PII: SOO22-36!37(97)00020-6
Pergamon
COMPARISON
J. Phy. Chem So/ids Vol 58. No. 9. pp. 1399-1402. 1997 1997 Elsevier Science Ltd. All rights reserved Printed in Great Britain 0022-3697i97 117.00 + 0.00
0
OF NON-LINEAR OPTICAL SUSCEPTIBILITIES KNb03 AND LiNb03 DONGFENG
XUE
OF
and SIYUAN ZHANG*
Laboratory of Rare Earth Chemistry and Physics, Changchun Institute of Applied Chemistry, Chinese Academy of Sciences, Changchun, 130022, Jilin, Peoples Republic of China (Received 26 November 1996; accepted 6 January 1997)
Abstract-From the chemical bond viewpoint, second-order non-linear optical (NLO) tensor coefficients of KNb03 and LiNbOs crystals have been calculated. By using the bond-valence theory of complex crystals and the modified bond-charge model, we were able to determine contributions of each type of constituent chemical bond to the total second-order NLO susceptibility. The tensor values thus calculated are in good agreement with experimental data. From the comparison of NLO tensor coefficients of these two crystals, we found that the major NLO contributors are KO rzgroups and Li06 octahedra, not the distorted Nb06 octahedra. The difference between their NLO properties arises from their different structural characters, and the high coordination number of constituent elements in KNbO, makes its valence electrons become more delocalised compared with those of LiNbOJ. 0 1997 Elsevier Science Ltd. All rights reserved Keywork A. inorganic compounds, A. optical materials, D. dielectric properties, D. optical properties
1. INTRODUCTION
finally would lead to different
Potassium niobate (KNbOs) is a perovskite ferroelectric with the simplest electrics.
structure of oxygen octahedra
It is a promising
non-linear
material because of its high NLO coefficients; at the same time, its elastic, electro-optic, acoustic and acoustooptic properties
of the con-
In this paper, from the chemical bond viewpoint we calculate the linear as well as non-linear optical properties and chemical bond parameters of each type of constituent chemical bond in the two crystals, KNb03 and LiNbOs, by using the bond-valence theory of complex crystals [7] and the modified bond-charge model [8]. By comparison of their chemical bond properties, we can reveal the origin of the difference in their NLO properties.
ferro-
(NLO)
optical
properties
stituent chemical bonds in the two crystals.
have made this crystal one of the most
widely studied [ 1, 21. Lithium niobate (LiNb03) is used extensively in integrated and electro-optics; being one of the most popular NLO materials, it has also been investigated thoroughly with respect to its NLO properties and
2. THEORY
applications. It is interesting that both crystals with Nb06 octahedra in their structures have large NLO coefficients, and at room temperature orthorhombic coefficients efficiency
KNb03
single crystals
phase have higher phase-matchable and second
harmonic
than those of LiNb03
that consideration
generation
Chemical bond non-linearities have been evaluated on the basis of linear results by means of the bond-charge model of Levine [9]. The corresponding macroscopic property is the SHG coefficient dijk and the complete expression for the total non-linear susceptibility, diik, is written as
with NLO (SHG)
[3-61. It is obvious
of Nb06 octahedra alone cannot give
dvk.= Ed,:, P
a reasonable explanation for the large difference, and K0i2 groups and LiOs octahedra should be taken into
In known inorganic NLO crystals, KNbO3 and LiNbOj including
only two groups, KO rz and Nb06 groups in KNb03 and Li06 and NbO6 groups in LiNb03. Starting from their crystal structures,
F”d$(C) =
we find that all constituent elements
have their own coordination environments,
(1)
where d& is the total macroscopic non-linear contribution of constituent chemical bonds of type p. F” is the fraction of bonds of type ~1composing the crystal, d&(C) is the ionic fraction of the non-linear optical coefficient, and d&(Eh) is the covalent fraction,
account. However, to date, there has been no work offering a quantitative explanation for this difference. are the members with the simplest constituents
= z Fp~;k(C)+d;k(E,,)] P
and this
G$N~(14.4)bp exp( -k:$)
[(Zi)’ +“(z:)*]
(x~)‘CP
(Es)2V’)2q” (2)
*To whom all correspondence should he addressed. I399
D.XUE and% ZHANG
1400
CP = 14.4V exp( - kt.4) [(l/n).(Zi)*/$
-~(Z~)*/$]
(n< 1)
(8)
(3) where qk is the geometrical contribution of the bonds of type cc,which can simply be calculated from
in which the sum on X is over all n: bonds of type /1 in the unit cell, and $(A) is the direction cosine with respect to the ith coordinate axis of the Xth bond of type fl in this cell. N: is the number of bonds of type p per cubic centimetre. Levine found @’= p(Nc)p, and that the actual value for index p depends on the given crystal structure [9]. In our calculation of NLO crystals, it is found that the value of bP decreases with increasing value of NC. In crystals with H-O bonds, the coordination number of the H atom is always less than three, so the the average coordination number NC is less than three, and the value for p is three [IO]. In KNbOs the much higher coordination number of the K+ cation leads to a relatively higher coordination number of the oxygen, which makes KNbOs different from other NLO materials including LiNbOS. In NLO complex crystals the common expression for b” is that P = /3(Nc)2 when 3.0 < NC < 5.33, V = p(N~)‘.82s when NC = 6, and b” = /3(N[)‘.2 when NC > 7.67. /3 can be deduced from the index of refraction for the crystal: here for KNb03, Ra = 2.2576 [ 111,and for LiNb03, no = 2.23 [12], at 1.064 pm. N; = N&l( 1 + n) + n-N&/( I+ n)
(5)
In eqn (2) exp( - e.4) is the Thomas-Fermi screening factor, (2:)’ and (g)’ are the effective valence electrons of A and B ions, respectively [7]. n is the ratio of the number of atoms of element B to that of element A in the subformula, and $ is the susceptibility of a single bond of type ~1.If a crystal is composed of different types of chemical bonds (labelled a) then the total x can be resolved into contributions x” from the various types of bond ,=(~~-1)/4r=~F~x’=~N~XI: P
P
(6)
in which x” = (4~) - ’ 4,%~/E~)2is the total macroscopic susceptibility of a crystal composed entirely of bonds of type p (including local-field effects), and DPis the plasma frequency. C” and Ei are the heteropolar and homopolar parts of the average energy gap between the bonding and the antibonding states, respectively, and (Q2 = (E;)2 + (r?Y2. r? = 14.4P exp( -e$) (n-= 1)
[(Zi)‘/{
- (ZL)*/4] (7)
E”h = 39 . 74/(dp)2.48
(9)
4 = 612 is the average radius of elements A and B in A, and the core radius is P$= 0.35$. The fractional covalency fl of the individual bonds is calculated from fc” = (E;)2/(E;)2. The difference in the atomic sizes is given by pP = (6 - $)/(4 + 4;). where { and 4 are the covalent radii of atoms A and B, their values being taken from ]131. In eqns (2) and (3), q” is the bond charge of the pth bond [8, 101 ~=(n:)*[ll(x~+l)+K~]e
(10)
K is a function of the average covalency F, of all kinds of
bonds in a complex crystal and of the coordination number N,,, of its central cation. We can determine its value by using the following equation [8]: K= (2Fc - 1.1)/N,,
(11)
where F, is defined as
From the viewpoint of chemical bonds, complex (or multibond) crystals can be decomposed into a sum of constituent subformulae [7]. When the detailed crystal structure is known, the subformula of any kind of chemical bond A-B in the complex crystal AoBbDdGg.. . (crystal molecular formula) can be written as follows [N(B - A).a/Ncn] .A [N(A - ~).b/~cs] .B
(13)
in which A,B,D,G,. . . are different chemical elements or different symmetry positions of the same element in the crystal formula, and n,b,d,g,. . . are numbers of the corresponding element, N(I - J) is the coordination number of I ion by J ion, and Nc&ca,. . . are the nearest-neighbour coordination numbers of each element in the crystal. Once the subformula equation of a complex crystal is determined, the chemical bond parameters of bonds of type p in eqns (2) and (3) can be calculated by using the bond-valence theory of complex crystals [7], and then these calculated parameters can be used to evaluate the NLO tensor coefficient dvk of the crystal.
3. RESULTS AND DISCUSSION
At room temperature, KNbOs crystallises in the space group Bmm2 (point group mm2), with cell dimensions a = 5.697 A, b= 3.971 Aandc =5.722 A [14];LiNbOJ is rhombohedral with space group R3c (point group 3m), its
Comparison of non-linear optical susceptibilities of KNbOj and LiNbOJ
1401
cell constants being a = 5.14829 w and c = 13.8631 A
octahedra,
1151. Both crystals have the distorted Nb06 octahedral
acters of these constituent groups, i.e. geometrical
structure. The restrictions
and coordination environment. In KNbOj, oxygen atoms are linked with four potassium atoms and two niobium
imposed by the crystal sym-
metry and the Kleinman symmetry conditions [ 161 on NLO tensor coefficients result in both crystals having the three tensor coefficients djl, dj2 and dj3, and, for LiNb03, d3, = d32. According valence
which is due to the different structural char-
atoms, niobium atoms compared with potassium atoms have relatively more valence electrons, a lower coordination number and a stronger ability to offer these valence
to their crystal structure data, the bond-
equation
(subformula
equation)
for the two
KNb03 = 1/12KO(l),
electrons,
Thus Nb-0
+ 1/12KO(l’)r
have
less linear
+ 1/6KO(l”),
two niobium
+ 1/3Nb0(2) + 113Nb0(2’)
In LiNbO,,
(14)
atoms,
(15) lead to different
of valence electrons in each kind of consti-
atoms
Therefore,
lower value of fractional
LiNb03 = l/2LiO(1)3,2 + 1/2LiO(s),,, + 1/2NbO(1)s,2 + 1/2NbO(s)s,2
niobium
compared
with
lithium atoms have relatively more valence electrons, the same coordination number and a stronger ability to offer these electrons.
environments
contribution.
oxygen atoms are linked with two lithium atoms and
+ l/3KO(2)2 + 1/3K0(2’)r + 1/3NbO( 1)
The different coordination
bonds have the lower value of
fractional covalency f/ compared with K-O bonds and therefore
crystals is written as
distributions
factor
Nb-0
bonds have a
covalency f,” and can have
closer linear contributions when the same coordination environment is under consideration. The structural change in KNbO, makes K-O bonds the major contributor to the total linearity, surpassing the contribution
of
tuent chemical bond. For KNbOj, in K-O bonds, Zi = 1
Nb-0
and & = 1.5; in Nb-0
small, which finally lead to smaller d3, and dj2 compared
bonds, Z&, = 5 and G = 15. On
bonds in value. The values of G[;, and Gg3 are quite
the other hand, for LiNb03, in Li-0 bonds, Zti = 1 and & = 2.0; in Nb-0 bonds, Zzb = 5 and & = 10. From the
with djj. The signs of Gt are opposite in KOt2 groups or
refractive indices of KNb03 and LiNb03, we can calcu-
cancellation
late the detailed
values of GE lead to a strong cancellation, and this is KNbozis smaller than that of the reason why the value of dj3
bond parameters,
linearity
and non-
linearity of individual bonds, which are listed in Table 1 and Table 2. The final calculated NLO tensor coefficients of the two crystals are listed in Table 3. The discrepancy
Li06 octahedra
dLiNbO, 33
and in NbOh octahedra,
among
ds values;
,
the amount of electronic delocalisation
calculated
obviously
mental data; therefore,
with the experi-
these reasonable
reveal the origin of the difference
parameters
between
can
these two
NLO materials. From Tables 1 and 2 we can see the NLO behaviour in
absolute
The calculated results also show there is an increase in
results from the dispersion of its three principal of dy refractive indices, and it can be seen from Table 3 that the results agree satisfactorily
leading to a
the similar
value of the fractional KNb03
lower
covalency
in KNbO,. The
f,” in KNbOj
than that in LiNb03,
that the valance
electrons
weakly than those in LiNb03.
is
we find in
are bonded
This agrees
more
with the
conclusions obtained from the optical spectroscopy
[ 171.
KNbOj and LiNbOj is dominated by the distorted KOlr
In conclusion, the difference between the NLO properties of KNb03 and LiNbO,, both of which have NbOh
groups
octahedra in their crystal structures, is a consequence
and Lion
octahedra,
not the distorted
NbOh
of
Table I. Chemical bond parameters, linear and non-linear properties of each type of bonds in KNbOl and their contributions to the total linearity and non-linearity K-0(1’)
K-0( I) d” (A, EE (ev) CP (eV) R 4%X1 x;h q’le GP 31 d:,t d?’ GF $, d” 17
2.8671 2.9929 3.1767 0.4703 4.975 1 1.4231 0.0579 0.0000 0.0000 0.0000 -1.0000 187.6603
2.8829 2.8764 3.0633 0.4686 5.1197 1.4644 0.0566 0.0000 0.0000 0.0000
0.0000
1.0000 - 199.0666
tdr (10-9esu). $l’he value approaches 0.
K-0( 1”)
K-0(2)
2.8486 2.963 1 3.1473 0.4699 5.0112 1.4334 0.0576 0.0080 -3.0619 0.0000 0.0000 --oYk
2.7923 3.1135 3.2917 0.4722 4.8363 1.3834 0.0592 -0.1276 90.4087 -0.2403 170.2315 -0.1074 76.0973
K-0(2’)
Nb-O( I)
2.8734 2.9001 3.0863 0.4689 5.0895 1.4558 0.0568 0.1269 -99.8365 0.2547 -200.377 1 0.1520 - 119.5297
1.9891 7.2204 27.7035 0.0636 2.2473 0.6428 I .0322 0.0000 0.0000 0.0602 0.1025 0.0002 0.0004
Nb-O(2) 2.1798 5.7538 2 1.8875 0.0646 2.7498 0.7866 0.8947 0.3296 0.8765 O.OOOiI 0.0000 0.4 185 I.1131
Nb-O(2’) 1.8645 8.4768 32.595 1 0.0633 I .9635 0.5616 1.1305 -0.3727 -0.4726 0.0000 0.0000 -0.2871 -0.3641
D. XUE and S. WANG
1402
Table 2. Linear and non-linear properties and chemical bond parameters of each type of bond in LiNbGl and their contributions to the total linearity and non-linearity LiNbGJ Li-O(1)
Li-O(s) d” (A)
EC (eV) C (eV)
f:
4rti X: 4/e G:, d$ ( 1Od9esu)
r& 4, (10e9 esu)
2.0683 6.5540 4.6324 0.6669 2.7134 0.9543 0.1795 -0.1521 22.7847 -0.0412 6.1714
Nb-O(s)
Nb-O(l)
2.2385 5.3868 3.9379 0.6517 3.1129 I .0948 0.1621 0.1753 -34.2479 0.3625 -70.8352
2.1123 6.2206
11.6706 0.2212 5.5288 1.9445 0.5106 -0.1848 2.9738 -0.2997 4.8237
I .8886 8.2111 15.1450 0.2272 4.5365 I .5955 0.6021 0.1839 -2.2004 0.1071 -1.2814
Table 3. Comparisons between theoretical values and experimental data on non-linear tensor coefficients of KNbO j and LiNbG3 crystals, at 1.064 pm KNbO, Expt. -37.72$ 27.05.288 -43.69$ 30.6$,32Il -65.41$ 46.6$,48.8(
LiNbG3 talc.
Expt.
talc.
- 12.0858
- 10.311
- 10.6898
-30.043 1
-10.311
-10.6898
-54.08%
-64.511
-61.1214
tThe units of d, are 10m9esu. ssee [31. $The recommended absolute values of [6]. lIThe absolute data from [4]. ]iSee 15.61.
the different behaviours of their valence electrons, i.e. the higher amount of electronic delocalisation in KNb03 than in LiNbOs. The latter can be attributed to their different crystal structures, which lead to different values of the geometrical contributions of constituent chemical bonds and different coordination environments of the constituent elements. All of these differences can be found from the chemical bond parameters, the linear and non-linear properties of each type of constituent bond. This is why we start from a chemical bond search for the relationship between crystal structure and nonlinear optical properties. Acknowledgements-This work was supported by the State-Key Program of Basic Research of China. REFERENCES I. Zgnoik, M., Schlesser, R., Biaggio, L., Voit, E., Scheny, J. and Gtinter, P.. J. Appl. Phys., 1993.74. 1287. 2. Zgnoik, M., Nakagawa, K. and Gtinter, P., J. Opr. Sot. Am.,
1995, B12, 1416.
3. Biaggio. L., Kerkoc, P., Wu, L.-S.. Gbnter, P. and Zysset, B., J. Opt. Sot. Am., 1992, B9.507. 4. Uematsu, Y., Jpn J. Appl. Phys., 1974,13, 1362. 5. Miller, R. C., Nordland, W. A. and Bridenbaugh, P. M., J. Appl. Phys., 1971,42,4145. 6. Roberts, D. A., IEEE J. Quantum Electron., 1992, QE-28, 2057. 7. Zhang, S. Y., Chin. J. Chem. Phys., 1991.4. 109. 8. Xue, D. F. and Zhang, S. Y., J. Phys.: Condens. Matter, 1996,8,1949. 9. Levine, B. F., Phys. Rev., 1973, B7.2600. 10. Xue, D. F. and Zhang, S. Y ., J. Phys. Chem. Solids, 1996,57, 1321. 11. Zysset, B., Biaggio, L. and Giinter, P., J. Opr. Sot. Am., 1992. B9.380.
12. Boyd. G. D., Miller, R. C., Nassu, K., Bond, W. L. and Savage, A., Appl. Phys. Len.,1964,5,234; Bond, W. L., J. Appl. Phys., 1965,X&1674.
13. Dai, A. B. and Shen, M. C., Periodic Table of Element (in Chinese). Scientific Technology Press, Shanghai, 1981. 14. Katz, L. and Megaw. H. D., Acta Crystallog., 1%7, 22, 639.
15. Abrahams, S. C., Reddy. J. M. and Bernstein, J. L., J. Phys. Chem. Solids, 1966,27,977. 16. Kleinman, D. A., Phys. Rev., 1962,126, 1977. 17. Wiegal, M., Edmond, M. H. J., Stobbe, E. R. and Blame. G., J. Phys. Chem. Solids, 1994,55,773.