Optical pumping of charged excitons in unintentionally doped InAs quantum dots
G. Munoz-Matutano, B. Alen, J. Martinez-Pastor, L. Seravalli, P. Frigeri, S. Franchi
aa r X i v : . [ c ond - m a t . m e s - h a ll ] A p r Optical pumping of charged excitons in unintentionally doped InAs quantum dots
G. Mu˜noz-Matutano, B. Al´en, ∗ and J. Mart´ınez-Pastor ICMUV, Instituto de Ciencia de Materiales, Universidad de Valencia, P.O. Box 2085, 46071 Valencia, Spain.
L. Seravalli, P. Frigeri, and S. Franchi
CNR-IMEM Institute, Parco delle Scienze 37/a, 1-43100 Parma, Italy. (Dated: November 15, 2018)As an alternative to commonly used electrical methods, we have investigated the optical pumpingof charged exciton complexes addressing impurity related transitions with photons of the appropriateenergy. Under these conditions, we demonstrate that the pumping fidelity can be very high whilestill maintaining a switching behavior between the different excitonic species. This mechanism hasbeen investigated for single quantum dots of different size present in the same sample and comparedwith the direct injection of spectator electrons from nearby donors.
PACS numbers: 73.63.Kv, 81.07.Ta, 78.67.Hc
Nowadays, InAs/GaAs self-assembled quantum dots(QDs) are well known nanostructures with importantapplications envisaged within the quantum computationand cryptography fields.
The singly charged excitonstate (trion), either positive or negative, is of particularimportance because it lacks fine structure splitting, en-abling the efficient generation of single photons, and alsobecause, after radiative recombination, it leaves behind asingle charge with well defined spin. Therefore, there isan increasing interest in the electrical or optical controlof the exciton charge state as a necessary step for the spinmanipulation. The charge in QD states can be electri-cally controlled by tuning the gate voltage in field effectstructures embedding intrinsic QD layers.
However,this method can produce undesired effects like the reduc-tion of the oscillator strength induced by the externalfield. The charge state can also be controlled by opticalinjection, and different charging schemes have been pro-posed using above or below barrier excitation.
In this work, we demonstrate the selective formation ofcharged exciton complexes in initially empty QDs underthe presence of unintentional acceptor and donor impu-rities. Furthermore, the optical pumping mechanism isinvestigated for two ensembles of InAs QDs with very dif-ferent size present in the same sample: small QDs emit-ting below 970 nm and large QDs emitting at 1165 nmat 4 K.The MBE (molecular beam epitaxy) growth startswith a 100 nm-thick GaAs buffer grown at 600 C, followedby InAs deposition at 505 C and at very low growth rate(0.009 ML/s) and ends with a 100 nm-thick GaAs capgrown by atomic layer MBE at 360 C. During the InAsdeposition, the substrate was not azimuthally rotated,thus producing a continuous variation of InAs coverageson the sample surface. The combination of low growthrate (LGR) and graded coverage allowed us to obtainparticularly low surface density values, down to 2 µ m − ,suitable for optical investigation of isolated QDs. In par-ticular, the coverage of the sample under considerationhere is 2.5 MLs, with a density of about 16 µ m − , asestimated by atomic force microscopy (AFM) measure- L-QDS-QD (a) (D ,C )SQD I P L ( a r b . un i t s ) Energy (eV)
40 P WL ( X,A )(e,C ) LO (e,C ) S-QD
10 P P L-QD I P L ( a r b . un i t s ) Energy (eV) (c) (b)
FIG. 1: (a) 1x1 µm AFM image of a similar uncapped sam-ple. Representative QDs of the two families coexisting in thesample have been encircled; (b) PL spectra obtained at 10 Kand two excitation powers showing the emission bands corre-sponding to each family; (c) Detail of the PL spectrum in theenergy range between the WL and the GaAs edges. ments shown in Fig. 1(a) carried on uncapped samples.The AFM images also evidence a bimodal distributionof QD sizes, with most frequent values of 9 and 14 nmfor the heights and 36 and 54 nm for the diameters ofsmall (SQDs) and large (LQDs) quantum dots, respec-tively. Both the bimodal size distribution and the rela-tively large values for the QD dimensions have been re-ported for similar nanostructures grown by LGR.
Conventional photoluminescence (PL) characteriza-tion was carried out with the sample held in the cold fin-ger of a closed-cycle He cryostat. Single QD spectroscopywas performed by using an optical fiber based diffractionlimited confocal arrangement inserted in the He exchangegas chamber of an immersion cryostat. The PL signal,excited by a tunable Ti:sapphire, was dispersed by a 0.5(0.3) m focal length grating spectrograph and detectedwith a cooled InGaAs focal plane array (Si CCD) forwavelengths above (below) 1000 nm. The excitation ofthe PL spectrum ( PLE) is acquired using the same de-tectors while varying the excitation wavelength.Figure 1(b) shows the ensemble PL spectra recordedat 10 K using two different excitation power densities(P0 = 0.5 W/cm2) at 790 nm. Two relatively broademission bands are observed at 1.08 eV and 1.38 eV,corresponding to the two different QD families observedby AFM. Excitation above the GaAs band edge allowsalso for the observation of the WL line at 1.425 eV andthree other bulk related optical transitions, as shown inFig. 1(c). Three bands are clearly observed correspond-ing to the GaAs exciton bound to neutral acceptor at1.513 eV (X, C ), the free electron-neutral acceptor tran-sition at 1.493 eV (e, C As ), and its LO phonon replicaat 1.457 eV. The unintentional incorporation of impu-rities (such as carbon acceptors) coming from the growthenvironment is a general feature of MBE, as well as ofall growth techniques. In our case, Hall measurementsof similarly grown GaAs buffer layers reveal a residualn-type carrier concentration n = N D − N A cm − .Thus, the band centered at 1.493 eV is related to the(e, C As ) and ( D , C As ) recombination paths, and, inthermal equilibrium, a large number of ionized acceptorsare available due to the compensation process enablingthe efficient optical pumping of free electrons (and boundholes) as explained below.Figures 2(a) and 2(b) show the PL spectra recorded attwo different excitation energies for two individual QDsof the small and large QD ensembles, respectively. Uponexcitation at 1.53 eV, we find characteristic ”spectral linesets” throughout the sample surface. Neutral exciton (X)and biexciton (XX) features are easily identified by theslope of their integrated intensity dependence with ex-citation power ( I XX ∼ I X ) [Fig. 2(c)]. Yet, the addi-tional spectral lines observed at both sides of the neutralexciton and showing a linear behavior with power cancorrespond to either negatively charged ( X − n ) or posi-tively charged ( X + n ) excitons. In principle, the residualn-type doping of our sample should favor the captureof additional electrons by QDs. However, at low tem-peratures, this effect competes with the trapping of thesame electrons by ionized donors. This equilibrium canbe disrupted, and the population of free electrons canbe increased, by resonantly pumping the optical transi-tions related to ionized acceptors (( C − As , e), ( C − As , D + ))at 1.49 eV. In such situation, we observe that the lowenergy peaks in both spectral line sets are enhanced, asshown in Figs. 2(a) and 2(b). Peaks labeled A and A’are thus related to radiative recombination of negativetrions, X − , with binding energies E SQDX − = 7.5 meVfor the SQD, and E LQDX − = 3.7 meV for the LQD. Anadditional peak, not observed exciting above the GaAsbarrier, appears now at a lower energy than the negativetrion for the large QD [Fig. 2(b)]. Following our argu- X -1 XXX
Energy (eV) I P L ( a r b . un i t s ) X Energy (eV) X -1 X -2T I P L ( a r b . un i t s ) Excitation Intensity ( m W ) Excitation Intensity ( m W ) (a) (b)(c) (d) A~1 C~1.6B~0.8 A’~1B’~0.8 C’~0.8E exc =1.53 eV E exc =1.49 eV
D’ A’ B’C’CA B E exc =1.53 eVE exc =1.49 eV FIG. 2: The upper panels show typical PL spectra obtainedat 5 K using two different excitation energies for individualdots of the SQD (a) and LQD (b) families. Figures (c) and(d) represent the excitation power dependence of the mostimportant transitions under above barrier excitation for SQDand LQD, respectively. ment, it can be tentatively attributed to the emission ofnegative doubly charged excitons recombining on theirtriplet state with E SQDX − T = 5 meV. All the other peaksare partially (peaks B and B’) or totally (peaks C andC’) quenched upon resonant excitation on the impurityrelated optical absorption. In our pumping scheme, thisis the expected behavior for neutral excitons (B and B’),biexcitons (C) and positive trions (C’).The splitting energies just found, 7.5/0.0/-1.0 meV forthe X − /X/XX spectral line set, are typical of as-grownsmall quantum dots emitting at this energy. For thelarge QDs, we have found 5.0/3.7/0.0/-1.0 meV for the X − T /X − /X/X +1 set, which also agree with recent re-sults reported for this kind of large quantum dots (lessstudied in the recent literature). An enhancement of the negative trion luminescenceupon excitation below the GaAs barrier has been re-ported by Moskalenko et al , and by Chang et al , forQDs emitting around 1.3 eV.
To enable the resonantpumping of electrons to the conduction band, they con-sider the partial ionization of neutral acceptors by thesurface electric field. In our case, electron transfer to-wards QDs is warranted by the residual n-type dopingof our sample. On one hand, it implies a reservoir of N A ionized acceptors for the optical pumping scheme ex-plained above. On the other hand, even in absence oflight, N D − N A donors still contain an electron ready forbeing captured by the QDs if they were sufficiently closeto the investigated dot. To illustrate the difference be-tween both effects, in Fig. 3 we analyze the PLE spectracorresponding to the spectral line sets identified in Fig. 2(LQD and SQD1), and, for comparison, we also includethe PLE spectrum of a different dot (SQD2) which ex-hibit a clear signature of electron injection from a donorimpurity.First, in Fig. 3(a) we show the integrated PLE spectraobtained adding up the intensity of all different lines de-tected for each dot. All three spectra have been normal-ized to their maxima and exhibit spectral features clearlycorrelated with the emission bands shown in Fig. 1(c) andincluded, as a shadowed spectrum, in Fig. 1. Togetherwith the GaAs and heavy hole WL ( HH W L ) transitions,we found strong absorption at 1.49 eV and 1.46 eV,and most remarkably at 1.477 eV, which we assign tothe light hole WL transition ( LH W L ) reported at thisenergy. Yet, the most important conclusions regardingthe charge switching effect can be extracted from panels(b) and (c) of the same figure. We calculate the opti-cal pumping efficiency for the different charged excitoncomplexes by evaluating the intensity ratio η = X n − X X n + X as a function of the excitation energy. In Fig. 3(b), weobserve that for the negative trion η finds a clear max-imum at 1.493 eV for SQD1 (solid line), just where thegeneration of free electrons is expected through opticalpumping of ionized acceptors. The fidelity of the pro-cess is 85 % and spans over a spectral window of 24 meV(full width at half maximum) around the (e, C As ) band.A similar result is obtained for LQD as shown in Fig.3(c). Although, in this case, the effect is less pronouncedand occurs in a broader range around 1.485 eV. Out ofthe impurity window, the pumping efficiency for X − decreases and finds its minimum at the GaAs and WLband edges. The behavior of SQD2 is strikingly differ-ent as shown by dashed line in the same figure, and lessfrequent among the SQDs studied in the sample. Witha similar emission energy (1.294 eV) and binding energy(6 meV), the pumping efficiency of the negative trion forSQD2 exhibits an almost flat dependence with excitationenergy. The high fidelity (96 %) only drops appreciablybelow the HH W L transition and towards the GaAs bar-rier. The most likely explanation for this behaviour isthe continuous injection of a spectator electron from anearby neutral donor with a yield higher than the ra-diative rate of the neutral exciton. Our result indicatesthat in applications that would need the preparation ofcharged exciton complexes with high fidelity, modulationdoping of the active region can surpass other mechanismsin a broad excitation window, yet the optical pumpingscheme at the acceptor level can be more flexible whenmore than one complex has to be addressed in the samequantum dot.Finally, it should be noted that in both SQDs andthe LQD, the negative species are largely depleted nearthe band edges. Indeed, for the latter we can follow thepumping efficiency of the positive trion to find the oppo-site trend, as shown by the dotted line in Fig. 3(c). At theband edges, the local density of states is very large andexcitons are photocreated with nearly zero momentum. X +1 LQD X -2 X -1 X n - X / X n + X
62 % -1.0-0.50.00.51.0
SQD1SQD2 X - - X / X - + X
96 % 85 % -1.0-0.50.00.51.0 P L I n t . ( a r b . un i t s ) Photon Energy (eV)
LQDSQD1 m P L E I n t . ( a r b . un i t s ) SQD2 LH WL (a)(b)(c) FIG. 3: The integrated (see text) µ PLE spectra obtained at 5K for two small quantum dots (SQD1-2) and one large quan-tum dot (LQD) are shown. The optical pumping efficiencyof the different charged exciton complexes is represented forSQD1 and SQD2 in (b) and for LQD in (c). For compari-son, the ensemble PL spectrum shown in Fig. 1(c) has beenalso included in the background in logarithmic scale (solidspectrum).
On our sense, one possible explanation is that a largenumber of photocreated electrons could be trapped onthe ionized donors before relaxing into quantum dots faraway. This would lead to a decreased population of elec-trons inside the quantum dots. Assuming a shorter cap-ture time for electrons than for holes in their respectiveionized impurities, in average, this process will produce ahigher probability of neutral or positive trion recombina-tion at the band edges. This is a reasonable hypothesisgiven the larger concentration and shallower binding en-ergy of donors in our case.In summary, optical pumping of charged exciton com-plexes in single InAs QDs has been demonstrated. In thepresence of acceptor and donors in the surroundings ofthe QDs, exciton charge preparation can be nicely con-trolled by using photons of the appropriate energy. Thecharge mechanism has been compared for two kinds ofQDs, and for two different regimes of carrier injection,finding a consistent behavior in both. We demonstratethat the pumping fidelity can exceed 85 % enabling theprecise control of the charge state in quantum informa-tion applications.The authors gratefully acknowledge financial supportby the Spanish MEC through projects TEC-2005-05781-C03-03 and NAN 2004-09109-C04-04, by the ItalianMIUR through the FIRB project ”Nanotecnologie e Nan-odispositivi per la Societ dell’Informazione” and by theEuropean Commission through SANDiE Network of Ex- cellence (NMP4-CT-2004-500101). ∗ [email protected]; Permanent address: IMM, Insti-tuto de Microelectr´onica de Madrid (CNM, CSIC), IsaacNewton 8, 28760 Tres Cantos, Madrid, Spain. A. Imamoglu, D. D. Awschalom, G. Burkard, D. P. Di-Vincenzo, D. Loss, M. Sherwin, and A. Small, Phys. Rev.Lett. , 4204 (1999). R. M. Stevenson, R. J. Young, P. Atkinson, K. Cooper,D. A. Ritchie, and A. J. Shields, Nature , 179 (2006). M. Atat¨ure, J. Dreiser, A. Badolato, A. H¨ogele, K. Karrai,and A. Imamoglu, Science , 551 (2006). R. J. Warburton, C. Sch¨aflein, D. Haft, F. Bickel, A. Lorke,K. Karrai, J. M. Garc´ıa, W. Schoenfeld, and P. M. Petroff,Nature , 926 (2000). J. J. Finley, P. W. Fry, A. D. Ashmore, A. Lemaitre, A. I.Tartakovskii, R. Oulton, D. J. Mowbray, M. S. Skolnick,M. Hopkinson, P. D. Buckle, et al., Phys. Rev. B ,R161305 (2001). B. Al´en, J. Mart´ınez-Pastor, D. Granados, and J. M.Garc´ıa, Phys. Rev. B , 155331 (2005). B. Al´en, J. Bosch, J. Mart´ınez-Pastor, D. Granados, J. M.Garc´ıa, and L. Gonz´alez, Phys. Rev. B , 045319 (2007). A. Hartmann, Y. Ducommun, E. Kapon, U. Hohenester,and E. Molinari, Phys. Rev. Lett. , 5648 (2000). D. V. Regelman, E. Dekel, D. Gershoni, E. Ehrenfreund,A. J. Williamson, J. Shumway, A. Zunger, W. V. Schoen-feld, and P. M. Petroff, Phys. Rev. B , 165301 (2001). E. S. Moskalenko, K. F. Karlsson, P. O. Holtz, B. Mone- mar, W. V. Schoenfeld, J. M. Garc´ıa, and P. M. Petroff,Phys. Rev. B , 085302 (2001). W. H. Chang, H. S. Chang, W. Y. Chen, T. M. Hsu, T. P.Hsieh, J. I. Chyi, and N. T. Yeh, Phys. Rev. B , 233302(2005). N. I. Cade, H. Gotoh, H. Kamada, T. Tawara, T. Sogawa,H. Nakano, and H. Okamoto, Appl. Phys. Lett. , 172101(2005). F. Briones and A. Ruiz, J. Cryst. Growth , 194 (1991). M. Colocci, F. Bogan, L. Carraresi, R. Mattolini, A. Bosac-chi, S. Franchi, P. Frigeri, M. Rosa-Clot, and S. Taddei,Appl. Phys. Lett. , 3140 (1997). G. Costantini, C. Manzano, R. Songmuang, O. Schmidt,and K. Kern, Appl. Phys. Lett. , 3194 (2003). Y. Nakata, K. Mukai, M. Sugawara, K. Ohtsubo,H. Ishikawa, and N. Yokohama, J. Cryst. Growth ,93 (2000). L. Pavesi and M. Guzzi, J. Appl. Phys. , 4779 (1994). S. Rodt, A. Schliwa, K. P¨otschke, F. Guffarth, and D. Bim-berg, Phys. Rev. B , 155325 (2005). C. D. Savio, K. Pierz, G. Ade, H. U. Danzebrink, E. O.G¨obel, and A. Hangleiter, Appl. Phys. B , 84 (2006). N. I. Cade, H. Gotoh, H. Kamada, H. Nakano, andH. Okamoto, Phys. Rev. B , 115322 (2006). A. Winzer, R. Goldhahn, G. Gobsch, H. Heidemeyer,O. Schmidt, and K. Eberl, Physica E13