J. Phys. Chem. 1983, 87,5117-5120
5117
Nuclear Magnetic Resonance Study of Thallium-205 in Binary Molten Salt Mixtures Yoshlo Nakamura, Yuklharu Kltazawa, Mltsuo Shlmojl, and Shlgezo Shlmokawa' Department of Chemistry, Faculty of Science, and NMR Laboratory, Faculty of Engineering, Hokkaido University, 060 Sapporo, Japan (Received: January 3, 1983; In Final Form: April 15, 1983)
The chemical shifts of T1+in molten binary mixtures of nitrates as well as in those of chlorides have been measured as a function of composition and temperature. The shifts of z05Tl+increase in the paramagnetic direction with temperature and in the diamagnetic direction with decreasing the size of added foreign cations. The results are interpreted by the change in the degree of "overlap" between the thallous ion and anion with temperature and composition.
Introduction The isotopes of 205Tland z03Tlare spin ' / 2 nuclei and have high NMR receptivities. The chemical shifts of these nuclei have extensively been studied in various compounds and solutions.' The thallous ion (T1') has an outer atomic orbital of ( 6 ~ 1which , ~ is easily deformable or mixed with higher orbitals and consequently gives a pronounced chemical shift as large as lo4ppm. NMR studies a t high temperatuer with these nulcei, however, are rather limited. The first investigation of the chemical shifts of T1+ in molten thallous halides and their mixtures with alkali halides has been made by Hafner and N a ~ h t r i e b ,who ~,~ proved that the T1+ ion is a useful NMR probe for the study of chemical environments of ions in molten salts as well as in crystals4 and g l a ~ s e s . ~ In the present paper, we report some new measurements of the chemical shifts of z05Tl+in molten binary mixtures of nitrates and of chlorides with a high-resolution, hightemperature NMR apparatus: in order to obtain structural information on these molten salt mixtures and also to reexamine the simple anion polarization model proposed by Hafner and N a ~ h t r i e b . ~
-
Experimental Section Chemical shifts of 205Tlwere measured with a Bruker SXP 4-100 pulse spectrometer operating a t 20.8 MHz. The wide-gap magnet was modified by a high-resolution type and the magnetic field was stabilized with a proton external lock systema6 The schematic design of the hightemperature probe is shown in Figure 1. The heater consists of a pure nichrome wire wound noninductively on a quartz tube. A platinum radio-frequency coil was wound on an internal quartz tube, into which a sample cell of 8 mm 0.d. was inserted. The temperature was kept constant within fl "C with a temperature control system. The whole assembly of the probe was cooled by a water jacket. The temperature range of the measurements was from the melting point of the samples to 350 "C for nitrate systems and 550 "C for chloride systems. The fid signals for each experimental run were accumulated for 10 to 300 scans to improve the signal-to-noise ratio and then Fourier transformed with a Nicolet BNC-12 computer system. The chemical shift, 6, measured a t a fixed magnetic field is given by 6 =
(vR
- vS)/vR
Chemical Shifts o f M o l t e n Thallium Salts -A6
T,, TlNO, TlCl TlBr T1I
"C
206.5' 429 459 440
X
l o 4a t
-A6(T,)
X
lo4
500 "C
at T ,
O b (0)' 12.7 (9.4) 15.7 (12.7) 22.5 (18.9)
1 . 3 (1.7)' 5.4 (5.6) 5.2 (5.5) 3.3 (3.1)
a G. J. J a n z , "Molten Salts H a n d b o o k " , A c a d e m i c Press, Referred to pure TlNO, a t 500 "C New Y o r k , 1967. ( e x t r a p o l a t e d ) . ' Values in parentheses are t a k e n f r o m ref 2.
Reagent grade materials (MN03, MCl; M = T1, Ag, and alkali metals) were obtained commercially from Wako Chemical Ind. Ltd. and used without further purification. Sample mixtures were prepared by mixing a desired amount of each component, dried, and melted thoroughly in a sample cell with an electric furnace. The sample cell was then sealed after repeated evacuation. Results Figure 2 shows the observed temperature dependence of the shift of resonance frequency of zOT1+ in pure TlN03 referred to that a t 250 "C together with that for a (TlNa)N03 mixture. The temperature coefficient of the chemical shift, d6/dT, was (-0.36 f 0.01) X lo4 deg-l, which was different by a factor of 2 from the earlier data,2 (-0.70 f 0.05) X deg-l. We think that the present value is more reliable in view of the reproducibility of the data. It may be noted that the value of d6/6T is little affected by the slightly different choice of the reference frequency. The temperature dependence of the shifts for the mixtures studied was the same as that of the pure TlN03 within experimental errors. Thus, the chemical shifts referred to pure TlN03 at the same temperature, A6, was practically independent of temperature, as shown in Figure 3 for the (T1-Li)N03 system. Experimental errors in A6 are generally better than &2%. The results for the composition dependence of the chemical shifts at 350 "C for the six binary nitrate systems studied are shown in Figure 4. The observed shifts vary almost linearly with the mole fraction of the foreign cation or XyNOB. For comparison we have also measured the chemical shifts of T1+ in some molten mixtures of TlCl
(1)
where us and uR are the resonance frequencies of the sample and the reference, respectively. Since no "free ion" as an ideal reference is available, pure molten T1NO3 was provisionally taken as a reference. NMR Laboratory.
T A B L E I:
(1) J. F. Hinton and R. W. Briggs, "NMR and the Periodic Table", R. K. Harris and B. E. Mann, Ed., Academic Press, London, 1978, p 288. (2) S. Hafner and N. H. Nachtrieb, J. Chem. Phys., 40, 2891 (1964). (3) S. Hafner and N. H. Nachtrieb, J. Chem. Phys., 42, 631 (1965). (4) S. Hafner, J . Phys. Chem. Solids, 27, 1881 (1966). (5) R. K. Momii and N. H. Nachtrieb, J. Phys. Chem., 72,3416 (1968). (6) S. Shimokawa and E. Yamada, J. Mag. Reson., 51, 103 (1983).
0022-3654/83/2087-5 1 17$01.50/0 0 1983 American
Chemical Society
The Journal of Physical Chemistry, Vol. 87, No. 25, 1983
51 18
SF
n
Nakamura et al.
(TI-M
2 t
-
0
I Flgure 1. High-temperature NMR probe: (a) sample cell; (b) furnace; (c) rf coil; (d) external lock; (e) thermocouple; (f) water inlet; (9) water jacket of aluminum. h i g h field
,
I I
I
Cs
0
1.0
0.5 A
XMNO,
Flgure 4. Chemical shifts of 205TIin (TI-M)NO, mixtures referred to pure TINO, at 350 OC (400 OC for (TI-Cs)NO, mixtures).
I 3
Y
- 4 1
2
I
. I -
2 3w
d
I
-
0
:2 -
pure T I N 0 3
x o
aJ
I-
1 2 50
300 temperature
Q
7 -I
35OoC
Figure 2. Shifts of resonance frequency of '05TI with temperature for pure TINO, and (Tlo,5-Nao,,)N0, at 8.45 kG.
0
- 0.5
1.0
MCI
Flgure 5. Chemical shifts of 20?I in (TI-M)Cl mixtures referred to pure TIC1 at 550 'C.
with alkali metal chlorides or AgC1. The results of chemical shifts referred to pure TlCl at 550 "C are shown in Figure 5. The general trend is in reasonable agreement with previous data.3 The composition range was limited by the high melting temperature of alkali-halide-rich mixtures. In the chloride mixtures, the temperature coefficients of the mixtures were also nearly the same as that of the pure TlC1, for which d6/dT = (-0.76 f 0.01) X deg-' and in good agreement with the reported value,3 (-0.78 f 0.05) X lo4 deg-l. The changes of 6 at the melting temperature for TlNO,, TlC1, and some related compounds are given in Table I, which contains also the values of 6 referred to pure TlN03 at 500 "C. These values are compared with previous data,2 which should be corrected to some extent.
3
t
0 x 2
Q \
I
0
P
I
I
I
I
1
+
I L
LONO,
Figure 3. Chemical shifts of *',TI pure TINO,.
Discussion The chemical shift defined by eq 1 is written as the difference between the nuclear-screening parameter for the sample us and that for the reference ur: 6 = - '5, = AU = A'5d + A~J, = 6, 6, (2)
in (TI-Li)N03 mixtures referred to
The diamagnetic contribution AfJd or for Tl' is estimated , ~that i t can be neglibibly small in to be -10 ~ p m so
The Journal of Physical Chemistry, Vol. 87, No. 25, 7983 5 119
NMR of *05TI in Binary Molten Salt Mixtures
*
comparison with the observed large shifts of the order of io4 ppm. If we start with Ramsay’s general theory of magnetic shielding,s a simplified expression is given for the paramagnetic contribution, a, in ionic compoundsg (3)
where pB is the Bohr magneton and r is the electron radial coordinate. In the overlap model of Kondo and Yamashita: the constant B is given by ZA where A is a sum over the squares of overlap integrals of the wave functions of neighboring cations and anions, and Z is the coordination number. In this model, ( l / r 3 ) is an average over the excited-state orbitals of an ion and A E is the corresponding energy of the manifold of low-lying excited states of the ion above its ground state. Here, we shall adopt this overlap model, for the purpose of attempting to explain the present experimental results. The presence of covalent bonding in TlNO, and TlX has been suggested from thermochemical data. Kleppa and Hersh have pointed out that the non-Coulombic contribution to the lattice energy of T1N03 and AgNO, is of the order of 10% and they showed that the non-Coulombic contribution to the excess heats of these nitrates with alkali nitrateslO is significant. The covalency of molten TlNO, and AgN0, has also been demonstrated from measurements of the Raman spectra.’l Thus, we suppose that a small but significant overlap of the wave functions of T1+ and the anion is present in the molten thallium salts and their mixtures. In the overlap model, the change of the T1+ chemical shifts with temperature and composition may be interpreted by the change of various factors appearing in eq 3. Since we have no accurate knowledge of the wave functions of the excited states of T1+in the melts, we cannot correctly evaluate the values of ( l / r 3 ) and AE. We assume, therefore, that the main contribution to 6, arises from the change of the overlap integral or the factor A, neglecting the environmental changes of other factors, particularly in common anion mixtures. Hafner and Nachtrieb, have shown that a net electrostatic field a t the anion positions is established, if T1+and the substituted foreign cation have different ionic radii. This electrostatic field polarizes the anion and thus modifies the overlap of T1+and the anion. They discussed such a polarization effect on this overlap by considering a linear array of T1’ and a foreign cation M+ with an intervening anion X-. The net electrostatic field E, a t the center of X- in the direction of T1+ X- is then given by
-
where Ro and Rm are the interionic distances between Tl+ and X- and between M+ and X-, which are given by the sum of the respective ionic radii. Although the Taylor expansion to first order used in,eq 4 may not be sufficient for large ARCat,we use this linearized expression for the sake of simplicity in the present discussion. The electric (7) W. G. Schneider and A. D. Buckingham, Discuss. Faraday Sac., 34, 147 (1962). (8) N. F . Ramsey, Phys. Reu., 78, 699 (1950). (9) J. Kondo and J. Yamashita, J. Phys. Chem. Solids, 10, 245 (1959). (10) 0. J. Kleppa and L. S.Hersh, J. Chem. Phys., 36, 544 (1962). (11) D. W. James and Wah-Hing Leong, Trans. Faraday Soc., 66,1948 (1970).
I
I
0
0.5 d
1
I
I.o
(A )
Rcai
Figure 6. Dependence of chemical shifts on the difference in cationic = 0.20 at 350 O C . radii for (TI-M)NO, mixtures with X,,,,
1.0
1
I
‘
I
l-
-1.0
’
I
0
0.5
d Rcat,
I .o
i
(A
Figure 7. Dependence of chemical shifts on the difference in cationic radii for (TI-M)CI mixtures with X,,, = 0.20 at 550 OC.
dipole moment of the anion caused by this field is thus of the form3
P, = e a r = CYE, =
2eaMcat
(5) Ro3 where CY is the polarizability of the anion and Ar is the dipole length which is assumed to be equal to the change of the interionic distance between T1+ and X-, AR. We note that the overlap parameter A can be a function of the interionic distance R, such that A = p exp(-CR).12 Then with the use of eq 3, we may write for the change in the chemical shift A6 for mixtures with the foreign cation mole fraction XMx,referred to the pure T1X a t the same temperature A6/6o AA/Ao -(CoAR + R0AC)XMX (6)
where the subscript 0 denotes the pure state. Combining eq 5 and 6, we finally obtain
Here, only the competition of the nearest-neighbor cations of the intervening anion is taken into account and the effect of more distant cations is neglected. This result is in qualitative agreement with the observed linear composition dependence of A6 on XMX, shown in Figures 4 and 5. We next examine the dependence of A6 on the difference in the cationic radii, ARmt, at a given composition. Figure
5120
The Journal of Physical Chemistry, Vol. 87, No. 25, 1983
6 and 7 respectively show the results for the (TI-M)NO, and (T1-M)Cl mixtures a t XMx= 0.20. For the cationic radii, we have taken the values of Pauling,13except for T1+, for which the value of Kordes14 (1.59 A) was used, as recommended by Kleppa and Hersh’O in their analysis of the thermodynamic data. For the nitrate mixtures, a straight line passing through the origin is obtained, which implies that A C = 0 in eq 7. Cations smaller than T1+ (mat> 0) cause positive (diamagnetic) shifts (A6 > 0) with the decrease of the overlap of T1+ and X- and vice versa, in accordance with the prediction of eq 7 (note that 6, < 0). On the other hand, in the chloride mixtures, we obtain two separate lines as shown in Figure 7. The line for the mixtures with alkali ions deviates systematically toward the paramagnetic direction. This indicates that A C is negative and constant in these mixtures and that A C = 0 in the mixture with Ag+. This may be attributed to the fact that the substitution of a near-neighboring T1+ ion of the Cl- ion by an alkali ion increases the overlap of the remaining Tl+-Cl- bond (AC < 0), apart from the polarization effect, while the substitution by the Ag+ ion may not cause such an effect because of the similar capacity of Ag+ and T1+ to overlap with the anion. In the nitrate mixtures, on the other hand, the T1+ ion is coordinated possibly to one of the three oxygen atoms in the NO3- ion and the substitution of T1+ by a foreign cation with different ionicity may not cause any significant change of the overlap of the remaining Tl+-N03- bond and only the polarization effect should be responsible for the chemical shift as given by eq 7 with A C = 0. The X-ray diffraction data of Ohno and Furukawa15 show that the coordination of Ag+ and Li+ to a nitrate ion is different from that of other alkali ions: Ag+ and Li+ tend to locate a t the corner of the triangular NO3- and have a NaC1-like coordination, while the other alkali nitrate melts have a zinc-blend-like structure. The neutron diffraction data of Suzuki and Fukushima16reveal that the nitrate ion is an isosceles triangle in molten TlNO,, AgN03, and LiNO,, while it is a regular triangle in all the other alkali nitrates, consistent with the observation of vibrational spectroscopy16and also with the X-ray data.15 The present (12) P. 0. Lowdin, Adu. Phys., 5, 1 (1956). (13) L. Pauling, “The Nature of the Chemical Bond”, Cornel1 University Press, Ithaca, New York, 1960. (14) E. Kordes, 2. Phys. Chern. 48, 91 (1941). (15) H. Ohno and K. Furukawa, J , Chern. SOC.,Faraday Trans. 1 , 7 4 , 297 (1978). (16) K. Suzuki and Y. Fukushima, 2. Naturforsch. A , 32,1438 (1977).
Nakamura et al.
NMR results are, however, not sufficient to resolve such details of the coordination of the respective cations to the nitrate ion. It is that the interionic distance decreases upon melting in most of ionic crystals, while the coordination number decreases only slightly from its crystalline value. The average rate of the decrease of the ionic distance is about 5% for alkali ch10rides.l~ If this is the case for TlC1, for which no data are available,ls we can estimate the change of the shift on fusion A6(Tm) from the approximated relation A6/60 z -Corn. The constant C, is estimated to be -3.3 A-1 from the slope of the straight line given in Figure 7, using the values of 2.96 A3 for d9and -20 X for 6,. Thus, we obtain A6(Tm) zi which gives a correct order of magnitude of the observed shifts in Table I, although other factors in eq 3 should be changed to some extent on fusion. The negative (paramagnetic) temperature coefficient of the shifts in the molten state can also be interpreted from the increase in the overlap by increasing the amplitude of thermal motion of the ions around the equlibrium position with increasing temperature. The anion polarization model due to the foreign cations with different ionic radii, adopted in the present analysis, has been found applicable to the analysis of the quadrupolar relaxation rates of 7Li in molten (Li-K)N0320and ( L ~ - C S ) N O ,mixtures. ~~ In summary, we concur with the conclusions of Hafner and Nachtrieb, that the molten chemical shifts observed in 205Tl+NMR in molten (TlM)X mixtures arise from the varying degrees of overlap of the wave functions of the cation and anion resulting from the competition of T1+ and M+ for polarization of an intervening X- ion. Registry No. ‘05Tl, 14280-49-0; TlN03, 10102-45-1; LiNO,, 7790-69-4;NaNO,, 7631-99-4;AgNO,, 7761-88-8;KNOB,7757-79-1; RbNO,, 13126-12-0; CsN03, 7789-18-6; TlCl, 7791-12-0; LiC1, 7447-41-8;AgC1,7783-90-6;NaCl, 7647-14-5;KCl, 7447-40-7;RbC1, 7791-11-9; CSCl, 7647-17-8. (17) Y. Waseda, “The Structure of Non-crystalline Materials”, McGraw-Hill, New York, 1980. (18) A larger decrease of both the interionic distance and coordination number has been reported for AgCl (J. Y. Derien and J. Dupuy, Phys. Chern. Liquids,5, 71 (1976)). (19) J. R. Tessman, A. H. Kahn, and W. Shockley, Phy. Reu., 92, 890 (1953). (20) D. Harold-Smith, J. Chem. Phys., 60, 1405 (1974). (21) Y. Nakamura, S.Shimokawa, and M. Shimoji, to be submitted
for publication.