M

1DtcalApttitl

smn(igI

2066 J. Opt. Soc. Am. B/Vol. 23, No. 10 /October 2006 Rico et al.

Polarization and local disorder effects on theproperties of Er3+-doped XBi„YO4…2, X=Li or Na

and Y=W or Mo, crystalline tunablelaser hosts

auricio Rico, Antonio Méndez-Blas, Vladimir Volkov, María Ángeles Monge, Concepción Cascales, and Carlos Zaldo

Consejo Superior de Investigaciones Científicas, Instituto de Ciencia de Materiales de Madrid,c/Sor Juana Inés de la Cruz 3, Madrid 28049, Spain

Andreas Kling

Instituto Tecnológico e Nuclear, Estrada Nacional No. 10. 2686-953 Sacavém and Centro de Física Nuclearda Universidade de Lisboa, Avenida Professor Gama Pinto 2, Lisbon 1649-003, Portugal

María Teresa Fernández-Díaz

Institut Laue Langevin, ILL Boîte Postale 156X, Grenoble 38042, France

Received December 19, 2005; revised May 3, 2006; accepted May 16, 2006; posted June 2, 2006 (Doc. ID 66670)

Pure and Er-doped ��Er�crystal�2.4�1020 cm−3�, NaBi�WO4�2 (NBW), NaBi�MoO4�2, and LiBi�MoO4�2 crystalshave been grown by the Czochralski method. The three crystal hosts have similar structural and optical prop-erties. The noncentrosymmetric space group I4̄ (No. 82) crystallographic structure has been establishedthrough 300 K single-crystal x-ray and neutron (for NBW only) diffraction. Er3+ energy levels were determinedexperimentally and simulated in the S4 symmetry through a crystal-field analysis. With this background, thelarge spectroscopic bandwidths observed were ascribed to the presence of two 2b and 2d sites for Er3+ and todifferent short-range Na+ and Bi3+ distributions around both sites. The radiative properties of Er3+ are de-scribed by the Judd–Ofelt theory achieving branching ratios and radiative lifetimes for transitions useful aslaser channels. The 4I13/2→ 4I15/2 laser channel ���1.5 �m� shows a peak emission cross-section �EMI ���1530 nm��0.5�10−20 cm2 and a quantum efficiency ��0.68 to 0.74. The laser emission is envisaged to betunable by ���100 nm. © 2006 Optical Society of America

OCIS codes: 140.3500, 160.3380, 160.4760, 160.5690, 300.1030, 300.2140.

TtsalwshN(sic

Dt“ccto

. INTRODUCTIONouble tungstate (DT) and molybdate (DM) crystals with

he chemical formula XT�YO4�2, where X is a monovalentation, T is a trivalent one (often a trivalent rare earth),nd Y=Mo6+ or W6+ are crystalline hosts accepting aarge concentration of laser active trivalent lanthanides.t temperatures close to their melting or decompositionoints most of these compounds show the tetragonal crys-alline structure of the scheelite, CaWO4, with the cen-rosymmetric space group (SG) I41/a (No. 88). Upon cool-ng to RT, phase transitions (polymorphicransformations) to other crystalline structures withower symmetry may occur.

DT and DM compounds with tetragonal crystallinetructure are found at RT when no polymorphic transfor-ation occurs upon cooling. In a limited number of cases,

amely, NaT�WO4�2 �T=Bi,La–Er�, NaBi�MoO4�2NBM), and KT�MoO4�2 �T=La–Nd�, a congruent meltings also found, and therefore the efficient Czochralski (Cz)rowth technique can be applied to obtain single crystals.n the tetragonal lattice, the monovalent X+ and trivalent

0740-3224/06/102066-13/$15.00 © 2

3+ ions randomly share the same cationic sublattice. Inhis case, a variety of crystal-field (CF) potentials on theites occupied by T ions occur, and therefore the opticalbsorption (OA) and emission bands of optically activeanthanides are broad. The large bandwidths associatedith the X /T disorder allow laser tunability in broad

pectral ranges. In fact, about 40 nm range of tunabilityas been recently reported for the Yb3+ laser emission inaGd�WO4�2 (Ref. 1), NaLa�WO4�2, and NaLa�MoO4�2

Ref. 2). This broad emission band allowed sub-100 fs la-er pulses in NaGd�WO4�2 by mode locking,

3 and it prom-ses even shorter ��50 fs� laser pulses by using pulseompression techniques.

These outstanding applications of disordered DT andM crystals have been recognized only recently, and

herefore the preparation and characterization of thesedisordered” hosts are little developed. In particular, Liompounds have been very scarcely investigated. In theontext of exploitation of crystals with large bandwidths,his work studies the crystal growth and the relationshipf crystal structure with optical properties of LiBi�MoO �

4 2

006 Optical Society of America

(cehc1csawtotep

2Pmn±

bmfslttptdlli

iTe=wot

tr

apScftecEederlawStlRM

todGtw4ae(a

bmbtm=tcFtTEtcpwRlupc=RcT

Rico et al. Vol. 23, No. 10 /October 2006 /J. Opt. Soc. Am. B 2067

LBM), and its doping with Er3+ in comparison to thelose-related NaBi�WO4�2 (NBW), and NBM crystals. Thefficiency of the NaBi-based DT and DM crystals as laserosts was shown for Nd3+ (Ref. 4). Er3+ has several laserhannels operating at RT. Of particular interest are the.5 �m 4I13/2→

4I15/2 laser emission for long-distance fiberommunications and the 2.9 �m 4I11/2→

4I13/2 laser emis-ion for medical applications, but other laser channelslso involve all multiplets below 4S3/2 (Ref. 4). Previousorks reported some information of Er3+ emission spec-

roscopy in NBW5,6 but ignoring the polarized characterf the transitions already known.7 Here, the Er3+ spec-roscopy is discussed in terms of the host disorder withxplicit consideration of its polarized character, and theerspectives for laser operation are investigated.

. EXPERIMENTAL TECHNIQUESroduct synthesis and crystal growth experiences wereade using Pt crucibles and a Si2Mo heating element fur-

ace. The temperature was set with a resolution of0.2°C using an Eurotherm 818P temperature controller.The composition analyses of the hosts have been made

y ion-beam techniques. Rutherford backscattering (RBS)easurements were performed using 1.8 MeV H+ ions

rom a 3.1 MV van de Graaff accelerator at Sacavém. Aurface barrier detector with a small solid angle �1.7 msr�ocated at 160° recorded the spectrum of the backscat-ered protons for the verification of the concentration ofhe crystal constituents other than Li. Moreover, the articles produced in the nuclear reaction between pro-ons and Li were detected by an annular surface barrieretector located at 180° with respect to the beam. Thearge solid angle �14 msr� allowed for the recording of-particle spectra with sufficient statistics despite the

ow-reaction cross section and atomic concentration of Lin the sample.

The Er concentration in the crystals was determined bynductively coupled plasma (ICP) analyses conducted on ahermo Jarrel As IRIS Advantage axial plasma spectrom-ter. For the Er analysis, the spectral line at �337.271 nm was selected, and the signal backgroundas established and discounted. For each crystal host,nly the most concentrated sample was analyzed by ICP;he Er concentration for other crystals was calculated by

Table 1. Er Concentration [Er] and SegregationCoefficient S of the Crystals Used in This Work

Crystal�Er�meltmol.%

�Er�crystalmol.% S

Ercrystal�1020 cm−3

NBW: Er3+ 0.02 0.03 1.5 0.020.44 1.47 3.3 0.920.88 3.85 4.3 2.41

NBM: Er3+ 0.16 1.2 7.5 0.750.33 1.68 5.1 1.050.73 3.45 4.7 2.15

LBM: Er3+ 0.03 0.11 3.7 0.070.16 0.88 5.5 0.560.3 1.22 4.0 0.78

he comparison of their integrated OA. Table 1 summa-izes the Er concentrations in the crystals used.

X-ray diffraction (XRD) analyses of pure NBW, NBM,nd LBM as well as Er-doped NBW crystals have beenerformed from data collected at RT by a SiemensMART CCD diffractometer equipped with a normal fo-us 3 kW sealed tube. Small prismatic single crystals cutrom each grown material were selected in order to reducehe possibility of twinning and to minimize the absorptionffect. Data were collected over a quadrant of the recipro-al space by a combination of three sets of exposures.ach set had a different angle for the crystal, and eachxposure of 20 s covered 0.3° in �. The crystal-to-detectoristance was 5.08 cm. For each crystal, unit cell param-ters were determined by a least-squares fit of about 35eflections with I�20��I�. Further details are describedater (see Table 3). Neutral-atom scattering factors for alltoms were used, and anomalous dispersion correctionsere applied.8 All calculations were performed using theHELXTL program,9 and ATOMS software for the struc-ure plots. Also, scan intensity measurements of some se-ected individual hkl reflections have been performed atT on an Enraf-Nonius CAD4 diffractometer, witho–K radiation.Very precise and accurate measurements of Bragg in-

ensities for a considerably more larger, �80 mm3 crystalf NBW were also performed in the hot neutron four-circleiffractometer D9, at the Institut Laue Langevin (ILL),renoble, France. The monochromator is a Cu crystal in

ransmission geometry using the (220) planes. The usedavelength was 0.84 Å and the data collection requireddays. Further details concerning the geometry, detector,

nd possible environments for the sample in the D9quipment can be consulted at the ILL web pagewww.ill.fr). Specific details of the current experiment arelso described later (see Table 3).For optical measurements, the crystals were oriented

y using Laue XRD patterns, cut, and polished with dia-ond paste. Refractive indices were measured at 300 K

y the minimum-deviation-angle method using prisms;he estimated error is ±0.002. OA measurements wereade in a Varian spectrophotometer model Cary 5E ��200 to 3000 nm�. The sample temperature was varied in

he 10–300 K range by using a He closed-cycle Oxfordryostat equipped with the proper temperature controller.or polarized measurements, the sample was inserted be-

ween a Glan–Taylor polarizer and a depolarizer sheet.he spectra are labeled as � �E�c ,H �c� or �E �c ,H�c�;and H are the electric and magnetic field components of

he light, respectively. Photoluminescence (PL) was ex-ited in the continuous mode with a Ti-sapphire laser, dis-ersed in a SPEX spectrometer �f=34 cm� and measuredith a 77 K cooled Ge photodiode or a cooled Hamamatsu928 photomultiplier. Lifetime measurements of 4S3/2

evel were excited with a dye laser (LSI, model DUO-220)sing Couramin-540A. For the lifetime of the 4I13/2 multi-let, we excited the 4I11/2 multiplet at ��980 nm with ahopped Ti:sapphire laser and the emission at �1550 nm was measured by a 77 K cooled Hamamatsu5509-72 photodetector with 3 ns of rise time. In bothases, the signal was stored and averaged by a 500 MHzektronix TDS520 digital oscilloscope.

3ATTospDdarc=asp=cT

BWcLl

Rtbpspsntont

�

e1twotgc

r0

CN(nertatrsttcvpsaa

F(tu

FLc

2068 J. Opt. Soc. Am. B/Vol. 23, No. 10 /October 2006 Rico et al.

. EXPERIMENTAL RESULTS. Crystal Growthable 1 shows the Er-doped crystals used in this work.he crystal growth details of NBW and NBM have beenffered in a previous work.10 The LBM phase was synthe-ized using Li2CO3, Bi2O3, and MoO3 initial products androcedures similar to those described for NBM.10 TheTA of LBM showed a congruent melting at 650°C. Un-oped and Er-doped crystals were pulled from the melt atrate of 1.8–2.8 mm/h using undoped LBM seeds. The

otation speed was 4–6 rpm, and the melt was allowed toool during the growth process at a rate of dT /dt0–0.4°C/h. Undoped LBM crystals show a yellow colorssociated with an absorption edge in the blue. Figure 1hows a LBM crystal achieved and a transparent polishedlate of NBW. Er segregation coefficients S�Er�crystal / �Er�melt�1 are found under all growthonditions for the considered Bi-based DT and DM (seeable 1).

. Compositional Analysishile the host composition of NaBi�YO4�2, Y=W or Mo,

rystals are well established in the previous work,10 theBM composition must be assessed particularly for the

ight Li ion.Figure 2(a) shows the RBS spectra of a LBM crystal.

BS is a very sensitive method to determine the concen-ration of elements with medium and large atomic num-ers �Z�8� in a sample, and thus the results of Fig. 2(a)rovide reliable Bi/Mo=2 and O/Mo=4 molar ratios, con-idering that the elastic scattering for 1.8 MeV incidentrotons is about 3.2 times larger than the RBS crossection.11 However, RBS fails in the detection of Li, anduclear-reaction analysis (NRA) has to be used as an al-ernative ion-beam method. The 7Li�p ,�4He reaction12

ffers, owing to the high abundance of 7Li (92.5%) inatural Li, a convenient way for detection, and it hasherefore already been frequently applied.13,14

This reaction has a high positive Q valueQ= +17.4 MeV� yielding two particles, each with an en-

ig. 1. (Color online) (a) As-grown Er-doped LBM crystal boule.b) Polished NBW plate grown by the CZ method. (c) ac view ofhe X /BiO8-YO4 coordinations containing dimeric �X /Bi�2O14nits along the c axis (vertical direction).

rgy of approximately 7.6 MeV for the reaction with.8 MeV protons. The high energy of the released par-icles enables the study of the Li content in the materialithout any overlapping RBS signals from any of thether constituents up to 8 �m depth. Figure 2(b) showshe -particle spectra recorded for the LBM crystal to-ether with a simulation performed using the SENRASode.15 A best fit to the spectrum has been obtained for a

7Li:Bi ratio of 0.89 that, taking into account the 7Li natu-al abundance of 92.5%, provides a molar Li:Bi ratio of.96±0.08 in the crystal.

. Crystalline StructureBW and NBM were initially ascribed to the SG I41/a

No. 88).16 Later studies described these crystals in theoncentrosymmetric SG I4̄ (No. 82) because of the pres-nce of nonnegligible (006, 006̄, 002, 1̄10, and 11̄0) x-rayeflections.17 Recently the same authors have examinedhe RT crystalline structure of Cz-grown NBW (Ref. 18)nd NBM crystals grown by the Stöber method.19 At thatime, they did not find the aforementioned 00l and hh̄0eflections; therefore, they concluded that the correctymmetry was I41/a instead of I4. Moreover, it was foundhat NBM undergoes, at T=241 K, a phase transition tohe monoclinic SG I2/a (No. 15). Although independentomposition analyses of all these crystals were not pro-ided, previous I4 assignments have been attributed to aossible departure of the crystal compositiontoichiometry.18 The crystalline structure of LBM waslso ascribed to the symmetry of the SG I41/a (Ref. 20),nd a later work16 also assumes this SG.

ig. 2. (a) RBS spectrum and (b) NRA spectrum of an undopedBM single-crystal. Points are the experimental results and theurves are the fits using the SENRAS simulation code.

itautisslogtotttpEiE

shIstatwspttoaa

ecospptwt

wIfrdsopaMfiihgmar

fcOtIrOpiihf

tX

D

1Ats

�sa

f(domf

btuu�T

0

h

Rico et al. Vol. 23, No. 10 /October 2006 /J. Opt. Soc. Am. B 2069

Whichever of the two indicated tetragonal symmetriess the correct one for the XBi�YO4�2 hosts, they are struc-urally disordered crystals owing to the distribution of thelkaline X+ (Li or Na) and Bi3+ cations either at thenique 4b crystallographic site, with S4 symmetry, whenhe material is described in the SG I41/a, or in the twondependent 2b and 2d point sites, also S4, for crystals de-cribed in the SG I4̄. In the latter case, although in bothites Bi3+ is coordinated to eight oxygens, owing to theoss of symmetry the Bi–O distances and bond angles arebviously different in each BiO8 polyhedra. The inhomo-eneous broadening observed at low temperature for op-ical bandwidths of Er3+ embedded in XBi�YO4�2 has itsrigin in the multiple local environments around Er3+ inhe host, with small site-to-site CF differences. However,he magnitude of the broadening will be obviously relatedo the envisaged number of crystal sites for Er3+, and ariory is expected to enlarge with increasing number ofr sites. Therefore ascertaining the crystalline structure

s relevant for the correct interpretation of the observedr3+ optical bandwidths.The structural determination procedure used here con-

isted of an initial evaluation of the significance of somekl reflections, which are systematic absences in SG41/a, through their individual psi-scan intensity mea-urements. The reflections selected for this purpose werehose constituting violations of the a plane, hh̄0, h�2n,nd hence the SG I41 would be still possible, as well ashose forbidden for the 41 axis, 00l, l even but l�4n,hich leads to the SG I4̄. After collecting each reflection

everal times, for every crystal, the average value is com-ared to the estimated background (EB) intensity ob-ained averaging 00l reflections, l�2n, forbidden in allhese SGs. It can be observed in Table 2 that the intensityf these reflections, and especially 002 and 110 are wellbove the EB level for the three matrices, NBW, NBM,nd LBM, considered here.Second, the analysis of single-crystal x-ray data for

ach pure host and for Er-NBW and the refinement of therystal structure, that is, the determination of atomic co-rdinates, occupancy factors (OF) for shared cationic po-itions, 2d�I� and 2b�II�, and thermal anisotropic dis-lacements for all atoms, have also allowed for theresence of a certain number of systematic absence excep-ions to be determined, for the a plane and for the 41 axis,ith intensity I�3��F�, that is, well above the I�2��F�

hreshold for reflections considered in the refinement,

Table 2. Assessment of the I41/a\I4̄ Distortionfrom Individual Scan Intensity Measurements

Ihkl versus Background Intensity EB

hkl X-Ray ReflectionsForbidden in I41/a NBW NBM LBM

0l, l even and�4n:I002 /EB 34/12 42/5 28/9I006 /EB 19/12 5/5 10/9

h̄0, h�2n:I1 10 /EB 25/12 25/5 15/9I3 30 /EB 15/12 7/5 14/9

hich undoubtedly indicated the noncentrosymmetric SG4̄. Furthermore, besides the assessment of the distortionrom the I41/a to the I4̄ symmetry, the crystal structureefinements also provide precise estimations of the X /Biistribution over the sites 2d�I� and 2b�II� that they areharing in each studied disordered crystal, which more-ver lead to positive values of the anisotropic thermal dis-lacement for all atoms and adequately low R1 discrep-ncy factors. All these results are included in Table 3.ore precise results have been obtained for the NBW re-

nement from single-crystal D9 neutron data, also shownn Table 3, and concerning the occupancy factors, theyave turned out to be exactly the same whatever the ori-in of the diffraction data. Table 4 provides the derivedain interionic distances in the crystals, which for NBW

re the more precise, corresponding to neutron diffractionesults.

It is worth noting that a slight alkali deficiency derivedrom the current x-ray analysis is observed for the LBMrystal, which confirmed NRA values shown previously.n the other hand, the LBM crystal appears as the crys-

al with lowest distortion from the centrosymmetric SG41/a, i.e., the lowest number of observed reflections cor-esponding to systematic absences of I41/a and cationicF closer to random 0.5–0.5 values. The other crystalsresent larger departures from this situation, and the Erncorporation develops a more intense I4̄ character, thats, more observed reflections not allowed in I41/a and aigher degree of local cationic ordering derived of the dif-erent X /Bi OFs.

Figure 1 shows an ac projection of the I4̄ crystal struc-ure showing the alternated 2b and 2d positions for the/Bi and the tetrahedral YO4 groups.

. Optical Characterization

. Crystal Birefringences expected for a birefringent crystal, the OA of LBM in

he blue region is anisotropic. The OA thresholds ob-erved at 300 K are approximately 416 nm for-polarized light and approximately 427 nm for-polarized light. Similar anisotropy in the UV optical ab-orption edges have been previously observed in NBWnd NBM hosts.10

On the basis of the tetragonal unit cell determinedrom the XRD analysis in Subsection 3.C, the ordinaryno, �c axis) and extraordinary (ne, �c axis) refractive in-ices of LBM were determined. Figure 3 shows the resultsbtained and the Sellmeier coefficients achieved for ainimum least-squares deviation fitting to the following

ormula:

n2 = A +B

1 − �C/��2− D�2. �1�

Similar results for NaBi�YO4�2, Y=W and Mo, haveeen already reported by us.10 It is interesting to notehat the three Bi-based single-crystal hosts examined bys, namely, the NBW, NBM, and LBM, are negativeniaxial crystals with refractive index and birefringence

�n= �no−ne�� increasing from NBW to NBM and to LBM.herefore LBM appears to be a promising material for

N

U

V

C

A

F

C

�

h

R

D

G

F

R

E

A

2070 J. Opt. Soc. Am. B/Vol. 23, No. 10 /October 2006 Rico et al.

Table 3. Crystal Data at 296(2) K and Structure Refinement details of XBi„YO4…2, X=Li,Na YMo, W, and Er-doped NBW „†Er‡crystal=2.4Ã1020 cm−3…

ominal formula NBWa–e,i,j NBW,Era,c,e,i NBMa,c,e,i LBMa,c,e,i

nit cell dimensions a=b (Å) 5.2752(6) 5.2827(7) 5.2676(10) 5.2178(8)

c 11.503(2) 11.514(2) 11.552(12) 11.469(2)

olume �Å3� 320.11(9) 321.32(9) 320.54(12) 312.25(9)alculated density �Mg/m3� 7.549 7.521 5.718 5.699bsorption coefficient �mm−1� 63.338 63.100 31.307 32.066(000) 612 612 484 468

137

rystal size �mm3� 0.09�0.07�0.074�4�5

0.08�0.10�0.12 0.05�0.08�0.12 0.05�0.07�0.10

range for data collection (°) 3.54–26.40 3.54–28.81 3.53–28.36 3.55–28.15

4.19–54.39

kl Limiting indices 3̄–6, 6̄–1, 1̄4–141̄0–10, 1̄0–10, 2̄2–8

6̄–6, 4̄–7, 8̄–15 6̄–2, 6̄–6, 1̄0–11 6̄–6, 3̄–5, 1̄5–8

eflections collected/unique [R(int)] 671/314 [0.0414]1806 /901 [0.0235]

783/371 [0.0339] 704/268 [0.0397] 872/350 [0.0373]

ata/restraints/parameters 314/0/30901 /2 /32

371/0/30 268/0/30 350/0/30

oodness-of-fit on F2 1.069 0.889 1.095 1.176

inal R1, wR2 indices [I�2sigma�I�] 0.1009, 0.2443 0.0488, 0.1583 0.0839, 0.2229 0.0389, 0.11430.0397, 0.0783

1, wR2 indices (all data) 0.1070, 0.2515 0.0579, 0.1723 0.0867, 0.2283 0.0481, 0.1211

0.0756, 0.0795

xtinction coefficient 0.06(3) 0.035(8) 0.13(3) 0.141(12)

tom, site and coordinate

X(I)/Bi(I) 2d x 1212

12

12

y 0 0 0 0

z 1414

14

14

U(eq)g 13(6) 29(2) 10(3) 16(1)

X(II)/Bi(II) 2b x 1212

12

12

y 1212

12

12

z 0 0 0 0

U(eq)g 8(7) 5(4) 8(2) 17(1)

Y(1) 2a x 0 0 0 0

y 0 0 0 0

z 0 0 0 0

U(eq)g 5(3) 17(2) 4(2) 12(1)

Y(2) 2c x 0 0 0 0

y 1212

12

12

z 1414

14

14

U(eq)g 6(5) 10(2) 2(2) 10(1)

O(1) 8gg x 2420(40) 2411(10) 2420(30) 2410(30) 2406(15)

y 8540(40) 8483(8) 8480(30) 8500(30) 8496(15)

z 830(20) 831(4) 860(15) 820(15) 838(7)

U(eq)g 14(6) 17(1) 23(4) 12(5) 23(2)

O�2�8gf x 2390(40) 2384(6) 2350(30) 2380(30) 2414(15)y 3490(40) 3440(6) 3510(30) 3530(30) 3481(15)

z 1650(20) 1637(4) 1663(16) 1635(19) 1660(7)

U(eq)g 18(7) 14(1) 28(4) 10(5) 23(2)

OFh X(I)/Bi(I) 2d 0.35(2)/0.65(2) 0.34(2)/0.66(2) 0.55(1)/0.45(1) 0.480(5)/0.520(5)

0.37(2)/0.63(2)

OFh X(II)/Bi(II) 2b 0.54(3)/0.46(3) 0.58(2)/0.42(2) 0.47(1)/0.53(1) 0.492(5)/0.508(5)

0.54(3)/0.46(3)

osLailiNf

mo�

2Taew

etetmm(wtTl

d2atwtbsafetntl�wf

BBBBYY

Fta

Rico et al. Vol. 23, No. 10 /October 2006 /J. Opt. Soc. Am. B 2071

ptical waveguide fabrication on NBW or NBM sub-trates. In fact, the lattice mismatches for theBM/NaBi�W or MoO4�2, �aLBM-asub� /asub=1.04�10−2nd 0.89�10−2, respectively, are small and therefore lownterfacial stress is expected. LBM layers can be grown byiquid phase epitaxy, as the melting point of LBM �650°C�s lower than the melting points of NBW �935°C� andBM �871°C�. It is also interesting to note that the bire-

ringence ��n�0.1� of these three Bi-based crystals is

Table 4. Selected Bond Lengths (Å) for Pure NBW,NBM, and LBM and Er-Doped NBW Crystalsa

Ligands NBWb NBW: Er NBM LBM

i�I� /X�I�–O�1� �4 2.489(4) 2.463(16) 2.50(2) 2.466(8)i�I� /X�I�–O�2� �4 2.487(3) 2.514(16) 2.512(16) 2.459(8)i�II� /X�II�–O�1� �4 2.481(3) 2.491(13) 2.481(18) 2.466(8)i�II� /X�II�–O�2� �4 2.475(4) 2.500(17) 2.47(2) 2.465(7)�1�–O�1� �4 1.781(4) 1.807(14) 1.768(17) 1.765(8)�2�–O�2� �4 1.801(4) 1.758(18) 1.778(16) 1.773(8)

aX=Na or Li.bDistances corresponding to results from the neutron diffraction refinement.

ig. 3. Refractive index dispersion of uniaxial LBM single crys-al measured at 300 K . The points are the experimental resultsnd the curves the Sellmeier fits.

Table 3.

Systematic absence exceptions:

hkl I�3� for 41 axis (Total I�2�) 2(4)

hkl I�3� for a plane (Total I�2�) 1(13)

aXRD wavelength 0.71073 ��.bNeutron diffraction �ND� wavelength 0.84 ��.cXRD absorption correction SADABS.dND absorption correction DATAP.eRefinement method: full matrix least squares on F2.fAtomic coordinates �104.gU�eq� is defined as one third of the trace of the orthogonalized Uij tensor.hOccupancy factor for the indicated site. They were refined with no restrains betiPositive anisotropic thermal displacements for all atoms have been obtained andjItalics indicate results from neutron D9-ILL, Grenoble data.

uch larger than those found in lanthanide-based tetrag-nal DT crystals, such as NaLa�WO4�2 and NaGd�WO4�2�n�0.005�.21

. Er3+ Energy Levelshe analysis of the observed energy levels of Er3+ in NBWnd NBM crystals and the parameterization of their CFffects have been studied by some of us in a previousork.7 The analysis is extended now to the LBM host.The energy level Stark sequence of the fundamental

4I15/2 multiplet was determined from the polarized4I13/2

mission shown in Fig. 4. These spectra were obtained af-er excitation in the 4F7/2 ��exc=488 nm� multiplet and themission �EMI�1.5 �m has been referred to E=0 cm−1 forhe largest wavelength peak observed. The Stark levels ofultiplets ranging from 4I13/2 to

2H9/2 have been deter-ined directly from the ground-state optical absorption

GSA) at 10 K. These latter spectra show some hot bandsith small intensity due to the thermal population of the

wo first excited 4I15/2 Stark levels with energy �40 cm−1.

ransitions from the fundamental 4I15/2�0� Stark level areabeled with arrows in Fig. 4.

The �- and -polarized PL and OA spectra clearly showifferences as expected from the S4 point symmetry of theb and 2d sites in the SG I4̄. A good example of this situ-tion are the 4I15/2�0�→

4S3/2 transitions. The possibilityhat the Er3+ low-temperature spectroscopic propertiesere determined by a monoclinic local symmetry, such as

hat described in NBM for T�241 K, (Ref. 19) is ruled outy the presence of the well-defined polarization featureshown in Fig. 4 and in previous works.7 In the monoclinicssumption, a unique �3,4 irreducible representation IRor the 4f11 Er3+ configuration would exist, with no differ-nces between the � and -polarized spectra. Contrary tohis, the Er3+ bands are seen either only in � or simulta-eously in � and configurations. Table 5 summarizeshe energy of the levels determined, the experimental po-arization of the observed ground-state transition (� or

), and the corresponding experimental FWHM band-idth. -polarized spectra (not shown for brevity) were

ound to be very similar to �-polarized ones; this shows

tinued)

4(6) 3(3) 2(8)

7(22) 9(22) 0(11)

and 2d sites.

ilable from the authors.

(Con

ween 2b

are ava

tp

tdogtststetabdIlrmttf

l

NhH6baTrm

3Aiftttactc

o��gsal

bplma

wet

wmtc

p(chpe

F�v

2072 J. Opt. Soc. Am. B/Vol. 23, No. 10 /October 2006 Rico et al.

hat the observed transitions are dominated by electric di-ole (ED) contributions.In the absence of a magnetic field, the electric field of

he lattice splits the 2S+1LJ multiplets into J+12 Kramers

oublets. The first step for the proper assignment of thebserved levels to the IR (�5,6 and �7,8) of the S4 pointroup is to determine the IR of the ground Stark level ofhe 4I15/2 multiplet. The simplest case to analyze is theplitting of J=5/2 manifold. The 4F5/2 manifold splits inwo �5,6 levels and one �7,8 level. The experimental re-ults in Fig. 4 show that the level with lowest energy inhis multiplet has clearly a � character, whereas the high-st one is observed in � and configurations. The level inhe middle seems to be mainly � polarized, however thisssignment is not free of uncertainty. To reach an unam-iguous conclusion, we have performed the CF simulationescribed later in this section assuming either �5,6 or �7,8R for the ground Stark level. The assumption of a �7,8 IReads to an energy sequence for the 4S3/2 multiplet in di-ect conflict with the polarization characters found experi-entally. Therefore, the only possibility consistent with

he experimental results is the assumption of �5,6 IR forhe ground Stark level of 4I15/2 and a mainly � characteror the experimental spectra of the middle level of 4F5/2.

The CF simulation has been performed in a way simi-ar to that described previously for Er3+ in NBW and

ig. 4. PL and OA at 10 K of Er3+ in LBM crystal. �Er�=0.071020 cm−3. -OA and �-PL spectra appear arbitrary displaced

ertically.

BM crystals.7 The free ion (FI) and S4 CF interactionsave been treated simultaneously through very completeFI and HS4 Hamiltonians (see Ref. 7) with up to 20 andin FI and CF parameters, respectively, using the entire

asis set of wave functions. The initial values for the FInd CF parameters were those obtained for Er3+ in NBW.able 6 summarizes the phenomenological FI and CF pa-ameters resulting of the best simulation of the experi-ental sequence of energy levels.

. Er3+ Radiative Processesrecent work5 described the radiative processes of Er3+

n NBW through the application of the Judd–Ofelt (JO)ormalism.22,23 Unfortunately, the polarized character ofhe Er3+ transitions in this latter crystal was ignored, andherefore the reported integrated absorption cross sec-ions are uncertain. In this section, we shall revise the JOnalysis of Er3+ in NBW taking into account the uniaxialharacter of the crystal, and we shall compare this withhose also performed here for Er3+ in the NBM and LBMrystals.

To apply the JO theory, the 300 K polarized OA spectraf Er3+ in NBW ��Er�crystal=0.92�1020 cm−3�, NBM�Er�crystal=1.05�1020 cm−3�, and LBM ��Er�crystal=0.56

1020 cm−3� have been recorded. Table 7 shows the inte-rated absorption cross sections, �JJ� / �Er�, for � and pectra. These experimental results have been averageds �2�+� /3 and used to achieve the experimental oscil-ator strengths for each excited multiplet as

fexp =4�0mc

2

e2�̄2

�JJ�

�Er�. �2�

The experimental average oscillator strengths obtainedy us for Er3+ in NBW are larger than those reported in arevious work for the same NBW host.5 This is mostikely due to the different conditions used for the experi-

ental determination, in our case, explicitly taking intoccount the polarized character of the OA.The JO theory has been described in a large number of

orks, and details of the treatment can be foundlsewhere.22–24 We shall just mention that the ED oscilla-or strengths can be obtained as

fED,cal = �82mc

3h�̄�2J + 1�SJJ�, �3�

SJJ� = k=2,4,6

�k�4fn�L,S�J�Uk�4fn�L�,S��J���2, �4�

here �k are the JO parameters to be obtained from theinimization of the f̄exp− fcal differences. Table 7 includes

he �k sets obtained for Er3+ in the Bi-based DT or DMonsidered here. In this calculation, we have used the�Uk�� matrix elements provided previously25 and theroper refractive indices for NBW and NBM 10 and LBMthis work, Subsection 3.D.1). Also, magnetic dipole (MD)ontributions to the experimental oscillator strengthsave been taken into account to calculate the ED fexp. Ex-ressions for these contributions can be foundlsewhere.24

resi�rraacttfs

Pn

tis1lNr

w

aw�s2lt

w

Rico et al. Vol. 23, No. 10 /October 2006 /J. Opt. Soc. Am. B 2073

The �k sets obtained for Er3+ in the three hosts areather similar. This could be anticipated from the similarxperimental values of the oscillator strengths. The �ket corresponding to Er3+ in NBW is slightly larger thann the molybdates. However, in the three cases the ratios

2/�6 and �4 /�6 fall in the 8.1–9.4 and 0.75–1.25 ranges,espectively. Using the calculated �k sets, the radiativeates, AJJ� were obtained as described in the literature

25

nd from them the branching ratios, �JJ�=AJJ� /J�AJJ�,nd the radiative lifetimes, �r=1/J�AJJ� of the lumines-ence were calculated. Table 8 shows relevant results ob-ained for Er3+ in the NBM single crystal as a represen-ative of the three compounds considered. Although theull set of multiplets was used in the calculations, Table 8hows only transitions with ��4%.

Figure 5 shows an overview of relevant facts of the Er3+

L emissions observed at 10 and 300 K in NBM. Although

ot shown for the sake of brevity, similar results were ob- s

c

ained for NBW and LBM single crystals. It is worth not-ng that the polarized character of the 4I13/2→

4I15/2 emis-ion of interest for lasing, particularly the band at about520 nm, is more intense in configuration. The nonpo-arized spectral distribution reported previously inBW:Er (Ref. 5) is intermediate between the now-

esolved � and PL spectra.To evaluate the laser efficiency of Er3+ in these hosts

e have determined the experimental lifetimes of the4S3/2 and

4I13/2 multiplets in the most diluted samplesvailable. The experimental lifetime of the 4S3/2 multipletas excited resonantly at �=542 nm and recorded at �EMI

4S3/2→4I15/2��552 nm. A single exponential was ob-

erved for all hosts and temperatures. The value was0 �s at 300 K and 22 �s at 7 K. The 4S3/2 experimentalifetime obtained is very short in comparison to the radia-ive value �r�267 �s calculated from the JO treatment;

ee Table 8. This short lifetime must be attributed to the

Table 5. Observed „E0… at 10 K and Calculated for an Average S4 Site „Ec… Energy Levels „cm−1…of Er3+ in LBM Single Crystala

2S+1LJ � IR Eo Ec �Eo �Ec2S+1LJ � IR Eo Ec �Eo �Ec

4I15/2 � �5,6 0 8 04S3/2 � �5,6 18 349 18 350 13 2

� �7,8 18 8 0 �2P3/2� � �7,8 18 409 18 413 20 11�5,6 33 41 5�7,8 54 58 3

2H211/2 � �5,6 19 063 19 071 12 2�7,8 208 209 4 �4G11/2� � �7,8 19 081 19 096 14 6�7,8 251 242 7 � �7,8 19 129 19 142 13 10�5,6 292 286 14 � �5,6 19 199 19 192 7�5,6 310 313 20 � �7,8 19 215 19 204 18 8

� �5,6 19 225 19 210 74I13/2 � �7,8 6527 6524 6 3

� �5,6 6538 6524 9 34F7/2 � �7,8 20 462 20 451 14 3

� �5,6 6576 6574 9 4 �2G17/2� � �5,6 20 474 20 468 14 7� �7,8 6646 6640 18 4 � �5,6 20 533 20 545 15 8� �5,6 6675 6666 10 � �7,8 20 566 20 567 28 15� �7,8 6691 6694 23 9� �5,6 6710 6714 26 15

4F5/2 � �5,6 22 149 22 142 13 5�2D15/2� � �5,6 22 170 22 158 13 3

4I11/2 � �7,8 10 199 10 184 5 3 � �7,8 22 194 22 193 16 9�2H211/2� � �5,6 10 202 10 194 7 5

� �5,6 10 217 10 240 10 54F3/2 � �5,6 22 506 22 503 21 4

� �7,8 10 232 10 260 10 6 �2D13/2� � �7,8 22 547 22 549 28 12� �5,6 10 254 10 269 10� �7,8 10 286 10 282 18 11

2G19/2 � �7,8 24 466 24 450 14 3�4F9/2 ,

2H29/2� � �5,6 24 534 24 544 15 14I9/2 � �7,8 12 352 12 342 11 3 � �7,8 24 567 24 568 15 10

� — 12 448 1 � �7,8 24 660 24 661 20 5� — 12 483 13� �7,8 12 518 12 504 20 3� �7,8 12 594 12 598 17 13

4F9/2 � �7,8 15 226 15 227 13 1�4I9/2� � �5,6 15 252 15 263 13 10

� �7,8 15 266 15 267 18 1� �7,8 15 328 15 336 16 8� �5,6 15 379 15 392 22 13

a� and IR indicate the OA observed polarization character and corresponding irreducible representation of the transition, respectively. �Eo is the observed OA FWHM line-idth. �E is the energy separation simulated using SOM CF parameters �Ref. 28� for the nonequivalent 2b and 2d sites in the host.

i�Pt

c�tTmrMiis

4Tba

b

waptaaa

onttsfcss�

cd�pstle

wlatvcal7

t�po�arcp

pOpfi

=

b

2074 J. Opt. Soc. Am. B/Vol. 23, No. 10 /October 2006 Rico et al.

nteraction with the higher close-lying 2H11/2 level withr�29 �s. In fact, the thermal intensity evolution of theL at �=520 to 560 nm, see Fig. 5, shows the strong in-eraction at RT between the 2H11/2 and

4S3/2 multiplets.For 4I13/2 lifetime determination, the 300 K PL was ex-

ited at �EXC�4I15/2→

4I11/2��980 nm and measured atEMI�

4I13/2→4I15/2��1540 nm using the lowest concentra-

ion available for each host. The results are included inable 8. This excitation scheme should produce realisticeasurement in view of the short experimental lifetime

eported for the 4I11/2 multiplet, namely, �=116 �s.6

oreover, the emission reabsorption �4I15/2→4I13/2� lead-

ng to artificial delay enlargement of the intensity decay,s negligible owing to the very low absorption of the

Table 6. Phenomenological FI and CF Parameters„cm−1… for Er3+ in LBM Single Crystala

Parameter Value Value in 2b Value in 2d

E0 35 003(1)E1 6538.3(8)E2 32.45(3)E3 664.96(8) 18.97(9)� �−536�� [1790]� 2357.1(9)

M0 b [5.9]P2 c [825]T2 [400]T3 [32]T4 [238]T6 �−289�T7 [336]T8 [299]B0

2 412 470 401B0

4 −610�56� −580 −564B4

4 ±929�28� ±730 ±715B0

6 −170�35� −85 −30B4

6 ±460�20� ±576 ±556S4

6 ±113�59� ±123 ±110S2

d 184 210 179S4 483 395 386S6 192 232 226ST 318 291 277L 50�e 11.3

Residue 4939.3

aItalic values are CF parameters calculated using the SOM model �Ref. 28� withreviously determined 2b and 2d crystal data ��=0.08; effective charge for=−0.8�. Values in parentheses refer to estimate standard deviations in the indicated

arameter. Values in square brackets were not allowed to freely vary in the parametertting.

bM2=0.56M0; M4=0.32M0.cP4=0.75P2, P6=0.50P2.dThe CF strength parameter Sk and ST are defined as ST= ��1/3��kSk

2�1/2, Sk�1/ �2k+1���B0

k�2+2�q�Bqk�2+ �Sq

k�2�1/2.e�= ����i�2 / �1− p��1/2 , �i=Eo−Ec; 1 is the number of levels and p is the num-

er of parameters.

amples used.

. DISCUSSIONhe laser efficiency of a 2S+1LJ multiplet is related to theranching ratio distribution, emission cross section �EMI,nd quantum efficiency �, among other parameters.For a given transition, the absolute value of �EMI can

e estimated by the reciprocity principle26 as

�EMI = �GSAZl

Zue�Ezl−h��/kBT, �5�

here �GSA=GSA/ �Er� is the GSA cross section; Zu and Zlre the partition functions of the upper and lower multi-lets, respectively; and Ezl is the energy difference be-ween the lowest Stark levels of both multiplets. Zu, Zl,nd Ezl, can be obtained from previous results for NBWnd NBM7 and from those reported in Table 5 for LBM. Inll the hosts considered here, Zl /Zu�1 was found for the

4I13/2→4I15/2 laser channel.

Figure 6 shows the experimental spectral dependencef the 300 K 4I13/2→

4I15/2 Er3+ emission in NBW after

ormalization to the absolute peak value calculated ob-ained by using Eq. (5). Within the experimental uncer-ainty, Er-doped NBM and LBM have similar spectralhapes and peak values similar to those shown in Fig. 6or Er-doped NBW. At �=1535 nm, the peak emissionross sections in both polarization configurations areimilar, �EMI�0.5�10−20 cm2, but in general, �EMI islightly larger in configuration, particularly for �1500 nm.The Er peak �EMI value around 1530–1550 nm in NBW

an be compared to those of Er in YAG.27 For this stan-ard laser crystal, a peak 300 K �EMI ���1530 nm�0.7�10−20 cm2 has been found, but laser action takes

lace for �EMI�0.2�10−20 cm2. This shows that the crossections of Er3+ in XBi�YO4�2 crystals are of the magni-ude required for laser operation. The expected spectralaser tunability of Er in NBW can be estimated from theffective emission cross-section �EFF,

�EFF = ��EMI − �1 − ���GSA, �6�

here � is the inversion coefficient. To achieve laser oscil-ation �EFF�0 is required. Figure 7 shows that a moder-te inversion coefficient ���0.3� is needed to achieve thehreshold laser condition. For a medium value of the in-ersion coefficient �=0.5, the spectral range of tunabilityalculated is up to approximately 88 nm with a FWHM ofpproximately 60 nm. This bandwidth would allow mode-ocked laser pulses of 50 fs for Gaussian pulse shape or0 fs for sech2 pulse shape.The 4I13/2 quantum efficiencies �=�exp/�r obtained for

he disordered DT and DM hosts considered here are �0.68 to 0.74. A more efficient �=0.84 value has been re-

orted for Er in NBW.5 In our view, this latter result isverestimated owing to the recording conditions of theexp�4750 �s lifetime, i.e., pumping the

4S3/2 multipletnd detecting with cooled Ge photodiode. Moreover, theadiative lifetime previously calculated5 is uncertain be-ause the MD contribution to the 4I13/2→

4I15/2 transitionrobability was ignored. Another determination of the

4 3+

I13/2 Er experimental lifetime in NBW reported �exp

�tiltttttb

eccpltcm

NNL

wa

Rico et al. Vol. 23, No. 10 /October 2006 /J. Opt. Soc. Am. B 2075

3400 �s,6 this is closer to the �exp values obtained inhis work. Although detailed information of the character-stics of the previously used detectors is not available, thearger lifetime values previously reported are likely dueo detector rise-time limitations or to emission reabsorp-ion. In our case, all these artifacts are excluded among tohe fast detector rise time �3 ns� and very low Er concen-ration ��Er��1018 cm−3� used, which also excludes a life-ime reduction by energy transfer between Er–Er neigh-ors.In view of the �EFF and � values achieved here, laser

mission at 1.5 �m in Er-doped NBW, NBM, and LBMrystals seems feasible. The broad bandwidths of theserystals allow an efficient absorption of the radiationumping even using laser diodes not specifically wave-ength matched and as mentioned above a broad laserunability range can be expected. To optimize these appli-ations, the physical contributions to the optical linewidthust be identified. In Subsection 3.C, it was shown that

Table 8. ED and MD TransitionProbabilities, AED and AMD; Branching Ratios, �;Radiative, �r, and Experimental, �exp, Lifetimes of

Er3+ in NBM Calculated from the �2=7.37,�4=0.64, �6=0.78 Set

a

Transition�

(nm)AED+AMD

�s−1��ij(%)

�r ��exp���s�

4G9/2→ 4I13/2 481 84,906 81.84 9.64I15/2 367 9881 9.52

4G11/2→ 4I15/2 382 179,592 93.81 5.22H9/2→ 4I11/2 698 1088 11.24 103

4I13/2 557 4491 46.414I15/2 410 3757 38.83

4F3/2→ 4I9/2 993 408 7.00 1714I11/2 813 1745 29.924I13/2 628 256 4.384I15/2 447 3358 57.58

4F5/2→ 4I13/2 643 2219 34.03 1534I15/2 454 3689 56.57

4F7/2→ 4I11/2 973 335 4.39 1244I13/2 719 650 8.034I15/2 491 6755 83.45

2H11/2→ 4I15/2 527 33,322 97.15 294S3/2→ 4I13/2 850 927 24.77 267 (20

b)4I15/2 549 2643 70.65

4F9/2→ 4I11/2 1976 153 5.96 3894I13/2 1151 130 5.064I15/2 660 2273 88.39

4I9/2→ 4I13/2 1709 88 33.03 37374I15/2 812 173 64.54

4I11/2→ 4I13/2 2758 36+14 87.88 2415 (116b)

4I15/2 991 364 12.124I13/2→ 4I15/2 1547 100

BW �Er�=2�1018 cm−3 213+94 3254 (2400b)BM �Er�=7.5�1019 cm−3 204+108 3200 (2300b)BM �Er�=7�1018 cm−3 211+117 3048 (2060b)

aOnly transitions with ��4% are shown.b300 K experimental 4S3/2,

4I11/2, and4I13/2 lifetimes of Er in NBM crystals doped

ith the lowest concentration available 4I13/2 experimental lifetimes of Er in NBWnd LBM are also included.

Tab

le7.

Ave

rage

Wav

elen

gth

�̄,I

nte

grat

edO

AC

ross

Sec

tion

�/†

Er‡

,Ave

rage

dE

xper

imen

tal

ED

Osc

illa

tor

Str

engt

hf̄ E

D,e

xp,a

nd

Cal

cula

ted

ED

Osc

illa

tor

Str

engt

hf̄ E

D,c

al,f

or2S

+1 L

J,M

ult

iple

tsof

Er3

+in

NB

W,N

BM

,an

dL

BM

.a

4 I15

/2→

�̄

(nm

)

NB

W:E

rN

BM

:Er

LB

M:E

r

��/�

Er�

�10−

27cm

3 ��

/�

Er�

�10−

27cm

3 �f̄ E

D,e

xp��

108 �

ḟ ED

,cal

��10

8 ��

�/�

Er�

�10−

27cm

3 ��

/�

Er�

�10−

27cm

3 �f̄ E

D,e

xp��

108 �

f ED

cal

��10

8 ��

�/�

Er�

�10−

27cm

3 ��

/�

Er�

�10−

27cm

3 �f̄ E

D,e

xp��

108 �

f ED

,cal

��10

8 �

4 G9/

236

51.

33.

517

315

416

822

24 G

11/2

380

5538

3861

3683

4340

4394

2 H9/

240

70.

61.

459

9694

954 F

5/2,

3/2

451

0.8

2.2

7012

11.

02.

074

113

11.

769

110

4 F7/

249

03.

65.

319

622

92.

34.

113

721

42.

64.

415

122

42 H

11/2

522

4041

1673

1812

4941

1922

1921

4545

1866

1865

4 S3/

254

71.

42.

263

600.

92

4854

0.9

1.6

4352

4 F9/

266

06

1019

019

16.

19.

418

717

77.

211

220

211

4 I9/

280

01.

03.

432

251.

52.

532

242

3.3

4334

4 I11

/298

02.

86.

246

894.

57.

163

843.

45.

849

794 I

13/2

1520

2450

160b

144

2342

144b

129

2143

139b

123

�10

6rm

s��

f�1.

30.

400.

42�

k�1

0−20

cm2 �

�2=

7.57

,�

4=

0.71

,�

6=

0.93

�2=

7.37

,�

4=

0.64

,�

6=

0.78

�2=

6.57

,�

4=

0.87

,�

6=

0.70

a The

fitqu

ality

obta

ined

with

the

�k

JOpa

ram

eter

sets

isde

scri

bed

byrm

s��

f�.

b Mag

netic

dipo

leco

ntri

butio

ns,

f DM

,w

ere

eval

uate

dan

ddi

scou

nted

asfo

llow

s:N

BW

���

,0.6

8�

10−

6an

d�

,0.6

6�

10−

6 �,

NB

M� �

,0.7

0�

10−

6an

d�

,0.6

8�

10−

6 �,

and

LB

M� �

,0.7

2�

10−

6an

d� �

,0.7

0�

10−

6 �.

Lcdauttcsmfvtt

tasetBTtsarEttt

cwsc

Fts�-polarized spectra. (b) -polarized spectra.

Ft

Fig. 5. Er PL

2076 J. Opt. Soc. Am. B/Vol. 23, No. 10 /October 2006 Rico et al.

BM is isostructural to NBW and NBM, and all of theman be described with the symmetry of the SG I4̄ with twoifferent lattice sites for Er3+, thus we refer to this fact assite contribution to the bandwidth. The transformation

pon cooling of the SG I41/a to the SG I4̄ cannot be jus-ified on the basis of an off-stoichiometry crystal composi-ion, rather it seems likely that the thermal history of therystal plays a major role. Moreover, since the 2b and 2dites are shared by the alkaline, Bi3+ and Er3+ cations,ultiple environments around Er3+ are associated to dif-

erent cationic distributions, we refer to this fact as an en-ironment contribution to the bandwidth. The purpose ofhe following discussion is to account for the influence ofhe first of these two contributions.

The Er3+ CF analysis made in Subsection 3.D.2 yieldedhe phenomenological FI and CF parameters of an aver-ge S4 Er3+ center. The FI parameters are roughly theame for similar hosts, while individual sets of CF param-ters in both 2b and 2d sites can be calculated throughhe simple overlap model28 (SOM), by using establishedi(I)–O and Bi(II)–O crystallographic distances given inable 5. Table 6 (SOM column) includes these two collec-ions of semiempirical calculated CF parameters in the S4ymmetry. Using each one of these CF parameter setslong with the common optimized FI parameters, sepa-ated sequences of energy levels have been derived forr3+ in the two sites. The energy differences, �Ec, be-

ween these two sequences for each Stark level provideshe site contribution to the bandwidth observed at lowemperature.

Table 5 shows a comparison of the calculated �Ec siteontributions and the experimental �Eo FWHM band-idths for Er3+ in LBM. Results for NBW and NBM are

imilar and not shown for the sake of brevity. From thisomparison, it can be observed that the site contributions

in NBM.

ig. 6. Comparison between the experimental PL spectral dis-ribution (�EXP, line) and calculated

4I13/2→ 4I15/2 emission crossection (�emi, points) of Er3+ in NBW single crystal. (a)

ig. 7. Spectral dependence of the effective emission cross sec-ion of Er in NBW.

3+

awaTrTes(d

htofcpc

5Nnscpb

hts

tltcta

Lilssscibst

ATtMpGiEC

ce

R

Fdf

Rico et al. Vol. 23, No. 10 /October 2006 /J. Opt. Soc. Am. B 2077

re generally slightly lower than the experimental band-idths, i.e., they are inside the experimental linewidthsnd not resolved under the experimental conditions used.his implies that the environment contribution can beoughly estimated as half of the experimental linewidth.his is still much larger than the inhomogeneous broad-ning observed in ordered DT or DM hosts. Figure 8hows a comparison with monoclinic KGd�WO4�2-KGW) where Er sits in a unique site and lattice disor-er does not exist.Finally, it should be mentioned that although these

osts can be grown with very high Er concentration,29

heir disordered nature favors the presence of a fractionf Er–Er pairs that will introduce emission losses. There-ore the demonstration of laser action is a possible buthallenging task that likely should balance absorption ofumping (likely by codoping with Yb in the diode pumpedase) and Er concentration.

. CONCLUSIONSBW, NBM, and LBM crystals used in this work areegative uniaxial isostructural crystalline hosts with theymmetry of the SG I4̄. Even when doped with a signifi-ant concentration of Er �Er�melt�1 mol.%, the com-ounds have a congruent melting, and therefore they cane grown by the Czochralski method.Er3+ is placed in two nonequivalent lattice sites of the

osts and experiences a distribution of CFs associated tohe Li (or Na) and Bi random occupancy of these latticeites. As a consequence of this crystallographic feature,

ig. 8. Comparison of the low-temperature OA of Er3+ in “or-ered” (-KGW) and “disordered” DT and DM. (a) Bandwidthsor a 4S3/2 Stark level. (b) CF splitting of

4F5/2 multiplet.

he peak absorption and emission Er3+ cross sections areower than in the ordered monoclinic DT laser hosts, buthe balance between the 4I13/2 emission and absorptionross sections is positive even at moderate inversion ra-ios ��0.4, which suggests possible laser action at RTlong with some degree of tunability.The spectroscopic properties of Er in NBW, NBM, and

BM hosts are rather similar and show polarized featuresn accordance with the uniaxial character of the crystalattices and with the S4 local symmetry of the 2b and 2dites of the tetragonal SG I4̄. The polarized nature of thepectroscopy indicates that the Er3+ centers keep con-tant relative orientations and it excludes a monoclinic lo-al symmetry. The energy difference for a given Er3+ leveln the two lattice sites is lower than the experimentalandwidth; this difference was not experimentally re-olved in the spectroscopic measurements performed inhis work.

CKNOWLEDGMENTShe authors acknowledge the financial support through

he Spanish projects MAT2002-04603-C05-05 and CAMAT/0434/2004 and also from the European Commission,

roject DT-CRYS, UE NMP3-CT-2003-505580. TheRICE (Portugal)–CSIC (Spain) cooperation agreement

s gratefully acknowledged. M. Rico is supported by theducation and Culture Ministry of Spain under Ramón yajal programme.

Corresponding author C. Cascales’s e-mail address iscascales@icmm.csic.es. Corresponding author C. Zaldo’s-mail address is cezaldo@icmm.csic.es.

EFERENCES1. M. Rico, J. Liu, U. Griebner, V. Petrov, M. D. Serrano, F.

Esteban-Betegón, C. Cascales, and C. Zaldo, “Tunable laseroperation of ytterbium in disordered single crystals ofYb:NaGd�WO4�2,” Opt. Express 12, 5362–5367 (2004).

2. J. Liu, J. M. Cano-Torres, C. Cascales, F. Esteban-Betegón,M. D. Serrano, V. Volkov, C. Zaldo, M. Rico, U. Griebner,and V. Petrov, “Growth and continuous-wave laseroperation of disordered crystals of Yb3+:NaLa�WO4�2 andYb3+:NaLa�MoO4�2,” Phys. Status Solidi A 202, R29–R31(2005).

3. S. Rivier, M. Rico, U. Griebner, V. Petrov, M. D. Serrano, F.Esteban-Betegón, C. Cascales, C. Zaldo, M. Zorn, and M.Weyers, “Sub-80 fs pulses from a mode-lockedYb:NaGd�WO4�2 laser,” in Conference on Lasers andElectro-Optics/Europe, CLEO/Europe-EQEC 2005Conference Digest CD-ROM (2005), paper CF4-4-THU.

4. A. A. Kaminski, Crystalline Lasers (CRC Press, 1996).5. D. K. Sardar, C. C. Russell III, R. M. Yow, J. B. Gruber, B.

Zandi, and E. P. Kokanyan, “Spectroscopic analysis of theEr3+ �4f11� absorption intensities in NaBi�WO4�2,” J. Appl.Phys. 95, 1180–1184 (2004).

6. K. A. Subbotin, E. V. Zharikov, and V. A. Smirnov, “Yb andEr-doped single crystals of double tungstates NaGd�WO4�2,NaLa�WO4�2, and NaBi�WO4�2 as active media for lasersoperating in the 1.0 and 1.5 �m ranges,” Opt. Spectrosc.92, 601–608 (2002) [Opt. Spektrosk. 92, 657–664 (2002)].

7. M. Rico, V. Volkov, C. Cascales, and C. Zaldo,“Measurement and crystal field analysis of Er3+ energylevels in crystals of NaBi�MoO4�2 and NaBi�WO4�2 withlocal disorder,” Chem. Phys. 279, 73–86 (2002).

1

1

1

1

1

1

1

1

1

1

2

2

2

2

2

2

2

2

2

2

2078 J. Opt. Soc. Am. B/Vol. 23, No. 10 /October 2006 Rico et al.

8. International Tables for Crystallography (Kynoch, 1974),Vol. 4.

9. SHELXTL Version 6.10 software package, Siemens Energyand Automation Incorporated Analytical Instrumentation(2000).

0. V. Volkov, M. Rico, A. Méndez-Blas, and C. Zaldo,“Preparation and properties of disordered NaBi�XO4�2, X=W or Mo, crystals doped with rare earths,” J. Phys. Chem.Solids 63, 95–105 (2002).

1. M. Luomajarvi, E. Rauhala, and M. Hautala, “Oxygendetection by non-Rutherford proton backscattering below2.5 MeV,” Nucl. Instrum. Methods Phys. Res. B 9, 255–258(1985).

2. Y. Cassagnou, J. M. F. Jeronymo, G. S. Mani, A. Sadeghi,and P. D. Forsyth, “The 7Li�p ,� alpha reaction,” Nucl.Phys. 33, 449–457 (1962).

3. D. Heck, “Three-dimensional lithium microanalysis by the7Li�p ,�,” Nucl. Instrum. Methods Phys. Res. B 30,486–490 (1988).

4. A. Sagara, K. Kamada, and S. Yamaguchi, “Depth profilingof lithium by use of the nuclear reaction 7Li�p ,�4He,”Nucl. Instrum. Methods Phys. Res. B 34, 465–469 (1988).

5. G. Vizkelethy, “Simulation and evaluation of nuclearreaction spectra,” Nucl. Instrum. Methods Phys. Res. B 45,1–5 (1990).

6. J. Hanuza, M. Maczka, and J. H. van der Mass, “PolarizedIR and Raman spectra of tetragonal NaBi�WO4�2,NaBi�MoO4�2 and LiBi�MoO4�2 single crystals withscheelite structure,” J. Mol. Struct. 348, 349–352 (1995).

7. J. Hanuza, A. Haznar, M. Maczka, A. Pietraszko, A.Lemiec, J. H. van der Maas, and E. T. G. Lutz, “Structureand vibrational properties of tetragonal scheeliteNaBi�MoO4�2,” J. Raman Spectrosc. 28, 953–963 (1997).

8. M. Maczka, E. P. Kokanyan, and J. Hanuza, “Vibrationalstudy and lattice dynamics of disordered NaBi�WO4�2,” J.Raman Spectrosc. 36, 33–38 (2005).

9. A. Waskowska, L. Gerward, J. Staun Olsen, M. Maczka, T.Lis, A. Pietraszko, and W. Morgenroth, “Low-temperatureand high-pressure structural behaviour ofNaBi�MoO4�2—an X-ray diffraction study,” J. Solid StateChem. 178, 2218–2224 (2005).

0. P. V. Kletsov, V. A. Vinokurov, and R. F. Klevtsova, “Doublemolybdates and tungstates of alkali metals with bismuth,M+Bi�TO4�2,” Z. Kristallogr. 18, 1192–1197 (1973).

1. J. Liu, J. M. Cano-Torres, F. Esteban-Betegón, M. D.Serrano, C. Cascales, C. Zaldo, M. Rico, U. Griebner, and V.Petrov, “Continuous-wave diode-pumped operation of anYb:NaLa�WO4�2 laser at room temperature,” Opt. LaserTechnol. (available online December 9, 2005).

2. B. R. Judd, “Optical absorption intensities of rare-earthions,” Phys. Rev. 127, 750–761 (1962).

3. G. S. Ofelt, “Intensities of crystal spectra of rare-earthions,” J. Chem. Phys. 37, 511–520 (1962).

4. C. Görller-Walrand and K. Binnemans, “Spectralintensities of f– f transitions,” in Handbook on the Physicsand Chemistry of Rare Earths, K. A. Gschneider, Jr. and L.Eyring, eds. (Elsevier Science, 1998), Vol. 25, p. 101.

5. M. J. Weber, “Probabilities for radiative and nonradiativedecay of Er3+ in LaF3,” Phys. Rev. 157, 262–272 (1967).

6. D. E. McCumber, “Einstein relations connecting broadbandemission and absorption spectra,” Phys. Rev. 136,A954–A957 (1964).

7. J. Koetke and G. Huber, “Infrared excited-state absorptionand stimulated emission cross sections of Er3+ dopedcrystals,” Appl. Phys. B 61, 151–158 (1995).

8. P. Porcher, M. Couto dos Santos, and O. Malta,“Relationship between phenomenological crystal fieldparameters and crystal structure. The simple overlapmodel,” Phys. Chem. Chem. Phys. 1, 397–405 (1999).

9. Z. Cheng, Q. Lu, S. Zhang, J. Liu, X. Yi, F. Song, Y. Kong, J.Han, and H. Chen, “Growth and properties of NaEr�WO4�2crystals,” J. Cryst. Growth 222, 797–800 (2001).