Spectroscopic properties of K5Li2UF10

Chemical Physics 310 (2005) 239–248 www.elsevier.com/locate/chemphys

Spectroscopic properties of K5Li2UF10 M. Karbowiak

a,*

, Z. Gajek b, J. Dro_zd_zyn´ski

a

a

b

Faculty of Chemistry, University of Wrocław, ul. F. Joliot-Curie 14, 50-383 Wrocław, Poland W. Trzebiatowski Institute of Low Temperature and Structure Research, Polish Academy of Sciences, P.O. Box 1410, 50-950 Wrocław, Poland Received 30 June 2004; accepted 26 October 2004

Abstract A new uranium (III) fluoro-complex of the formula K5Li2UF10 has been synthesised and characterised by X-ray powder diffraction and electronic absorption spectra measurements. The compound crystallises in the orthorhombic system, space group Pnma, ˚ , V = 1121.89 A ˚ 3, Z = 4 and is isostructural with its K5Li2NdF10 and K5Li2LaF10 analowith a = 20.723, b = 7.809, c = 6.932 A gous. The absorption spectrum of a polycrystalline sample of K5Li2UF10 was recorded at 4.2 K in the 3500–45,000 cm
1 range and is discussed. The observed crystal-field levels were assigned and fitted to parameters of the simplified angular overlap model (AOM) and next to those of a semi-empirical Hamiltonian, which was representing the combined atomic and one-electron crystal-field interactions. The starting values of the AOM parameters were obtained from ab initio calculations. The analysis of the spectra enabled the assignment of 71 crystal-field levels of U3+ with a relatively small r.m.s. deviation of 37 cm
1. The total splitting of 714 cm
1 was calculated for the 4I9/2 ground multiplet.
2004 Elsevier B.V. All rights reserved.

1. Introduction Although in the scientific literature the identification of more than 150 uranium (III) compounds were reported [1], only a few fluoro-compounds of the ion could be so far prepared and characterized. Among them, the most thoroughly studied was uranium trifluoride, due to its stability in atmospheric conditions and a relatively easy synthesis by reduction of UF4 with metallic uranium [1]. Hitherto, magnetic susceptibility data are available for UF3 [2,3], UZrF7 and UZr2F11 [4] and absorption spectra for UF3 only, by means of the teflon disk technique [5]. Besides, the identification of a number of fluorouranates (III) in the MF–UF3 (M = Na, K, Rb, Cs) systems have been reported, but physico-chemical characterisations of the phases were limited to analyses of powder diffraction data [1]. Crystal-field investigations

have been for the most part performed for U3+ doped chloride single crystals [6–10]. Optical analyses of the ion in a fluorine surroundings were restricted to the MF2 (M = Ca, Sr and Ba) and LiYF4 host single crystals. From absorption and fluorescence measurements of the U3+:MF2 single crystals a number of energy levels could be derived [11] but so far, only for U3+:LiYF4 a complete crystal-field analysis has been performed [12]. This paper presents the preparation as well as a characterisation by X-ray powder diffraction, magnetic and spectroscopic methods of a new uranium (III) fluorocomplex of the formula K5Li2UF10. The crystal-field energy levels were determined from a low temperature absorption spectrum and have been analysed in terms of the AOM and parametric Hamiltonian models.

2. Experimental *

Corresponding author. Tel.: +48 71 3757304; fax: +48 71 3282348. E-mail address: [email protected] (M. Karbowiak). 0301-0104/$ - see front matter
2004 Elsevier B.V. All rights reserved. doi:10.1016/j.chemphys.2004.10.040

For the preparation of K5Li2UF10 a well ground stoichiometric mixture of KF, LiF and UF3 was put in a

240

M. Karbowiak et al. / Chemical Physics 310 (2005) 239–248

graphite crucible, which was next placed in a quartz tube and sealed under vacuum. The mixture was heated at 450 C for 2 days. UF3 was obtained by reduction of UF4 with metallic U powder, obtained by thermal decomposition of UH3. UF4 was precipitated in anhydrous methanol by dropping an excess of a NH4F solution to that of uranium tetrachloride. KF, LiF and UF4 were dried by heating under high dynamic vacuum at 400 C. X-ray powder diffraction measurements of the compound were carried out on a ‘‘Stoe Fast Powder Diffractometer’’, using Cu Ka radiation. The unit cell parameters were obtained by least-squares refinement of all observed reflections in the 10–60 2h range. For this purpose, the Crysfire Powder Indexing System and Chekcell Graphical Powder Indexing Cell and Spacegroup Assignment software were applied [13]. Absorption spectra of thin film of the compound were recorded on a Cary-50 UV–Vis–NIR spectrophotometer in the 3500–50,000 cm
1 range. A well ground mixture of the compound with some chlorinated naphthalene oil (Halowax, index of refraction 1.635) was placed between two quartz windows, approximately 1 cm in diameter, pressed to obtain a transparent layer, and placed in an Oxford Instrument model CF-1204 cryostat, mounted in the cell compartment of the spectrophotometer.

3. Phenomenological model for description of the electronic transitions Energy-level calculations were carried out by applying a phenomenological model, based on the effective Hamiltonian approach [14,15]. The eigenvectors and eigenvalues of the crystal-field levels were obtained by simultaneous diagonalization of the combined free-ion and crystal-field energy matrices. The applied Hamiltonian includes the following terms: ^ ¼H ^A þ H ^ CF : H

ð1Þ

^ A contains the spherically symmetric (atomic) parts of H ^ H and is defined as: X ^ SO þ aLð ^ A ¼ Eave þ ^ L ^ þ 1Þ H F k ðnf ; nf Þf^ k þ f5f A k¼2;4;6

^ 2 Þ þ cGðR ^ 7Þ þ þ bGðG þ

X j¼0;2;4

^j þ M jm

X

X

T i^ti

i¼2;3;4;6;7;8

P k^ pk ;

ð2Þ

parameters are associated with the two-body correction ˆ (R7) are Casimir operators for the G2 ˆ (G2) and G terms. G and R7 groups and Lˆ is the total orbital angular momentum. The three-particle configuration interactions are expressed by T iti ði ¼ 2; 3; 4; 6; 7; 8Þ, where Ti are parameters and ˆti are three-particle operators. The electrostatically correlated spin–orbit perturbation is represented by the Pk parameters and those of the spin–spin and spin-other-orbit relativistic corrections by the Mj parameters. The operators associated with these paramˆ j and pˆk, respectively. eters are designated by m The H CF term of the Hamiltonian represents the one electron crystal-field interactions and is for the Cs site symmetry defined as [14,16],  X X
ðkÞ q ðkÞ ^ CF ¼ H Bkq C ðkÞ Bk0 C 0 ; ð3Þ q þ ð
1Þ C
q þ k;q>0

k

where C ðkÞ q ðiÞ is the spherical tensor operator of rank k, which acts on all the i electrons in the open shell, X C ðkÞ C ðkÞ ð4Þ q ¼ q ðiÞ; i

Bkq

while are crystal-field parameters. For the Cs symmetry the crystal-field Hamiltonian includes 15 real Bkq parameters, out of which 14 are independent (see Table 2). The B21 parameter was therefore arbitrary set to zero. In the fitting procedure the full (364 · 364) SLJMJ matrix for the f3 electronic configuration was diagonalized. The calculations have been performed by applying the f-shell empirical programs written by Reid [17] and running on a PC under the Linux Mandrake operating system.

4. Results and discussion 4.1. Chemical properties K5Li2UF10 forms a fine crystalline redish-brown solid and is one of the most stable uranium (III) compounds, considerably less hygroscopic and sensitive to oxidation by atmospheric oxygen than, e.g., chloro- or bromocomplexes of uranium (III). Nevertheless, after a few hours of exposition to air it becomes green, due to hydrolysis and oxidation to uranium (IV). The compound is insoluble in methanol, while in water it undergoes decomposition within several hours. In concentrated HCl, it is soluble with the formation of the characteristic unstable UCln3
n ð3 < n 6 6Þ complex anions.

k¼2;4;6

where Eave is the spherically symmetric one-electron part of the Hamiltonian, F k(nf,nf) and f5f represent the radial integrals of the electrostatic and spin–orbit interactions, ˆ SO are the angular operators correspondwhile fˆk and A ing to these interactions, respectively. The a, b and c

4.2. X-ray powder diffraction analysis and crystal structure The powder pattern of K5Li2UF10 was indexed on the basis of an orthorombic cell with a = 20.723(1),

M. Karbowiak et al. / Chemical Physics 310 (2005) 239–248

241

Table 1 Observed and calculated d spacings and observed relative intensities for K5Li2UF10 hkl

dobs

dcalc

I/I0

hkl

dobs

dcalc

I/I0

210 201 011 211 410 401 020 411 002 202 221 601 212 611

6.259 5.775 5.185 4.643 4.325 4.155 3.906 3.670 3.467 3.290 3.232 3.090 3.031 2.875

6.235 5.763 5.181 4.636 4.317 4.151 3.903 3.665 3.467 3.288 3.232 3.092 3.030 2.874

84 100 91 12 22 11 30 36 64 24 98 70 60 36

421 701 620 222 602 031 231 422 013 431 820 622 413

2.835 2.719 2.584 2.511 2.442 2.423 2.369 2.316 2.263 2.200 2.159 2.076 2.041

2.843 2.723 2.587 2.515 2.447 2.436 2.371 2.318 2.252 2.205 2.158 2.073 2.038

18 34 35 92 60 90 20 42 18 40 18 19 56

˚ and V = 1121.89 A ˚ 3. The b = 7.809(2), c = 6.932(2) A observed and calculated d spacings and observed relative intensities are listed in Table 1. There are four molecules per unit cell. The calculated and pyknometrically measured densities are equal to 3.77 and 3.69 g cm
3, respectively. The diffraction data reveal, that the compound is isostructural with the analogous K5Li2NdF10 [18] and K5Li2LaF10 [19] fluorides, which have been assigned to the space group D16 2h
Pnma. Hence, one may presume that the characteristic feature of the crystal structure of K5Li2UF10 are also sheets perpendicular to the a-axis formed by isolated UF8 dodecahedra and LiF4 tetrahedra [19]. The U polyhedra are isolated from each other and one may assume, that the distance be˚, tween uranium ions should be not shorter than 6.72 A which is the Nd–Nd distance determined for K5Li2NdF10 [18]. The relatively large distance between adjacent uranium ions limits interion interactions and reduces the probability of cooperative effects induced by direct metal–metal interactions. The point symmetry for the central metal ion sites was determined to be Cs.

(a)

4000

6000

8000

10000

12000

14000

Energy (cm-1)

(b)

4.3. Absorption spectra and CF calculations 14000

A survey absorption spectrum of the compound is presented in Fig. 1. In the 4000–18,200 cm
1 range the spectrum consists of sharp and well separated absorption lines, which arise from intraconfigurational 5f3 ! 5f3 transitions from the lowest Stark component of the ground multiplet to excited states of the 2S + 1LJ multiplets. Above this spectral region, the 5f3 ! 5f3 transitions are obscured by strong and broad 5f3 ! 5f26d1 bands. Fig. 2 presents a high resolution spectrum in the 4I9/2 ! 2H9/2 and 4I9/2 ! 4F5/2 absorption ranges. As one may expect for an undiluted sample, the lines are somewhat broader, e.g., the line at 9425 cm
1 has a half width of 7 cm
1. The number of observed lines in the 5f3 ! 5f3 transition range do not exceeds the (2J + 1)/2 value, which is indicative for single

16000

18000

20000

22000

24000

26000

Energy (cm-1) Fig. 1. Survey absorption spectra of thin films of K5Li2UF10 recorded at 4.2 K.

U3+ sites in the crystal. As a result of the low site symmetry of the U3+ ion, the CF levels are non-degenerate, except of KramerÕs degeneracy and are not accompanied by distinct vibronic sidebands. Fig. 3 shows the absorption spectrum in the 4I9/2 ! 4I11/2 transition range recorded at 298 and 4.2 K. The indicated by arrows hot bands have been assigned as transitions from the second and the third excited component of the 4I9/2 ground multiplet to the two lowest components of the 4I11/2

242

M. Karbowiak et al. / Chemical Physics 310 (2005) 239–248 4

I9/2

4

*

F5/2

(b)

* 4

I9/2

2

*

H9/2

*

(a)

9400

9600

9800

10000

10200

10400

4200

10600

4400

Energy

Energy (cm-1) Fig. 2. A detailed view of the 4I9/2 ! 2H29/2 and 4I9/2 ! 4F5/2 absorption transitions at 4.2 K. Arrows indicate absorption lines assigned as transitions of U3+ and included in crystal-field calculations.

multiplet. On this basis one could establish the energies of the two lowest Stark components of the 4I9/2 ground multiplet at 153 and 295 cm
1. Since emission transitions are completely quenched in undiluted uranium (III) compounds, the energies of the two highest energy components of the ground multiplet could not have been determined experimentally. Figs. 4–6 present details of absorption spectra recorded in 11,000–12,500, 13,200–14,300 and 14,500– 18,500 cm
1 energy range, respectively. The lines assigned as absorption transitions of U3+ and included in crystal-field analysis are in Figs. 2–6 marked with arrows.

11000

11200

4600

11400

11600

4800

5000

(cm-1)

Fig. 3. The 4.2 K (a) and 298 K (b) absorption spectrum of K5Li2UF10 in the 4I9/2 ! 4I11/2 transition range. Asterisks indicate hot bands assigned as transitions from the second and third excited level of the ground 4I9/2 multiplet to the two lowest energy components of the 4 I11/2 multiplet.

In the 0–18,200 cm
1 range, the theory predicts 80 intraconfigurational f ! f transitions. From analysis of the absorption spectrum we could determine and include in the CF calculations as many as 71 experimental crystal-field levels, originating from 20 2S + 1LJ multiplets. Since, for U3+ ions at the low symmetry sites of polycrystalline K5Li2UF10 one cannot obtain information on the irreducible representation of the levels, they were assigned by adjusting the experimental wavenumbers with the calculated ones within a particular multiplet, in terms of increasing energies. In cases where the number of levels was smaller than that predicted by theory,

11800

12000

12200

12400

12600

Energy (cm-1) Fig. 4. Absorption spectra in the 4I9/2 ! (4I15/2 + 4G5/2 + 4S3/2 + 4F7/2) transition range of K5Li2UF10 at 4.2 K. Arrows indicate absorption lines assigned as transitions of U3+ and included in crystal-field calculations.

M. Karbowiak et al. / Chemical Physics 310 (2005) 239–248

13200

13400

13600

13800

Energy 4

14000

243

14200

14400

(cm-1)

4

Fig. 5. Absorption spectra in the I9/2 ! G7/2 transition range of K5Li2UF10 at 4.2 K. Arrows indicate absorption lines assigned as transitions of U3+ and included in crystal-field calculations.

the experimental levels were assigned to the nearest calculated values. To the best of our knowledge, so far CF calculations of any lanthanide or actinide ion in compounds of the K5Li2MF10 type, have not been performed. Thus, due to the absence of reference data for initial values of the CF parameters, we have utilized those obtained from ab initio calculations. For this purpose we have applied a perturbation approach, comprising leading terms of direct electrostatic, exchange, multipole polarization, and inter-ionic renormalization contributions. This type of calculation has already been employed, among others, in our previous papers [7,8,20]. The

14500

15000

15500

16000

details and reliability of the method is discussed elsewhere [16,21–24]. Owing to the lack of symmetry at the U3+ site in the K5Li2UF10 compound as many as 14 independent parameters are describing the CF potential. A leastsquares procedure with so many parameters may lead to a false minimum of no physical meaning. In order to reduce the space of possible solutions we have applied one of the commonly accepted among the simplified phenomenological CF models – the angular overlap model (AOM), known to be especially efficient in cases of low symmetry systems [7,20,23]. In this approximation, the whole CF effect is described by the three er, ep and

16500

17000

17500

18000

18500

Energy (cm-1) Fig. 6. Absorption spectra of K5Li2UF10 at 4.2 K. Arrows indicate absorption lines assigned as transitions of U3+ and included in crystal-field calculations.

244

M. Karbowiak et al. / Chemical Physics 310 (2005) 239–248

ed AOM parameters, only. Thus, in the early phase of calculations, the three AOM parameters with the initial values of er = 800, ep = 400 and ed = 100 cm
1, were freely varied. In addition the free-ion Fk and f5f parameters were also varied. As initial values for the free-ion parameters were taken those obtained for U3+:LaCl3 [9], while the remaining parameters of the Hamiltonian (Eq. (2)) were fixed at values characteristic for the U3+ ion [25]. The final values of the AOM parameters were equal to er = 1192 (±66), ep = 403 (±67) and ed = 28 (±53) cm
1. During the fitting procedure the AOM parameters were stable and except of the last one, they were determined with relatively small errors. A small value for ed was expected, since this parameter has been proved to be lattice-dependant and usually less important [23]. One may state that the AOM model has described the energy levels structure relatively well. The r.m.s error was 55 cm
1 and differences between experimental and calculated energy values were not larger than 100 cm
1. The next stages of the fitting procedure were performed with the application of the Bkq crystalfield parameters. The obtained, final values of the AOM parameters, were used for the determination of the initial values of the Bkq parameters by applying the equation [16]: X l Bkq ¼ W kq el ; ð5Þ l¼0;1;2

where W

l kq

 
1 3 k 3 2k þ 1 l ¼ ð
1Þ ð2
dl0 Þ 7 0 0 0    al 3 k 3 X k  R  C q ðHt ; Ut Þ Rt
l 0 l t

ð6Þ

are the coefficients determined by the ligand coordinates Rt, Ht, Ut and the power exponents al are characterizing the distance dependence of the intrinsic parameters for the U3+–F
linear ligator. We accepted the values of al from [23]. The el in Eq. (6) is defined for the average metal–ligand distance R. In the last stage of the fitting procedure fourteen Bkq crystal field parameters as well as the Fk, f5f, a, b and P2 free-ion parameters were freely varied and simultaneously determined. In this step of calculations the AOM parameters have served as a constraint on the physical reasonableness of the Bkq parameters, such as the basic features of the metal–ligand bond [16,23]. The final parameter values are shown in Table 2, whereas the experimental and calculated energy levels are listed in Table 3. The final r.m.s. deviation (37 cm
1) is relatively small as for an actinide ion in a strong crystal field. For comparison, in the CF analysis of U3+:LiYF4 forty energy levels were fitted to six Bkq parameters with a r.m.s. deviation of 50 cm1 [12]. The largest difference (82 cm
1) between the experimental and calculated CF

Table 2 Free-ion and crystal-field parameters for K5Li2UF10 (all values in cm
1, except n) Parameter Eave F2 F4 F6 a b c f5f P2 B20 B22 B40 B41 B42 B43 B44 B60 B61 B62 B63 B64 B65 ad ne

AOMa

705 139
482 185 30 1699 2550 172 964
330 843 157
642

Fitted value

bc

19,752 (25) 38,407 (75) 34,185 (102) 20,031 (95) 27.4 (6.6)
1005 (47) [1317] 1627 (15) 1618 (76) 922 (79) 783 (60)
750 (106) 26 (122) 41 (135) 1424 (149) 2463 (116) 241 (138) 321 (97)
49 (148) 1293 (125) 631 (165)
114 (148) 37 71

a

The Bkq parameter values calculated from the AOM parameters: er = 1192, ep = 403 and ed = 28 cm
1. b The Ti parameters were during a fitting procedure kept at the constant values: T2 = 306, T3 = 42, T4 = 188, T6 =
242, T7 = 447 and T8 = 300 cm
1. The Mj as well as P4 and P6 parameters were constrained by the Hartree–Fock determined fixed ratios: M0 = 0.67, M2 = 0.55M0, M4 = 0.38M0, P4 = 0.5P2, P6 = 0.1P2. c Numbers in parentheses indicate errors in determination of the parameter values. P d Root mean square deviation: r ¼ i ½ðDi Þ2 =ðn
pÞ1=2 , where Di is the difference between the observed and calculated energies, n is the number of levels fitted and p is the number of parameters freely varied. e Number of levels included in the fitting procedure.

level energy values has been found for the second component of the 4G7/2 multiplet. The calculated splitting of the 2H29/2 multiplet is equal to 463 cm
1, which is significantly lower than the experimentally observed value of 601 cm
1. However, in an one-electron Hamiltonian model, the splitting value of this mutiplet is usually underestimated. A similar tendency has been noticed for U3+:LaCl3, where the calculated splitting value is more than by 100 cm
1 [9] smaller than the experimental one whereas after inclusion of the two-particle correlation crystal-field (CCF) operators into the semiempirical Hamiltonian, a considerable improvement in the fitting results could be noticed. However, in the case of K5Li2UF10 this procedure may be not applied, as the number of one-electron CF parameters is so large, that the inclusion of additional adjustable parameters would make the calculations untracteable. The values of the obtained free-ion parameters are typical for the U3+ ion. The Bkq parameters retain stable

M. Karbowiak et al. / Chemical Physics 310 (2005) 239–248

245

Table 3 Calculated and experimental energy levels for K5Li2UF10 2S + 1

LJa

4

I9/2

4

I11/2

4

F3/2

4

I13/2

2

H29/2

4

F5/2

4

I15/2 + 4S3/2 + 4G5/2 + 4F7/2

4

G7/2

4

F9/2

2

H211/2

2

K13/2 + 4D3/2

Eigenvectorsb 4

2

84 I9/2 + 14 H29/2 85 4I9/2 + 112H29/2 834I9/2 + 142H29/2 814I9/2 + 132H29/2 804I9/2 + 142H29/2 914I11/2 + 42H211/2 924I11/2 + 22H211/2 924I11/2 + 32H211/2 944I11/2 + 32H211/2 944I11/2 + 32H211/2 924I11/2 + 52H211/2 604F3/2 + 222D13/2 564F3/2 + 212D13/2 864I11/2 + 62K13/2 864I11/2 + 52K13/2 864I11/2 + 72K13/2 874I11/2 + 42K13/2 874I11/2 + 72K13/2 904I11/2 + 62K13/2 904I11/2 + 72K13/2 282H29/2 + 182G19/2 282H29/2 + 152G19/2 312H29/2 + 162G19/2 292H29/2 + 152G19/2 332H29/2 + 162G19/2 554F5/2 + 224G5/2 634F5/2 + 124G5/2 584F5/2 + 84G5/2 544G5/2 + 194F5/2 274G5/2 + 164I15/2 494I15/2 + 124S3/2 344I15/2 + 84F7/2 434I15/2 + 134F7/2 264I15/2 + 184F7/2 384I15/2 + 164F7/2 264I15/2 + 154G7/2 434I15/2 + 134F7/2 344I15/2 + 194F7/2 574I15/2 + 114F7/2 454I15/2 + 84F7/2 264G5/2 + 184I15/2 474I15/2 + 214F7/2 344I15/2 + 174F7/2 624I15/2 + 112K15/2 424I15/2 + 204F7/2 674G7/2 + 174F7/2 654G7/2 + 164F7/2 624G7/2 + 214F7/2 594G7/2 + 204F7/2 574F9/2 + 212H29/2 494F9/2 + 162H29/2 504F9/2 + 202H29/2 454F9/2 + 212H29/2 484F9/2 + 202H29/2 272H211/2 + 362K13/2 372H211/2 + 172K13/2 362H211/2 + 164G11/2 212H211/2 + 352K13/2 262H211/2 + 382K13/2 382H211/2 + 164G11/2 492K13/2 + 202H211/2 562K13/2 + 152H211/2 212K13/2 + 362H211/2

Calculated energy (cm
1)

Experimental energy (cm
1)

12 165 337 584 715 4426 4522 4564 4669 4925 4947 7222 7372 8124 8219 8266 8386 8493 8785 8801 9499 9613 9702 9811 9962 10,198 10,401 10,490 11,121 11,270 11,316 11,364 11,514 11,601 11,641 11,720 11,803 11,872 12,064 12,139 12,246 12,310 12,374 12,448 12,548 13,628 13,726 13,900 13,989 14,793 14,867 14,960 15,051 15,111 15,296 15,465 15,581 15,636 15,698 15,737 15,830 15,867 15,925

0 153 295 – – 4464 4507 4577 4615 4950 4980 7220 7430 8147 8223 8261 8417 8473 8775 – 9425 9614 9701 9828 10026 – 10,433 10,461 11,158 11,279 11,309 11,357 11,507 11,581 11,644 11,773 11,807 11,848 12,071 12,101 12,231 – 12,380 12,430 – 13,601 13,644 13,933 13,973 14,786 14,851 14,956 14,980 15,096 15,316 15,465 15,547 – 15,702 15,768 15,841 15,890 15,912

Eexp
Ecalc (cm
1)
12
12
42 – – 38
15 13
54 25 33
2 58 23 4
5 31
20
10 –
74 1
1 17 64 – 32
29 36 9
7
7
7
20 3 53 4
24 7
38
15 – 6
18 –
27
82 33
16
7
16
4
71
15 20 0
34 – 4 31 11 22
13 (continued on next page)

246

M. Karbowiak et al. / Chemical Physics 310 (2005) 239–248

Table 3 (continued) 2S + 1

4

LJa

4

D1/2 G9/2 + 2G17/2

4

D5/2 + 2L15/2 a b

Eigenvectorsb

Calculated energy (cm
1)

Experimental energy (cm
1)

Eexp
Ecalc (cm
1)

252K13/2 + 352H211/2 144D3/2 + 132H211/2 124D3/2 + 134H211/2 702K13/2 + 64I13/2 722K13/2 + 54I13/2 692K13/2 + 44I13/2 354D1/2 + 332P1/2 574G9/2 + 62G17/2 594G9/2 + 42G17/2 374G9/2 + 122G17/2 464G9/2 + 92G17/2 334G9/2 + 162G17/2 594G9/2 + 82G17/2 264G9/2 + 172G17/2 252G17/2 + 164G9/2 262G17/2 + 154F7/2 314D5/2 + 162D25/2 282L15/2 + 222K15/2

16,020 16,071 16,194 16,256 16,474 16,535 16,869 17,082 17,206 17,292 17,346 17,469 17,516 17,550 17,671 17,779 17,870 18,152

16,025 16,095 16,170 16,219 16,430 16,545 16,832 17081 – 17,347 17,386 17,495 17,530 17,561 17,685 – 17,839 18,197

5 24
24
37
44 10
37
1 – 55 40 26 14 11 14 –
32 45

Nominal quantum numbers for the atomic state associated with group. The leading eigenvector components, in percentage.

in the fitting procedure and are in general accordance with the calculated ones on the basis of the er, ep and ed parameters, obtained from the AOM step. Therefore, we may conclude, that the achieved values retain their physical meaning and that we have avoided in the calculations a pitfall of a local minimum. Since, the errors in the determination of those of the Bkq parameters which are displaying small values, i.e., B41 , B42 and B62 , are larger than the parameter values itself, one ought to consider the fixing of the values, either at zero or at those resulting from the AOM step. However, the value obtained from the fitting procedure for B42 parameter corresponds well with the calculated one from the AOM step, and should be therefore regarded as a well determined, in spite of the relatively large error. The absolute values of the B41 and B62 parameters are lower than those calculated in AOM step, but the inclusion of these parameters in the fitting procedure has been found important for the stability of the overall fit and they were therefore freely varied in the adjustment, also. The total splitting of the 4I9/2 ground multiplet could not be experimentally determined from emission spectra, since these cannot be recorded for an undiluted sample. However, as has been shown in [7], a relatively accurate determination of a possible large number of CF level results also in a relatively accurate determination of the energy level structure of the ground multiplet. Furthermore, for K5Li2UF10 only the two highest energy levels of the 4I9/2 multiplet could not have been determined experimentally. The two lowest energy levels of the ground multiplet could be determined from the observation of hot absorption bands. The calculated value of the total splitting is 715 cm
1 which is, as expected, lar-

ger than those reported for U3+ doped in chloride host crystals for which the CF is weaker (451 cm
1 for U3+:LaCl3 [9] and 626 cm
1 for U3+:Cs2NaYCl6 [26]). In fluoride matrices the total splitting of the ground level of U3+ ions have been determined for CaF2, SrF2 and LiYF4 crystals, and are equal to 667, 460 and 1113 cm
1, respectively [11,12]. The splitting of the ground multiplet can be directly related to the magnitude of the total crystal-field strength, which may be expressed by the scalar parameter Nv, defined as [27], " #1=2 X 4p k 2 ðBq Þ : ð7Þ Nv ¼ ð2k þ 1Þ k;q For K5Li2UF10, Nv = 4334 cm
1, while for U :LiYF4, Nv = 5529 cm
1 [12]. The difference in the crystal-field strength results mainly from differences in U–F distances in both compounds. In LiYF4, the U3+ ions substitute for Y3+ and the Y–Cl distances are equal ˚ [28]. For K5Li2UF10 the to 2.244(4x) and 2.297(4x) A U–F distances have not been determined, but assuming that they are similar to those of La–F in K5Li2LaF10 (the ionic radius of La3+ is very close to that of U3+) one may expect that they adopt values between ˚ [19]. 2.30–2.48 A It has been shown [27] that one may expect a linear dependence between the scalar crystal-field strength parameter (Nv) and the maximum Stark splitting (DEmax) of J-terms with small J-mixing. It is the case of the ground multiplet. Fig. 7 shows the maximum splitting of the 4I92 ground multiplet of U3+ ions in hitherto studied hosts, versus the Nv parameter. In this figure, the dash line presents a linear fit to all experi3+

M. Karbowiak et al. / Chemical Physics 310 (2005) 239–248

247

LiYF4

1200

Cs2NaYCl6

Cs3Lu2Cl9

K5Li2UF10 CsCdBr3

Ba2YCl7

RbY2Cl7 (site 2) Cs3Y2I9

K2LaCl5

K2LaBr5

400

LaCl3

600

RbY2Cl7 (site 1)

800

K2LaI5

∆Εmax(4I9/2)(cm-1)

1000

200 2000

3000

4000

5000

6000

Nv(cm-1) Fig. 7. Maximum Stark splitting of the 4I9/2 ground multiplet versus Nv parameter for the U3+ ion in different matrices. The dash line resulted from the linear fit to all experimental points, whereas the solid line presents the best linear fit obtained for chloride hosts, only.

mental points. As one may notice, the experimental points do not match very well the linear dependence and the most significant deviations from the predicted trend are observed for hosts with relatively large CF strength. If one takes into account the chloride hosts only, the variation of the Nv parameter with DEmax (4I9/2) may be well approximated by the linear dependence, shown with the solid line in Fig. 7. The experimental points for fluoride hosts are located above, and those for bromide and iodide hosts below the line determined for chlorides, and the deviations are larger for iodides than for bromides. This may indicate that the predicted linear relation is obeyed only for hosts with a similar covalency. However, due to a limited experimental data set additional analyses are required in order to state whether the observed regularity is not an accidental trend. 5. Summary This paper presents the synthesis as well as the results of X-ray powder diffraction, magnetic and spectroscopic investigation, of a new uranium (III) fluoro-complex of the formula K5Li2UF10. The analysis of low temperature absorption spectra allowed the assignment of 70 observed 5f3 ! 5f3 transitions. For the determination of the initial values of the Bkq parameters, the AOM model has been an useful tool. The electrostatic, spin–orbit, two-body correlation atomic parameter as well as fourteen Bkq crystal-field parameters have been determined from a least-squares fit of the calculated energy level values to those of experimental ones. For the 4I9/2 ground multiplet a total splitting of 714 cm
1 has been established.

Acknowledgements The authors thank Dr. J. Janczak of the W. Trzebiatowski Institute of Low Temperature and Structure Research of the Polish Academy of Sciences in Wrocław, for the X-ray powder diffraction measurements as well Dr. Michael Reid (University of Canterbury, New Zealand) for providing his f-shell programs.

References [1] J. Dro_zd_zyn´ski, Chemistry of tervalent uranium, in: A.J. Freeman, C. Keller (Eds.), Handbook on the Physics and Chemistry of the Actinides, vol. 6, Elsevier Science Publishers B.V., Amsterdam, 1991, pp. 282–336. [2] M. Berger, M.I. Sienko, Inorg. Chem. 6 (1967) 324. [3] H. Nguyen-Nghi, A.J. Dianoux, H. Marquet-Ellis, Compt. Rend. 259 (1964) 811. [4] G. Fonteneau, J. Lucas, J. Inorg. Nucl. Chem. 36 (1974) 1515. [5] H. Schmieder, E. Dornberger, B. Kanelakopulos, Appl. Spectrosc. 24 (1970) 499. [6] M. Karbowiak, J. Dro_zd_zyn`ski, K.M. Murdoch, N.M. Edelstein, S. Hubert, J. Chem. Phys. 106 (1997). [7] M. Karbowiak, Z. Gajek, J. Dro_zd_zyn´ski, Chem. Phys. 261 (2000) 301. [8] M. Karbowiak, N. Edelstein, Z. Gajek, J. Dro_zd_zyn´ski, Spectrochim. Acta A 54 (1998) 2035. [9] M. Karbowiak, J. Dro_zd_zyn´ski, M. Sobczyk, J. Chem. Phys. 117 (2002) 2800. [10] M. Karbowiak, A. Mech, Z. Gajek, N. Edelstein, J. Dro_zd_zyn´ski, New J. Chem. 26 (2002) 1651. [11] W.A. Hargreaves, Phys. Rev. 156 (1967) 331. [12] E. Simoni, M. Louis, J.Y. Gesland, S. Hubert, J. Lumin. 65 (1995) 153. [13] Collaborative Computational Project Number 14 for Single Crystal and Powder Diffraction CCP14, EPSRC Research Computing Initiative. Available from: .

248

M. Karbowiak et al. / Chemical Physics 310 (2005) 239–248

[14] B.G. Wybourne, Spectroscopic Properties of Rare Earths, Interscience, New York, 1965. [15] W.T. Carnall, H. Crosswhite, H.M. Crosswhite, J.P. Hessler, N.M. Edelstein, J.G. Conway, G.V. Shalimoff, R. Sarup, J. Chem. Phys. 72 (1980) 5089. [16] J. Mulak, Z. Gajek, The Effective Crystal Field Potential, Elsevier, Amsterdam, 2000. [17] M.F. Reid, University of Canterburry, New Zealand (private information). [18] B.C. McCollumn, A. Lempicki, Mater. Res. Bull. 13 (1978) 883. [19] G. Dominiak-Dzik, S. Goła˛b, M. Bałuka, A. Pietraszko, K. Hermanowicz, J. Phys.: Condens. Matter 11 (1999) 5245.

[20] M. Karbowiak, J. Dro_zd_zyn`ski, Z. Gajek, J. Alloys. Comp. 323– 324 (2001) 678. [21] D.J. Newman, Adv. Phys. 20 (1971) 197. [22] D. Garcia, M. Faucher, J. Chem. Phys. 82 (1985) 5554. [23] Z. Gajek, J. Mulak, J. Phys.: Condens. Matter 4 (1992) 427. [24] Z. Gajek, J. Alloys.Comp. 219 (1995) 238. [25] W.T. Carnall, A Systematic Analysis of Trivalent Actinides Chlorides in D3h site symmetry, in ANL Report 89/39, Argone National Laboratory, 1989. [26] M. Karbowiak, J. Dro_zd_zyn`ski, S. Hubert, E. Simoni, W. Streßk, J. Chem. Phys. 108 (1998) 10181. [27] F. Auzel, O.L. Malta, J. Phys. 44 (1983) 201. [28] E. Garcia, R.R. Ryan, Acta Cryst. C 39 (1993) 2053.