,6-diphenyl-
1,3,5-hexatriene (DPH) the measured lifetime near 10 ns is
much longer than that calculated from eq. 1.4, which is near
1.5 ns.17 In this case the calculation based on the absorption
spectrum of DPH is incorrect because the absorption transi-
tion is to a state of different electronic symmetry than the
emissive state. Such quantum-mechanical effects are rarely
seen in more complex fluorophores with heterocyclic
Figure 1.13. Emission spectra of eosin and erythrosin B (ErB). atoms.
PRINCIPLES OF FLUORESCENCE SPECTROSCOPY 11

1.4.1. Fluorescence Quenching

The intensity of fluorescence can be decreased by a wide
variety of processes. Such decreases in intensity are called
quenching. Quenching can occur by different mechanisms.
Collisional quenching occurs when the excited-state fluo-
rophore is deactivated upon contact with some other mole-
cule in solution, which is called the quencher. Collisional
quenching is illustrated on the modified Jablonski diagram
Figure 1.14. Jablonski diagram with collisional quenching and fluo-
in Figure 1.14. In this case the fluorophore is returned to the rescence resonance energy transfer (FRET). The term Σki is used to
ground state during a diffusive encounter with the quencher. represent non-radiative paths to the ground state aside from quenching
The molecules are not chemically altered in the process. For and FRET.
collisional quenching the decrease in intensity is described
by the well-known Stern-Volmer equation: allowing fluorescence measurements to detect dynamic
processes in solution or in macromolecules. The basic idea
F0 is that absorption is an instantaneous event. According to
! 1 " K %Q& ! 1 " kq ∏0 %Q & (1.6)
F the Franck-Condon principle, absorption occurs so fast that
there is no time for molecular motion during the absorption
In this expression K is the Stern-Volmer quenching
process. Absorption occurs in the time it takes a photon to
constant, kq is the bimolecular quenching constant, ∏0 is the
travel the length of a photon: in less than 10–15 s. As a result,
unquenched lifetime, and [Q] is the quencher concentration.
absorption spectroscopy can only yield information on the
The Stern-Volmer quenching constant K indicates the sen-
average ground state of the molecules that absorb light.
sitivity of the fluorophore to a quencher. A fluorophore
Only solvent molecules that are immediately adjacent to the
buried in a macromolecule is usually inaccessible to water-
absorbing species will affect its absorption spectrum.
soluble quenchers, so that the value of K is low. Larger val-
Absorption spectra are not sensitive to molecular dynamics
ues of K are found if the fluorophore is free in solution or
and can only provide information on the average solvent
on the surface of a biomolecule.
shell adjacent to the chromophore.
A wide variety of molecules can act as collisional
In contrast to absorption, emission occurs over a longer
quenchers. Examples include oxygen, halogens, amines,
period of time. The length of time fluorescent molecules
and electron-deficient molecules like acrylamide. The
remain in the excited state provides an opportunity for
mechanism of quenching varies with the fluoro-
interactions with other molecules in solution. Collisional
phore–quencher pair. For instance, quenching of indole by
quenching of fluorescence by molecular oxygen is an excel-
acrylamide is probably due to electron transfer from indole
lent example of the expansion of time and distance provid-
to acrylamide, which does not occur in the ground state.
ed by the fluorescence lifetime. If a fluorophore in the
Quenching by halogen and heavy atoms occurs due to
excited state collides with an oxygen molecule, then the flu-
spin–orbit coupling and intersystem crossing to the triplet
orophore returns to the ground state without emission of a
state (Figure 1.5).
photon. The diffusion coefficient (D) of oxygen in water at
Aside from collisional quenching, fluorescence quench-
25EC is 2.5 x 10–5 cm2/s. Suppose a fluorophore has a life-
ing can occur by a variety of other processes. Fluorophores
time of 10 ns. Although 10 ns may appear to be a brief time
can form nonfluorescent complexes with quenchers. This
span, it is in fact quite long relative to the motions of small
process is referred to as static quenching since it occurs in the
molecules in fluid solution. The average distance (Θx2)1/2 an
ground state and does not rely on diffusion or molecular col-
oxygen molecule can diffuse in 10–8 s or 10 ns is given by
lisions. Quenching can also occur by a variety of trivial, i.e.,
the Einstein equation:
non-molecular mechanisms, such as attenuation of the inci-
dent light by the fluorophore itself or other absorbing species.
Θx2 ! 2D∏ (1.7)
1.4.2. Timescale of Molecular Processes
in Solution The distance is about 70 Å, which is comparable to the
The phenomenon of quenching provides a valuable context thickness of a biological membrane or the diameter of a
for understanding the role of the excited-state lifetime in protein. Some fluorophores have lifetimes as long as 400
12 INTRODUCTION TO FLUORESCENCE

ns, and hence diffusion of oxygen molecules may be relative angle between these moments determines the max-
observed over distances of 450 Å. Absorption measure- imum measured anisotropy [r0, see eq. 10.19]. The fluores-
ments are only sensitive to the immediate environment cence anisotropy (r) and polarization (P) are defined by
around the fluorophore, and then only sensitive to the
instantaneously averaged environment. I|| $ I!
Other examples of dynamic processes in solution r! (1.8)
I|| " 2I!
involve fluorophore–solvent interactions and rotational dif-
fusion. As was observed by Stokes, most fluorophores dis-
I|| $ I!
play emission at lower energies than their absorption. Most P! (1.9)
fluorophores have larger dipole moments in the excited I|| " I!
state than in the ground state. Rotational motions of small where I|| and Iz are the fluorescence intensities of the verti-
molecules in fluid solution are rapid, typically occurring on cally (||) and horizontally (z) polarized emission, when the
a timescale of 40 ps or less. The relatively long timescale of sample is excited with vertically polarized light. Anisotropy
fluorescence allows ample time for the solvent molecules to and polarization are both expressions for the same phenom-
reorient around the excited-state dipole, which lowers its enon, and these values can be interchanged using eqs. 10.3
energy and shifts the emission to longer wavelengths. This and 10.4.
process is called solvent relaxation and occurs within 10–10 Several phenomena can decrease the measured
s in fluid solution (Figure 1.14). It is these differences anisotropy to values lower than the maximum theoretical
between absorption and emission that result in the high sen- values. The most common cause is rotational diffusion
sitivity of emission spectra to solvent polarity, and the (Figure 1.15). Such diffusion occurs during the lifetime of
smaller spectral changes seen in absorption spectra. Solvent the excited state and displaces the emission dipole of the
relaxation can result in substantial Stokes shifts. In proteins, fluorophore. Measurement of this parameter provides infor-
tryptophan residues absorb light at 280 nm, and their fluo- mation about the relative angular displacement of the fluo-
rescence emission occurs near 350 nm. rophore between the times of absorption and emission. In
fluid solution most fluorophores rotate extensively in 50 to
100 ps. Hence, the molecules can rotate many times during
1.5. FLUORESCENCE ANISOTROPY the 1–10 ns excited-state lifetime, and the orientation of the
Anisotropy measurements are commonly used in the bio- polarized emission is randomized. For this reason fluo-
chemical applications of fluorescence. Anisotropy measure- rophores in non-viscous solution typically display
ments provide information on the size and shape of proteins anisotropies near zero. Transfer of excitation between fluo-
or the rigidity of various molecular environments. Anisotro- rophores also results in decreased anisotropies.
py measurements have been used to measure protein–pro- The effects of rotational diffusion can be decreased if
tein associations, fluidity of membranes, and for immuno- the fluorophore is bound to a macromolecule. For instance,
assays of numerous substances. it is known that the rotational correlation time for the pro-
Anisotropy measurements are based on the principle of tein human serum albumin (HSA) is near 50 ns. Suppose
photoselective excitation of fluorophores by polarized light. HSA is covalently labeled with a fluorophore whose life-
Fluorophores preferentially absorb photons whose electric time is 50 ns. Assuming no other processes result in loss of
vectors are aligned parallel to the transition moment of the anisotropy, the expected anisotropy is given by the Perrin
fluorophore. The transition moment has a defined orienta- equation:18
tion with respect to the molecular axis. In an isotropic solu- r0
tion, the fluorophores are oriented randomly. Upon excita- r! (1.10)
1 " (∏ / θ )
tion with polarized light, one selectively excites those fluo-
rophore molecules whose absorption transition dipole is where r0 is the anisotropy that would be measured in the
parallel to the electric vector of the excitation (Figure 1.15). absence of rotational diffusion, and θ is the rotational cor-
This selective excitation results in a partially oriented pop- relation time for the diffusion process. In this case binding
ulation of fluorophores (photoselection), and in partially of the fluorophore to the protein has slowed the probe's rate
polarized fluorescence emission. Emission also occurs with of rotational motion. Assuming r0 = 0.4, the anisotropy is
the light polarized along a fixed axis in the fluorophore. The expected to be 0.20. Smaller proteins have shorter correla-
PRINCIPLES OF FLUORESCENCE SPECTROSCOPY 13

The extent of energy transfer is determined by the dis-

tance between the donor and acceptor, and the extent of
spectral overlap. For convenience the spectral overlap (Fig-
ure 1.16) is described in terms of the Förster distance (R0).
The rate of energy transfer kT(r) is given by

kT (r ) !
1 R0
∏D r
( ) 6
(1.11)

where r is the distance between the donor (D) and acceptor

(A) and ∏D is the lifetime of the donor in the absence of
energy transfer. The efficiency of energy transfer for a sin-
gle donor–acceptor pair at a fixed distance is
Figure 1.15. Effects of polarized excitation and rotational diffusion on
the polarization or anisotropy of the emission.
R60
E! (1.12)
R60 " r6
tion times and are expected to yield lower anisotropies. The
anisotropies of larger proteins can also be low if they are Hence the extent of transfer depends on distance (r).
labeled with long-lifetime fluorophores. The essential point Fortunately, the Förster distances are comparable in size to
is that the rotational correlation times for most proteins are biological macromolecules: 30 to 60 Å. For this reason
comparable to typical fluorescence lifetimes. As a result, energy transfer has been used as a "spectroscopic ruler" for
measurements of fluorescence anisotropy will be sensitive measurements of distance between sites on proteins.20 The
to any factor that affects the rate of rotational diffusion. The value of R0 for energy transfer should not be confused with
rotational rates of fluorophores in cell membranes also the fundamental anisotropies (r0).
occur on the nanoscale timescale, and the anisotropy values The field of RET is large and complex. The theory is
are thus sensitive to membrane composition. For these rea- different for donors and acceptors that are covalently
sons, measurements of fluorescence polarization are wide- linked, free in solution, or contained in the restricted
ly used to study the interactions of biological macromole- geometries of membranes or DNA. Additionally, depending
cules. on donor lifetime, diffusion can increase the extent of ener-
gy transfer beyond that predicted by eq. 1.12.
1.6. RESONANCE ENERGY TRANSFER
Another important process that occurs in the excited state is
resonance energy transfer (RET). This process occurs
whenever the emission spectrum of a fluorophore, called
the donor, overlaps with the absorption spectrum of anoth-
er molecule, called the acceptor.19 Such overlap is illustrat-
ed in Figure 1.16. The acceptor does not need to be fluores-
cent. It is important to understand that RET does not
involve emission of light by the donor. RET is not the result
of emission from the donor being absorbed by the acceptor.
Such reabsorption processes are dependent on the overall
concentration of the acceptor, and on non-molecular factors
such as sample size, and thus are of less interest. There is
no intermediate photon in RET. The donor and acceptor
are coupled by a dipole–dipole interaction. For these rea-
sons the term RET is preferred over the term fluores-
cence resonance energy transfer (FRET), which is also in Figure 1.16. Spectral overlap for fluorescence resonance energy
common use. transfer (RET).
14 INTRODUCTION TO FLUORESCENCE

1.7. STEADY-STATE AND TIME-RESOLVED

FLUORESCENCE

Fluorescence measurements can be broadly classified into

two types of measurements: steady-state and time-resolved.
Steady-state measurements, the most common type, are
those performed with constant illumination and observa-
tion. The sample is illuminated with a continuous beam of
light, and the intensity or emission spectrum is recorded
(Figure 1.17). Because of the ns timescale of fluorescence,
most measurements are steady-state measurements. When
the sample is first exposed to light, steady state is reached
almost immediately. Figure 1.17. Comparison of steady-state and time-resolved fluores-
cence spectroscopy.
The second type of measurement is time-resolved,
which is used for measuring intensity decays or anisotropy
decays. For these measurements the sample is exposed to a Substitution of eqs. 1.13 and 1.14 into 1.15 yields the Per-
pulse of light, where the pulse width is typically shorter rin equation, 1.10.
than the decay time of the sample (Figure 1.17). This inten- Perhaps a simpler example is how the steady-state
sity decay is recorded with a high-speed detection system intensity (ISS) is related to the decay time. The steady-state
that permits the intensity or anisotropy to be measured on intensity is given by
the ns timescale. (
It is important to understand the relationship between
steady-state and time-resolved measurements. A steady-
ISS ! "I e0
$t/∏
dt ! I0∏ (1.16)
0
state observation is simply an average of the time-resolved
phenomena over the intensity decay of the sample. For The value of I0 can be considered to be a parameter that
instance, consider a fluorophore that displays a single decay depends on the fluorophore concentration and a number
time (∏) and a single rotational correlation time (θ). The of instrumental parameters. Hence, in molecular terms,
intensity and anisotropy decays are given by the steady-state intensity is proportional to the lifetime.
This makes sense in consideration of eqs. 1.1 and 1.2,
which showed that the quantum yield is proportional to
I(t) ! I0 e $t/∏ (1.13)
the lifetime.

r(t) ! r0 e $t/θ (1.14)

1.7.1. Why Time-Resolved Measurements?
where I0 and r0 are the intensities and anisotropies at t = 0, While steady-state fluorescence measurements are simple,
immediately following the excitation pulse, respectively. nanosecond time-resolved measurements typically require
Equations 1.13 and 1.14 can be used to illustrate how complex and expensive instrumentation. Given the relation-
the decay time determines what can be observed using flu- ship between steady-state and time-resolved measurements,
orescence. The steady-state anisotropy (r) is given by the what is the value of these more complex measurements? It
average of r(t) weighted by I(t): turns out that much of the molecular information available
from fluorescence is lost during the time averaging process.
$0( r(t)I(t)dt For example, anisotropy decays of fluorescent macromole-
r! (1.15) cules are frequently more complex than a single exponen-
$0( I(t)dt
tial (eq. 1.14). The precise shape of the anisotropy decay
In this equation the denominator is present to normalize the contains information about the shape of the macromolecule
anisotropy to be independent of total intensity. In the and its flexibility. Unfortunately, this shape information is
numerator the anisotropy at any time t contributes to the lost during averaging of the anisotropy over the decay time
steady-state anisotropy according to the intensity at time t. (eq. 1.15). Irrespective of the form of r(t), eq. 1.15 yields a
PRINCIPLES OF FLUORESCENCE SPECTROSCOPY 15

single steady-state anisotropy. In principle, the value of r 4',6-diamidino-2-phenolindole (DAPI). There is a wide
still reflects the anisotropy decay and shape of the mole- variety of fluorophores that spontaneously bind to DNA.
cule. In practice, the information from r alone is not suffi- A great variety of other substances display significant
cient to reveal the form of r(t) or the shape of the molecule. fluorescence. Among biological molecules, one can observe
The intensity decays also contain information that is fluorescence from reduced nicotinamide adenine dinu-
lost during the averaging process. Frequently, macromole- cleotide (NADH), from oxidized flavins (FAD, the adenine
cules can exist in more than a single conformation, and the dinucleotide, and FMN, the mononucleotide), and pyri-
decay time of a bound probe may depend on conformation. doxyl phosphate, as well as from chlorophyll. Occasional-
The intensity decay could reveal two decay times, and thus ly, a species of interest is not fluorescent, or is not fluores-
the presence of more than one conformational state. The cent in a convenient region of the UV visible spectrum. A
steady-state intensity will only reveal an average intensity wide variety of extrinsic probes have been developed for
dependent on a weighted averaged of the two decay times. labeling the macromolecules in such cases. Two of the most
There are numerous additional reasons for measuring widely used probes, dansyl chloride DNS-Cl, which stands
time-resolved fluorescence. In the presence of energy trans- for 1-dimethylamino-5-naphthylsulfonyl chloride, and fluo-
fer, the intensity decays reveal how acceptors are distributed rescein isothiocyanate (FITC), are shown in Figure 1.19.
in space around the donors. Time-resolved measurements These probes react with the free amino groups of proteins,
reveal whether quenching is due to diffusion or to complex resulting in proteins that fluoresce at blue (DNS) or green
formation with the ground-state fluorophores. In fluores- (FITC) wavelengths.
cence, much of the molecular information content is avail- Proteins can also be labeled on free sulfhydryl groups
able only by time-resolved measurements. using maleimide reagents such as Bodipi 499/508
maleimide. It is frequently useful to use longer-wavelength
1.8. BIOCHEMICAL FLUOROPHORES probes such as the rhodamine dye Texas Red. Cyanine dyes
are frequently used for labeling nucleic acids, as shown for
Fluorophores are divided into two general classes—intrin-
the labeled nucleotide Cy3-4-dUTP. A useful fluorescent
sic and extrinsic. Intrinsic fluorophores are those that occur
probe is one that displays a high intensity, is stable during
naturally. Extrinsic fluorophores are those added to a sam-
continued illumination, and does not substantially perturb
ple that does not display the desired spectral properties. In
the biomolecule or process being studied.
proteins, the dominant fluorophore is the indole group of
tryptophan (Figure 1.18). Indole absorbs near 280 nm, and
emits near 340 nm. The emission spectrum of indole is 1.8.1. Fluorescent Indicators
highly sensitive to solvent polarity. The emission of indole Another class of fluorophores consists of the fluorescent
may be blue shifted if the group is buried within a native indicators. These are fluorophores whose spectral proper-
protein (N), and its emission may shift to longer wave- ties are sensitive to a substance of interest. One example is
lengths (red shift) when the protein is unfolded (U). Sodium Green, which is shown in Figure 1.20. This fluo-
Membranes typically do not display intrinsic fluores- rophore contains a central azacrown ether, which binds Na+.
cence. For this reason it is common to label membranes Upon binding of sodium the emission intensity of Sodium
with probes which spontaneously partition into the nonpo- Green increases, allowing the amount of Na+ to be deter-
lar side chain region of the membranes. One of the most mined. Fluorescent indicators are presently available for a
commonly used membrane probes is diphenylhexatriene variety of substances, including Ca2+, Mg2+, Na+, Cl–, and
(DPH). Because of its low solubility and quenched emis- O2, as well as pH. The applications of fluorescence to
sion in water, DPH emission is only seen from membrane- chemical sensing is described in Chapter 19.
bound DPH (Figure 1.18). Other lipid probes include fluo-
rophores attached to lipid or fatty acid chains.
While DNA contains nitrogenous bases that look like 1.9. MOLECULAR INFORMATION FROM
fluorophores, DNA is weakly or nonfluorescent. However, FLUORESCENCE
a wide variety of dyes bind spontaneously to DNA—such
1.9.1. Emission Spectra and the Stokes Shift
as acridines, ethidium bromide, and other planar cationic
species. For this reason staining of cells with dyes that bind The most dramatic aspect of fluorescence is its occurrence
to DNA is widely used to visualize and identify chromo- at wavelengths longer than those at which absorption
somes. One example of a commonly used DNA probe is occurs. These Stokes shifts, which are most dramatic for
16 INTRODUCTION TO FLUORESCENCE

Figure 1.18. Absorption and emission spectra of biomolecules. Top, tryptophan emission from proteins. Middle, spectra of extrinsic membrane probe
DPH. Bottom, spectra of the DAPI bound to DNA (– – –). DNA itself displays very weak emission. Reprinted with permission by Wiley-VCH, STM.
From [21].

polar fluorophores in polar solvents, are due to interactions 6-(p-toluidinyl)naphthalene-2-sulfonate (TNS), which dis-
between the fluorophore and its immediate environment. plays the favorable property of being very weakly fluores-
The indole group of tryptophan residues in proteins is one cent in water (Figure 1.21). The green emission of TNS in
such solvent-sensitive fluorophore, and the emission spec- the absence of protein is barely visible in the photographs.
tra of indole can reveal the location of tryptophan residues Weak fluorescence in water and strong fluorescence when
in proteins. The emission from an exposed surface residue bound to a biomolecule is a convenient property shared by
will occur at longer wavelengths than that from a trypto- other widely used probes, including many DNA stains. The
phan residue in the protein's interior. This phenomenon was protein apomyoglobin contains a hydrophobic pocket that
illustrated in the top part of Figure 1.18, which shows a shift binds the heme group. This pocket can also bind other non-
in the spectrum of a tryptophan residue upon unfolding of a polar molecules. Upon the addition of apomyoglobin to a
protein and the subsequent exposure of the tryptophan solution of TNS, there is a large increase in fluorescence
residue to the aqueous phase. Prior to unfolding, the residue intensity, as well as a shift of the emission spectrum to
is shielded from the solvent by the folded protein. shorter wavelengths. This increase in TNS fluorescence
A valuable property of many fluorophores is their sen- reflects the nonpolar character of the heme-binding site of
sitivity to the surrounding environment. The emission spec- apomyoglobin. TNS also binds to membranes (Figure
tra and intensities of extrinsic probes are often used to 1.21). The emission spectrum of TNS when bound to
determine a probe's location on a macromolecule. For model membranes of dimyristoyl-L-"-phosphatidylcholine
example, one of the widely used probes for such studies is (DMPC) is somewhat weaker and at longer wavelengths
PRINCIPLES OF FLUORESCENCE SPECTROSCOPY 17

Figure 1.20. Effects of sodium on the emission of Sodium Green.

From [22].

concept is illustrated in Figure 1.22, which shows the emis-

sion intensity of a protein- or membrane-bound fluorophore
in the presence of the water-soluble quencher iodide, I–. As
shown on the right-hand side of the figure, the emission
intensity of a tryptophan on the protein's surface (W2), or on
the surface of a cell membrane (P2, right), will be decreased
in the presence of a water-soluble quencher. The intensity
Figure 1.19. Fluorophores for covalent labeling of biomolecules.
of a buried tryptophan residue (W1) or of a probe in the
membrane interior (P1) will be less affected by the dis-
compared to that of apomyoglobin. This indicates that the solved iodide (left). The iodide Stern-Volmer quenching
TNS binding sites on the surface of the membrane are more constant K in eq. 1.6 will be larger for the exposed fluo-
polar. From the emission spectrum it appears that TNS rophores than for the buried fluorophores. Alternatively,
binds to the polar head group region of the membranes, one can add lipid-soluble quenchers, such as brominated
rather than to the nonpolar acyl side chain region. Hence, fatty acids, to study the interior acyl side chain region of
the emission spectra of solvent-sensitive fluorophores pro- membranes, by measurement from the extent of quenching
vide information on the location of the binding sites on the by the lipid-soluble quencher.
macromolecules.

1.9.3. Fluorescence Polarization or Anisotropy

1.9.2. Quenching of Fluorescence

As described in Section 1.4.1, a wide variety of small mol- As described in Section 1.5, fluorophores absorb light along
ecules or ions can act as quenchers of fluorescence, that is, a particular direction with respect to the molecular axes. For
they decrease the intensity of the emission. These sub- example, DPH only absorbs light polarized along its long
stances include iodide (I–), oxygen, and acrylamide. The axis (Figure 1.18). The extent to which a fluorophore
accessibility of fluorophores to such quenchers can be used rotates during the excited-state lifetime determines its
to determine the location of probes on macromolecules, or polarization or anisotropy. The phenomenon of fluores-
the porosity of proteins and membranes to quenchers. This cence polarization can be used to measure the apparent vol-
18 INTRODUCTION TO FLUORESCENCE

Figure 1.21. Emission spectra of TNS in water, bound to apomyoglo-

bin, and bound to lipid vesicles.

ume (or molecular weight) of proteins. This measurement is

possible because larger proteins rotate more slowly. Hence,
Figure 1.22. Accessibility of fluorophores to the quencher (Q).
if a protein binds to another protein, the rotational rate Reprinted with permission by Wiley-VCH, STM. From [21].
decreases, and the anisotropy(s) increases (Figure 1.23).
The rotational rate of a molecule is often described by its
rotational correlation time θ, which is related to chains. Hence, an increase in the amount of unsaturated
fatty acid is expected to decrease the anisotropy. The appar-
ent microviscosity of the membrane is determined by com-
)V
θ! (1.17) paring the polarization of the probe measured in the mem-
RT
brane with that observed in solutions of known viscosity.
where ) is the viscosity, V is the molecular volume, R is the Anisotropy measurements are widely used in biochem-
gas constant, and T is the temperature in EK. Suppose a pro- istry, and are even used for clinical immunoassays. One rea-
tein is labeled with DNS-Cl (Figure 1.23). If the protein son for this usage is the ease with which these absolute val-
associates with another protein, the volume increases and so ues can be measured and compared between laboratories.
does the rotational correlation time. This causes the
anisotropy to increase because of the relationship between 1.9.4. Resonance Energy Transfer
the steady-state anisotropy r to the rotational correlation
time θ (eq. 1.10). Resonance energy transfer (RET), sometimes called fluo-
Fluorescence polarization measurements have also rescence resonance energy transfer (FRET), provides an
been used to determine the apparent viscosity of the side opportunity to measure the distances between sites on
chain region (center) of membranes. Such measurements of macromolecules. Förster distances are typically in the range
microviscosity are typically performed using a hydrophobic of 15 to 60 Å, which is comparable to the diameter of many
probe like DPH (Figure 1.23), which partitions into the proteins and to the thickness of membranes. According to
membrane. The viscosity of membranes is known to eq. 1.12, the distance between a donor and acceptor can be
decrease in the presence of unsaturated fatty acid side calculated from the transfer efficiency.
PRINCIPLES OF FLUORESCENCE SPECTROSCOPY 19

The use of RET to measure protein association and dis-

tance is shown in Figure 1.24 for two monomers that asso-
ciate to form a dimer. Suppose one monomer contains a
tryptophan residue, and the other a dansyl group. The
Förster distance is determined by the spectral overlap of the
trp donor emission with the dansyl acceptor absorption.
Upon association RET will occur, which decreases the
intensity of the donor emission (Figure 1.24). The extent of
donor quenching can be used to calculate the donor-to-
acceptor distance in the dimer (eq. 1.12). It is also impor-
tant to notice that RET provides a method to measure pro-
tein association because it occurs whenever the donor and
acceptor are within the Förster distance.

1.10. BIOCHEMICAL EXAMPLES OF

BASIC PHENOMENA

The uses of fluorescence in biochemistry can be illustrated

by several simple examples. The emission spectra of pro-
teins are often sensitive to protein structure. This sensitivi-
ty is illustrated by melittin,23 which is a 26-amino-acid pep-
tide containing a single tryptophan residue. Depending Figure 1.23. Effect of protein association and membrane microviscos-
ity on the fluorescence anisotropy. From [21].
upon the solution conditions, melittin can exist as a
monomer or self-associate to form a tetramer. When this
occurs the tryptophan emission maximum shifts to shorter RET can also be used to study gene expression. One
wavelengths (Figure 1.25). These spectra show that a pro- approach is to use donor- and acceptor-labeled DNA
tein association reaction can be followed in dilute solution oligomers (Figure 1.28). If a complementary mRNA is
using simple measurements of the emission spectra. present, the oligomers can bind close to each other.26 The
Association reactions can also be followed by RET will result in increased emission intensity from the
anisotropy measurements. Figure 1.26 shows anisotropy acceptor. The lower panels show fluorescence microscopy
measurements of melittin upon addition of calmodulin.24 images of human dermal fibroblasts. The cells on the left
Calmodulin does not contain any tryptophan, so the total were not stimulated and did not produce the K-ras gene
tryptophan emission is only due to melittin. Upon addition product. The cells on the right were stimulated to cause
of calmodulin the melittin anisotropy increases about expression of the K-ras oncogene. Mutations in the gene are
twofold. This effect is due to slower rotational diffusion of associated with human colorectal cancer.27 The images
the melittin–calmodulin complex as compared to melittin show the emission intensity of the acceptor. Significant
alone. The emission spectra of melittin show that the tryp- acceptor intensity is only found in the stimulated cells
tophan residue becomes shielded from the solvent upon where the mRNA for K-ras is present. RET is now widely
binding to calmodulin. The tryptophan residue is probably used in cellular imaging to detect many types of gene
located at the interface between the two proteins. expression and the intracellular proximity of biomolecules.
Resonance energy transfer can also be used to study
DNA hybridization.25 Figure 1.27 shows complementary
DNA oligomers labeled with either fluorescein (Fl) or rho- 1.11. NEW FLUORESCENCE TECHNOLOGIES
damine (Rh). Hybridization results in increased RET from 1.11.1. Multiphoton Excitation
fluorescein to rhodamine, and decreased fluorescein inten-
sities (lower panel). Disassociation of the oligomers results During the past several years technological advances have
in less RET and increased fluorescein intensities. The provided new uses of fluorescence. These technologies
extent of hybridization can be followed by simple intensity have been quickly adopted and are becoming mainstream
measurements. methods. One of these technologies is two-photon or multi-
20 INTRODUCTION TO FLUORESCENCE

Figure 1.26. Effect of melittin–calmodulin association on the

anisotropy of melittin. Inset: Emission spectra of melittin in the pres-
Figure 1.24. Energy transfer between donor- (D) and acceptor- (A) ence and absence of calmodulin (CaM). Modified from [24].
labeled monomers, which associate to form a dimer. In this case the
donor is tryptophan and the acceptor DNS.
images are otherwise distorted from fluorescence from
above and below the focal plane. Figure 1.30 shows an
photon excitation and multiphoton microscopy.28–30 Fluo- example of the remarkably sharp images obtainable with
rescence is usually excited by absorption of a single photon multiphoton excitation. The cells were labeled with three
with a wavelength within the absorption band of the fluo- probes: DAPI for DNA, Patman for membranes, and
rophore. Pulse lasers with femtosecond pulse widths can tetramethylrhodamine for mitochondria. A single excitation
excite fluorophores by two-photon absorption. Such lasers wavelength of 780 nm was used. This wavelength does not
have become easy to use and available with microscopes. If excite any of the fluorophores by one-photon absorption,
the laser intensity is high, a fluorophore can simultaneous- but 780 nm is absorbed by all these fluorophores through a
ly absorb two long-wavelength photons to reach the first multiphoton process. The images are sharp because there is
singlet state (Figure 1.29). This process depends strongly no actual phase fluorescence that decreases the contrast in
on the light intensity and occurs only at the focal point of non-confocal fluorescence microscopy. Such images are
the laser beam. This can be seen in the photo, where emis- now being obtained in many laboratories.
sion is occurring only from a single spot within the sample.
Fluorophores outside the focal volume are not excited.
Localized excitation from two-photon excitation has 1.11.2. Fluorescence Correlation Spectroscopy
found widespread use in fluorescence microscopy. Multi-
Fluorescence correlation spectroscopy (FCS) has rapidly
photon excitations allow imaging from only the focal plane
become a widely used tool to study a wide range of associ-
of a microscope. This is an advantage because fluorescence
ation reactions.31 FCS is based on the temporal fluctuations
occurring in a small observed volume. Femtoliter volumes
can be obtained with localized multiphoton excitation or
using confocal optics (Figure 1.31, top). Bursts of photons
are seen as single fluorophores diffuse in and out of the
laser beam. The method is highly sensitive since only a few
fluorophores are observed at one time. In fact, FCS cannot
be performed if the solution is too concentrated and there
are many fluorophores in the observed volume. Less fluo-
rophores result in more fluctuations, and more fluorophores
result in smaller fluctuations and a more constant average
signal.
FCS is performed by observing the intensity fluctua-
tions with time (Figure 1.31, middle panel). The rate of
Figure 1.25. Emission spectra of melittin monomer and tetramer.
Excitation was at 295 nm. In the schematic structure, the tryptophans fluctuation depends on the rate of fluorophore diffusion.
are located in the center between the four helices. From [23]. The intensity increases and decreases more rapidly if the
PRINCIPLES OF FLUORESCENCE SPECTROSCOPY 21

number of fluorophores being observed. The position of the

curve on the time axis indicates the fluorophore's diffusion
coefficient.
Figure 1.32 shows how FCS can be used to study lig-
and bindings to a single neuronal cell. The ligand was an
Alexa-labeled benzodiazepine.32 Benzodiazepines are used
to treat anxiety and other disorders. FCS curves are shown
for Alexa-Bz in solution and when bound to its receptor.
The receptors were present on the membrane of a single
neuronal cell where the laser beam was focused. The FCS
curve clearly shows binding of Alexa-Bz to the receptor by
a 50-fold increase in the diffusion time. Such measurements
can be performed rapidly with high sensitivity. FCS will be
used increasingly in drug discovery and biotechnology, as
well as biochemistry and biophysics.

1.11.3. Single-Molecule Detection

Observations on single molecules represent the highest

Figure 1.27. DNA hybridization measured by RET between fluores-
cein (Fl) and rhodamine (Rh). The lower panel shows the fluorescein obtainable sensitivity. Single-molecule detection (SMD) is
donor intensities. Revised from [25]. now being performed in many laboratories.33–34 At present

Figure 1.28. Detection of mRNA with RET and molecular beacons.

The lower panels show acceptor intensity images of control cells and
stimulated fibroblasts expressing the K-ras gene. Revised from [26].

fluorophores diffuse rapidly. If the fluorophores diffuse

slowly they remain in the observed volume for a longer
period of time and the intensity fluctuations occur more
slowly. The amplitude and speed of the fluctuations are
used to calculate the correlation function (lower panel). The Figure 1.29. Jablonski diagram for two-photon excitation. The photo
height of the curve is inversely proportional to the average shows localized excitation at the focal point of the laser beam.
22 INTRODUCTION TO FLUORESCENCE

Figure 1.32. FCS of an Alexa-labeled benzodiazepine (Alexa-Bz) in

solution and on a single neuronal cell. Revised from [32].

Figure 1.30. Fluorescence image of RBL-3H3 cells stained with

DAPI (blue), Patman (green), and tetramethylrhodamine (red).
most single-molecule experiments are performed on immo-
Courtesy of Dr. W. Zipfel and Dr. W. Webb, Cornell University. bilized fluorophores, with fluorophores chosen for their
Reprinted with permission from [30]. high quantum yields and photostability (Chapter 23). A typ-
ical instrument for SMD consists of laser excitation through
microscope objective, a scanning stage to move the sample
and confocal optics to reject unwanted signals. SMD is now
being extended to include UV-absorbing fluorophores,
which was considered unlikely just a short time ago. The
probe 2,2'-dimethyl-p-quaterphenyl (DMQ) has an absorp-
tion maximum of 275 nm and an emission maximum of 350
nm. Figure 1.33 (left) shows intensity images of DMQ on a
quartz cover slip.35 The spots represent the individual DMQ
molecules, which can yield signals as high as 70,000 pho-
tons per second. The technology for SMD has advanced so
rapidly that the lifetimes of single molecules can also be
measured at the same time the intensity images are being
collected (right). The individual DMQ molecules all display
lifetimes near 1.1 ns.
Without the use of SMD, almost all experiments
observe a large number of molecules. These measurements
reveal the ensemble average of the measured properties.
When observing a single molecule there is no ensemble
averaging, allowing for the behavior of a single molecule to
be studied. Such an experiment is shown in Figure 1.34 for
a hairpin ribozyme labeled with a donor and acceptor.
Steady-state measurements on a solution of the labeled
ribozyme would yield the average amount of energy trans-
fer, but would not reveal the presence of subpopulations
showing different amounts of energy transfer. Single-mole-
cule experiments show that an individual ribozyme mole-
Figure 1.31. Fluorescence correlation spectroscopy. Top: observed cule fluctuates between conformations with lower or high-
volume shown as a shaded area. Middle: intensity fluctuations.
er amounts of energy transfer.36 These conformational
Bottom: correlation function.
changes are seen from simultaneous increases and decreas-
PRINCIPLES OF FLUORESCENCE SPECTROSCOPY 23

Figure 1.33. Single-molecule intensities and lifetimes of DMQ in a

quartz slide. Excitation at 266 nm. Figure courtesy of Dr. S. Seeger,
University of Zürich. Reprinted with permission from [35].

es in donor and acceptor emission. The single-molecule

experiments also reveal the rate of the conformational
changes. Such detailed information is not available from Figure 1.34. Single-molecule RET between a donor (green) and
acceptor (red) on a hairpin ribozyme. Revised from [36].
ensemble measurements.
At present single-molecule experiments are mostly
performed using cleaned surfaces to minimize background. tein (YFP). Upon activation Ras binds GTP (Figure 1.35).
However, SMD can be extended to intracellular molecules In this experiment activated Ras binds GTP labeled with a
if they are not diffusing too rapidly. One such experiment is fluorophore called Bodipy TR. Single molecules of YFP-
detection of single molecules of the oncogen product Ras Ras were imaged on a plasma membrane (lower left). Upon
within cells.37 Ras was labeled with yellow fluorescent pro- activation of Ras, fluorescent red spots appeared that were

Figure 1.35. Binding of Bodipy TR-labeled GTP to activated YFP-Ras. The lower panels show single-molecule images of YFP-Ras and Bodipy TR-
labeled GTP. The overlay of both images shows the location of activated Ras from the plasma membrane of KB cells. Bar = 5 µm. Revised from [37].