Large stark tuning of InAs/InP quantum dots

Epitaxial InAs quantum dots embedded in InP (InAs/InP) are excellent sources of telecom single photons.^1,2 Previous studies have demonstrated single-photon generation in these materials with large Purcell enhancement³ and record high single-photon purity.⁴ InAs/InP quantum dots can also act as indistinguishable single-photon sources at telecom wavelengths suitable for quantum communications.⁵ Many applications in quantum information processing require multiple identical quantum emitters. However, the emission wavelength of individual quantum dots varies due to small differences in their shape, size, strain environment, and chemical composition. This makes it difficult to integrate multiple identical quantum emitters on the same chip. To overcome this spectral randomness, we need a method to tune the emission wavelength of individual dots over a broad spectral range without degrading the coherence properties of the emitter.

Stark tuning is an effective technique to tune the wavelength of quantum dots. This technique uses an electric field to tune the emission wavelength using the quantum-confined Stark effect. In InAs/GaAs quantum dots emitting at wavelengths below 1 μm, Stark tuning has produced wavelength shifts of up to 25 meV with minimal degradation of emission properties,^6,7 enabling two different dots to show two-photon interference.^8,9 Large Stark tuning at telecom wavelengths has been demonstrated in InAs quantum dots embedded in GaAs¹⁰ or InAlGaAs.¹¹ However, the extension of this approach to InAs/InP has not been studied carefully. Previous Stark tuning demonstrations of InAs/InP quantum dots suffered from a limited tuning range of <0.8 meV^12,13 because at high electric field intensities, the dot gains additional charges that cause discrete jumps in the optical spectrum. Achieving large wavelength tuning ranges requires new device geometries that limit this charging effect.

In this letter, we demonstrate a large DC Stark shift in the emission wavelength of InAs/InP quantum dots at telecommunication wavelengths. We place the InAs quantum dots embedded in InP between two metallic plates that are separated by a SiO₂ buffer layer to avoid introduction of charges. An applied DC electric voltage shifts the emission of the quantum dot by more than 5 meV without any observed spectral jumps due to charging, providing a much larger tuning range than previously available for this material. Steady-state and time-resolved photoluminescence measurements show a clear tunneling effect introducing non-radiative decay processes that limit the emitter brightness at high voltages. Photon correlation measurements show robust anti-bunching behavior over a wide voltage range. Our results pave the way for developing large identical arrays of quantum emitters at telecom wavelengths. Such identical quantum emitters are highly desirable for quantum information applications, such as boson sampling,¹⁴ quantum communication,¹⁵ and optical quantum computing.^16–18 Furthermore, our technique is agnostic to the nature of the substrate and could, thus, be incorporated in more complex hybrid photonic device structures that combine quantum dots with silicon photonics¹⁹ or lithium niobate photonics.²⁰ 

Figure 1(a) shows the proposed device, in which the InAs/InP quantum dots are located between two metal plates that provide the DC electric field for Stark tuning. We patterned the bottom contact on a silicon carrier chip with electron beam lithography followed by metal deposition (5 nm Cr/50 nm Au) and lift-off. Next, we used a microprobe to transfer an InP waveguide containing InAs quantum dots onto the bottom contact and covered the sample with 500 nm SiO₂ deposited by plasma enhanced chemical vapor deposition. Then, we formed the top contact by electron beam lithography, followed by 20 nm Cr deposition and lift-off. We chose a thin Cr layer to have optical access through the contact. Finally, we opened a window in the SiO₂ by buffer oxide etching to electrically access the buried metal pad connected to the bottom contact. Figure 1(b) shows a false-color scanning electron microscopy image of the fabricated device.

We performed photoluminescence measurements on the quantum dots in a closed-cycled cryostat at 4 K. To measure the photoluminescence spectrum of the quantum dots, we used a 780 nm continuous-wave laser for excitation and collected the photoluminescence signal directly through the top metal contact. Figure 2(a) shows the photoluminescence spectrum of multiple quantum dot emission lines at different DC electric voltages. All quantum dot emission lines start to redshift as we introduce the voltage due to the quantum confined Stark effect.

We extracted photoluminescence emission properties (center wavelength, intensity, and linewidth) from Lorentzian fits to the photoluminescence spectrum at each applied voltage. The relative change in the center wavelength of a quantum dot (QD1) that we labeled in Fig. 2(a) is shown in Fig. 2(b). We observed a large shift in the emission wavelength of up to 8 nm (5.1 meV) and attribute it to the quantum confined Stark effect. We also measured shifts of similar size for 9 different quantum dots on the same chip as demonstrated in the histogram plot in Fig. S1. A quadratic fit of the form $ΔE=pF+βF2$ to this shift allows us to determine the polarizability (β) and dipole moment (⁠$p$⁠) of the InAs/InP quantum dots. Here, ΔE is the relative change in emission energy and F is the electric field intensity. From the fit, we obtained $β=0.19±0.01 μeVcm2kV2$ and $p/e=0.14±0.02 nm$⁠, where e is the charge of an electron. These values are slightly lower than the previously reported values for InAs/InP quantum dots.^12,13 Different growth conditions and shape of the quantum dots, parasitic resistance in the device,²¹ and non-ideal thickness of the insulator could explain the lower measured values of dipole moment and polarizability compared to the previous works.

Figures 2(c) and 2(d) show the integrated intensity and linewidth of the QD1 emission as a function of the applied voltage. The integrated intensity of the QD1 emission at 26 V drops to ∼10% of its initial value at 0 V applied voltage. At large electric field intensities [see Fig. 2(e)], the created carriers are no longer confined in the quantum dot potential, which leads to tunneling.^22,23 Carrier tunneling reduces the efficiency of emission and therefore explains the drop in integrated intensity. The linewidth of QD1 [Fig. 2(d)] remains constant until very high electric field intensities. We observe linewidth variations among different quantum dots. The particular dot (QD1) we studied has a broad linewidth, but other quantum dots in the sample can feature much narrower linewidths and can exhibit two-photon interference.^24,25 Therefore, narrow linewidth quantum dots combined with our Stark tuning approach are promising for multi-photon interference applications.

The carrier tunneling effect also represents itself in the single exciton lifetime of the quantum dots. Time-resolved photoluminescence measurements reveal changes in the lifetime due to oscillator strength modification as well as tunneling effects.⁷ To measure the lifetime of the quantum dot emission, we used a 785 nm pulsed laser with a 40 MHz repetition rate. Figure 3(a) shows the evolution of the lifetime as we changed the applied voltage. Initially, at 0 V, the lifetime is 1.44 ns, and at higher electric fields (V = 28 V), the lifetime increases to 1.71 ns [see Fig. S2(a)]. We attribute the longer lifetime to the decrease in the oscillator strength of the quantum dot caused by larger separation of the electron-hole pair.²⁶ At even larger electric field intensities, tunneling effects dominate the lifetime and reduce it to 1.39 ns [see Fig. S2(b)].^7,26 The rich dynamics of these two competing processes were absent in the photoluminescence intensity measurement [Fig. 2(c)] because both the tunneling effect and oscillator strength reduction lead to a decrease in the emission intensity.

Single-photon emission from quantum dots is vital for their application in quantum information processing. Therefore, we performed photon correlation measurements at two different voltages to confirm the single-photon nature of the quantum dot emission and its stability while applying an electric field for Stark tuning. We used a 780 nm continuous-wave laser to excite the quantum dots and sent the photoluminescence signal to a grating spectrometer to filter the emission from QD1. The filtered signal was then sent to a fiber-based beam splitter followed by two superconducting single photon detectors. Figures 3(b) and 3(c) show the second-order photon correlation measurements for QD1 at 0 V and 19 V, respectively. Both measurements show clear antibunching behavior, confirming the single photon emission from QD1 even when the emission wavelength was shifted by ∼1.8 meV. From exponential fits to the raw data without any deconvolution or background subtraction, we obtained $g20=0.12±0.18$ and $g20=0.31±0.18$ for applied voltages of 0 V and 19 V, respectively. We measured the background (including the detector dark counts) for both voltages by disconnecting the quantum dot signal from one of the single photon detectors.⁵ We subtracted the background from the measured coincidences and fitted the subtracted data to an exponential function without any deconvolution. From the fits, we obtained $g20=0.02±0.2$ and $g20=0.09±0.22$ for applied voltages of 0 V and 19 V, respectively. To the extent of our error bounds, the $g20$ at 19 V does not degrade from its original value at 0 V and remains below the classical limit of 0.5, indicating the robustness of single-photon emission against Stark tuning.

In summary, we have demonstrated a large Stark shift in the InAs/InP quantum dot emission wavelength, which is promising for the realization of multiple identical telecom single-photon emitters on a chip. Photon correlation measurements demonstrate robust single-photon emission of the quantum dots before and after Stark tuning. The introduction of wide bandgap barrier layers during the quantum dot growth can reduce the carrier tunneling rates. Therefore, Stark tuning range will only be limited due to reduction in oscillator strength and should further increase.⁷ Photon collection in this device is based on out-of-plane radiation of the quantum dot through the top metal contact. Using fully transparent top contacts, such as graphene²⁷ and indium tin oxide,²⁸ can prevent photon reflection at the top contact and improve the photon collection efficiency. Furthermore, integrating these wavelength-tunable single-photon emitters with integrated photonic waveguides will enable an on-chip optical network with multiple identical quantum emitters for studying multi-photon interference effects,¹⁴ quantum simulation,^29,30 and linear optical quantum computation.^16,17

See supplementary material for the histogram plot of the Stark shift for 9 different quantum dots as well as time-resolved photoluminescence data corresponding to the measurement in Fig. 3(a).

The authors would like to acknowledge support from the Center for Distributed Quantum Information at the University of Maryland and Army Research Laboratory, and the Physics Frontier Center at the Joint Quantum Institute.