ELECTRONIC SUPPLEMENTARY INFORMATION

Electronic Supplementary Material (ESI) for Journal of Materials Chemistry C. This journal is The Royal Society of Chemistry 17 Design of Yb 3+ optical bandwidths by crystallographic modification of disordered Calcium Niobium Gallium laser garnets M. D. Serrano, J. O. Álvarez-Pérez, C. Zaldo, J. Sanz, I. Sobrados, J. A. Alonso, C. Cascales*, M. T. Fernández-Díaz and A. Jezowsi ELECTRONIC SUPPLEMENTARY INFORMATION

CZOCHRALSKI GROWTH OF CaNbGa GARNET CRYSTALS Figure ESI.1. Single crystals grown by the Czochralsi method. (a) Undoped CNGG, (b) 11.at% Yb:CNGG, (c) undoped CLNGG ( parallel sides polished), crystal grown with 99.5% purity CaCO 3, and (d) 8at% Yb:CLNGG ( parallel sides polished), crystal grown with 99.99% purity CaCO 3. The scale in the figures is in centimeters.

CRYSTALLOGRAPHIC INFORMATION Table ESI.1 Structure refinement details and crystal data at 9± K for CLNGG and 8at% Yb:CLNGG crystals by using single crystal X-ray diffraction data. Wavelength (Å) Crystal system, space group, Z Unit cell dimension (Å), Volume (Å 3 ) Calculated density (Mg/m 3 ) Absorption coefficient (mm -1 ) Crystal size (mm 3 ) range for data collection (deg) Limiting indices Reflections collected Independent reflections Completeness to, % Absorption correction Refinement method Goodness-of-fit on F R indices [I>(I)], R1, wr R indices (all data), R1 wr Extinction coefficient Largest diff. pea and hole (eå -3 ) CLNGG a.7173 Cubic, Ia-3d (No. 3), 8 1.5(1), 195.13(3).57 11.79..1.1 3.99 to 3. -17h17, -1717, -17l17 397 5 [R(int =.8] 3., 1 Multi-scan SADABS Full-matix least squares on F 1.11.11.1.135..13(). and -.58 8at% Yb:CLNGG b 1.85(1), 19.93(3).911 1.15.3.1.11. to 3.38 -h, -17, 18l 5183 [R(int =.89] 3.38, 1 1.1.1.71.9.9.9(7).3 and -.3 a The total exposure time for data collection was 5.5 h. The frames were integrated with the Bruer SAINT software pacage using a narrow-frame algorithm. The starting cell constant a is derived upon the refinement of the XYZ-centroids of 718 reflections I> (I) with 7.98 deg < <.5 deg. b The total exposure time for data collection was 3.1 h. The frames were integrated with the Bruer SAINT software pacage using a narrow-frame algorithm. The starting cell constant a is derived upon the refinement of the XYZ-centroids of 98 reflections I> (I) with 7.997 deg < < 7. deg. Table ESI. Atomic coordinates (1 ), occupancy factor, OF, and equivalent isotropic displacement parameters U(eq) (Å 1 3 ) for CLNGG and 8at% Yb:CLNGG crystals from scxrd data. U(eq) is defined as one third of the trace of the orthogonalized Uij tensor. Note that atomic coordinates for O atoms are given with one order lower accuracy than those obtained from pnd data (Table in the main text). Ion site CLNGG 8at% Yb:CLNGG x y z OF U(eq) x y z OF U(eq) Ca Yb Nb(1) Ga(1) Ga() Nb() c c 1a 1a d d - 5-5 5 15-875 875.98() -.(3).3().81().5() 9(1) - (1) (1) (1) (1) 5 5 5 5 15 15 875 875.937(1).3(1).7(3).35(3).77().88() 8(1) 8(1) 5(1) 5(1) (1) (1) O 9h 5(1) 175(1) 995(1) 9(1) 51(1) 178(1) 99(1) 8(1) 3

HYPOTHETICAL STRUCTURAL MODELS FOR Li DISTRIBUTION IN THE CNGG-TYPE GARNET STRUCTURE: RESULTS OF CORRESPONDING PRELIMINARY SINGLE CRYSTAL XRD REFINEMENTS FOR THE CLNGG CRYSTAL Taing into account the used amounts of reagents for preparing the CLNGG crystal (before the synthesis of the garnet phase, see the main text),.75 of Li per CLNGG formula will be considered as the starting quantity of Li in next structural refinements. Since unfortunately the neutron bound coherent scattering lengths (in fm) of Nb and Ga are very similar, 7.5 and 7.88 respectively, and thus any reliable refinement of their relative populations over a same crystal site is hampered using neutron diffraction data, different models for Li distribution in the CNGG-type garnet structure were preliminarily evaluated by refinements using single crystal XRD data with the SHELXTL program. Three hypothetical structure models were considered: A) Li located only in octahedral 1a sites, B) Li located in both octahedral 1a and tetrahedral d sites, and C) Li located only in tetrahedral d sites. In A and B models the starting population factors of Li in 1a sites were.1375 and.88, respectively, with a further Li starting population factor of.58 in d for B model (i.e. 5% of nominal Li in each ind of site). If no restraint over the population of 1a sites is applied (to test if 1a sites are fully occupied by Nb, Ga and Li cations or if there are vacancies), the refinements were unstable and they do not converge. On the other hand, when the restraint over the population of 1a is imposed (i.e. the site 1a is considered to be fully occupied by Nb, Ga and Li, with no octahedral vacancies), the refinement of the A model yields cationic populations of Nb.3±.3, Ga.39±. and Li -.±.1 over 1a sites, with R 1 =.117, being the populations of Ca over dodecahedral c, and of Ga and Nb over tetrahedral d almost the same that those in the scxrd refinement previously reported (with results in Table ESI.). The slightly negative Li population obtained from refinement of the A model indicates that Li is not located in octahedral 1a sites, and in fact after removing Li of 1a sites, the refinement converges to the Table ESI. population values over c, 1a and d sites, with about 1% of vacancies over tetrahedral d sites. In the refinement of the B model with 1a sites restrained to be fully occupied, after some cycles that yield populations of Nb.59±., Ga.3±.3 and Li -.3±. over 1a sites, the refinement became unstable and it does not converge. On the other hand, any scxrd refinement with all nominal Li only in tetrahedral d sites, the C model, has no physical meaning since previous (E. Castellano-Hernández et al., Cryst. Growth Des. 1, 1, 18) and current refinements of CNGG-type garnets reveal the existence of an important level of vacancies in these d sites, indistinguishable of Li in scxrd refinements. A more realistic approach for analyzing A, B and C hypothetical structure models was based on the analysis of vacancies over 1a and d sites, in such a way that if the

refinement yields vacancies, these latter should be an indication of the possible Li presence in these sites. Considering again the model A, only Nb and Ga were located in 1a sites, and no restraint is applied over the population of these sites. The resulting 1a populations were Nb.5±. and Ga.3±., and R 1 =.119, with populations of Ca over dodecahedral c and Ga and Nb (and vacancies) over tetrahedral d sites nearly the same as in Table ESI.. Results of the refinement indicate, once again, that the octahedral 1a sites are fully occupied by Nb and Ga, with no vacancies, thus no Li is located in 1a sites. In the model B, with Nb and Ga located in 1a and in d sites, and no restraints applied over the population of both inds of sites, the refinement yielded population factors of Nb.±.3 and Ga.±. in 1a sites, and Ga.83±. and Nb.±. in d sites, R 1 =.11, that is, a full Nb+Ga occupation of octahedral 1a sites, thus excluding the location of Li in these sites, while up to 13% of vacancies over tetrahedral sites are disclosed. Some of these vacancies may correspond to Li cations in these d positions, thus it seems clear that the only possibility for Li location is that of model C, which has been then tested with pnd data, as indicated in the main text. In summary, every of the attempted scxrd refinements to locate Li has provided the same clear indication about the CLNGG crystal structure: the 1a octahedral positions are fully occupied by Nb and Ga, in a ratio of about :1, and only about 87% of d tetrahedral sites are occupied by Ga (predominantly) and by Nb (a very small amount) being the remaining 13% either vacant or occupied by Li, as it has been tested and demonstrated by the current refinement using pnd data presented in the main text. 5

OPTICAL PROPERTIES (cm -1 ) 15 1 5 a) (cm -1 ) 15 1 5 99.99% purity CaCO 3 85 9 95 1 15 (nm) 99.5% purity CaCO 3 (cm -1 ) Li 3K K 3 5 7 8 9 1 11 CNGG b) 7 8 9 3 31 3 (cm -1 ),, c) (1) () (3) () (5), 55 5 7 75 8 Figure ESI. a) General view of the RT optical absorption of 8at% Yb:CLNGG prepared with 99.5% purity CaCO 3 (blac line) showing the bandgap absorption in the ultraviolet (< 9 nm), a broad pre-edge absorption from = 5 nm to the absorption bandgap edge, absorption bands associated with crystal coloration (= 57-8 nm) and Yb 3+ absorption (= 85-15 nm). The 8at%Yb:CLNGG crystal grown with 99.99% CaCO 3 (blue line) is free of pre-edge absorption. b) Detailed comparison of the ultraviolet absorption edge of undoped CLNGG ( µm thic plate) and CNGG (8 µm thic plate) crystals: at 3 K, CLNGG (blac line), CNGG (red line), and at K, CLNGG (blue line), CNGG (magenta line). (c) RT optical absorption of CNGG-type crystals leading to greenish coloration: (1) 8at%Yb:CMNGG modified with extra 7.5% (with respect to Nb) of Mg. ().3at%Yb:CNGG. (3) Undoped CLNGG. () 8 at% Yb:CLNGG grown with 99.5% CaCO 3. (5) 8 at% Yb:CLNGG grown with 99.99% CaCO 3.

REFRACTIVE INDEX,,1,1,8,, 1,9 1,9 CLNGG CLNGG Sellmeier parameters A=,15 B=1,581 C=19,817 nm D=1,3531 1-8 nm - CNGG 8 1 1 1 1 18 Figure ESI.3 Refractive index dispersion, n(λ), of undoped CLNGG single crystal ( ) and its fit (continuous line) to the Sellmeier law with the parameter set given in the inset. For comparison n() of undoped CNGG single crystal () has been also included. Yb 3+ F 5/ LIFETIME (s) 9 8 7 5 3 1 5 1 15 5 3 TEMPERATURE (K) Figure ESI. Thermal evolution of the F 5/ fluorescence lifetime of Yb 3+ in the.3at% Yb:CNGG crystal, EXC = 971 nm, EMI = 11 nm. CRYSTAL FIELD SIMULATIONS In the Simple Overlap Model (SOM) the crystal field parameters (CFPs) are calculated from the atomic positions in the structure. It is assumed that the interaction energy of trivalent lanthanide Ln 3+ in a chemical environment is produced by an electrostatic potential of charges uniformly distributed over small regions centered around the midpoint of the R L distance from Ln 3+ to the ligand L, R being the shortest distance. The charge in each region is proportional to the total overlap integral between the f and the s and p orbitals of the Ln 3+ and L, respectively. The CFPs are written as B q r L 1 L 1 A q (L), L R R L 3.5 [1]. 7

The sum over L is restricted to the first neighbors, i.e. over all ligands of the first coordination sphere, consequently the required crystallographic data are restricted to the closest ligand positions and thus r radial integrals are not corrected from the spatial expansion. A q is the lattice sum and it taes into account the symmetry properties of the Ln 3+ site, including the effective charge attributed to L. The sign of the denominator stands for differentiating the type of L: when a single type of L is considered, a minus sign corresponding to the normal shift of the charge barycenter from the mid-point of the R L bonding distance should be taen, and when different L are present the minus sign corresponds to the most covalent one. The SOMPLUS model, an extension of SOM, allows estimating CFPs from the crystallographic positions and distances of all atoms, oxygens and first sphere of nearest cations, around the central Ln 3+. Effects from different cations at a same crystal site arise by considering individual effective charges. Charge compensation and vacancies at a given site can be also taen into account in the SOMPLUS model. Table ESI.3 Effect of the substitution of one tetrahedral d Ga 3+ (at 3.1 Å) on the crystal field potential of c Yb 3+ in disordered CaNbGa garnets: SOMPLUS calculated crystal field parameters (cm -1 ). Effective charges:.,., 1.,., and 3.7 for Li +, Mg +, Ga 3+, Ge + and Nb 5+, respectively, and =.8. 1 T 1Li + T 1Mg + T 1Ga 3+ T 1Ge + T 1Nb 5+ T B - -9-597 -531-5 -3 B 1 595 555 7 379 55 B 15 13 8 599 588 573 B 11 13 17 3 7 B 38 31 18 1-18 B 9 95 9 99 73 77 B -58-58 -59 - -58 - B 5 7 9 51 7 B -1-1 -599-577 -515-31 ib 71 73 8 9 117 175 ib 8 7 3 17 3 ib -7-73 -8-9 -11-175 ib 5 11 13 17 39 ib -97-35 -331-37 -5-9 By using SOMPLUS, CFPs calculations for c Yb 3+ in disordered CaNbGa garnets tae into account the effect from nearest d sites when one Ga 3+ (at 3.1 Å from central c Yb 3+ ) is replaced by one vacancy T, or by one Li +, Mg +, Ge + or Nb 5+ cation. Table ESI.3 gives derived CFPs (real B q and complex ib q ) in each case, which 8

correspond to a local symmetry lower than the D symmetry of the YbO 8 dodecahedron in the garnet. Yb 3+ SPECTROSCOPY Figure ESI.5 shows the decomposition of the - Yb 3+ K OA in Mg-modified 8at%Yb:CMNGG crystal by following the procedures described in the Discussion section of the main text, with the following I o, E o, parameters: Ga 3+ - T.55, 17.8, 5.5 Ga 3+ -Mg +.18, 19.,.9, +.18, 183.5,.7, +.13, 17.3,. Ga 3+ -Ga 3+.91, 199.87,.8, +.15, 193.,.8, +.53, 18, 3.5 Ga 3+ -Nb 5+.35, 131,.5 OA K (cm -1 )..5..3..1. 8%Yb:CMNGG Mg contribution -.1 13 131 13 19 18 17 1 ENERGY (cm -1 ) Figure ESI.5 Fit of the K - Yb OA of a 8at%Yb:CMNGG crystal (3at% extra Mg with respect to Nb)., experimental results. Red, orange, blue and green lines are the contributions of the nearest tetrahedra (3.1 Å) to central Yb 3+ occupied by vacancy T, Mg +, Ga 3+ or Nb 5+, respectively. The blac line is the fit and the bottom gray line is the fit error. Table ESI. Probabilities P of Ga, Nb, vacancy ( T ), and Li cationic distributions on the two (n=) edge sharing nearest neighbors at 3.1Å d tetrahedral sites around a dodecahedral c Yb 3+ site in.3at%yb:cngg and 8at% Yb:CLNGG crystals..3at%yb:cngg 8at%Yb:CLNGG Ga:Nb: T P Ga:Nb:Li P :: 1:1: 1::1 :1:1 :: ::.7.1.1.1.1. :: 1::1 1:1: :: :1:1 ::.5.178.113..15.5 9

Table ESI. gives the probabilities P calculated according the description made in the main text of the more liely sharing edges tetrahedral cationic distributions around c Yb 3+ for.3at%yb:cngg and 8at% Yb:CLNGG crystals, using OF given in Table. From Table ESI. it is obvious that the most probable cationic configuration over the edge sharing tetrahedra, i.e. Ga 3+ -Ga 3+, is much more probable than the two next Ga 3+ - Nb 5+ or Ga 3+ - T, being probability of the rest one order of magnitude lower than the latter. Consequently, a large band and two minor ones are expected, that agrees the experimental result shown in Figure 13a. Figure ESI. shows the 3 K photoluminescence emission spectra of 8at%Yb:CNGG and 8at%Yb:CLNGG under excitation at different wavelengths inside the OA of Yb 3+. It is worth noting that the shape of the luminescence spectra is insensitive to the excitation wavelength. This corresponds to strong Yb-Yb energy transfer and thus the excitation of all coexisting Yb centers. PL INTENSITY NORMALIZED PL INTENSITY,,,,,3,,1, 8at%Yb:CNGG (a) EXC (nm) 9.3 97.5 97.9 971. 97. 973. 973.7 1 15 11 (b) EXC (nm) 9.3 97.5 97.9 971. 97. 973. 973.7 1 15 11 PL INTENSITY NORMALIZED PL INTENSITY..3..1..3..1. 8at%Yb:CLNGG (c) 1 15 11 (d) EXC (nm) 9.3 97.5 97.9 971. 97. 973. 973.7 1 15 11 EXC (nm) 9.3 97.5 97.9 971. 97. 973. 973.7 Figure ESI. 3 K luminescence spectra of Yb 3+ in CNGG-type crystals for different excitation wavelengths. (a,b) 8at%Yb:CNGG. (c,d) 8at%Yb:CLNGG. In the bottom row (b and d) the luminescence intensity was normalized to the integrated area. 1