Lanthanide ions as luminescence probes


Material Information

Lanthanide ions as luminescence probes
Physical Description:
v, 113 leaves : ill. ; 28 cm.
Hirschy, Linda Moore, 1953-
Publication Date:


Subjects / Keywords:
Phosphorescence   ( lcsh )
Chemiluminescence   ( lcsh )
Rare earth metals   ( lcsh )
bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )


Thesis (Ph. D.)--University of Florida, 1983.
Includes bibliographical references (leaves 109-112).
Statement of Responsibility:
by Linda Moore Hirschy.
General Note:
General Note:

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 001029738
notis - AFB1840
oclc - 10031278
System ID:

Full Text








I would like to express my gratitude to those people

in my life who have made graduate school a memorable exper-

ience. First, I thank Dr. J.D. Winefordner, whose kindness

and support smoothed the rough spots and created a good

situation in which to learn. For this, I also acknowledge

the JDW group members. I have learned many lessens while

in this group, and I will never forget the experience.

I would also like to thank the other faculty members

at the University of Florida who contributed to the

learning experience. These include my committee members,

as well as many others with whom I had contact throughout

my graduate career.

Finally, and most importantly, I would like to thank

my friends and family. Without their support I certainly

would have floundered along the way. Special credit goes

to my husband, John, whose unbending faith and infinite

patience carried me through the catastrophes, both

perceived and real.



ACKNOWLEDGMENTS ..................................... ii

ABSTRACT ............................................ iv


1 INTRODUCTION ................................. 1

2 BACKGROUND AND THEORY ..................... 10

The Lanthanides ........................ 10
Chemical and Physical Properties ...... 10
Analytical Chemistry of the
Lanthanides ........................ 13
Energy Transfer ...................... 17
Tetracyclines .......................... 28

3 EXPERIMENTAL SECTION ....................... 31

Reagents ............................... 31
Apparatus .............................. 31
Methods ................................ 40

4 RESULTS AND DISCUSSION .................... 43

Initial Studies ........................ 43
Tetracycline Analysis ................... 46
Study of pH Dependence ................. 77
Applications ........................... 89
Other Compounds ........................ 99

5 CONCLUSIONS AND FUTURE WORK ............... 102

REFERENCES .......................................... 109

BIOGRAPHICAL SKETCH .............................. 113

Abstract of Dissertation Presented to
the Graduate Council of the University of Florida
in Partial Fulfillment of the Requirements
for the Degree of Doctor of Philosophy




April 1983

Chairman: James D. Winefordner
Major Department: Chemistry

This work is concerned with the development of a room

temperature, fluid solution phosphorescence (RTP) techni-

que. It involves the use of lanthanide ions (specifically

Eu3+) as luminescence probes for analytes that associate

with them in solution. Tetracycline (TC) and several of

its analogues are used as model analytes for the study.

In solution, the TCs chelate the Eu3+ ions. The

organic ligand absorbs a broad range of wavelengths of

visible light to achieve an excited singlet state. The

proximity of the heavy metal ion promotes intersystem

crossing to the lowest triplet. The triplet state

energy is transferred to the ion, which luminesces in a

characteristic narrow, line-like band.

This offers several advantages for analyte quantita-

tion. First, since the Eu3+ bands are so narrow, the

complex luminescence is relatively more intense than that

of the free organic ligand. The peak intensity of the

TC:Eu3+ luminescence is tenfold greater than that of the

free TC. Since the probe ion luminescence occurs at a

relatively long wavelength (617 nm for Eu3+), spectral

discrimination against the pernicious background of

biological samples is possible. The extended

lifetimes of lanthanide luminescence (20 us for TC:Eu3+)

provide a means of time resolution against scatter and

background fluorescence. Finally, lanthanide luminescence

is in no way diminished by the presence of 02, making it

more convenient than other RTP techniques.

Quantitation of TC and its analogues by this method

gives detection limits in the nanogram range. It is shown

that, with careful control of variables such as pH and

probe ion concentration analytical calibration curves that

are linear over 4 decades can be produced.

The TC:Eu3+ system offers some information about the

complex pH dependence of TC metal binding. Studies of

luminescence lifetimes and intensities as a function of

pH are presented.


Room temperature phosphorimetry is becoming increas-

ingly popular, primarily because it offers several advan-

tabes over other analytical techniques. First, it is

selective for those few compounds which are physically

capable of appreciable luminescence from the triplet state.

Second, since phosphorescence occurs in a relatively long

time frame, one is able to discriminate against shorter-

term interfering signals such as scatter and background

fluorescence. All luminescence techniques have the advan-

tage of being extremely sensitive and relatively cheap.

In order for phosphorescence to occur within a parti-

cular molecular system, that system must meet certain

requirements. It must be capable of absorbing energy to

attain an excited singlet state, undergoing intersystem

crossing to the triplet state, and then undergoing radia-

tional deexcitation from the triplet to the ground state in

preference to other deactivation pathways. These processes,

as well as several others that a molecule may undergo, are

outlined in Figure 1.

The first requirement is rather easily met. Most

organic molecules contain some functional group or unsatur-

ation which allows for excitation via an n-r* or 7T-*



1 0
0 Ln 7

- I I I
4_>0 0 0 0 0

S4r-- -

0 0

4-) C3 E .

0 m
) 41I-4

0 U 01
"i (L II i rl

oC 0
OO -

(M 4- -4 F

-4 CM U 0
0iO O 0

m u u 4-J
0-1 10 1 C) 1)
.14 r-i o
4- 4c-J 5. 0
0 C c0 Q) 4-
'1 0 C 4-O 0 1- fU
(1) 4Jl (f) C U 0
344 4d 4- c II 0

"0 u II II t &,
OX e- U




0 r
o >



transition. Intersystem crossing (ISC) from the excited

singlet to the corresponding triplet state is not so easily

achieved, since it requires a highly forbidden spin flip.

The probability of this process occurring is greatly

enhanced by the addition of heavy atoms or paramagnetic

species, which cause mixing of the multiplicities of

states, and, thus, promote ISC (1).

Once the triplet is formed, deactivation may occur

through a number of pathways. The presence of paramag-

netic species,such as 02, may efficiently quench the trip-

let through charge transfer complexes as well as promote

ISC initially. Internal conversion (IC) and vibrational

relaxation (VR) also deactivate the excited triplet

molecule radiationlessly. For these reasons, analytical

applications of phosphorescence have, until recently, been

limited to rigid systems (2-4), such as low temperature

glasses or solid substrates (5-10). These rigid systems

not only limit the internal motions of the molecules, but

they also reduce the diffusion rate of triplet state


Until very recently, it has been fairly well accepted

that in liquid solutions at room temperature, phosphores-

cence intensities are too low to be analytically useful.

However, in 1980 Cline-Love and coworkers (11) showed that

room temperature phosphorescence in micellar solutions

was a viable analytical technique. Analytes were dis-

solved in micellar solutions containing thallium ions.

The heavy atoms enhanced ISC, and the resulting triplet

state was protected from diffusional quenchers by the


Donkerbroek et al. (12) demonstrated that with

rigorous deoxygenation of room temperature solutions,

luminescence from the triplet states of some highly halo-

genated molecules may be intense enough to be analytically

useful. They were able to achieve detection limits of

5.4 x 109 M for 1,4 dibromonaphthalene in hexane. They

went on to show that 1,4 dibromonaphthalene makes a good

triplet state acceptor for such donors as benzophenone,

thus introducing sensitized room temperature phosphores-

cence (i.e., phosphorescence resulting from energy trans-

fer) as an analytical technique. By exciting within the

absorption spectrum of benzophenone and monitoring the

sensitized phosphorescence of 1,4 dibromonaphthalene, they

were able to achieve detection limits as low as 4 x 10- M

(for benzophenone). In a later publication, Donkerbroek

et al. (13) expanded their description of this new techni-

que which involved intermolecular triplet-triplet energy
transfer. They reported detection limits as low as 108 M

for a number of substituted biphenyls and benzophenone.

Since this is an indirect emission technique, the absorp-

tion properties will be characteristic of the analyte

energy donor, but the emission properties will be charac-

teristic of the energy acceptor. Since the acceptor

emission must be at a longer wavelength (lower energy)

than the analyte emission would have been, some spectral

discrimination against background is possible.

One major disadvantage of room temperature phosphores-

cence still exists with this technique. Since it involves

intermolecular energy transfer and is diffusionally con-

trolled, other diffusional quenching processes still

compete effectively. It requires such rigorous deoxygen-

ation that workers in this laboratory (14) had difficulty

repeating the work.

The original purpose of this study was to develop a

system that combined the requirements for a strong phos-

phorescence signal with the simplicity of a room

temperature solution technique. Metal complexes seemed

to offer several advantages: 1) organic ligands are held

rigidly to metal ions, especially if they are chelated

at more than one site (i.e., bi- or polydentate); 2)

the central metal atom can itself provide the pertur-

bations required to enhance intersystem crossing to the

triplet state. Further literature search revealed that

lanthanide metal ion complexes offer an additional advan-

tage: excited chelates transfer their energy to the metal

ion, which then luminesces in a narrow, intense band,

very similar to the ion line. Since the energy transfer

within these lanthanide chelates is intramolecular it

should be more efficient than the intermolecular techni-

que previously described by Donkerbroek et al. The trans-

fer is not limited by diffusion so diffusional quenching

processes should not compete.

For these reasons, the more specific purpose of this

study became the evaluation of lanthanide sensitized

luminescence as an analytical technique. Tetracycline

(TC) and some of its analogues were chosen for this study,

and their structures appear in Figure 2. These compounds

were chosen for several reasons. Their ability to chelate

metal ions is well documented (15,16) and they are widely

used as antibiotics (17), which makes them analytically

interesting drug molecules.

This study includes an evaluation of the system for

TC quantitation in water as a function of parameters such

as pH and Eu3+ (probe ion, or energy acceptor) concentra-

tion. Quantitation was also carried out in blood serum.

Some more detailed information about the lifetimes and

structures of the complex as a function of pH are also

included. Finally, attempts to extend the method to other

compounds are described.











- -

I: 0












The Lanthanides

Chemical and Physical Properties

The unique properties of the lanthanide elements are

directly attributable to their electron configuration (18,

19). The filling of atomic orbitals as predicted by the

aufbau principle is the basis for the construction of the

periodic table. Looking at the sixth row of elements, one

can see that the filling proceeds as predicted through

atomic number 57 lanthanumm), which has the configuration
2 1
[Xe]6s25d1. Following La, the energy levels of the 4f

orbitals fall below the 5d's, and the next 14 elements

(atomic numbers 58-71, that is cerium through lutetium)

involve the filling of the 4f orbitals. Since these orbi-

tals place the electrons nearer the center of the atom

than the already present 5s25p66s25d1 electrons, the

addition of the 4f electrons has little effect on the

chemistry of these elements. On the other hand, magnetic

and spectroscopic properties are highly affected.

One interesting result of this unique electron arrange-

ment is the lanthanide contraction. Going from left to

right across the periodic table, the size of the lanthanide

elements decreases slightly. The addition of electrons to

the shielded, inner 4f orbitals does not compensate for the

increase in nuclear charge, which acts to "contract" the


The chemistry of the lanthanides is quite similar to

that of the Group III elements. One would expect this

since the outermost, or valance, electrons are the 6s25d1

for all 14 elements. This means that they all have stable

+3 oxidation states, and that they are all very similar to

each other, presenting a considerable challenge to those

wishing to chemically separate them.

In general, they resemble Ca2+ in their coordination

chemistry. The lanthanide carbonates, phosphates, and

oxalates are all insoluble. Their tendency to form com-

plexes is slight when compared with the transition metals,

but they do complex with oxygen donor ligands such as

B-diketones. Commonly, 6-coordinated species are formed,

but some 8-coordinated complexes are known (18).

Although the 4f electrons are too strongly bound and

shielded to be involved in the chemistry of the lanthanide

elements, their presence is directly responsible for the

spectral and magnetic properties. Almost all of the

lanthanides have unpaired electrons in the 4f levels (in

Lu, 4f14, the 4f's are completely filled) even in the +3

oxidation state, making them all paramagnetic. There is

considerable spin-orbit coupling and ISC is enhanced in

organic ligands associated with these metal ions.

Trivalent lanthanide ions have sharp-line fluorescence

and absorption features, even when they are present in

condensed phases (20). The electronic transitions take

place within the 4f levels, and, since these are shielded

from outside chemical and electrical influences they remain

relatively sharp and unshifted. These sharp, line-like

spectroscopic bands have been useful in characterizing and

quantitating individual lanthanides, since the wavelengths

at which the bands occur are determined by 4f configuration,

and so the spectrum of each lanthanide element is unique.

Although the f-f transitions are not appreciably

broadened or shifted by environment, the intensities of

some of the bands are particularly sensitive to changes in

ligands. Intensity increases of up to an order of magni-

tude may be observed when the usual aqueous solvent cage

is displaced by other ligands. This change in ligands

causes a change in the dielectric environment experienced

by the ion, and increases the transition probability for

some of the spectroscopic bands. Jorgensen and Judd (21)

were the first to describe this phenomenon, which they

called "hypersensitivity."

Chelation of lanthanide ions by organic species

which are capable of light absorption has an even greater

effect on the signal intensity than just increasing the

transition probability of the hypersensitive transitions

within the ion. If the chelating agent has an emission

spectrum that overlaps with the excitation spectrum of the

ion, energy will be transferred from the excited chelate

to a 4-f level within the ion (22). The luminescence from

the chelate is the narrow, line-like emission of the ion.

(The theory of energy transfer will be treated in a later

section in greater detail).

Analytical Chemistry of the Lanthanides

The increased intensity of luminescence caused by

chelation, both due to hypersensitivity of the bands them-

selves, and due to sensitization by energy transfer from

the chelate, has been widely exploited to improve the sen-

sitivity for fluorometric detection of the lanthanides. A

number of publications have discussed chelation of lan-

thanide ions with B-diketones to achieve detection limits

as low as 30 ng/mL (for Eu3+) (23,24). Several groups

(25,26) showed that the addition of Tri-n-octylphosphine

oxide (TOPO) as a synergic agent to form mixed complexes,

improved detection limits (to as low as 0.4 ng/mL for

Eu3+). Taketasu and Sato (27) and Taketasu (28) showed

that mixed lanthanide TOPO, 3-diketonate complexes could

be dissolved in aqueous micellar solution further enhancing

the signal.

By employing laser excitation, Yamada and coworkers

(29) were able to achieve detection limits for europium

(as the tris(l,l,l-trifluoro-4(2-thienyl)-2,4-butanediono)

complex or TTA) as low as 2 fg/mL. The same group

pushed Eu3+ detection limits even lower by taking advan-

tage of the relatively long luminescence lifetime of the

Eu(TTA)3 complex (30). By time resolving the lanthanide

signal away from shorter lifetime interfering signals,

they were able to detect Eu3+ at 0.4 fg/mL.

It should be noted here that in addition to the excel-

lent sensitivity and detection limits obtainable by quanti-

tating fluorescent chelates, selectivity is also available.

Since the luminescence characteristics of these chelates

are so similar to those of the ions, themselves, the lines

are narrow and distinctive. Thus Tb3 signals (at 545 nm)

are easily resolved from Eu3+ signals (at 615 nm). The

reader should recall that due to the extremely similar

chemistry of all the lanthanides, selectivity is not

readily available through strictly chemical techniques (i.e.,


Wright (20) and Gustafson and Wright (31) have used

selective excitation .of probe ion luminescence (SEPIL) to

quantitate lanthanides that have been coprecipitated with

calcium fluoride. Laser excitation of the lanthanides

within the precipitate is used not only to quantitate these

ions (detection limit for Eu3+ is 0.4 pg/mL) but also to

quantitate non-luminescent ions that have associated with

them. In a recent Analytical Chemistry (32) other workers

have presented a paper describing the use of LaF3 as a

matrix for similar SEPIL work.

In one of his publications (20) Wright suggests that

the use of lanthanide probe ions might also be a viable

method for quantitating analyte species that would associ-

ate with them in solution. In fact, lanthanide ions are

already widely used as luminescent and paramagnetic probes

in biochemistry (33-39). Approximately one third of all

naturally occurring proteins contain bound metal ions.

Ca2+ is one of the most important ions biologically, but

since it is neither paramagnetic nor luminescent, qualita-

tive information regarding its binding sites is quite

difficult to obtain. Fortunately, lanthanide ions sub-

stitute effectively for Ca2+ ions in these sites, often

with little or no effect on their biological activity.

Rare earth ions have very similar ionic radii to Ca(II)

(.12 A for Ca(II), while the trivalent lanthanide radii
0 0
range from 1.18 A to 0.97 A (35)), and they exhibit very

similar coordination chemistry. As previously mentioned,

they both prefer oxygen donor ligands and have coordina-

tion numbers greater than six. Luminescence decay life-

times provide a direct measure of the number of water mole-

cules coordinated to the metal ion. Studies of the energy

transfer between ions can offer information about the

distances between binding sites, and sensitization of the

probe ion luminescence by specific nucleic acid moieties

within the protein may indicate the identity of the

chelating groups. Tb + and Eu3+ are, by far, the most

widely used lanthanide probe ions because they are the

brightest emitters of the lanthanide series.

Probably the most widely used analytical application

of lanthanide ions is in nuclear magnetic resonance (NMR)

spectroscopy (40). Presently, a number of organic lantha-

nide chelates, soluble in the usual NMR solvents (CC14 or

CDC13) are available for use as chemical shift reagents.

The coordinately unsaturated chelates expand their coordi-

nation numbers to accommodate nucleophilic analytes in

solution. Once the analytes are associated their magnetic

environment is affected by the lanthanide ion causing a

shift in the NMR signal and often providing a means of

resolving overlapping signals. Recently, Bryden and'

Reilley (41) studied some water soluble lanthanide chelates

in the hope of extending this technique to aqueous NMR


The new analytical technique described in this disserta-

tion involves the principles of several techniques described

above. Here lanthanide ions are used as luminescence

probes for the quantitation of analytes that associate

with them in solution. Scheme 1 outlines the steps in-

volved in this technique.

Scheme 1:

1. A + Ln3+ --- A-Ln3+ Chelation

2. ALn3 h A*-Ln3+ Excitation

3. A*-Ln 3+ A-Ln3+* Energy Transfer
3. A -Ln --+ A-Ln Energy Transfer

--- A-Ln3+ + hv2

4. A-Ln3+*


where A = analyte

Ln3 = some lanthanide ion

hvi = excitation energy

hv2 f(hvl)= emission energy

The probe ion (Ln3+) is added to a solution already

containing the analyte, which chelates the ion. Next, the

analyte-ligand absorbs energy in a manner that is charact-

eristic of the ligand (i.e., it has a broad-banded visible

absorption spectrum, slightly red-shifted from that of the

free ligand). The energy is then transferred to the ion,

which luminesces in a manner characteristic of it (i.e.,

it has a narrow, line-like spectrum and the wavelength is

virtually unshifted from that expected for the particular


Energy Transfer

Weissman (22) was the first to show that energy trans-

fer from an organic ligand to a lanthanide ion occurred.

He studied europium salicylaldehyde crystals and noticed

that they had an intense broad absorption spectrum charact-

eristic of the organic ligand, and that the spectrum

remained unchanged when lanthanum or gadolinium ions were

substituted for the europium ion. He observed also that

the emission spectra of these complexes consisted of the

narrow line-like bands characteristic of the electronic

transitions within the 4f system of the ions.

Crosby et al. (42), and Whan and Crosby (43) estab-

lished the path of the energy migration within lanthanide

complexes. In their experiments, they prepared a number

of chelates of different ions within the lanthanide series.

A comparison of the triplet state energy levels for the

chelates with the lanthanide ion resonance levels revealed

that the requirement for line emission from a chelate is

that "the lowest triplet state energy of the complex must

be nearly equal to or must lie above the resonance energy

level of the rare earth ion." They concluded that 1)

excitation of the ions occurs indirectly via intramolecular

energy transfer; 2) energy transfer from the excited

singlet state of the complex does not occur; and 3) the

pathway for energy migration of the ion is via the lowest

triplet of the complex (or another excited state within

500 cm-1). This means that no change in multiplicity

occurs on energy transfer.

The energy transfer mechanism described above is

depicted in Figure 3. The chelated organic ligand is

excited to its first singlet state. Intersystem crossing

to the triplet must then occur. Although this is a spin

forbidden process, the relatively low quantum yield of

fluorescence in these chelates indicates that intersystem

crossing competes favorably with fluorescence. This is

probably due to the heavy atom effect of the closely

associated lanthanide ion.







n In

0 -------


-5, ---------



After intersystem crossing to a triplet state, the

molecule may undergo radiational deactivation (phosphoresce)

or it may transfer its energy to a low-lying state within

the metal ion. Since phosphorescence is a slow process,

energy transfer to the lanthanide competes favorably, so

that phosphorescence is effectively quenched. If the

energy is transferred to a resonance level within the 4f

system of the ion, the characteristic line-like emission


For those ions with closely spaced levels below

the energy-accepting level, radiationless deactivation

pathways may be fairly efficient and, therefore, the

quantum efficiency for luminescence is fairly low. For

both Tb3+ and Eu3+, the competing radiationless pathways

are few, making them the brightest emitters and thus the

best choices for probe ions.

Two different mechanisms for radiationless energy

transfer are currently accepted (44). These include a

coulombic or through space mechanism, described by Forster

(45) as well as an electron exchange mechanism described

by Dexter (46).

Forster's mechanism is described as a dipole-dipole

coupling of the energy donor (D) and acceptor (A). A

resonance interaction occurs via an electromagnetic field

set up by the oscillating dipoles of the donor and acceptor,

so actual contact of D and A is not required.

The rate of Forster energy transfer can be described

by the following equation

K2 kD
kET = K 6 JA) ()

where ko = rate of deactivation of donor in absence
of acceptor;

kET = rate constant for energy transfer;

K2 = orientation factor;

R = actual distance between the donor and


K = a constant determined by refractive index

and concentration;

J(OA) = spectral overlap integral; extinction
coefficient is included in the integration.

The energy transfer efficiency is described by the

following relationship (33)


OET = [1 + (R /Ro)6] (2)

where OET = efficiency of energy transfer;

R = "critical separation" for which energy

transfer efficiency is 50% (kET=kD);

R = actual separation between the donor and


This is directly related to some experimental observables

where I = intensity of donor luminescence in absence

of acceptor;

I = intensity of donor luminescence with

acceptor present;

T = natural radiative lifetime of donor with

acceptor absent;

T = lifetime of donor with acceptor present.

There are several important points to glean from

these equations. First, one can estimate the energy trans-

fer efficiency for a Forster system by knowing the change

in luminescent lifetime or intensity of the donor upon

addition of the acceptor. This transfer efficiency can

be used to determine R, the distance between D and A.

The Dexter, or electron exchange mechanism might be

thought of as a double electron substitution reaction,

such as the one described in Figure 4. A ground state

(HO, or highest occupied orbital) electron on the acceptor

is exchanged for the excited (LU, or lowest unoccupied

orbital) electron on the donor. This occurs when the

electron clouds overlap and so this mechanism requires

that D and A make physical contact.

The rate of energy transfer via the electron exchange

mechanism can be described by the following equation

-l 0


4-1 D

o z
0 M

O) r

0 40

U) 0

S0 -4

0 Q

0) 3 -4
o 80


C) a) -4

X 0
CO ao

(U 0




I I0

O r
'- ^ r*



kET = K J exp (-2R/L) (4)

where K = a value related to the specific coulombic

interactions between overlapping orbitals;

J = spectral overlap integral;

R = distance between D and A;

L = the van der Waal's radii.

The efficiency of energy transfer by this mechanism

cannot be related to an experimentally observed quantity


There are some interesting comparisons that can be

made between these two mechanisms. Whereas, the rate of
dipole-dipole energy transfer drops off as R6, the rate

of electron exchange drops off exponentially (as

exp(-2R/L)). This means that for R > 5-10 A, the electron

exchange mechanism is inoperable. While the rate for

Forster energy transfer is dependent on the oscillator

strengths of both the donor and acceptor, Dexter's electron

exchange mechanism is independent of both parameters.

Turro (44), Crosby et al. (42), and Bhaumik and

El-Sayed (47) suggested that triplet-triplet energy trans-

fer is forbidden by the Forster dipole coupling mechanism,

since both transitions involved are forbidden (D3 -D1

and A1-A3 ), and the energy transfer is dependent on the

oscillator strengths of both. The above named authors

all concluded that for triplet-triplet energy transfer

such as that occurring in lanthanide chelates, the mechan-

ism must be electron exchange.

Horrocks and Sudnick (33), on the other hand, have

presented experimental evidence that the energy transfer
from tryptophan moieties in proteins to Tb ions is a

Forster type transfer. Calculations based on Forster

theory to determine R values in proteins from experi-

mentally observed energy transfer efficiencies agreed

remarkably well with X-ray diffraction data.

Usual experimental methods for distinguishing the

two mechanisms are based on the fact that electron exchange

requires a collision and is, therefore, diffusion controlled.

Since the Forster dipole-dipole mechanism does not require

a collision, the energy transfer is not limited by diffu-

sion. The mechanism for intermolecular energy transfer

systems is easily assigned simply by observing whether

or not the transfer is diffusion controlled. Unfortunately

for intramolecular energy transfer systems such as the one

in this study, there is no way to experimentally determine

the mechanism. However, since Horrocks has presented

some experimental evidence that a very similar system

(Tb3+-protein) undergoes energy transfer via the Forster

mechanism, this author will treat the model system as a

Forster system.

In either case, it's obvious that the best possible

situation for energy transfer requires keeping the D-A

separation to a minimum. For this reason, we chose a

model compound that would chelate the lanthanide ion.


As mentioned above, tetracycline (TC) was chosen as

a model compound primarily because of its well-documented

(15,16) ability to chelate metal ions. Also, since

tetracycline antibiotics are widely used in human and

veterinary medicine (17), they are analytically interest-

ing molecules.

Tetracyclines have been used for the treatment of

a large variety of bacterial infections for over 30

years (48). These compounds inhibit the growth of micro-

organisms by blocking protein synthesis at the ribosomes.

Their metal chelating ability is somewhat of a

problem pharmacologically. The serum level of TC anti-

biotics may be reduced by 90% if they are administered

simultaneously with ferric sulfate (49). TCs are

incorporated into tissues rich in Ca2+ ions, such as the

growing bones and teeth of embryos and newborns (17). For

this reason, TC's are not usually administered to pregnant

women or children under 8 years of age.

A lot of research has been devoted to the study of

the coordination chemistry of the tetracyclines (15-17,

48-58) and binding constants for many different types

of metals have been calculated (15,16,48,50). This aspect

of TC chemistry is quite complex because of the existence

of 3 or more binding sites on each molecule, and because

the affinity of these sites for metal ions changes markedly

with pH.

As one can see by referring again to Figure 2,

tetracyclines are all derived from a system of four six-

membered rings arranged linearly with characteristic

double bonds. This conjugated system makes up 2 distinct

chromophores, separated by an sp3 carbon. The A ring

system represents one chromophore while the BCD system

forms the other.

Four proton dissociations have been observed for the

tetracyclines, with pKa's at 3.3, 7.7, 9.7, and 10.7 (48).

The first ionization is from the A ring. Most studies

agree that below pH 3, and before the first ionization

(loss of the enolic H on the A ring), no complexation occurs.

Complexation below pH 3 has been observed only with

lanthanide species (52). The second and fourth pKa's

arise from the BCD enolic groups. Many studies have shown,

as one would expect, that after the second pKa, metal

binding by the BCD B-diketone system becomes important

(52,54,55). The third pKa at 9.7 represents deprotonation

of the A-ring dimethylamino function. This may play some

role in chelation at higher pH.

Because tetracycline metal complexes are highly

fluorescent, Kohn developed a fluorometric detection method

for TCs in biological fluids (59). This method involved the

extraction and quantitation of the highly fluorescent

Ca2+:TC complex. This is still the most widely used

method for clinical quantitation of TCs.

While the Ca2+ complexes used in Kohn's method are

fluorescent, they still have the disadvantage of most

organic chelates: they have broad-banded spectra. While

other authors have improved on Kohn's method (60,61),

none have utilized the unique advantage offered by

lanthanide chelates. That is, the normally broad-banded

emission of the organic is transferred into a narrow,

more intense luminescence band, providing improved sensi-


As one would expect, the luminescence characteristics

of TCs and their metal complexes are highly pH dependent

(51). This is in part due to the variation in complex

stability with pH, but also due to the change in lumines-

cence characteristics of the chromophores, themselves,

upon deprotonation. A good deal of time in this study

was devoted to the analysis of the pH dependence of the

TC:Eu complex.



Reagents used in this study wer all of spectroscopic

or reagent grade. The compounds and their sources are

listed in Table 1. The serum studies were done using

quality assurance serum (level 1), obtained from the

General Diagnostics Division of Warner-Lambert Company,

Morris Plains, NJ. The S&S 903 filter paper used in the

room temperature phosphorescence studies was obtained from

Schleicher and Schuell, Keene, NH.


Absorbance studies were carried out using a Varian

(Palo Alto, CA) 634S double beam spectrophotometer. All

fluorescence measurements were made using a laboratory-

constructed fluorimeter, which is schematically represented

in Figure 5. The components and their sources are described

in Table 2. This is a moderate resolution instrument

with a continuum source. Detection limits were con-

firmed using a Perkin-Elmer (Oak Brook, IL) LS-5 spectro-

fluorimeter in the fluorescence mode. Lifetimes were

measured by another laboratory-constructed fluorimeter

whose components are listed in Table 3, and schematically

represented in Figure 6. The lifetime system has a pulsed

Table 1.


Chemicals Used in This Study
and Their Sources


Europium nitrate Alfa Products
Europium oxide Thiokol/Ventron Division
Terbium nitrate Danvers, MA

Phenobarbital Applied Science
State College, PA

p-Amino benzoic acid Eastman Kodak Company
Carbon tetrachloride Rochester, NY

Acetic acid Fisher Scientific
Nitric acid Fair Lawn, NJ
Sodium hydroxide
Tartaric acid
methane (THAM)

Diazepam Hoffman-LaRoche, Inc.
Nutley, NJ

Water House-distilled
(at Chemistry Dept., UF)

Ammonium chloride Mallinckrodt, Inc.
Ascorbic acid St. Louis, MO
Disodium hydrogen phosphate
Potassium citrate
Sodium acetate
Sodium sulfite

Europium diprivalomethane Merck and Co., Inc.
Rahway, NJ

Barbituric acid Nutritional Biochemicals
Cleveland, OH

Table 1. continued


Benzoic acid
Cortisone acetate
Nicotinic acid
Penicillinic acid


Sigma Chemical Co.
St. Louis, MO

US Industrial Chemicals Co.
New York, NY









m O



= 0)






2 I

Table 2. Experimental Equipment Used for
Conventional Fluorimeter in Figure 5

Item Model Source

Eimac Xenon Arc Lamp

Eimac Illimunator Power

Excitation Monochromator

Monochromator Scan


PS 300-1



Sample Housing

Emission Monochromator

Photomultiplier Module

Photomultiplier Tube







Eimac, Division
of Varian
San Carlos, CA

Eimac, Division
of Varian
San Carlos, CA

Heath Company
Benton Harbor,

Heath Company
Benton Harbor,

American Instru-
ment, Co.
Silver Springs,

Acton, MA

Heath Company
Benton Harbor,

Hamamatsu Corp.
Middlesex, NJ


Houston Instru-
Austin, TX









+- rn


Table 3. Instrument Components for the Pulsed
System Depicted in Figure 6

Item Model Source

N2 laser


Research Associates
London, Ontario,


Sample housing




Current to
voltage amplifier

Boxcar average



San Diego, CA


JY H-10

Jobin Yvon
American ISA, Inc.
Metuchen, NJ






Hamamatsu Corp.
Middlesex, NJ

Princeton Applied
Princeton, NJ

Princeton Applied
Princeton, NJ

Houston Instrument
Austin, TX

laser source and gated detection system. The filter

paper room temperature study was carried out using an

Aminco Bowman spectrofluorimeter (American Instrument, Co.,

Inc., Silver Spring, MD) equipped with a rotating can

phosphoroscope. A Perkin-Elmer IV pH meter was used to

measure pH.


Absorbance and fluorescence studies were carried out

in Tris buffer solution at pH 7.5. The buffer was prepared

by dissolving 1.21 g of tris(hydroxymethyl)aminomethane

(THAM) in 1 L of distilled water and adjusting the pH to

7.5 with HNO3. Other buffers that were tested (but found

inferior to the Tris) included phosphate buffer (pH 6.5,

prepared with KH2PO4 and NaOH), acetate buffer (pH 4.9,

prepared with NaOAc and HOAC), ammonia buffer (pH 8.9,

prepared with NH4NO3 and NaOH), and bisulfite buffer (pH-

7.1 prepared from Na2SO3 and HNO3).

Complexometric titrations were carried out using

50 mL of buffered 6 x 10-5 M TC and titrating with 1 x 10-2

M Eu3+ in a 100 mL beaker. Aliquots of %4 mL were removed,

placed in a cuvette, and monitored by absorbance or fluores-

cence (in separate experiments) after each addition of

titrant. The aliquot was then returned to the beaker,

so that no substantial loss of reagents occurred.

The pH titrations were carried out in unbuffered,

distilled water. A solution of 1:1 (TC:Eu, 6 x 104 M)
was titrated with 1.2 x 10 M NaOH solution and aliquots

were periodically analyzed for fluorescence and pH. No

attempt was made to control ionic strength.

Detection limit studies were measured by making

serial dilutions of 10-3 M TC stock solution with the 102

M buffer containing 10- M Eu3+. These were measured by

the laboratory fluorimeter as well as the LS-5. Detection

limits were calculated by taking 3 times the RMS blank

(Eu3+ in buffer) signal and dividing by the slope of the

calibration curve.

The luminescence lifetimes of the complexes were

determined for several points during both complexometric

and pH titrations carried out as described above. This

time, however, the aliquots in quartz cuvettes, were placed

in the pulsed laser fluorimeter. While the sample was

illuminated with the laser, the boxcar gate delay was

scanned from 0 to 100 is at a rate of 10 us/minute. The

resulting chart recorder output, showing the exponential

decay of the signal was used to calculate the lifetimes

(by hand).

Serum studies were performed by first scanning the

serum background on the conventional laboratory-constructed

fluorimeter. The serum was reconstituted as per manu-

facturers instructions. An aliquot of 100 pL of 10-2 M
facturers instructions. An aliquot of I00 L of i0 M

Eu3+ was added to 10 mL of the serum and another scan
was carried out. After the addition of 17 pL of 10 M

tetracycline, another scan was made. Detection limits

were determined in a similar manner (continuous addition

of small aliquots of TC to a Eu3+ serum solution), and

calculated as described above.

The solid substrate room temperature phosphorescence

study was conducted using the Aminco rotating can phos-

phorimeter. Scans of the paper (S&S 903) background, paper

and H20 background, paper with 2 pL of 102 M Eu3+ added,

and the paper with Eu3+ and 2 pL of 3 x 10-4 M TC were


Finally, studies for the screening of other compounds

by lanthanide luminescence were performed by adding
approximately 5 mg of the compound to a 10- M solution

( 10 mL) of Eu3+ in tris buffer. The excitation mono-

chromator was scanned while the emission monochromator

was held at the emission maximum for the probe ion being

studied (545 nm for Tb3+ and 615 nm for Eu3+).


Initial Studies

The initial studies were performed because it was

felt that a better alternative for obtaining room temper-

ature phosphorescence was needed. As mentioned in the

introduction, one generally needs a rigid medium to

minimize the radiationless deexcitation pathways available

to analyte species both through vibration and through

quenching by species in solution. Another important

requirement is that there be a heavy atom present to

provide the necessary perturbations to induce intersystem

crossing in the analyte. The lanthanide chelates used in

NMR spectroscopy as chemical shift reagents seemed ideally

suited to our purpose. Because they have been designed

to interact with many different types of polar organic

molecules, it was felt that they might be as generally

useful in luminescence spectrometry as they are in NMR.

Much study has been devoted to understanding the inter-

action that occurs in solution between a lanthanide shift

reagent (LSR) and a nucleophilic analyte (62). The most

widely used chelates were designed to be soluble in

nonpolar organic solvents and to function as Lewis acids.

In solution, a complex is formed between the organic

nucleophile and the coordinately unsaturated LSR. Since

many biologically active compounds, such as drugs and

pesticides, contain nucleophilic groups, many of these

analytically interesting compounds should interact with

any LSR. It was hoped that the complexes formed in this

manner would provide rigidity, efficient intersystem

crossing, and possibly a means of sensitizing the lumines-

cence of the central lanthanide ion.

Initial attempts to observe phosphorescence in this

type of chelate system were made using the commercially

available LSR europium diprivalomethane, Eu(dpm)3. It was

observed that the complex itself luminesces at the usual

Eu3+ wavelengths;excitation maximum for the complex was

315 nm. The reader should recall from Chapter 2, that the

excitation spectrum should be characteristic of the

chelating agent, while the emission spectrum should be

that of the metal ion. This was what we observed for

Eu(dpm)3, itself: a broad excitation spectrum and narrow,

line-like emission spectrum.

Benzoic acid (BA), para-amino benzoic acid (PABA), and

aspirin (acetylsalicylic acid, ASA) were the analytes used

in these initial attempts. It was discovered that the

major effect of adding the LSR to a solution containing

one of these drugs was to partially quench their lumines-

cence. In all cases, luminescence was stronger in the

absense of the LSR. However, it was also observed, that an

excitation spectrum of aspirin could be obtained by scan-

ning the excitation monochromator and monitoring Eu3+

luminescence. Although the Eu3+ signal was much weaker

than the free-aspirin luminescence and so would not be

analytically useful, it indicated that energy transfer

to the LSR complex did occur.

There was another problem with this first approach

that made it rather unsuccessful. The LSRs that are

commercially available have limited solubility. They are

made to be used in chloroform and carbon tetrachloride

and are quite insoluble in other solvents. This made the

method very impractical because most interesting analytes

are soluble in biological fluids, i.e., aqueous systems.

In fact, the three organic acids used in this initial study,

never fully dissolved in the prepared solutions and so the

concentrations were unknown and detection limits could not

be determined. It was for this reason that we abandoned

the LSR experiments, and decided to pursue the develop-

ment of other methods in aqueous systems.

Since, at the present time,no water soluble LSRs

are commercially available (40), it was decided that the

next approach would be to use water soluble lanthanide

salts in the hope that the aqueous ions would associate
with nucleophilic analytes in solution. Since Eu is

the brightest emitter of the lanthanide series and since

it is most widely used as a probe ion, we chose to continue

our studies with it. Again, several organic acids were

chosen as "analyte" ligands. Citric acid, ascorbic acid,

ASA, and PABA were all combined with Eu(N03)3 in basic

aqueous solution. There were no signals due to energy

transfer observed, except a weak sensitization of Eu3+

luminescence by citric acid (for solutions greater than

about 102 M in concentration), and by ASA. Later, after

the methodology had been developed using tetracycline, the

detection limit for ASA by this method was found to be

0.4 ppm. This is slightly better than the detection limit

reported by Latz for ASA quantitation by low temperature

phosphorescence (63).

However, at this early stage in the work, no appre-

ciable sensitization of Eu3+ luminescence had been ob-

served. Any signals that were obtained only occurred

when both the probe ion and the analyte were present in

large quantities (>10-2 M). It was felt that in order to

give this method the best chance of succeeding, a strongly

chelating model compound was needed. Upon consultation

with Dr. Eric Dose, it was decided that further attempts

would be made using tetracycline.

Tetracycline Analysis

There was good reason to believe that the TC:Eu+ system

would have a fairly high binding constant, since the

literature abounds with information on TC's excellent

chelating capability. In fact, Table 4 lists the binding

constants for Eu:TCx complexes, determined at pH 5.50 by

Saiki and Lima (64). Each was determined by 3 different

methods, and one can see that the values agree fairly well.

As expected, tetracycline chelated the Eu3+ ions in

aqueous solution. The dramatic effect that the probe ions

have on tetracycline luminescence is illustrated in Figure

7. The dotted line represents the luminescence of a 10-

M solution of tetracycline. The solid line represents the

signal from the same solution after addition of an equal

concentration of Eu 3+. As one can see, the TC lumines-

cence is completely quenched, and only the bands due to

Eu3+ luminescence are present. These bands are compara-

tively quite narrow and intense.

If one considers this to be a Forster energy transfer

system (see Chapter 2), then the energy transfer efficiency

can be related to the change in intensity of the donor

luminescence (TC) upon addition of the energy acceptor

(Eu ) by Equation (3)

ET = 1 I/o (3)

From Figure 7, it's obvious that the intensity of

luminescence of the TC without the Eu3+ present (I ) is

much greater than the intensity of TC signal after Eu3+

is added (I). In fact, I approaches zero, showing that

the energy transfer efficiency, ET,must approach unity

for this system.

Table 4. Binding Constants for Eu(TC)x

Method log B1 log 62 log B3

1 3.53 6.38 9.91

2 3.35 6.70 10.05

3 3.97 5.93 9.90

Ave 3.62 6.34 9.95

Figure 7. Luminescence spectra of 10- M solution of free
TC (---) and the same solution after the
addition of 1 equivalent of Eu3+ (-). The
excitation wavelength for both spectra was 398

Tetracycline --

Eu3 complex

/ -

0 0 0 0
O 0 W 0
IT to to W


The relative quantum efficiency of the system can be

estimated by comparing the areas under the luminescence

curves of the two species. Table 5 outlines such an esti-

mation. If one considers the spectra in Figure 7 and

calculates the areas under these curves in terms of

frequency units rather than wavelength, one obtains a

reasonable estimate of the relative energy emitted in

each case. Table 5 shows that the sum of the areas for

the two TC:Eu peaks is very nearly equal to that of the

free TC. This indicates that the quantum efficiencies of

the two systems are approximately equal since, as later

data will show, their absorbances are comparable. Thus,

there are no additional radiationless energy losses

introduced by adding the probe ion. The energy that is

normally distributed over a broad range of emission wave-

lengths in the free TC is effectively channelled into the

narrow line emission of the Eu3+ ion. The obvious result

is a dramatic increase in signal intensity, and improved

sensitivity for luminescent detection.

In the words of H. Laitinen, each analytical method

has seven ages: conception, verification, (instrument)

development, maturity, applications, broad acceptance, and

senescence (65). It was our hope that once the method

had been conceived we would be able to verify and develop

it, and possibly find some applications before senescence

occurred. Since TC has such complex solution chemistry,


;e a)


C4 o(+




(U C: *
)> 0
C)C -

90 C



*- 1Jl U
g a)

0 0

a) ^^


"- /^

o 0
C0 oC

in -i

) 0
0 0




U .0
U .0


0 >

U -












it was important first to verify that reproducible results

could be obtained if most solution parameters affecting

the complex equilibrium could be controlled. Obviously

the one parameter that could not be controlled is the

analyte, TC, concentration, since it was hoped that the

technique would give reproducible results over a wide

range of concentrations.

Attempts to discover the stoichiometry of the complex

forming in aqueous solution by absorbance studies failed

at first. Erratic results were obtained probably because

not enough other parameters were controlled during these


It was discovered fairly early that pH had a huge

effect not only on the complex formation, but also on its

luminescence. Figure 8 illustrates this pH dependence.
The data were taken during a titration of 6 x 10 M

complex (no excess of either Eu3+ or TC) with 102 M NaOH.

It shows that there is a small signal at very low pH, a

minimum around pH 5, and the signal reaches a maximum at

around pH 9. Unfortunately, the solution turns cloudy

beyond pH 8, and so it was decided that the best pH for

the quantitation of TC would be around pH 6 pH 8. Since

the signal has such a large pH dependency, it was neces-

sary to use a buffer to control pH and to assure that all

solutions (standards and samples) would have the same pH.






"( 0





o o o o o o o o O o
_0 OO r- -o ) -wt0
aouaosajonij |aI

The pH of blood serum is 7.5, and since we were hoping

to use the method for serum analyses, this was another

reason to develop the technique for the central pH region.

The search for a buffer was made more difficult

because lanthanide phosphates, carbonates, and oxalates

are all insoluble. This was discovered the hard way, when

2 L of pH 7.5 phosphate buffer was used to dilute Eu(N03)3

that had been prepared from Eu203 and EuPO4 precipitated.

The solution was discarded.

Solutions were prepared with acetate buffer (pH 4.9),

ammonia buffer (pH 8.9), and bisulfite buffer (pH 7.1).

As expected, the bisulfite buffer solution of Eu:TC complex

gave the most intense luminescence signal, but with time

a precipitate was formed with this solution, too. Finally,

a pH 7.5 buffer was prepared using tris(hydroxymethyl)-

aminomethane, (abbreviated "Tris" or THAM).. This effec-

tively maintained solution pH at 7.5 even though it is the

lower limit of its buffer range (pKa = 8.1) and did not

cause any unwanted side reactions, such as precipitation,

so it was used throughout the study.

With the pH controlled we were able to perform

absorbance studies to verify the stoichiometry of the

complex. Figure 9 shows absorbance spectra made for (a)

free TC, and (b) fully completed Eu:TC. These spectra and

the others appearing in Figure 9 were made during the
-5 -M
titration of 6 x 10 M TC with a 1 x 10 M solution


+ .2
M 0



-.) C
-4 M

'- 0 c


H0 -

ri rU

S0 .- 0
u as

r 1 -0 -4
44-1 4-
C) O
U c

U *H oM

3- +1
U U i


08,0 09'0 0O'0







of Eu3+ in buffered aqueous solution. Curve a was made

before any europium was added. Curves were made with each

incremental addition of Eu3+, until curve b which repre-

sents a solution containing equimolar quantities of Eu3+

and TC.

Several important conclusions can be drawn from

looking at this figure. The bathochromic shift in the

absorption spectrum indicates that complexation does occur.

The presence of 2 isobestic points indicates that only 2

species are present in solution, i.e., free TC and com-

plexed TC. This led us to believe that only one complex

was forming under these conditions. Also, although the

absorption maximum shifts on complexation, the molar

absorptivity only changes about 12%. This is not great

enough to account for the large change in luminescence

characteristics observed in Figure 7, and it supports

the estimated relative quantum efficiencies from Table 5.

Since molar absorptivities for the free and completed TC

are similar, and since their total energy emission is

approximately equal, they must have nearly equal quantum


Figure 10 shows another similar absorbance experiment.

In this case, spectra were made of free TC (a) and of

solutions containing an excess of the probe ion. This

shows that, even for the last curve (d) in which a 20 fold

excess of Eu3+ is present, there is very little change





+ 11

0 *o0 0i C

O +







in the absorbance characteristics from when the total Eu

present equals TC. In fact, when the absorbance signal

at 400 nm is corrected for dilution and plotted against

the ratio of Eu:TC in solution (see Figure 11), the curve

clearly indicates that the stoichiometry of the complex

is 1:1.

Also in Figure 11, one can see that although excess

Eu3+ does not apparently affect absorbance characteristics

of the complex, it does have a large effect on complex

fluorescence. If more than one equivalent of Eu3 is

present, it causes a substantial depression in the signal.

There are several ways to interpret this signal

depression. One is to consider the Eu3+ ions to be totally

uncomplexed and free to act as collisional quenchers in

solution. This type of bimolecular quenching is diffusion

controlled, and can be described by the Stern-Volmer

relationship (44)

[Q=O] = 1 + kq T1[Q] (5)

where I[Q=] = intensity when no quencher is present;

I Q] = intensity with quencher cone = [Q];

k = quenching rate constant;

[Q] = quencher concentration;
T = luminescence lifetime of the analyte.



* -

cn 3


zt 0

0 -4

LH 4-J


u ct

(U (






O 4




x ,


0r( 0

7 I

a~uoqiosqf / -ua3saJofld la d

One can test the validity of this relationship
I[Q=O] If the quenching
for a system by plotting I[Q] vs [Q]. If the quenching
mechanism is a Stern-Volmer collisional quenching, this

plot should give a straight line. Figure 12 shows that

such a plot of the fluorescence data presented in Figure

11 is not straight line, indicating that some other

mechanism is probably involved.

Some authors (66,67) have suggested that a combined

static and dynamic (collisional) mechanism might be

indicated by a Stern-Volmer plot of [Q=] vs [Q]2 or [Q]3
We used an Apple computer with a curve fitting program

to attempt to fit the Figure 11 data to just such a higher

order relationship. Again, no reasonable results were


Further evidence suggesting that a dynamic or

collisional mechanism was not involved surfaced when we

measured the luminescence lifetimes of the complex in Tris

buffer as a function of excess [Eu3+]. This experiment

was an exact duplication of the one that produced the data

in Figure 11, except that a pulsed laser fluorimeter

equipped with a gated detection system was used so that

lifetimes as well as intensities could be measured. It

was discovered that, although the intensity decreased as

before, the lifetime of the complex did not change. Table

6 shows the ten lifetime values that were obtained for

[Eu3+] excesses from zero to four times the complex



(0 = [0] )I

Table 6. Eu:TC Complex Luminescence Lifetimes
as a Function of Excess Eu3+

Moles excess Eu3+ Peak Intensity
Moles 1:1 Complex Lifetime (ps) (Relative Scale)































mean lifetime 19.95 us
Std. deviation 1.62 us

concentration (10- M). All values were within 8% of

the mean value 19.59 ps (standard deviation = 1.61 us).

If the signal depression were caused by any type of

collisional quenching, the lifetime should have decreased

with increasing [Q], according to Equation (6) (44).

k + kq[Q]6)

where T = lifetime of complex;

k, = natural rate of deactivation

kq[Q] = rate of quenching.

The possibility that the excess Eu3+ ions will

absorb the 617 nm radiation of the complex was also eval-

uated. This was done by placing a 10-2 M solution of Eu3+

in a cuvette between the sample solution (1:1, TC:Eu)

and the emission monochromator. No effect on the signal

was noted, indicating that the signal depression in Figure

11 is not due to such a post filter effect.

The remaining possibility is that excess Eu3+ ions

chemically react with the complex in some way. It is pos-

sible that a second Eu3+ ion is bound by the TC molecule,

but its still unclear why this would cause a decrease in

luminescence. Looking again at Figure 10, one can see

that there is a very slight bathochromic shift with a

large excess of probe ion. Also the absorbance peak

at 388 nm is diminished. This supports the conclusion

that possibly a second, less luminescent complex is forming.

The large effect that excess probe ion has on the

luminescence of the Eu:TC complex is an important consid-

eration if this is to be an analytically useful technique.

In a real situation, the concentration of the TC will be

unknown, so an excess of the probe ion must be used.

Figure 13 shows that, as long as the probe ion concen-

tration is held constant, linear reproducible calibration

curves for TC quantitation can be obtained. Each curve

represents a different concentration of probe ion. In

each case, total concentration of Eu3+ was held constant
-3 -2
throughout the [TC] range. For both 103 M and 102 M

Eu3+ the curves become nonlinear at the high concentra-

tion range. Again, this may be due to formation of a

less luminescent complex with higher [Eu3+]. Obviously,

for the best range of linearity at higher TC concentrations,

the probe ion concentration should be minimized.

This is also true at the low end of the analytical

calibration curves. Higher [Eu3+] caused increased back-

ground, which made detection limits poorer. Only with

the probe ion at 10-4 M was the detection limit lowered

below 10-7 M.

Since 02 quenching plays an important role in

decreasing phosphorescence signals of organic in solution,

we felt it was important to test its effect on our system.

A solution of Eu:TC complex was deoxygenated for 15 min

by bubbling with N2 and no improvement in signal was noted.


Figure 13. Calibration curves for TC obtained by
sensitized Eu3+ luminescence. Each curve was
obtained using a constant [Eu3+], as shown.

[Eu 3[ ] = 10-4 M

Euj10=o-3 M
[Eu =10-2M

-5 -4 -3

log [TC]



The solution was then aerated for 15 minutes longer

and still the signal intensity was unchanged. We con-

cluded that there was no need to go to the trouble of

deoxygenating this system.

Once the optimum conditions for the analysis had been

determined, analytical calibration curves for TC and

several of its analogues, chlortetracycline (CTC) and

doxycycline (DC), could be determined. Table 7 shows the

detection limits for these compounds, determined in tris

buffer at pH 7.5 with [Eu3+] = 10-4 M. The analytical

calibration curves were obtained using the laboratory-

constructed fluorimeter with conventional source described

in Chapter 3, and confirmed using the Perkin-Elmer LS-5

in the fluorescence mode. Detection limits were calculated

by taking 3 times the RMS blank noise and dividing by

the slope of the calibration curve. Detection limits for

all three compounds were in the nanomolar range.

The long luminescence lifetime of the Eu:TC complex

(refer again to Table 6) offers a big advantage for improv-

ing detection limits. Since the lifetime is longer than

most fluorescence lifetimes (usually in the nanosecond

timeframe), shorter term interfering signals can be time

resolved away from unwanted background fluorescence or

scatter. This technique is illustrated in Figure 14. A

pulsed source is used with a gated detection system,

capable of being turned on only after faster interfering

Table 7. Limits of Detection for Tetracycline
and Analogues by Lanthanide Sensitized Luminescence
(Xemission = 615 nm for all compounds)

Compound Excitation (nm) LOD
M ng/mL

Tetracycline 398 2 x 10 1

Chlortetracycline 400 4 x 10 2

-9Doxycycline 385 7 x 10
Doxycycline 385 7 x 10 3


i tU
' HO


o u

(U 0
*4 0 C
C )+


U I )
*r- L
+- r

^ e **

U")r E-i




/ /
I -

-- -- N3


signals have decayed away. In Figure 14, the shorter

dashed line represents the lifetime of the pulsed flash-

lamp source in the Perkin-Elmer LS-5. Since scatter and

fluorescence have relatively short lifetimes, their signals

will fall off with the source pulse. The longer lumines-

cence signal (longer dashes) of the complex will decay

more slowly and can be detected after the other signals

are gone.

The gated system of the LS-5 did not improve the

detection limits for the TC compounds in aqueous solution

primarily because the limiting blank signal was also of

longer lifetime. It is expected, however, that in real

applications such as serum analysis, time resolution

technique will be useful. In such cases, the limiting

signals are often background fluorescence and scatter.

Study of pH Dependence

In the course of the literature search carried out

in an attempt to understand the complex pH dependence of

the Eu:TC complex luminescence, it became obvious that the

complex solution equilibria of TC had caused some confusion

for other workers. The purpose of this section of the

study was to attempt to elucidate the nature of the changes

that occur with pH for the Eu:TC system, and possibly

to clear up some of the confusion regarding TC metal



The major confusion in the literature arises from the

fact that TC has several sites that potentially might

chelate metal ions. (See Figure 2, again.) Some of these

binding sites become substantially more nucleophilic

after deprotonation, so that chelating capability might

be expected to vary with pH.

In the earliest studies, Albert (15,16) determined

binding constants for a number of transition metal ions.

He concluded that only the group having pKa%7 underwent

ionization on chelation. It is now known that pKa=7.7

is due to the 3-diketonesystem of the B-C rings. Doluisio

and Martin (57) thought that the group having pKa=7.7

was the dimethylamino group on the A ring, and so they

wrongly concluded that A-ring binding occurs. Papers by

Gulbis and Everett (54), and Gulbis, Everett, and Frank

(53) concluded that for Na in 50:50 DMSO:H20, no binding

occurred anywhere but in the A-ring. Circular dichroism

studies by Mitscher et al. (55) showed that at very low

pH no binding occurs, at pH 3-7 A-ring binding occurs,

and at above pH 7.5, binding is at the BC B-diketone. The

most reasonable explanation for this confusion is probably

that the binding site does change with pH, but there are

still some discrepencies. Durckheimer (17) agrees with

Mitscher et al. that no binding occurs below pH 3, while

Celotti and Fazakarl, (52) have shown through NMR studies

that TC complexes Gd3+ ions at pH 2.

The NMR study done by Celotti and Fazakarly is

probably the most relevant to this work, since it

involves Gd3+ ions which have chemistry very similar to

Eu3+ ions (see Chapter 2). Figure 15 depicts the struc-

tures of the two complexes that they proposed from their

NMR data. As one can see, at pH 2 the ion is bound by the

A-ring. There is no need for deprotonation to occur before

this group is available for chelation. After the second

proton is removed from the TC molecule, the binding site

shifts to the B-diketone of the BC rings, resulting in a

more planar complex.

In fact, our studies support the conclusions made by

Celottie and Fazakarly that several different complexes

form over the pH range. In Figure 16, the pH titration
curve of a 6 x 10 molar solution of Eu:TC complex is

shown. Also, plotted in Figure 16 are the relative lumines-

cence intensities of the complex during titration. From

the titration curve (solid line), one can see that the first

equivalence point is reached after 2.8 equivalents of

NaOH have been added. Prados et al. (68) have shown that

aqueous solutions of lanthanide ions require 2.8 equiva-

lents of base to titrate the hydrolysate. Equivalence

points are also evident after 3.80 and 4.80 equivalents.

Figure 17 presents some data taken with the help of

Frank Van Geel, a postdoctoral fellow in the group. It

shows the lifetimes of the various complexes as a

function of pH. In the central pH range,

Figure 15. Structures of the TC:Gd3+
and pH 6.0 (52).

complex at pH 2.0


H/ / NH2
OH N(CH3)2

pH 2.0


/ 0

pH 8.6









0 *H

(u o




IDubIS la|

Otil 0o1 01 011 001

08 OL 09 Og Ot O C 0 01




n0 N

O -c



. 0


6 8 L 9

0 L)





the luminescence lifetimes are the longest, and they

agree with the values shown in Table 6. These data were

obtained by running a titration identical to the one

from Figures 8 and 16, but this time using the pulse laser

fluorimeter with gated detector to measure lifetimes.

In looking at the luminescence decay curves for the

extremes (both high and low) of the pH range, it became

apparent that they were actually a convolution of two dif-

ferent curves. There were apparently two species of

vastly different lifetime in equilibrium with each other.

In the central region of the pH range, only the long life-

time component was present. At very low pH, only the

faster species was present.

A deconvolution of the decay curves enabled us to

determine the relative intensities of the short and long

lifetime species as a function of pH. These data are plot-

ted in Figure 18. It clearly shows that the dip in the

luminescence vs pH curve (Figures 8 and 16) is caused by

the equilibrium between the faster and slower luminescers.

A similar equilibrium was evident at the high end of the

pH range, but the curves were too difficult to deconvolute.

No luminescence from the free TC was observed at any pH.

From these data, we have concluded the following:

(1) at low pH (2-4), a complex is formed between the A-

ring amide and ketone oxygen atoms; (2) as deprotonation

of the A-ring proceeds, this complex becomes more








C- 0


(=! 3
0 o






(0 I


0 O w v N 0 Ct N
cN .

(STr) wu!la|!l

Figure 18. Relative intensities of the short (-) and
long (---) lifetime species as a function of
OH- added to a 6 x 10-5 M solution of 1:1
Eu:TC complex.


-*- lifetime = lIs
-*- lifetime = 20is




3 4
ml NaOH added





luminescent due to an increase in the molar absorbtivity

of the A-ring chromophore; this complex has a relatively

short luminescence lifetime; (3) as deprotonation of the

B-ring enol proceeds, the binding site shifts to the BC

6-diketone. This complex is more rigid, more highly

luminescent, and has a longer lifetime than the A-ring com-

plex. Signal intensity increases as more of the complex is

deprotonated; and (4) in the high pH range, when the

dimethylamino group of the A-ring is deprotonated

(pKa=9.7) the A-ring becomes a better chelating region and

the binding site shifts back to the A-ring. This produces

a complex of lower intensity and shorter lifetimes, again.

There is one other important factor that might explain

the change in lifetimes that occurs with pH. Water

molecules associated with lanthanide complexes in solution

are known to increase the rate of their radiationless

deactivation (33). Figure 19 shows how the energy levels

of Tb3+ and Eu3+ are coupled with the vibrational modes

of the 0-H and (0-D) bonds in water. Its quite possible

that the number of H20 molecules associated with the Eu:TC

complex changes with pH.


Eu3+ sensitized luminescence can be used to discrimi-

nate against unwanted intefering signals in two ways.

First, since Eu3+ luminesces at 617 nm, it is spectrally

Figure 19. Coupling of Eu3+ and Tb3+ energy levels with
O-H and O-D vibrational bands (33).


Europium (If)


v(O-H) v(O-D) Terbium (m)


- v=7

v=-6 ^~
v=4 --'

v= 5

v=3 -

v=2-- c

V = I-

v=O -



- v=O




15 -

10 -




5 -




removed from the fluorescence background of biological

samples. Second, since the signal has a relatively long

lifetime, it may be time resolved away from unwanted


An application of the spectral discrimination is

depicted in Figure 20. This shows some work done with a

colleague, Leigh Ann Files, in which TC detection limits

were determined by quantitating the Eu3+ luminescence in a

control (standard) serum solution. Figure 20 shows the huge

background fluorescence of the serum, itself, as well as

the background caused by the same serum after the probe

ion has been added. Its interesting to note that there

is no sensitization of Eu+ luminescence by the complex

serum matrix. This would not be true for Tb3, since its

luminescence is sensitized by tryptophan moieties within

proteins (33). Only after the TC has been added, do the
Eu lines appear. This figure clearly shows that the

617 lines is well removed from the serum background sig-

nal. Free TC has a peak luminescence at 515 nm, which

is a spectral region having the highest serum background.

Detection limits for TC in the serum matrix are

given in Table 8. The table includes detection limits

by other methods. The most sensitive method as far as

detection limits (1 ng) is a radioimmunoassay technique

developed by Faraj and Ali (69). However, this is an

extremely expensive and complicated technique, requiring

Figure 20. Fluorescence spectra of serum background,
serum background after Eu3+ (3 x 10-4 M) has
been added and the same serum after addition
of TC (6 x 10-6 M).


Serum and Eu3+

Serum Background


480 540 600 660
Wavelength (nm)



Table 8. Comparison of Limits of Detection
for Tetracycline by Lanthanide Luminescence
and Other Clinical Techniques

Method LOD (ng) Reference

Fluorescence of 150 59
Ca2+ complex

RIA 1 69

Lanthanide 150 this
Luminescence work

Fluorescence 2300 this
of TC (no complex) work