Large-scale optical characterization of solid-state quantum emitters Madison Sutula1Ian Christen1 Eric Bersin12 Michael P. Walsh1 Kevin C. Chen1 Justin Mallek2 Alexander Melville2 Michael Titze3 Edward S. Bielejec3

2025-05-03 0 0 9.39MB 17 页 10玖币

侵权投诉

Large-scale optical characterization of solid-state quantum emitters

Madison Sutula1,∗Ian Christen1, Eric Bersin1,2, Michael P. Walsh1, Kevin C. Chen1,

Justin Mallek2, Alexander Melville2, Michael Titze3, Edward S. Bielejec3,

Scott Hamilton2, Danielle Braje2, P. Benjamin Dixon2, and Dirk R. Englund1

1Research Laboratory of Electronics, Massachusetts Institute of Technology, Cambridge, MA 02139, USA

2Lincoln Laboratory, Massachusetts Institute of Technology, Lexington, MA 02421, USA

3Sandia National Laboratories, Albuquerque, NM 87185, USA

(Dated: October 26, 2022)

Solid-state quantum emitters have emerged as a leading quantum memory for quantum network-

ing applications. However, standard optical characterization techniques are neither eﬃcient nor

repeatable at scale. In this work, we introduce and demonstrate spectroscopic techniques that en-

able large-scale, automated characterization of color centers. We ﬁrst demonstrate the ability to

track color centers by registering them to a fabricated machine-readable global coordinate system,

enabling systematic comparison of the same color center sites over many experiments. We then

implement resonant photoluminescence excitation in a wideﬁeld cryogenic microscope to parallelize

resonant spectroscopy, achieving two orders of magnitude speed-up over confocal microscopy. Fi-

nally, we demonstrate automated chip-scale characterization of color centers and devices at room

temperature, imaging thousands of microscope ﬁelds of view. These tools will enable accelerated

identiﬁcation of useful quantum emitters at chip-scale, enabling advances in scaling up color center

platforms for quantum information applications, materials science, and device design and charac-

terization.

Keywords: color centers, diamond, wideﬁeld, large-scale, quantum information

Quantum emitters play a central role in quantum in-

formation science and technology [1, 2]. Color centers

in solids have emerged as a leading platform for quan-

tum information processing, with applications in sens-

ing, computation, and communication. Their electron

spin degree of freedom can store quantum states for mil-

liseconds to seconds [3, 4], and can be eﬃciently trans-

duced into a ﬂying qubit via a coherent spin-photon

interface. These light-matter interactions can be en-

gineered with cavity quantum electrodynamics [5, 6],

accessible through nanofabrication [7–12] and heteroge-

neous integration techniques [13–17]. Nuclear spin an-

cilla qubits provide additional degrees of freedom [18],

enabling longer storage times [19], local multi-qubit logic

protocols such as error detection and correction [20, 21],

and potential for more robust computation such as bro-

kered entanglement [22] or cluster state generation [23].

In spite of these promising properties and early demon-

strations, most approaches rely on the integration and

control of a single color center at each node [24–26]. How-

ever, quantum information processing nodes will require

many individually-addressable quantum emitters, each

with long-lived spin states and a high-quality photonic

interface [27]. A critical challenge faced by color center

qubits, especially when compared to neutral atom and

trapped ion qubit architectures, is spectral inhomogene-

ity [28]. Although a number of techniques have been

demonstrated to practically or theoretically overcome

this inhomogeneity, including retuning emitter transi-

tions via Stark [29] or strain tuning [30], compensat-

ing for diﬀerential phase accumulation [31], and shift-

∗mmsutula@mit.edu

ing emitted photons by optical frequency conversion [32]

or frequency modulation [33], each of these solutions re-

lies on pre-characterization of the optical transitions in-

volved. Additionally, up to now, approaches to under-

stand the range of color center performance and unifor-

mity have been limited to ensemble-level statistics, ob-

scuring individual color center properties. Scaling up the

number of solid-state quantum memories accessible in

quantum information processing systems requires large-

scale characterization techniques to identify the most vi-

able qubits for practical use.

Here, we report on a high-throughput approach for

spectroscopy and synthesis of quantum emitters, while

retaining single-emitter resolution, on diamond color cen-

ters (Fig. 1). We track color centers by registering them

to fabricated, machine-readable global coordinate sys-

tem, enabling the systematic comparison of individual

center sites over many experiments that is critical for

understanding the role of materials processing [34]. We

then implement resonant photoluminescence excitation

in a wideﬁeld cryogenic microscope to realize two orders

of magnitude speed-up over confocal microscopy. Finally,

we demonstrate automated chip-scale characterization of

color centers and devices at room temperature, imag-

ing thousands of microscope ﬁelds of view (see Extended

Data Videos).

With the large datasets accessible through these novel

approaches, we identify a sample with an exceptionally

narrow inhomogeneous distribution of silicon-vacancy

centers in diamond (SiVs), and we verify in a sample

implanted via focused ion beam (FIB) that the opti-

cal properties of highly strained SiVs do not degrade.

These techniques provide a paradigm in which the opti-

cal properties of individual color centers can be identi-

arXiv:2210.13643v1 [quant-ph] 24 Oct 2022

charge

traps

ɛxx

ɛyy

ɛzz∥⟨111⟩

ɛxx

ɛyy

ɛzz∥⟨111⟩

Vconfocal

QR code

Vwideeld

ɛxx

ɛyy

ɛzz∥⟨111⟩

ωZPL/2π

ω/2π

FIG. 1. Techniques for chip-scale characterization. a, Standard confocal microscopy enables spectroscopy on a diﬀraction-

limited volume, Vconfocal .b, Our wideﬁeld microscopy technique discussed in Section II allows for parallelized spectroscopic

measurements over many color centers simultaneously in a larger sample volume, Vwideﬁeld .c, We fabricate QR-style codes

onto the surface of our samples as a spatial reference frame which we decode convolutionally in real time, enabling us to track

and revisit color centers. Together, these techniques allow us to probe d, the optical transitions of color centers, including

perturbations such as e, local strain environment, which the silicon-vacancy center in diamond (pictured) is particularly sensitive

to, f, electric ﬁelds, which the nitrogen-vacancy center in diamond (pictured) is particularly sensitive to, and g, the time-varying

occupancy of charge traps in the crystal host, which is thought to cause inhomogeneous line broadening.

ﬁed, tracked, and categorized. We anticipate that the re-

ported high-throughput tools will close the loop between

materials processing and spectroscopy, enabling develop-

ment of scalable quantum information processors.

RESULTS

I. REGISTERING COLOR CENTERS TO

FABRICATED MARKERS

Scanning confocal microscopy is a standard technique

used to identify color centers, revealing bright spots

located relative to the coordinates set by the scan-

ning axes of the microscope. However, returning to a

previously-characterized set of emitters on a chip can

be challenging due to the vast diﬀerence between the

O(100 µm×100 µm) size of the ﬁeld of view and the

O(5 mm ×5 mm) size of the chip, especially after the

frame of reference is lost to sample removal. Previ-

ous work has used bright ‘lodestar’ emitters or image-

stitching techniques to return to a given set of emit-

ters [35], though these techniques are not general, as not

all samples possess suﬃciently unique features for each

ﬁeld of view of interest.

We demonstrate the ability to register emitters to a

global coordinate system using a standard home-built

cryogenic confocal microscope setup. This technique en-

ables us to revisit the same set of nitrogen-vacancy cen-

ters in diamond (NVs) over the course of multiple cool-

ing cycles, and therefore to track emitter properties over

time. We ﬁrst fabricate quick response-style (QR) codes

on the surface of the diamond chip (see Methods Sec-

tion 6). These are decoded in real-time with custom im-

age processing tools by establishing a coordinate trans-

form between the global sample coordinates and the local

microscope coordinates using the region of sample space

in view of the scanning confocal microscope (Fig. 2a).

We use oﬀ-resonant scanning ﬂuorescence microscopy

under 515 nm excitation to image a low-density of NV

centers on the surface of the chip. To create these ﬂuo-

rescent centers, the chip was implanted with 1×109/cm2

of 15N ions at 185 keV, and subsequently annealed at

1200◦C in ultra-high vacuum, tri-acid cleaned in a 1:1:1

mixture of nitric, perchloric, and sulfuric acids by boil-

ing for 1 hour [36], and ﬁnally cleaned in a 3:1 sulfu-

470.474 470.476 470.478 470.48 470.482

ω/2π (THz)

100

150

200

Intensity (a.u.)

2 3 4 5

Experiment iteration

100

150

Number of NVs

re-observed

10 μm

5 μm

a d

1 μm

-10 -5 0 5 10

ω/2π (GHz)

Number of NV sites

Control

Acid clean

-20 -15 -10 -5 0 5

(GHz)

Number of NV sites

Control

Acid clean

FIG. 2. Tracking color centers in a sample with fabricated QR codes. a, Confocal microscope image of a diamond

patterned with QR codes and implanted with 15N. The full ﬁgure shows a confocal scan of the sample under incoherent

illumination. The corners of four QR codes are emphasized with white circles. The insets show i, ﬂuorescence from part of the

full ﬁeld of view under 515 nm excitation, with single nitrogen-vacancy centers in diamond (NVs), and ii, an SEM micrograph

of a QR code. b, Typical photoluminescence excitation (PLE) spectrum for an NV center in this sample. c, Histogram of the

number of experiments in which an NV was observed for a sample cooled down four times, with processing steps in between

subsequent experiments. Errors in clustering sites are indicated in bin 5. d, Visualization of the shift in mean optical transition

frequency ∆¯ω/2π= ¯ωi/2π−¯ωj/2πof the two ms= 0 spin transitions between experiments j= 1 & i= 2, which served as a

control, and between experiments j= 1 & i= 3, before and after tri-acid cleaning. After tri-acid cleaning, two populations of

¯ω/2πare observed, attributable to polarization due to the presence of an electric ﬁeld caused by a change in surface termination.

e, Visualization of the shift in the splitting ∆δ=δi−δjbetween the two ms= 0 transitions between experiments j= 1 &

i= 2, which served as a control, and between experiments j= 1 & i= 3, before and after tri-acid cleaning.

ric acid:hydrogen peroxide piranha solution. We clas-

sify bright, diﬀraction-limited spots as candidate emitter

sites, as shown in the inset of Fig. 2a. We then determine

the spectral positions and linewidths of the zero-phonon

line (ZPL) optical transitions at each candidate site with

photoluminescence excitation (PLE) spectroscopy. To do

this, we measure the ﬂuorescence response of each emit-

ter by scanning a resonant laser over the ZPL transi-

tions and collecting light emitted into the phonon side-

band (PSB) on an avalanche photodiode. We label the

spin conserving transitions associated with ms= 0 by

mean ZPL frequency ¯ω= 1/2 (Ex+Ey) and splitting

δ=|Ey−Ex|[29]. Further information about the ex-

perimental setup and NV centers is provided in the Sup-

plementary Information. A typical NV PLE spectrum is

shown in Fig. 2b. Details regarding data analysis and QR

encoding are provided in the Methods Sections 2 and 4.

We track NV centers over four experiments by regis-

tering emitters to the sample rather than to the micro-

scope coordinates, in spite of the stochastic movement

of the sample caused by remounting along with warm-

ing and cooling the system between experiments. Ex-

periments 1 and 2 served as a control, as we thermally

cycled and remounted the sample in between, but per-

formed no additional materials processing. Between ex-

periments 2 and 3, the sample was tri-acid cleaned in a

1:1:1 mixture of nitric, perchloric, and sulfuric acids by

boiling for 1 hour [36] and between experiments 3 and

4, the sample was annealed in an oxygen environment at

450◦C for 4 hours [37]. In each experiment, NVs and QR

codes were registered to the local microscope coordinate

system, and then transformed to the global sample co-

ordinate system. After registering the locations of NVs

in all four experiments, we identiﬁed the NVs in each

experiment that belonged to the same confocal site on

the sample using a clustering algorithm constrained by

a diﬀraction-limited Euclidean threshold distance. The

number of experiments in which a color center was found

at the same global sample coordinate site is provided in

Fig. 2c: we observe that some NVs disappear over mul-

tiple experiments.

By tracking individual color centers over multiple cryo-

stat cooling cycles, we observe that tri-acid cleaning the

diamond leads to shifts in the mean frequency at a given

NV site ∆¯ω/2π= ¯ωi/2π−¯ωj/2π(for experiments i, j)

of the ms= 0 optical transitions exceeding those ob-

served in the control. Conversely, the change in splitting

∆δ=δi−δjbetween the two ms= 0 states does not de-

viate from the control. These observations, as shown in

Fig. 2d-e, are consistent with the addition of an electric

ﬁeld normal to the surface [29], attributable to a changed

surface termination [38]. Notably, the population-level

statistics for ¯ω/2πalone do not reveal this phenomenon;

tracking individual emitters uniquely enabled us to ob-

serve these minute spectral shifts.

10 μm

406.8136 406.814 406.8144

ω/2π (THz)ω/2π (THz)

406.7 406.8 406.9

ab c d

406.8148

ω/2π(THz)

200

400

600

800

406.813 406.814 406.815

100

150

200

Number of

SIV sites

Number of

SIV sites

Number of

SIV sites

45.5 46 46.5 47 47.5

GS (GHz)

200

400

600

257 257.5 258 258.5

ES (GHz)

100

150

Number of

SIV sites

FIG. 3. Wideﬁeld photoluminescence excitation of silicon-vacancy centers. a, Fluorescence image of Sample A

reconstructed from wideﬁeld PLE as the resonant laser frequency ω/2πwas swept over 10 GHz, using the same ﬁeld of view

as [39]. b, Strain map indicating the spatial distribution of the mean zero-phonon line (ZPL) transition ¯ω/2π(a proxy for

axial strain) at 584 sites that had one PLE peak at each of the four SiV transition frequencies. We observe that the SiVs

are split into two classes of axial strain. c, Histograms of the locations of the four optical transitions for 984 groups of SiV

peaks, where each group of SiV peaks consists of a peak at each of the four SiV transition frequencies. d, Mean ZPL transition

frequency ¯ω/2πfor 984 groups of SiV peaks: the split into two classes of axial strain is apparent in the distribution. Fitting

the empirical probability density function (σ= 10 MHz, see Methods Section 1) shown in red reveals a population centered at

406.8141 THz with standard deviation 59 MHz, and a population centered at 406.8136 THz with standard deviation 48 MHz.

e, Ground state splitting (∆GS ), and f, Excited state splitting (∆ES ), for 984 groups of SiV peaks. The ground and excited

state splittings closely approximate the splitting due to spin-orbit coupling alone, revealing the low-strain environment.

II. PARALLELIZED PHOTOLUMINESCENCE

EXCITATION VIA CRYOGENIC WIDEFIELD

MICROSCOPY

Although the ability to register the positions of sin-

gle emitters to the surface of a sample enables individual

color centers to be tracked and compared over time, the

characterization rate for the previous experiments is lim-

ited by the time to obtain one resonant PLE scan, as

each site must be measured sequentially. To overcome

this bottleneck, we modify our standard confocal micro-

scope setup to excite an entire ﬁeld of view with a tunable

resonant laser in wideﬁeld mode (see Supplementary In-

formation), and thus parallelize PLE measurements to

realize a dramatic speed-up. This is made possible by

the introduction of an electron-multiplying charge cou-

pled device (EMCCD) camera in the imaging path, which

enables us to resolve individual diﬀraction-limited color

centers via image processing, as detailed in Methods Sec-

tion 3. The largest speedup that can be achieved with

this techniques is set by the number of diﬀraction-limited

emitter sites in a given ﬁeld of view; we provide a detailed

discussion of the regimes in which our wideﬁeld technique

is faster than confocal microscopy in Methods Section 5.

We demonstrate wideﬁeld PLE for two systems contain-

ing silicon-vacancy centers in diamond (SiVs): one with

an exceptionally narrow inhomogeneous distribution of

optical transitions (Sample A), and one with a broad

distribution (Sample B).

For Sample A, SiVs were incorporated in-situ during

chemical vapor deposition (CVD) overgrowth on a low-

strain substrate. The sample was subsequently tri-acid

cleaned and annealed at 1200◦C in ultra-high vacuum.

Fig. 3a shows a ﬂuorescence map reconstructed from

wideﬁeld PLE measurements over the SiV zero-phonon

line (ZPL) (speciﬁcally using the ground spin-orbit con-

serving C transition). By measuring PLE over all four

SiV ZPL transitions (see Extended Data Video 1), we can

determine the strain environment in the overgrown layer

throughout the entire ﬁeld of view, revealing two classes

as shown in Fig. 3b-d. We ascribe this to two populations

of strain along the axis of the SiV. We generate an empiri-

cal probability distribution (PDF) function (σ= 10 MHz,

see Methods Section 1) to determine the inhomogeneous

distribution. A Gaussian ﬁt to the empirical PDF repre-

senting ¯ω/2πreveals a population centered at 406.8141

THz with standard deviation 59 MHz, and a population

centered at 406.8136 THz with standard deviation 48

MHz. Previous work has demonstrated that CVD growth

is a promising path towards generating a narrow inhomo-

geneous distribution of SiVs [40], but to our knowledge,

our measurement represents the largest number of SiVs

with center frequencies that lie within the bandwidth

given by the lifetime-limited linewidth, enabling work re-

quiring high spectral homogeneity [39]. In a single ﬁeld

of view, we measure 257 SiV sites with ground spin-orbit

文档加载中……请稍候！
如果长时间未打开，您也可以点击刷新试试。

下载文档到电脑，查找使用更方便

10 玖币 0人已下载

立即下载

摘要：

Large-scaleopticalcharacterizationofsolid-statequantumemittersMadisonSutula1,IanChristen1,EricBersin1;2,MichaelP.Walsh1,KevinC.Chen1,JustinMallek2,AlexanderMelville2,MichaelTitze3,EdwardS.Bielejec3,ScottHamilton2,DanielleBraje2,P.BenjaminDixon2,andDirkR.Englund11ResearchLaboratoryofElectronics,Mass...

展开>> 收起<<

Large-scale optical characterization of solid-state quantum emitters Madison Sutula1Ian Christen1 Eric Bersin12 Michael P. Walsh1 Kevin C. Chen1 Justin Mallek2 Alexander Melville2 Michael Titze3 Edward S. Bielejec3.pdf

共17页,预览4页

还剩页未读，继续阅读

声明：本站为文档C2C交易模式，即用户上传的文档直接被用户下载，本站只是中间服务平台，本站所有文档下载所得的收益归上传人(含作者)所有。玖贝云文库仅提供信息存储空间，仅对用户上传内容的表现方式做保护处理，对上载内容本身不做任何修改或编辑。若文档所含内容侵犯了您的版权或隐私，请立即通知玖贝云文库，我们立即给予删除！

Large-scale optical characterization of solid-state quantum emitters Madison Sutula1Ian Christen1 Eric Bersin12 Michael P. Walsh1 Kevin C. Chen1 Justin Mallek2 Alexander Melville2 Michael Titze3 Edward S. Bielejec3

相关推荐

开通VIP享超值会员特权

作者详情

相关内容

热门标签

举报选择: