Volume
116. number 5
CHEMICAL
PHYSICS
LE-iTERS
17 May 1985
TIME EVOLUTION OF THE DECAY OF THE *Do LEVEL OF Eu3+ IN GLASS MATERIALS DOPED WITH SMALL SILVER PARTICLES 0-L.
hlALTA,
PA.
SANTA-CRUZ,
G-F_ DE SA and F. AUZEL’
Deparxantento de Quimica Fundamental e Deportamento de F~sica da U.F. PE, Ctdade Umversibin~ Rec?/e. PE- _XMOO_ Brazil Received 26 January 1985; in final form 7 March 1985
We study the effect of small silver particles, embedded in Eu 3+-doped materials, on the decay curve of the sD, level of EL?_ In contrast with the results obtained for Eu3’ complexes deposited on a rough silver surface, we find that its decay curve is exponential and its lifetime remains unchanged.
1. Introduction We have recently observed the enhancement of the fluorescence from Eu3*doped materials due to the presence of small silver particles embedded in the medium [ 1 J_In connection with the SERS (surfaceenhanced Raman scattering) phenomenon, we interpreted this experimental result by means of an electromagnetic model based on the effect of the localized silver plasmons on the medium. A similar effect was observed by Weitz et al. [Z] in the case of Eu3+ complexes deposited on a rough silver surface. The fluorescence enhancement was of the same order of magnitude as obtained in ref. 11J and the lifetime of the SD, level of the Eu3+ ion was reduced by three orders of magnitude with a nonexponential decay curve. From the point of view of the electromagnetic model, this result indicates that the localized plasmons in the silver islands affect the 4f +s 4f transitions through a distancedependent interaction with the 4f electrons. Nevertheless, apart from quenching effects, since the zeroth-order 4f states have the same parity, this interaction is restricted to the even terms of its multipole expansion, i.e. the quadrnpole term, the hexadecapole term, . . .. which shouId be important,_therefore. only for those 1 On leave of .ZXXZXX from CNET. Bagneux, France.
396
196 rue de Paris, 92220
Eu3+ ions very near the silver islands. The dipole term of this expansion, which becomes allowed due to the action of the crystal field terms, can be shown to contriinte less tl an l/l0 of the allowed quadrnpole term. In ref. [l ] , the influence of the small silver particles on the fluorescence decay time of the Eu3+ ions was not analysed. The importance of carrying out such an analysis, in the case of a volumetric distribution of silver particles, is emphasized in the present paper. The experimental results, which are in contrast with the results of Eu3+ complexes deposited on a rough silver surface [Z] , are rationalized in terms of the electromagnetic model_
2. Measurements The Ag-doped glass samples used in the present work were the same used to obtain the optimum fluorescence enhancement in ref_ [l]_ It is worth quoting this result here (fig_ 1) together with the corresponding plasmon absorption band which is peaked at 3 14 nm (fig_ 2)_ Scanning transmission electron microscope (STEM) measurements have shown silver particles with an average diameter of x40 Jkand a particle concentration of xl.O1 8 cm- 3. For the lifetime measurements, the excitation source was a xenon lamp (OSRAM, XBO; 450 w) 0 009-2614/85/$03.30 0 Elsevier Science Publishers B-V_ (North-Holland Physics Publishing Division)
Volume 116, number 5
17 May 1985
CHEMICAL PHYSICS LETTERS
(b) 570
590
610 W AVELENCTH
630 (nm
650
1
Fig. 1. Fluorescence spectrum of the Eu” ion in the presence (a), and in the absence (b), of silver particles_ The Ag concentration is 7.5 in weight percent.
----I
I : I
t0 pulsed by a MK/2 electronic chopper associated with an electronic frequency controller. The pulse was used to trigger a boxcar system consisting of (a) a scan delay generator (model 882) opening a measurement gate (width = 5 p) after an adjustable delay, and (b) a linear gate type detector (model 881) measu&g the signal. The output was displayed on a XY recorder (HP 7034A). The fluorescence output wavelength (612 nm) was selected by an interferential filter. Fig. 3 shows the decay curves of the SD, level in the absence (a), and in the presence (b), of silver
IC
5
IS
Fig. 3. Decay curves of the ‘De level of Eu3 -+ in the absence (a), and in the presence(b), of small silver particles. In both cases the decay after the pulse is exponential with a lifetime of =2 Ins
particles,
respectively.
In both
cases the decay
curve,
after the pulse, is exponential and the measured lifetime is a2 ms. The excitation wavelength was 314nm. By exciting at this wavelength, we may also obtain information about the lifetime of the absorbing level from which the 5Do level is populated.
3. Discussion and conclusions
280
I
Fig. 2. Absorption spectrum of amaIl silver particles embedded in a Eu3 *-doped glassmateriaL
In contrast with the results obtained in the case of Eu3+ complexes deposited on a rough silver surface [2], in the present case of a volumetric distribution of silver particles, our lifetime measurements show that no distanc&ependent interaction between the localized plasmons and the 4f electrons could be detected. Let us consider the four-level scheme shown in fig. 4. The absorbing level 1 is assumed to be in resonance with the absorption by the plasmons in the silver particles. The levels 2 and 3 correspond to the ?Do and 7F, 4f states respectively_ The absorption transition rate, W, and the spontaneous ?,mnsition rate A,, are assumed to be electric&pole allowed while 397
Vdume
CHEh5ICAL
116. number 5
PHYSICS
17 May 1985
.LEITERS
matching the solutions of eq. (1) at t = to _The average population 92(t) is then given by q2(t) = 47rN /” ’ 7j2(R) R 2 dR ,
(5)
a
where N is the concentration of Eu3+ idns and R. is halfthe average distance between the silver particles
Fig. 4. Four-level scheme used to describe the absorption and emission by the system Eu3 + plus ligands.The plasmon absorption band is indicated on the left, in resonancewith level 1.
A,, is in first order forbidden by Laporte’s rule_ However, due to the action of the odd crystal-field terms, this transition rate becomes, in second order, aIlowed by the forced electricdipole mechanism. Since there are indeed many levels between those labelled I and 2, we assumed that K corresponds to an effective transition rate mainly composed by nonradiative de~ys. The silver particles are considered as spheres of radius Q. Then, for Eu3* i&S situated at a distance R from the center of the nearest silver particle, the time of the populations, 91 and 772, of the levels 1 and 2 is given by the equations dq2(R)jdz = - r&R) drllC_RWr
772(R) +Kql(RI
= -Til(R)ql(R)
+ W(R),
, OGt
(1)
(2)
and d~l(R)/~=-Til(R)
)
v~(R) 9
t>to
3
(3)
where rO is the duration of the pulse and we have as-
sumed that, at moderate excitation intensities, the ground state is little depleted_ 71 and 72 are the lifetimes of levels 1 and 2. respectively. Thus, solving for v2(R) we find 772(R) = B(R) exp I---t/T1 @)I
+ c(R) =P I-r/72CR11s (4)
where the constants B(R) and c(R) are found by 398
PI. From eqs. (4) and (5) we may note that in qrder to have exponential fluorescence -de&y_frok the 5Do level, 72 should be independent of F and, further, 7l should be much smaller than r2_ These two coridiiioni seems 36 be satisfied in our-c*, as indicated by our experimental results. This difference from the etiperirnentalresults of Weitz et al. [2] may be interpreted as follows: (i) In the present case the emission from the 5DO level is completely out of resonance with the p&kmon oscillations_ This minimizes both the macroscopic and the distance-dependent corrections to the external field. (ii) As mentioned in section 1, due to parity selection rules the interaction between thk 4f eIectrons and external fieIds is, in first order, restricted to even multipole terms which are more sekitive.than the dipole interaction to the disk&e from the silver particle to tl& Eu3+. Thus, the fraction of neared Eu3+ ions in the bulk being very small, as compared to the case of a coating of Eu3+ complexes on a rough Surface 123, one expects that the nonexponential contnlmtion to the total decay should be negligible. This interpretation is also supported by the fluorescence spectrum shown in fig. 1. The two transitions 5Qo + 7F, and 5p. + 7 F2 are of magnetic and electric-dipole origin respectively. Th&, the interaction between the plasmons and the 4f eIectrons would preferentially affect the latter transition. However, Corn the fluorescence spectrum we may note that these two transitions are enhanced by the same factor, suggesting that -thisinteraction was no! detectable. Finally, we conclude that the degree to which the localized piasmons may affect the decay of 4f states can be satisfactorily understood by the electromagnetic model.
Volume 116, number S
CHEMICAL PHYSICS LEM?ERS
Acknowledgement
References
The authors thank h&s- A. Izrael for carrying out the STEM measurements, Mme M. Faucher for helpfbl ~rxnrnents on the manus&pt, Ms. Sihba C. de Mendon- and Ms. Margarete F. da Silva for helping with the preparation of the sam$es and the CNPq (Brazilian agency) for fnalcial support.
[I 3 0-L.
17
May
1985
hfalta, P-A_ SantsHkuz, G-J?_ de Z&f and F, AuzeE, J. Luminescence33 (1985), to be published. [Zf D.A. weltz, S_ Garoff, CD. I-hson, TJ. Gram% and J. ciefiten, J. Luminescence24/2S (1981) 83.
399