centers in alkali halides - Springer Link

1 downloads 0 Views 839KB Size Report
AgBr or AgI with single crystals of CsBr and CsI, respectively, (Karl. Korth, Kiel, Germany) at 550 .... Energy Difference to Ema x. Fig. 5. C band of CsI: Ag- at 78 K ...

Z. Physik 234, 362--378 (1970)

Optical Transitions of Ag- Centers in Alkali Halides W. KLEEMANN Erstes Physikallsehes Institut der Universitat, G~Sttingen Received March 16, 1970 Ag- centers in alkali halides give rise to a strong absorption band in the 300 nm region (formerly called "B band"). Its resolved triplet structure in CsC1 suggests that it corresponds to the C band of the isoelectronic In + center. Two very weak bands are found in several alkali halides in the 400 nm region. These new bands are assigned to the A and B transitions of the In+-type centers. This is supported by the doublet structure in the A band, and by the temperature dependence of the oscillator strength of the B band. In KCI:Ag- the ratios of the oscillator strengths are found to be fc/fa= 610 and fc/fB= 3,400 at low temperatures. The energy parameters of Ag- centers are computed and compared with those of other sZ-type centers. The electron-lattice coupling parameters are estimated from the Jahn-Teller splitting of the C band in CsCI and of the A band in KC1. The temperature dependence of the lifetime of the visible fluorescence suggests that a metastable state is involved in the emission process after a C band excitation.

w1. Introduction Experimental evidence for the existence of negative heavy metal ions like C u - , A g - , and A u - in alkali halides has been given in preceding papers a'z. T h e y f o r m anionic centers with the electronic configuration ns 2 (n = 4 , 5, 6). The properties of isoelectronic cationic s2-type centers like G a +, I n +, and T1 + are t h o r o u g h l y investigated a. They give rise to three U V absorption bands A, B, and C, which correspond to the electronic transitions of the free ions f r o m the g r o u n d state ~So to the excited states 3P1, 3P2, and aP~. A further band, called D band, in the exciton region of the crystal is assumed to be due to a disturbed exciton transition. The absorption spectra of A u - centers agree well with this picture. As expected they reveal the A, B, C, and one or two D bands 2'4. In contrast to this only one A g - b a n d has been f o u n d s, 2. F o r historical reasons this b a n d was n a m e d B b a n d 6. To avoid confusion with the B 1 Kleemann, W.: Z. Physik 214, 285 (1968). 2 Fischer, F.: Z. Physik 231, 293 (1970). 3 See for example Fowler, W. B., edit. Fowler, W. B.: Physics of color centers, p. 53. New York and London: Academic Press 1968. 4 Mabuchi, T., Yoshikawa, A., Onaka, R.: J. Phys. Soc. Japan 28, 805 (1970). 5 Degrieck, W., Jacobs, G.: phys. stat. sol. 18, 279 (1966). -- Topa, V.: Rev. Roum. Phys. 12, 781 (1967). 6 Etzel, H. W., Sehulman, J. H.: J. Chem. Phys. 22, 1549 (1954).

Optical Transitions of Ag- Centers in Alkali Halides


band of the other s2-type centers this name will refer only to the absorption band of the 1So ~ 3P2 transition in this paper. Fischer 2 suggests a correspondence of the Ag- band to the C band rather than to the A band 1. This conclusion is drawn from a systematical comparison of the relative band positions and their height ratios for series of isoelectronic centers. Further evidence is given by the large oscillator strength of the Ag- band and its unresolved triplet structure in RbC11. The A and B bands are expected to have very small oscillator strengths. For this reason they have not been discovered in absorption until now. Weak A and B bands should also be found for the Cucenters besides their strong UV bands 7,1, which are likely to be assigned to the C and D bands 2 A first confirmation of the supposition outlined above has been given quite recently by measurements of the excitation spectrum of KCI:Agby Kojima etal. 8. The emission band at 2.86 eV can be excited by irradiation into the well-known 4.35 eV band as well as into two very weak bands at 3.26 and 3.11 eV. The latter part of the excitation spectrum is interpreted as the B and A bands. In this paper we intend to confirm these results by measurements of the absorption. The A and B bands of Ag- centers are dicovered at wavelengths in the 400 nm region. Heavily doped crystals are necessary to find the extremely weak transitions. They are compared with those of other sZ-type centers with respect to their band shapes and oscillator strengths. In another part of this paper we study the Ag- absorption in CsC1 type crystals. We hope to find the typical triplet structure of the C band 9 to be more pronounced than in NaC1 type crystals. This is expected because of the increasing tendency of band splitting in the series NaC1, KCI, and RbC11, and the extraordinarily distinct fine structure of the C band of In + in cesium halides 1~ Finally, lifetime measurements of the visible fluorescence are to reveal certain properties of Ag- centers in their relaxed excited state.

w Experimental Procedure a) Preparation of the Crystals Single crystals of KC1, KBr, and KI are grown from reagent grade powders (Merck) by the Kyropoulos method. The melt is doped with 5 • -2 mole ~ of the respective silver halide. After growing, the 7 Fukuda, K., Nakagawa, T. : Bull. Inst. Chem. Res., Kyoto Univ. 39, 158 (1961). 8 Kojima, K., Shimanuki, S., Maki, M., Kojima, T,: To be published in J. Phys. Soc. Japan. 9 Fukuda, A. : J. Phys. Soc. Japan 27, 96 (1969). 10 Fukuda, A., Makishima, S. : Phys. Letters 24A, 267 (1967). 25*


W. Kleemann:

crystals are annealed for 24 hours in a quartz tube in an inert gas atmosphere. After this treatment the Ag + ions are nearly uniformly distributed in the crystal with a concentration of 3 - 1 0 x 1017 cm -3. CsBr:Ag + and CsI:Ag + are prepared by a diffusion reaction of AgBr or AgI with single crystals of CsBr and CsI, respectively, (Karl Korth, Kiel, Germany) at 550 ~ in a Supremax glass ampoule. The crystals are homogeneously doped after about 24 hours. The preparation of single crystals of CsCI:Ag + is more difficult because of the transformation of its lattice structure at 469 ~ A modified Bridgman method is applied 11, in which the quartz ampoule passes the furnace two times. In the first run the melt crystallizes, whereas in the second run the lattice transformation takes place. We find the crystal consisting of small monocrystalline regions. The conversion from Ag § to Ag- centers is performed by electrolytical reduction with F centers as described previously 1. From our experience the above mentioned amounts of Ag + ions are the upper limit for complete conversion. Silver colloids are avoided by rapidly quenching the samples. As for CsC1 the F centers are produced at 450 ~ in order to avoid the lattice transformation at higher temperatures. Freshly cleaved samples are used in the case of NaCl-type crystals. As for the cesium halides, samples have to be cut from the bulk crystal. They are ground and polished for optical measurements. The thickness of the samples amounts to 0.2-1 mm for measuring the C bands, and 4 - 1 0 mm for measuring the A and B bands.

b) Absorption Measurements The optical absorption is measured by a spectrophotometer CARY 14 R. Temperatures from 2.6 to 450 K are achieved by cooling the inner tank of an optical cryostat with liquid helium, hydrogen, or nitrogen, or by warming up with warm air. Temperatures are measured by means of a Au + 0.03 ~ Fe - chromel thermocouple, which is clamped directly onto the sample.

c) Lifetime Measurements Radiative lifetime is measured after a short pulse of light exciting the Ag- centers. Two light sources are available. A nanosecond lightpulser TRW 88 A provides pulses of about 10 ns halfwidth. The other source is a spark gap in nitrogen atmosphere, which is fed by a voltage supply of 5,000 V over a 5 nF capacitor. Its pulse width is 0.3 gs. It is used for the measurement of long decay times (10 -3 s) because of its higher intensity. The fluorescence signals are detected by a photomulti11 Avakian,P., Smakula,A.: J. Appl. Phys. 31, 1720 (1960).

Optical Transitions of Ag- Centers in Alkali Halides


plier tube (EMI 6094 B). The signal from the load resistor is fed through an emitter follower into an oscilloscope (Tektronix 536). A sampling technique is used in connection with an X - Y-recorder for lifetimes up to 50 ps. For longer lifetimes the photographic method is employed.

w3. Experimental Results a) C Bands in Cesium Halides The absorption bands of A g - centers in CsC1, CsBr, and CsI are shown in Figs. 1, 2, and 3 a, respectively, for different temperatures. The band areas are constant with respect to temperature up to 450 K within the experimental accuracy. Some characteristical data of the bands are compiled in Table 1. The absorption peaks are situated in the 300 nm region. Above 100 K they shift to smaller energies linearly with temperature (Table 1, column 5). The temperature dependence of the halfwidths amounting to 0.11 - 0.16 eV at low temperatures is described quite well by H(T)=H(O) [cth(hv,/2k T)] m . (1) From (1) the effective frequency of the interacting lattice vibrations is found to be ~,,=1.8 x 1012 s -1 in all three crystals. Regarding all these properties there is no distinct difference between the A g - bands in CsCl-type crystals and in NaCl-type crystals.


310 nm 330 1-

~2.6 K ~45K







300 nrn




} CsCl:AgC Band

2.6K /



-E30 O ~J



///k t/I _O


< 10



; /





ttt\\ t7t l+.3





3.9 eV 3.7 4.3 4.2 Photon Energy







4.1 eV


Fig. l a--c. C band of CsCI:Ag- at different temperatures (a), at 2.6 K (b), and at 78 K (c). The broken line in (c) is the computed curve by CHO 25


W. Kleemann: 300


320 i

C. C~ C1 "~A ~



nm 340


/ ~.,I/20K


/7OK o~ r



/150 g /350 K



Photon Energy

3.7 eV

Fig. 2. C band of CsBr:Ag- at differenttemperatures Table 1. C band of A g - centers in cesium halides. )'max,Emax, and H are extrapolated to T = O . --dEmax/dT is valid for T > IO0 K )'max [am]

Emax [eV]





10 -12 S

[meV/K] CsC1 CsBr CsI

301.5/298 308,9 328,8

4.112/4.158 0.156 4.013 0.112 3.770 0,113

0.26 0.32 0.33

1.8 1.8 1.8

A remarkable fine structure, however, occurs in the Ag- band of CsCl. At 78 K we find three peaks at 4.098, 4.146, and 4.194 eV (Fig. 1c), which will be denoted by C 1, C 2 , and C3, respectively.

We obtain the following properties of the three subbands after a graphical decomposition into three Gaussian curves: 1. At 2.6 K the C1 peak is higher than the two other peaks. C1 decreases most rapidly and C3 most slowly with increasing temperature. At the same time the resolution of Cz and C3 is improved, whereas that of C 1 and C2 becomes poorer. 2. C1 and C3 are nearly symmetrically situated with respect to Cz. A small decrease of the distance between C a and C2 compared with that of C2 and Ca is found at higher temperatures. 3. The distance Ac between C2 and C a varies linearly with ~/T above 100 K. At low temperatures a constant value is attained (Fig. 4). The symmetrical structure and the ~/T dependence at high temperature agree with the observations made on the C bands of other s2-type centers

Optical Transitions of Ag- Centers in Alkali Halides

320 300

34Ohm 360 _ C 1.2 [ I1~20 K crn-,/



.78 K



450 nm 470 A2 /3'~1 / / j4.8 K

Csl :Ag-

I f



3.8 (a)


3.6 eV 3.4 2.9 Photon Energy


2.7 eV



Fig. 3 a and b. C band (a), and B and A bands (b) of CsI: Ag- at different temperatures /0


50 100



Ac O.

300K 500









0.05 / / / 0 0





(Temperature) l/2

Fig. 4. Separation Ac of the C2 and C3 peak of the C band in CsCI:Ag- plotted against ]/T

in alkali halides 9. Thus the designation of these A g - bands as C bands seems to be justified. The large oscillator strength in the order of unity supports this interpretation. A poor/y resolved triplet structure is found in the C band of CsBr at low temperatures (Fig. 2). A detailed analysis of the band shape reveals essentially the same features as for C s C I : A g - . The C band of CsI: A g - has a symmetrical shape without any structure. A comparison with a Gaussian curve, which has the same area and peak height, in


W. Kleemann: C3



Cl ~ G a uc(78K) Dan d &sian

~ crn-I

(.3 0


I~ _

+ o,1 o -o,l eV Energy Difference to Emax Fig. 5. C band of CsI: Ag- at 78 K (full line) compared with a Gaussian curve (broken line). The band areas and heights are normalized

Fig. 5, however, gives strong evidence for a complex structure within the band measured at 78 K. Its deviations from the Gaussian curve are interpreted as C1 and C3. b) A and B Bands in KCI, KBr, 1>1)

at intermediate temperatures. For KCI:Ag- we find agreement above 100 K and obtain c z =0.48 eV. 2. The temperature To of the band height reversal obeys the equation

To=O.298(EN-EA)2/kc 2


Inserting To =270 K for KC1 we obtain c2 =0.29 eV. In the above formulas the influence of the Eg and the Alg modes is neglected. This may be responsible for the difference of the two values obtained. They are both somewhat larger than c2,,~0.2 as estimated by Kojima et aL s. At low temperatures the low energy tails of the A bands are very steep. Presumably this is due to the weak coupling as discussed above in connection with the asymmetry of the C bands. This is supported by the existence of a particularly small Stokes shift involved in the A emission (0.16 eV for KCI:Ag-). In contrast to the A bands of the Aucenters 2, however, no zero-phonon line could be detected in the Agbands at low temperatures. I wouldlike to thank Prof. Dr. R. Hilschfor his kind interest in this investigation. The experimental work was supported by the "Deutsche Forschungsgemeinschaft". Dipl.-Phys. Dr. W. Kleemann I. PhysikalischesInstitut der Universit/it D-3400 G6ttingen,BunsenstraBe9

Suggest Documents