394OPTICS LETTERS /Vol.27,No.6/March 15,2002
Continuous-wave ultraviolet laser action in strongly scattering
Nd-doped alumina
B.Li,G.Williams,and S.
C.Rand
Division of Applied Physics,Randall Laboratory,University of Michigan,Ann Arbor,Michigan 48109–1022
T.Hinklin and R.M.Laine
Department of Materials Science &Engineering,University of Michigan,Ann Arbor,Michigan 48109
宿迁网络问政Received September 17,2001 We report electrically pumped,cw laser action near 405nm from Nd 31-doped d -alumina nanopowders.To our knowledge,this is the first report of stimulated emission from the high-lying 2F -
excited states,achieved through feedback from strong elastic scattering of light over transport path lengths shorter than half a wave-length.©2002Optical Society of America OCIS codes:140.0140,140.5680,140.3610.
Nd solid-state lasers customarily operate on the 4
F 3/24I 11/2transition of the Nd 31ion and provide reliable sources for metrology,quantum optics,high-field studies,high-power industrial,and other applications.For infrared operation,Nd lasers are sustained by population inversion of a relatively low energy state that is easily excited by f lash lamps or diodes.Operation at shorter wavelengths has been reported through the use of internal doubling 1and upconversion,2but to our knowledge laser action from the long-lived,high-lying 2F levels located above 38,000cm 21has never been reported,because of the difficulty of pumping them efficiently.Inversion of these levels opens up new possibilities,however,for visible and ultraviolet sources based on Nd 31.Here we report continuous-wave laser action near 405nm from the 2F 5/2level of Nd 31in d -Al 2O 3nanopowders as well as advances in the use of electrical pumping and strong electromagnetic localization 3,4that yield continuous nanoscopic laser action in a dielectric host without any external feedback structure.
Highly scattering powders are diff icult to pump op-tically because incident light does not penetrate the medium well,particularly in the regime of strongly localized electromagnetic transport,when the at
tenua-tion distance is less than a wavelength ͑l ͒.Then,irre-spective of incident intensity,light penetrates a short,fixed distance ͑ϳl ͒below the surface and is almost en-tirely ref lected.Instead,we used electron pumping to energize high-lying states of Nd 31in strong scattering dielectric powders.Electrons have penetration depths that are voltage dependent and typically exceed an op-tical wavelength in the range 1–10keV.Hence they offer control over the positioning of the pumped vol-ume with respect to the surface of the medium.In this Letter we show that mild electron excitation in doped oxide nanocrystals readily generates continuous stim-ulated emission at room temperature in samples that are free of absorption and yet have transport mean free paths shorter than l ͞2.
Our samples consisted of unaggregated,single-crys-tal d -alumina particles 5–7with a dopant level of 1000parts in 106͑ppm ͒Nd 31ions.The size distribution was log normal,with 240Nd 31ions per particle of mean diameter 27nm.Powders were packed (at
CMEϳ30psi)into the recesses of an oxygen-free copper platen in an ultrahigh-vacuum chamber at ,1029Torr.A steerable beam of 2–10-keV electrons was lightly focused to a spot diameter of 1–2mm on the sample.Luminescence was analyzed with a 1-m Czerny–Turner grating spectrometer.The beam current was monitored with a Faraday cup and precisely controlled with a calibrated grid voltage,yielding f luctuations of less than 1%͞h and long-term reproducibility better than 3%.
Cathodoluminescence (CL)spectra,recorded for various beam currents and voltages in as-pre-pared powder samples,were helpful in assigning Nd 314f 4f transitions and quenching dynamics,which are essential for understanding the mecha-nism of light emission.In Fig.1,two spectra are shown at different currents in one sample.At 3.3m A,the largest peaks in the spectrum occur above 25,000cm 21.At 16.9m A the largest peaks appear below 25,000cm 21and the original intense lines have vanished.Such dramatic spectral quenching is unexpected in phosphors and is difficult to understand without detailed transition assignments.Spectro-scopic studies of RE 31:d Al 2O 3were not possible in the past,owing to the difficulty of preparing doped,d -phase crystals by traditional growth methods.This forced us to make assignments in the powders by other means.RE 31emission wavelengths can be predicted within a few nanometers in most materials,
8
Fig.1.CL spectra of Nd :d 2Al 2O 3nanoparticles ͑f 27nm ͒at I 3.3m A and I 16.9m A at 8kV,showing intensity reversals near 25,000cm 21.0146-9592/02/060394-03$15.00/0
©2002Optical Society of America
March 15,2002/Vol.27,No.6/OPTICS LETTERS 395
but confusion with overlapping transitions can be avoided only by precise determination of additional
characteristics or splittings,for example,the interval between the 2F 5/2and 2F 7/2multiplets.An important determination of this type can be made by observation of cathodoluminescent transitions from 2F 5/2and 2F 7/2levels to one and the same state,for instance,singlet state 2P 1/2(Fig.2).2F 2P 1/2transitions can be iden-tified by comparison of CL with photoluminescence spectra obtained by use of UV excitation at 363nm that generates emission lines from all the radiative states lying below the 2F multiplets.Emission fea-tures that appear in CL but not in photoluminescence can then confidently be assigned as 2F emissions.Because Nd is Kramers degenerate,its Stark sub-levels are ͓J 1͑1͞2͔͒degenerate,so three 2F 5/22P 1/2transitions to the long-wavelength side of 660nm and four 2F 7/22P 1/2transitions to the short-wavelength side are expected.The figure shows how observation of all seven features,especially the lowest energy features of each set,reveals the interval D between the 2F 5/2and 2F 7/2multiplets.The value determined from the 2P 1/2features is D 644615cm 21.
道家导引术
In Fig.3we note that near 25,000cm 21,where the 2
F 5/24F 9/2,2F 5/24F 7/2and 2F 7/24F 9/2transitions are expected,the 2F 5/22F 7/2interval occurs again.In this range the observed interval is D 0664615cm 21,so that D D 0within experimental error.Although we cannot rule out assignments of some lines above 25,000cm 21to 2F 5/24F 7/2,the correspondence between these intervals provides independent cor-roboration that the upper levels of all the emission features in Fig.3are indeed 2F 5/2or 2F 7/2.A plot of the luminescent intensities observed on transitions from either of these two high-lying multiplets (where unambiguous assignments can be made)as a function of beam current is presented in Fig.4.Note that,regardless of which 2F multiplet is the upper state,all visible and UV 2F transitions quench rapidly at the threshold current of ϳ3.5m A,except one.The fact that 2F 7/2transitions quench when 2F 5/2transitions do should not be too surprising,since the Boltzmann factor (exp ͑2¯h D ͞k B T ͒ϳ0.04at room temperature)for occupation of the upper multiplet is significant,causing its population to be proportional to that of 2
F 5/2.However,the observation that the 2F 5/24F 9/2transition grows above a well-defined threshold
cur-rent while the intensities on other transitions from the same initial upper state (or its thermally connected companion)vanish is uniquely attributable to stimu-lated emission.Only stimulated emission can alter the effective branching ratios for radiative decay from a single level to several others.Consequently,Figs.3and 4provide compelling evidence for stimulated emission at 409.6,408.1,and 404.1nm above ϳ4m A.An important aspect of the current dependence of intensity on the 2F 5/24F 9/2transition (Fig.4)is its linearity above threshold.At 3.5m A,the intensity of this line begins to increase linearly until it satu-rates near 12m A.Linearity is characteristic of oscillators,whereas exponential growth typifies am-plifiers.This reveals that very effective feedback assists in the attainment of threshold in our samples.
Theoretically,feedback may arise from internal ref lec-tion at the vacuum –powder interface (a one-mirror laser),from ref lection among particles located on a macroscopic closed path (ring laser),or from strong multiple scattering (localization).However,Figs.3and 5show that,when electrons activate ions at depths greater than the transport distance of light,emission increases and threshold decreases.This improvement in laser characteristics as the
gain
Fig.2.Identification of 2F emissions to singlet state 2P 1/2of Nd 31.Top curve,CL including 2F multiplet emission;bottom curve,photoluminescence excluding 2F
emission.
Fig.3.Voltage dependence of CL spectral peaks in the 25,000-cm 21region.Features above 25,000cm 21grow rapidly up to 5kV and then quench.Features below 25,000cm 21are absent until 5kV.The splitting between these groups is labeled D 0
Fig.4.Current dependence of ultraviolet and visible CL intensities of the 2F 5/2and 2F 7/2states at 8kV.All curves (but one)quench rapidly above 3m A,where intensity on the 2F 5/24F 9/2transition undergoes an abrupt change in slope.
男峰论坛396OPTICS LETTERS/Vol.27,No.6/March15, 菜园保卫战
2002
Fig.5.Luminescent output at4and8kV in Nd:d2Al2O3.The current range is greater at8kV than at4kV because of extended linear grid control.Inset, penetration depth versus voltage,calculated with a low-energy Monte Carlo routine.
volume recedes(by several attenuation lengths)below
the surface indicates that surface ref lection cannot
account for the feedback.Since the particles are less than l͞15in diameter and are densely packed,it is
also evident that morphological resonances cannot pro-
vide efficient feedback.Unlike earlier experiments in which mode structure and coherence of pulsed powder
lasers were reported,9–11our observations are not con-sistent with light propagation around a closed path.
As we discuss below,however,the possibility that
multiple scattering gives rise to a nonpropagating, stimulated field through strong localization is entirely
consistent with our observations.
Scattering conditions were evaluated by coherent backscattering(CBS)techniques and have been re-
ported elsewhere,6so we only summarize key results
here.First,CBS data from our samples showed a broad backscattering cone,as expected for small
particles with a short mean free transport length,
lء.We deduced a value of lء174631nm,using standard analysis at l488nm.12This value is
somewhat less than l͞2.Second,the CBS cusp was
triangular at small angles.This permitted us to estimate a limit for losses in terms of a characteristic absorption length,l a.1cm.
The observations of spectral quenching(Figs.1and 3),emission thresholding,and linear output above
threshold(Fig.4)collectively furnish clear evidence of Nd31-laser action in the4–10-keV range.Since conventional cavities support cw laser action only with high-ref lectivity mirrors,these observations alone imply that light experiences high ref lectivity in all directions within our samples.However,this point is made more forcefully by our CBS measurements, which show that light in our samples propagates forward less than half a wavelength regardless of direction or polarization.Given that the coherence length is limited to a subwavelength value l c,lء,l in the absence of significant absorption͑l,,l a͒,light generated by impurity ions within our powders nec-essarily13acquires a spatial distribution resembling that of a three-dimensional evanescent wave.
We conclude that laser action in our samples results
from strong localization.In our experiments,emis-sion is evidently stimulated coherently over distance scales of tens of nanometers,but spatial random-ization on length scales of l͞2or more precludes directionality or mode selectivity.Backscattered light(and laser output)in our experiments is nearly speckle free,consistent with a subwavelength effective coherence length.All frequencies within the lumi-nescent linewidth experience comparable gain and feedback,independently of observation angle,just as expected for continuous,truly random laser action. Our experiments demonstrate cw laser action at new wavelengths near405nm in Nd31-doped nanopowder at room te
mperature for what is to our knowledge the first time.The mild electrical pumping conditions necessary to achieve this result lend support to the idea that the threshold for laser action is lowered by the onset of strong scattering.We reached similar conclusions earlier for Pr31-and Ce31-doped powders,6 at visible and ultraviolet wavelengths,making it apparent that doped,dielectric nanophosphors may provide an entire family of bright omnidirectional, speckle-free,monochromatic sources for improved lighting,chemical sensors,displays,short-range air-craft or space communication,nanoscale lithography, and other applications.
The authors gratefully acknowledge research sup-port from the U.S.Air Force Office of Scientific Research(F49620-99-1-0158),the National Science Foundation and DMR9975542),and the Army Re-search Office(DAAD19-99-1-0229).Monte Carlo calculations of electron penetration were performed by H.-Y.Chan.S. C.Rand’s e-mail address is scr@eecs.umich.edu.
References
1.T.Y.Fan,A.Cordova-Plaza,M.J.F.Digonnet,R.L.
Byer,and H.J.Shaw,J.Opt.Soc.Am.B3,140(1986).
2.R.M.Macfarlane, F.Tong, A.J.Silversmith,and
W.Lenth,Appl.Phys.Lett.52,1300(1988).
3.S.John,Phys.Rev.Lett.53,2169(1984).
4.P.W.Anderson,Philos.Mag.B52,505(1985).
5.G.Williams,S.C.Rand,T.Hinklin,and R.M.Laine,in
Conference on Lasers and Electro-Optics(CLEO/US) (Optical Society of America,W ashington,D.C.,1999), paper CTuG5.
6.G.R.Williams,B.Bayram,S.C.Rand,T.Hinklin,and
R.M.Laine,Phys.Rev.A65,013807(2001).
7.R.M.Laine,T.Hinklin,G.Williams,and S.C.Rand,
Mater.Sci.Forum344–346,500(2000).
8.C.Gorller-Walrand and K.Binnenans,in Spectral In-
tensities of f f Transitions,K.A.Gschneidner and L.Eyring,eds.,Vol.25of Handbook of Physics and Chemistry of Rare Earths(Elsevier,New Y ork,1998), pp.101–264.
9.H.Cao,Y.G.Zhao,S.T.Ho,E.W.Seelig,Q.H.Wang,
and R.P.H.Chang,Phys.Rev.Lett.82,2278(1999).
10.H.Cao,Y.Ling,J.Y.Xu,C.Q.Cao,and P.Kumar,
Phys.Rev.Lett.86,4524(2001).
11.R.K.Thareja and A.Mitra,Appl.Phys.B71,181
(2000).
12.J.X.Zhu,D.J.Pine,and D.A.Weitz,Phys.Rev.A44,
3948(1991).
13.J.W.Goodman,Statistical Optics(Wiley,New York,
1985),p.206.
豫公网安备41160202000603
站长QQ:729038198
关于我们
投诉建议