Letter pubs.acs.

org/NanoLett

Optical Pumping of a Single Electron Spin Bound to a Fluorine Donor in a ZnSe Nanostructure
Darin J. Sleiter,*, Kaoru Sanaka,, Y. M. Kim, Klaus Lischka, Alexander Pawlis,*,, and Yoshihisa Yamamoto,

E. L. Ginzton Laboratory, Stanford University, Stanford, California 94305, United States National Institute of Informatics, Hitotsubashi 2-1-2, Chiyodaku, Tokyo 101-8403, Japan Department of Physics, University of Paderborn, Warburger Strasse 100, 33098 Paderborn, Germany ABSTRACT: Here we demonstrate optical pumping of a single electron within a semiconductor nanostructure comprised of a single uorine donor located within a ZnSe/ ZnMgSe quantum well. Experiments were performed to detect optical pumping behavior by observing single photons emitted from the nanostructure when the electron changes spin state. These results demonstrate initialization and read-out of the electron spin qubit and open the door for coherent optical manipulation of a spin by taking advantage of an unconventional nanostructure. KEYWORDS: Optical pumping, donor, qubit, initialization, spin measurement

any proposed schemes for quantum information processing require scalable and homogeneous localized quantum bits (qubits) that can eciently interface with single photon ying qubits.15 Trapped ion qubits are extremely homogeneous and have a nearly perfect optical quantum eciency,68 but scalability is inherently dicult due to the need to trap and cool the ions. On the other hand, semiconductor nanostructures provide a much more scalable environment for quantum information. Electron spin qubits trapped in self-assembled InGaAs quantum dots (QDs) have a high optical quantum eciency and fast manipulation times, but suer from inhomogeneity due to their wide distribution in size when grown naturally.912 Charged nitrogen-vacancy (NV) centers in diamond are more homogeneous than quantum dots, but are more dicult to structure than more conventional semiconductors and have low optical quantum eciency.1316 Electron spins bound to individual donors in direct-bandgap semiconductors, however, are naturally homogeneous with a high-optical quantum eciency and in addition provide a scalable system.1721 Spin manipulation in QDs is typically carried out via a coupled optical lambda system provided by the ground-state transition of the QDs. A single electron trapped by a uorine donor in ZnSe forms a similar lambda system as used in QDs and has many attractive characteristics. The optical quantum eciency of the donor system is very close to unity21 and can be used to generate triggered single photons.19 Furthermore, it has recently been shown that photons from two independent donor nanostructures that each contain an isolated uorine atom can be indistinguishable and entangled through postselection.22 While coupling between the electron spin and the nuclear spin of the host crystal is the leading decoherence-causing mechanism in QDs,23 isotopic purication
2012 American Chemical Society

can be used to deplete the ZnSe host crystal of nuclear spins. This is not possible in IIIV semiconductors but has been very successful in extending the decoherence times of electrons in diamond and Si.25,26 Furthermore, F has a natural 100% abundance of spin-1/2 nuclei, which could be used as an additional qubit naturally coupled to the electron qubit.18,27,28 Finally, it has been recently shown that an ensemble of uorine donors in bulk ZnSe features long electron-spin dephasing 29 times T* 2 , greater than 30 ns for temperatures up to 40 K. Optically active donors have also been successfully implanted using ion implantation, which could eventually lead to deterministically placing single donors in specic locations.20 All of these characteristics make the ZnSe/F system an appealing qubit candidate, particularly for quantum repeater schemes. Here we present the experimental demonstration of optical pumping of the spin of a single electron trapped by a uorine donor placed within a ZnSe/ZnMgSe quantum well nanostructure. This can be used to initialize the spin qubit to a known state, which is the rst step of most quantum information processing schemes,1,2,5 and to provide a measurement (readout) of the spin state. Here, we rst describe the optical system of the donor-bound electron along with the nanostructure used to isolate the electron. Next, we present the evidence indicating that we address a single F-bound electron. Finally, we describe the optical pumping experiments we performed and discuss the experimental results that clearly indicate single electron optical pumping.
Received: October 2, 2012 Revised: November 20, 2012 Published: December 7, 2012
116
dx.doi.org/10.1021/nl303663n | Nano Lett. 2013, 13, 116120

Nano Letters When replacing a Se atom in ZnSe, a F atom will act as a donor (Figure 1a). The donor nucleus provides a shallow

Letter

hole. In our samples, the light-hole states are split o to higher energy by connement and strain in the nanostructure, leaving only the heavy-hole 3/2 spin states to be considered. The optical system of interest is thus composed of two D0 states and two D0X states. The orientation of the magnetic eld determines the allowed optical transitions.18 Unless otherwise noted, all of these results were obtained in Voigt geometry, where the magnetic eld is perpendicular to the growth direction and the optical access direction. Because of the fully connected pair of lambda systems in Voigt geometry, the spin of the D0 electron can be manipulated optically by transitions through the D0X state. To conrm the presence of isolated F donors, we observed the PL spectra of the nanodevices using a spectrometer and above-band (410 nm) pulsed excitation (Figure 2). We rst

Figure 1. (a) Crystal structure showing F acting as a donor when replacing a Se atom. (b) Schematic of the nanostructure, showing the QW with delta-doped F in the center. After etching into 100 nm mesas, there are only 2.4 donors per nanostructure on average. The passivation coating is added in order to protect the structure during thermal cycles. (c) Level diagram of the QW showing the D0 and D0X states. The dotted lines indicate the energies of the lowest QW particle-in-a-box states. (d) Zeeman splitting of the D0 and D0X states in Voigt geometry with a magnetic eld of 7 T. The dotted lines show the allowed optical transitions and H/V indicates the respective photon polarization.

potential well that is able to trap the donated electron with a binding energy of 29.3 meV in bulk,2,5,30 creating the neutral donor state (D0) shown in Figure 1c. At suciently low temperatures (all of the experimental data presented here were obtained at temperatures between 2 and 5 K), a F donor can trap an exciton in addition to the electron with a binding energy of 5 meV in bulk,24 creating the donor-bound exciton state (D0X). In bulk ZnSe, the D0X to D0 transition has an energy of 2.8 eV.24 Note that these emission and binding energies can be noticeably larger when the uorine donors are located within the excitonic connement of a quantum well (QW). Therefore, we combine the donor with a QW to form a donor-QW hybrid nanostructure that retains much of the desirable features of the donor system while allowing us to isolate a single donor-bound electron. The ZnSe/F nanostructure design used here (Figure 1b) is the same described in ref 18, where the isolation of a single donor-bound electron was conrmed. The sample used in the experimental results presented here consisted of a 2-nm ZnSe QW delta-doped with uorine and enclosed by Zn1xMgxSe (x 17%) cladding layers grown by molecular-beam epitaxy (MBE). The uorine areal -doping concentration was approximately 3 1010 cm2, resulting in 2.4 donors on average per 100 nm diameter mesa after structuring. The mesas were separated by 10 m, ensuring we could optically isolate photoluminescence (PL) from the individual mesa nanoemitters. When placed in a magnetic eld, both the D0X and D0 states split due to Zeeman splitting (Figure 1d). In the D0 state, the spin-1/2 electron splits into two states. In the D0X state, the two electrons combine together to form a spin singlet, and so the spin of the state is determined entirely by the spin of the
117

Figure 2. Simplied diagram of the experimental setup. The opticalpumping and spin-randomization lasers are incident on the nanostructure in a 4 K cryostat. Emitted photons are polarization ltered with a polarizing beam splitter (PBS) and frequency ltered using a pair of optical gratings and slits and then detected by either a single photon-counting module (SPCM) or a spectrometer.

looked for a mesa with a single peak in the PL spectra with the expected separation from the free exciton peak (515 meV) that is associated with the F-bound exciton transition in a QW. Saturation of this peak with increased pump power was observed, indicating that the excited state has a nite lifetime and number of occupiable states. Next, we used magnetophotoluminescence spectra to conrm the Zeeman splitting and polarization selection rules of the donor-bound exciton transition in both Faraday and Voigt geometry. We determined the electron and hole g-factors to be |ge| = 1.3 0.3, |g h | = 0.0 0.1 and |3g g | = 0.9 0.1, which is in good agreement h e with previous results.18 Finally, we observed a value of 0.22 0.06 in the normalized two photon correlation function g(2)() at zero delay (without background subtraction), well below the 0.5 threshold for a single emitter. Together, these pieces of evidence, all taken from the same mesa, give very high condence that this individual peak corresponds to the D0X to D0 transition of a single F-bound electron. Optical pumping was performed in Voigt geometry with a magnetic eld of 7 T by applying a continuous wave (CW) laser resonant with one of the D0X optical transitions, shown in Figure 3a. If the electron begins in state |1, it will be excited to |e1, where it will then decay back into either |1 or |0. If it falls into |1, it will be re-excited to |e1 and will eventually end in |0. If the electron begins in |0, the laser is far o resonance with any of the optical transitions connected to that state, and the electron will remain in state |0. In order to observe the optical pumping, we collected photons from the |e1 |0 transition. A single photon is emitted on this transition for each
dx.doi.org/10.1021/nl303663n | Nano Lett. 2013, 13, 116120

Nano Letters

Letter

Figure 3. Optical pumping behavior. (a) Level diagram of the nanostructure-bound electron showing which transition is optically pumped and which transition is monitored for single photon emission. (b) The optical pumping ratio versus the wavelength of the opticalpumping laser. is the ratio of the single photon emission count rate with the optical-pumping laser on to the count rate with the opticalpumping laser o. (c) versus the power of the optical-pumping laser with the spin-randomization laser power xed. (d) versus the power of the spin-randomization laser with the optical-pumping laser power xed.

optical pumping event. By accumulating photons from many individual optical pumping experiments, we determined the optical pumping rate and nal population of state |0. The timing of each individual optical pumping experiment was determined by the repetition rate of our pulsed spinrandomization laser. Every 13 ns, a picosecond above-band (410 nm) laser pulse puts the electron into either |e0 or |e1, randomly selected with approximately equal probability. The electron then decays into either |0 or |1, thus randomizing the spin state regardless of which state the electron was in prior to the spin-randomization pulse. The resonant spin-pumping laser is kept on continuously. Photons from the |e1 |0 transition are ltered out by polarization using a polarizing beam splitter (PBS) and by frequency using a pair of optical gratings and slits, as shown in Figure 2, and are then sent to either a spectrometer or a single photon counting module (SPCM). While the SPCM has timing resolution and better quantum eciency, the additional grating and slits of the spectrometer signicantly reduced the background noise due to photons scattered from the optical-pumping laser, which was separated by only 150 GHz from the signal photons. For each 13-ns experiment, assuming the electron within the nanostructure is excited every pulse, there is a 25% chance that a photon will be emitted from the |e1 |0 transition due to spontaneous emission just after the above-band pulse. In the absence of optical pumping, this is the only way a photon can be emitted from that transition. There is also a 50% chance that the electron will decay to |1 following the above-band pulse. If optical pumping is strong enough, an electron in state |1 will always be optically pumped within the 13-ns experimental window, emitting a photon along the |e1 |0 transition. Therefore, 75% of the experiments will emit a single photon with strong optical pumping, compared to only 25% in experiments without optical pumping. The ratio between
118

the count rate of single photons when the optical-pumping laser is on versus when it is o is a good measure of optical pumping, and it can take a value between 1 (no optical pumping occurring) and 3 (saturated optical pumping). This increase in the average number of photons emitted per experiment has been observed in our sample. In one set of experiments using the spectrometer, we measured 30.2 0.8 average counts/sec with only the spin-randomization laser on, 13.0 0.6 average counts/sec with only the optical-pumping laser on, 95.0 3.0 average counts/sec with both lasers on, and a background of 0.2 0.1 average counts/sec with both lasers o. Note that these measurements were taken within the linear response region of the spectrometer CCD and the count rate was calibrated against an optical power meter so that the spectrometer counts were proportional to the actual photon counts. In addition, the average measurement values and standard deviation were calculated statistically using multiple data sets, resulting in standard deviations smaller than the shotnoise expected for the CCD. Using the data from these experiments, we measure the optical pumping ratio to be (95.013.0)/(30.2 + 0.2) = 2.7 0.1, indicating nearsaturation optical pumping. We observed an optical pumping ratio larger than 2.3 when pumping on any of the four optical transitions, showing the fully connected nature of the optical system. The reason we observe 2.7 instead of 3 is due to the relationship between the power of the two laser incident on the nanostructure. At low spin-randomization laser powers, the D0X state is not created after every pulse, and so the electron state and the experiment are not reset. This serves to increase the average length of the experiment T and does not change the pumping rate from |1 |0. However, at low optical-pumping laser powers, the rate of optical pumping R is slow, reducing the probability of an electron in |1 being pumped into |0 within the experimental window. Together, R and T determine .
= 1 + 2(1 eTR ) R = P 1 T= 13[ns] 1 P2
(1) (2)

(3)

where P1 is the power of the CW optical-pumping laser and P2 is the average power of the pulsed spin-randomization laser. and are parameters that determine the relative eect of each laser (note that must be positive and less than 1). The experimental power dependence of is shown in Figure 3c,d. The theory curves are from eq 1 with and as tting parameters. Optical pumping is also a resonant eect, as seen in Figure 3b. In this set of data, the pulsed laser was held xed while the optical-pumping laser was set at dierent wavelengths with constant pumping power. The best-t curve is a Gaussian with a full-width at half-maximum (fwhm) of 58 GHz. This is 25 times larger than the lifetime-limited line width of 2.3 GHz (for a 70-ps lifetime19). The Gaussian shape and large line width suggest fast spectral diusion is occurring, resulting in timeensemble averaged measurements. The cause of the spectral diusion has not been systematically studied, but we hypothesize it is caused by Stark shifting from charge uctuations due to the combination of surface charge traps created through mesa structuring and the large number of free carriers created with the above-band laser pulses. The power
dx.doi.org/10.1021/nl303663n | Nano Lett. 2013, 13, 116120

Nano Letters density of the above-band laser pulses incident on the donor used in the results presented here was 104105 W/cm2. This is 23 orders of magnitude larger than the power density used in ref 19, where the line width was nearly lifetime-limited since indistinguishability was observed. Both experimental results are consistent with our hypothesis for the cause of the large line width seen here. We also observed the time-dependent behavior of optical pumping, as shown in Figure 4. The data show an initial pulse

Letter

Figure 4. Experimental (blue) and best-t simulated (red) photon arrival times during a 13 ns window in the case of (a) no optical pumping and (b) optical pumping behavior. The simulated data were averaged over 10 more experiments than the experimental data, resulting in the reduced variance.

due to photons collected from the spontaneous emission from the |e1 |0 transition after the spin-randomization pulse, followed by 13 ns of optical pumping. The signature of optical pumping is observed in the dierence between the data obtained with both lasers on, Figure 4b, and the data obtained with just the spin-randomization laser on, Figure 4a. The data were modeled using a Monte Carlo simulation which followed the state trajectory of the electron optical system both with and without optical pumping (Figure 4, red curves). The simulation rst initialized the electron into |e0 or |e1 with equal probability at t = 0. It was then transferred between the four states according to the allowed transition rates, and the time of jumps between |e1 and |0 was recorded as well as the nal electron state after 13 ns. The allowed transitions were (1) Rsp(|ei |j) (i,j = 0,1) spontaneous emission from either D0X state |ei into either D0 state |j with equal rate determined by the measured excited state lifetime of = 70 ps19 and (2) Rst(|ei |j) stimulated emission along all four transitions due to the optical-pumping laser. 1 R sp(|ei |j) = (4) 2
R st(|ei |j) = RXe /2
2 2

measured with just the resonant laser on, was added to the experimental data with no optical pumping and to the simulated data before tting. This Monte Carlo simulation was used to create a series of simulated data sets by varying two parameters: the total number of simulated experiments N, and the optical pumping rate R. The maximum-likelihood method was then used to nd the best t to the experimental data, and the 2 test was used to determine an interval of condence.31 From this analysis, we obtained a maximum likelihood pumping rate of 0.14 GHz with 95% condence that the pumping rate is between 0.008 and 0.75 GHz. The likelihood sharply drops at even lower pumping rates, giving 99% condence that the pumping rate is greater than 0.006 GHz. We can infer from the best-t simulation that the rate of transition from |1 to |0 was 1/15 ns1, and that after 13 ns of optical pumping, the electron ends in state |0 in 79% of the experiments. This rate and resulting population was limited by the available power of the optical-pumping laser incident on the sample in our setup and can likely be increased in future experiments. Simulations with higher optical pumping powers suggest this rate could be increased to greater than 1 ns1, which is eventually limited by the lifetime of the D0X state. In conclusion, we have experimentally demonstrated optical pumping of the single electron bound to a uorine donor in a ZnSe QW nanostructure through the D0X excited state to prepare one particular spin state of the electron qubit. The experiment shown here serves to initialize a spin qubit based on a single donor-bound electron in addition to indicating which state a qubit was in prior to optical pumping. Furthermore, the latter shows that electrons bound to a ZnSe/F QW nanostructure have the same state structure and optical transitions as electrons in InGaAs quantum dots, which have been successfully used to perform single-qubit operations in that system. The opportunities for nuclear spin depletion of the host crystal and donor ion implantation, and the possibility of using the donor nucleus as a naturally coupled long-lived spin qubit makes the ZnSe/F nanostructure a particularly appealing qubit alternative to QDs. Future experiments will be focused on coherent manipulation of the electron spin using fast optical pulses and measurement of decoherence times in both epitaxially doped and ion implanted nanostructures.

AUTHOR INFORMATION

Corresponding Author

*E-mail: (D.J.S.) dsleiter@stanford.edu; (A.P.) apawlis@mail. upb.de.

Notes

(5)

where R is the eective optical pumping rate and X accounts for the relative polarization of the optical-pumping laser and the optical transition. is the standard deviation associated with the 58 GHz Gaussian line width determined in Figure 3 and is the energy detuning between the laser and the optical transition. The laser was on resonance with the |e1 |0 transition and had a polarization of |H with an experimentally measured extinction ratio of 2000. The detuning from the other transitions was determined by the measured electron and hole g-factors. A large background count rate due to scattering of photons from the optical-pumping laser into the SPCM is present in the optical-pumping data, which could not be completely ltered out. Therefore, the mean of the background,
119

The authors declare no competing nancial interest.

ACKNOWLEDGMENTS This research was supported by the German Research Foundation (DFG), the NTT Basic Research Laboratories, the Commissioned Research of the Nation Institute of Information and Communications Technology (NICT), by the Ministry of Education, Culture, Sports, Science and Technology (MEXT), and through the Funding Program for World-leading Innovative Research and Development on Science and Technology (FIRST). We would also like to thank Zhe Wang for assistance in nding an appropriate mesa.
dx.doi.org/10.1021/nl303663n | Nano Lett. 2013, 13, 116120

Nano Letters

(1) DiVincenzo, D. Fortschr. Phys. 2000, 48, 771. (2) Ladd, T. D.; Look, P.; Nemoto, K.; Munro, W. J.; Yamamoto, Y. New J. Phys. 2006, 8, 184. (3) Cirac, J. I.; Zoller, P.; Kimble, H. J.; Mabuchi, H. Phys. Rev. Lett. 1997, 78, 3221. (4) Kok, P.; Munro, W. J.; Nemoto, K.; Ralph, T. C.; Dowling, J. P.; Milburn, G. J. Rev. Mod. Phys. 2007, 79, 135. (5) Childress, L.; Taylor, J. M.; Sorensen, A. S.; Lukin, M. D. Phys. Rev. A 2005, 72, 052330. (6) Maunz, P.; Moehring, D. L.; Olmschenk, S.; Younge, K. C.; Matsukevich, D. N.; Monroe, C. Nat. Phys. 2007, 3, 538. (7) Olmschenk, S.; Matsukevich, D. N.; Maunz, P.; Hayes, D.; Duan, L. M.; Monroe, C. Science 2009, 323, 486. (8) Seidelin, S.; et al. Phys. Rev. Lett. 2006, 96, 253003. (9) Greilich, A.; Yakovlev, D. R.; Shabaev, A.; Efros, A. L.; Yugova, I. A.; Oulton, R.; Stavarache, V.; Reuter, D.; Wieck, A.; Bayer, M. Science 2006, 313, 341. (10) Santori, C.; Fattal, D.; Vuckovic, J.; Solomon, G. S.; Yamamoto, Y. Nature 2002, 419, 594. (11) Press, D.; De Greve, K.; McMahon, P. L.; Ladd, T. D.; Friess, B.; Schneider, C.; Kamp, M.; Hofling, S.; Forchel, A.; Yamamoto, Y. Nat. Photonics 2010, 4, 367. (12) Patel, R. B.; Bennett, A. J.; Farrer, I.; Nicoll, C. A.; Ritchie, D. A.; Shields, A. J. Nat. Photonics 2010, 4, 632. (13) Santori, C.; Barclay, P. E.; Fu, K.-M. C.; Beausoleil, R. G.; Spillane, S.; Fisch, M. Nanotechnology 2010, 21, 274008. (14) Bernien, H.; Childress, L.; Robledo, L.; Markham, M.; Twitchen, D.; Hanson, R. Phys. Rev. Lett. 2012, 108, 043604. (15) Kurtsiefer, C.; Mayer, S.; Zarda, P.; Weinfurter, H. Phys. Rev. Lett. 2000, 85, 290. (16) Togan, E.; et al. Nature 2010, 466, 730. (17) Fu, K.-M.; Santori, C.; Stanley, C.; Holland, M. C.; Yamamoto, Y. Phys. Rev. Lett. 2005, 95, 187405. (18) De Greve, K.; Clark, S. M.; Sleiter, D.; Sanaka, K.; Ladd, T. D.; Panfilova, M.; Pawlis, A.; Lischka, K.; Yamamoto, Y. Appl. Phys. Lett. 2010, 97, 241913. (19) Sanaka, K.; Pawlis, A.; Ladd, T. D.; Lischka, K.; Yamamoto, Y. Phys. Rev. Lett. 2009, 103, 053601. (20) Kim, Y. M.; Sleiter, D.; Sanaka, K.; Yamamoto, Y.; Meijer, J.; Lischka, K.; Pawlis, A. Phys. Rev. B 2012, 85, 085302. (21) Steiner, T.; Thewalt, M. L. W.; Bhargava, R. N. Solid State Commun. 1985, 56, 933. (22) Sanaka, K.; Pawlis, A.; Ladd, T. D.; Sleiter, D. J.; Lischka, K.; Yamamoto, Y. Nano Lett. 2012, 12, 4611. (23) Ladd, T. D.; Press, D.; De Greve, K.; McMahon, P. L.; Friess, B.; Schneider, C.; Kamp, M.; Hofling, S.; Forchel, A.; Yamamoto, Y. Phys. Rev. Lett. 2010, 105, 107401. (24) Pawlis, A.; Sanaka, K.; Gotzinger, S.; Yamamoto, Y.; Lischka, K. Semicond. Sci. Technol. 2006, 21, 1412. (25) Balasubramanian, G.; et al. Nat. Mater. 2009, 8, 383. (26) Tyryshkin, A. M.; et al. Nat. Mater. 2012, 11, 143. (27) Childress, L.; Taylor, J. M.; Sorensen, A. S.; Lukin, M. D. Phys. Rev. Lett. 2006, 96, 070504. (28) Morton, J. J. L. arXiv:0905.4008v1 [quant-ph]. arXiv.org e-Print archive; http://arxiv.org/abs/0905.4008, 2009 (accessed Sep 20, 2012). (29) Greilich, A.; Pawlis, A.; Liu, F.; Yugov, O. A.; Yakovlev, D. R.; Lischka, K.; Yamamoto, Y.; Bayer, M. Phys. Rev. B 2012, 85, 121303(R). (30) Merz, J. L.; Kukimoto, H.; Nassau, K.; Shiever, J. W. Phys. Rev. B 1972, 6, 545. (31) Press, W. H.; Teukolsky, S. A.; Vetterling, W. T.; Flannery, B. P. Numerical Recipes in C++, 2nd ed.; Cambridge University Press: Cambridge, 2002; Chapter 15.

Letter

REFERENCES

120

dx.doi.org/10.1021/nl303663n | Nano Lett. 2013, 13, 116120