signature redacted - dspace.mit.edu
TRANSCRIPT
Chip-Scale Quantum Magnetometry via CMOS
Integration with Diamond Color Centers
by
Mohamed Ibrahim Mohamed Ibrahim
M.Sc., Electrical Engineering, Ain Shams University (2016)B.Sc., Electrical Engineering, Ain Shams University (2012)
Submitted to the Department of Electrical Engineering and ComputerScience
in partial fulfillment of the requirements for the degree of
Master of Science in Electrical Engineering and Computer Science
at the
MASSACHUSETTS INSTITUTE OF TECHNOLOGY
February 2020
@ Massachusetts Institute of Technology 2020. All rights reserved.
Signature redactedA uth or .. ... ...........................
Department of Electrical Engineering and Computer Science
Signature redacted January 30, 2020
C ertified by ........... ..................Ruonan Han
Associate Professor of Electrical Engineering and Computer ScienceThesis Supervisor
Signature redactedA ccepted by ....... ......................
MASS HSETS INSTITUTE Leslie A. KolodziejskiOFTECHNOLOGYrofessor of Electrical Engineering and Computer Science
MAR 13 2020 Chair, Department Committee on Graduate Students
LIBRARIES
Chip-Scale Quantum Magnetometry via CMOS Integration
with Diamond Color Centers
by
Mohamed Ibrahim Mohamed Ibrahim
Submitted to the Department of Electrical Engineering and Computer Scienceon January 30, 2020, in partial fulfillment of the
requirements for the degree ofMaster of Science in Electrical Engineering and Computer Science
Abstract
There has been increasing interest in spin-based quantum systems for a wide rangeof applications. In particular, the nitrogen-vacancy (NV) center in diamond hasdemonstrated outstanding sensing and imaging capabilities. However, previous controlapparatuses of these quantum systems have used discrete instrumentation to bothmanipulate and detect the NV's spin state. This limits potential applications. Inthis thesis the first chip-scale Complementary Metal Oxide Semiconductor (CMOS)platform that integrates the necessary components for NV quantum state preparation,control, and measurement is presented. A CMOS integrated system capable of thecontrol and readout of an ensemble of NV centers in diamond for magnetic fieldsensing is demonstrated. Scalar magnetic field sensing with a layer of nanodiamondparticles achieving 74 pT/VHz sensitivity is presented. In addition, vector magneticfield sensing with a slab of single crystalline diamond with enhanced sensitivityof 32.1 pT/v/IIz is also presented. Techniques for strong generation and efficientdelivery of microwave for quantum-state control, and optical filtering/detection ofspin-dependent fluorescence for quantum-state readout are introduced. This hybridarchitecture is a significant step towards a highly integrated quantum system withapplications in life sciences, tracking, and advanced metrology.
Thesis Supervisor: Ruonan HanTitle: Associate Professor of Electrical Engineering and Computer Science
3
4
Acknowledgments
First, I wish to express my most gratitude to my thesis supervisor, Prof. Ruonan Han,
for his guidance, encouragement, and useful discussions. He has been more than a
technical supervisor to me. I have learned and will continue to learn from him during
the rest of my stay at MIT.
I would like also to thank Prof. Dirk Englund for the collaboration through this
project of creating the first prototype of CMOS-diamond hybrid magnetometer. This
opportunity opened a whole research theme of creating CMOS integrated quantum
systems using NV centers in diamond. His help and advice are always invaluable.
Special thanks to my colleague and my friend Christopher Foy who has been
working closely with me in the last three year on this project and other projects. This
work would not be complete without his help. I would also like to thank Donggyu
Kim and Matthew Trushiem for working with Chris and me on this project. Their
input was essential and the discussions with them were very helpful.
Furthermore, thanks for the technical discussion and assistance from my labmates
including but not limited to: Cheng Wang, Xiang Yi, Muhammad Ibrahim Wasiq
Khan.
Special thanks to all my friends in Boston, the rest of the United States, and Egypt
for their help and continuous support. Last but not least, I would like to thank my
parents and sisters for their patience, care, and love that guided me. There are no
words or sentences that can express my appreciation towards them.
5
Contents
1 Introduction 11
2 Nitrogen Vacancy Center in Diamond Magnetometry 15
2.1 Overview of the Nitrogen Vacancy Center ............... .15
2.2 DC Magnetic Field Measurements Using NV .............. 17
3 Chip-scale CMOS-Diamond Magnetometer 21
3.1 CMOS-Diamond Hybrid Sensing Architecture ............. .21
3.2 CMOS-Diamond Hybrid Quantum Magnetometer ........... .22
3.3 On-chip Microwave generation ...................... 25
3.4 On-chip Optical Detection ........................ 30
3.5 Passivation Layer Etching . . . . . . . . . . . . . . . . . . . . . . . . 33
3.6 Diamond Sample Preparation . . . . . . . . . . . . . . . . . . . . . . 35
3.6.1 Nanodiamond Placement . . . . . . . . . . . . . . . . . . . . . 35
3.6.2 Bulk Diamond Placement . . . . . . . . . . . . . . . . . . . . 36
4 Experimental Results 39
4.1 Nanodiamonds Results . . . . . . . . . . . . . . . . . . . . . . . . . . 39
4.2 bulk diamond Results . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
5 Conclusion and Future Work 49
7
8
List of Figures
1-1 (a) A magneto-optical image of the a magnetic structure, recorded usinga single NV center on the AFM tip [6]. (b) Optical image of bacterialiving cells and the corresponding magnetic field image generated usingN V centers [7]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2-1 Nitrogen-vacancy centers in a diamond lattice. The blue, and redcircles represent carbon, and nitrogen atoms respectively. The whitecircles represent the vacancy. The projections (B2 1 , B,2 , B23, Bz 4) ofan external magnetic field Be,t along the four nitrogen-vacancy axesare also show n. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2-2 The energy-level diagram of a nitrogen-vacancy center in diamond. . . 16
2-3 The red fluorescence intensity of the NV centers at varying microwavefrequency (ODMR) under no external magnetic field bias . . . . . . . 17
2-4 The red fluorescence intensity of the NV center at varying microwave fre-quency (ODMR) under an external magnetic field bias with projectionsalong the four N-V axes. . . . . . . . . . . . . . . . . . . . . . . . . . 18
3-1 The basic schematic of the proposed CMOS-Diamond sensing architecture. 23
3-2 The schematic of the first CMOS-diamond quantum magnetometerwith a layer of nanodiamond on the top of the chip. . . . . . . . . . . 23
3-3 The schematic of the first CMOS-diamond quantum magnetometerwith a single crystalline slab on the top of the chip. . . . . . . . . . . 24
3-4 Schematic of the microwave generation circuitry. . . . . . . . . . . . . 25
3-5 The simulated and measured tuning curve of the on-chip ring VCO. . 26
3-6 The amplitude of the AC current flowing in the inductor as a functionof frequency. The plot shows a resonance behavior at 2.87GHz. .... 28
3-7 The simulated magnetic field profiles as a function of distance from thecenter of the proposed resonant inductor and a non-resonant one. . . 28
3-8 The simulated magnetic field profiles as a function of distance from theinductor center with and without capacitive parasitic loop. . . . . . . 29
3-9 The layout of the 3-turn on-chip inductor with parasitic capacitive loops. 29
3-10 The layout of the single-layer plasmonic grating filter implemented onM etal 8 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
9
3-11 The (a) real and (b) imaginary parts of the relative dielectric constantof copper. The calculated plots (solid lines) are based on the Drude-Brendel-Bormann model. The measured data (triangles and squares)are from 120, 21]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
3-12 FDTD simulated light transmission through the optical filter at (a)green light (A=532 nm) (b) red light (A=700 nm). . . . . . . . . . . . 32
3-13 (a) The cross section of the P+/N-well/P-sub (b) unpatterned photodi-ode layout with eddy current loops in the active area and possibly theanode and cathode connection. . . . . . . . . . . . . . . . . . . . . . 33
3-14 The proposed patterned photodiode layout (a) 2x2 active area layout(b) the radially connected anode connection that prevents any closedloops implemented in M1. (c) the full layout with cathode connection inM2 added. (d) the full layout with eddy current loops in the patternedactive area only. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
3-15 Optical image for the red fluorescence emitted by the chip surfaceunder green excitation (a) without passivation layer removed (b) withpassivation layer removed. Scale bar is 100tm. . . . . . . . . . . . . . 35
3-16 Top-view micrograph of (a) the fabricated CMOS chip sensing areaincluding the inductor, optical filter, and photodiode. (b) the samearea on the chip covered with nanodimaond particles. (c) the sensingarea covered by the single crystalline diamond slab. Scale bar is 100tm. 37
4-1 Optical micrograph of the CMOS chip (bottom) and photo of theprinted circuit board for testing (top). . . . . . . . . . . . . . . . . . 40
4-2 (a) The experimental setup of the ODMR experiment using off-chipcamera. (b) The corresponding measured and fitted ODMR spectrumat no external magnetic field and at 1.72 mT permanent magnet. . . 41
4-3 The measured and fitted ODMR spectrum using on-chip photodiode atno external magnetic field and at 2.2 mT permanent magnet. . . . . . 42
4-4 Frequency-modulated (FM) lock-in signal of NV spin-dependent fluo-rescence at zero external magnetic field . . . . . . . . . . . . . . . . . 44
4-5 FM lock-in signal with a permanent magnet (B = 6.27 mT). Thelinewidth of the ODMR is 7 MHz. Slopes dV/df at v_ = 2.8303 GHzand v+ = 2.9330 GHz are 42.969 nV/MHz and 42.450 nV/MHz, respec-tively . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 5
4-6 On-chip magnetometry (Blue) and temperature effect (Red) separationby detecting the effect of switching electromagnet on v± of the ODMRcurve of Fig. 4-5. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
5-1 NV-based magnetic imager with an array photodiodes below a singlemicrowave delivery structure. . . . . . . . . . . . . . . . . . . . . . . 51
5-2 NV-based individual Qubit control for scalable quantum informationprocessing applications. . . . . . . . . . . . . . . . . . . . . . . . . . . 51
10
Chapter 1
Introduction
Applications of complementary metal-oxide semiconductor (CMOS) integrated cir-
cuits in quantum apparatus are gaining increasing attention due to the prospect of
significantly increased hardware scalability and reduction of cost, size and power.
In particular, CMOS-based control and readout circuitry, operating at cryogenic
temperature, have been demonstrated recently towards the construction of large-scale
quantum computers [1, 2]. In comparison, the potential of CMOS circuits in another
vast application space of quantum systems - high-performance sensing - still awaits
more extensive investigations. Nitrogen-vacancy (NV) center in diamond stands as one
of the most promising technologies in this space. NV centers in diamond have emerged
as a leading room-temperature quantum sensor platform for temperature [3], electric
field [41 and magnetic fields [5-7]. The capabilities of NV-based quantum metrology
are based on its long spin coherence time [8] and its efficient optical interface for spin
polarization and readout. NV center in diamond has also demonstrated outstanding
imaging capabilities with sub-micron resolution. Imaging of a magnetic structure
and the magnetic fields generated by bacteria sample have been demonstrated (see
Fig. 1-1).
As a magnetometer, picotesla to sub-nanotesla sensitivity under ambient conditions
11
05 M
(a) (b)
Figure 1-1: (a) A magneto-optical image of the a magnetic structure, recorded usinga single NV center on the AFM tip 16]. (b) Optical image of bacteria living cells andthe corresponding magnetic field image generated using NV centers [7].
has been achieved using NV centers in diamond [9-11]. However, conventional
approaches for NV sensing involve bulky and discrete off-the-shelf instruments. These
instruments are required for spin state manipulation and readout. The NV systems
consist of a number of components among them are: (1) microwave signal generator,
and delivery interface to control the NV spin state. (2) optical filter to reject the
pump laser and a photodetector for NV spin-dependent fluorescence measurement.
(3) green pump laser. The current discrete realization of the above-mentioned system
limits practical applications and scalability.
In this thesis, a custom CMOS architecture that integrates the required components
in a single chip along with a hybrid integration with NV centers in diamond is reported.
This architecture stacks the microwave inductor, photonic filter, and photodiode into
a 200 pm x 200 pm footprint. We use this hybrid CMOS-diamond platform to
demonstrate ambient quantum magnetometry. The on-chip system is composed of
spin control and detection sub-systems. The on-chip magnetometer is tested using
nanodiamond particle layer and a single crystalline diamond slab. In chapter 2, the
basic principle of using NV centers in diamond for magnetic field sensing is introduced.
Next the details of the architecture and the design of the first prototype of the on-chip
hybird CMOS-diamond magnetometer is presented in chapter 3. The measurements
setup and results of the two diamond placement cases are given in chapter 4. Finally,
12
- M
chapter 5 provides a conclusion for this work while pointing out potential improvements
and perspectives for future work.
13
14
Chapter 2
Nitrogen Vacancy Center in Diamond
Magnetometry
2.1 Overview of the Nitrogen Vacancy Center
The negatively-charged nitrogen-vacancy (NV) center in diamond consists of a nitrogen
atom and a vacancy that substitute two adjacent carbon atoms in the diamond lattice
as shown in Fig. 2-1. For a single crystalline diamond, NV centers have four possible
orientations as shown in Fig. 2-1. The four orientations lay along the tetrahedral
axes of the host diamond. Figure 2-2 also illustrates the NV energy level diagram.
A spin magnetic triplet is formed at the ground state ( 3 A), consisting of a sub-level
ms= 0 at its lowest energy and another two degenerate sub-levels mS = ±1 raised
by -2.87 GHz, which is temperature dependant. Under green light (A~~532 nm)
excitation, the NV center spins are stimulated to their excitation states ( 3 E in Fig. 2-2)
and then relax back to the ground state ( 3 A). This is accompanied by spin dependant
red fluorescence, which is a unique proerty for NV centers. The relaxation of the
m, = 0 state is accompanied with a bright red fluorescence (A~600~800 nm). In
contrast, when the m, = ±1 states are excited and relax back, they can undergo a
15
Bz,~~~ ......-----.... Bs
BeB 2
NV Orientation I
Figure 2-1: Nitrogen-vacancy centers in a diamond lattice. The blue, and red circles
represent carbon, and nitrogen atoms respectively. The white circles represent the
vacancy. The projections (B_ 1 , B_ 2, B23, B 4 ) of an external magnetic field Bet along
the four nitrogen-vacancy axes are also shown.
04
600 nm800 nm
S=
It MS +1
P too14 11 2yeU
.7 G~z
'It% J M
114% MS=
I IDI I
% I
'SS.
S
m,=± 1~
m = 0 S'
Figure 2-2: The energy-level diagram of a nitrogen-vacancy center in diamond.
16
532 nm
non-radiative intersystem crossing into a metastable spin-singlet state ('A in Fig. 2-
2), and then transition into the m, = 0 ground level reducing the red fluorescence
intensity. Therefore, by applying a microwave (fo = -2.87 GHz) and measuring the
red fluorescence intensity of the NV center under green light excitation, the magnetic
resonance of the NV can be detected (shown in Fig. 2-3). Since this resonance is
detected optically, this is called optical detected magnetic resonance (ODMR) [6, 12].
As shown in Fig. 2-3, the amplitude of the dip at 2.87 GHz is called the contrast.
CContrast (C)
0
2.87Microwave Frequency (GHz)
Figure 2-3: The red fluorescence intensity of the NV centers at varying microwavefrequency (ODMR) under no external magnetic field bias.
2.2 DC Magnetic Field Measurements Using NV
The two unpaired electrons in each NV center form anti-parallel spins in m, = 0 and
parallel spins in m, = i1. When an external magnetic field Bext with a component
B, along the N-V axis (see Fig. 2-1) is applied, the m, = ±1 sub-levels are split apart
(i.e. Zeeman effect). The m, = +1 level with the two spins anti-parallel with B, has
17
higher energy than that of the m, = -1 level, where the two spins are parallel with
Bz, The photon frequency Af associated with such an energy gap is proportional to
|Bz|:
Af = f+ - f- = 2-/e|Bz|, (2.1)
where -ye is the gyromagnetic ratio and equals to 28 GHz/T, and f+ and f_ are the
frequencies for the transitions from m, = 0 to m, = +1 and m, = -1, respectively.
NV magnetometry is performed by determining f+ and f- via ODMR. We use
Af to derive Bz. Through detecting the magnetic resonance using ODMR, a static
or slowly-varying magnetic fields Bz, can be determined by sweeping a microwave
frequency fo around 2.87 GHz and monitoring the average intensity. The observed
resonances of Fig. 2-4 are f+ and f-, which give B,. Since NV centers have four possible
orientations in a single crystalline diamond as discussed in Section 2.1, this leads to four
D
(U)
0
2yeBzl f.1 - -.10 8
LL 2yeBz2 =f+2 - f-2
2YeBz = f+3 - f-32yeBz4 = f+4 - f-4
Microwave Frequency
Figure 2-4: The red fluorescence intensity of the NV center at varying microwave
frequency (ODMR) under an external magnetic field bias with projections along the
four N-V axes.
18
pairs of splitting in a single ODMR measurement (Fig. 2-4). The magnetometer based
on this principle, therefore, has vector-field measurement capability by monitoring the
different magnetic field projects and reconstructing Bez,. That is advantageous over
conventional Hall and fluxgate-based sensors [13, 14], where three devices in x- y- z-
axes are needed for vector detection.
An NV-center vector magnetometer normally has a static bias field to create the
aforementioned four ODMR spectral splittings (Fig. 2-4). The electronics of the
magnetometer then tracks the shifts of f+ and f_, so as to measure the variation of
the externally applied magnetic field. This can be done by recording the additional
change of the sensor output red fluorescence intensity measurements around each dip
in the magnetic resonance curve. Note this intensity change is caused by the shift of
each resonance frequency Afi due to the projection of the added field on the associated
NV-center axis ABzi (to be measured) on top of the bias static field. The sensitivity
obtained from an ODMR spectrum as shown in Fig. 2-4 is determined by two factors:
the linewidth and the signal-to-noise ratio (SNR) of each transition curve. The SNR
is proportional to the contrast of the experiment and inversely propositional to the
minimum detectable intensity signal. In Chapter 4, more details and an analytical
expressions of the sensitivity are provided.
19
20
Chapter 3
Chip-scale CMOS-Diamond
Magnetometer
3.1 CMOS-Diamond Hybrid Sensing Architecture
In the proposed CMOS-diamond sensing architecture, we use standard CMOS tech-
nology to implement most of the components required to manipulate and readout
the spin states of NV centers in diamond. A diamond layer full of NV centers is
attached on the top of the chip. A green laser source is used for optical excitation.
As shown in Fig. 3-1, a microwave signal generator, switches and current drivers that
feed a microwave coupling structure are needed to manipulate the spin state of the
NV centers. The microwave signal generator has to operate over a sufficient range (i.e.
2.6 GHz to 3.2 GHz) to enable the measurements of the ODMR curve. The microwave
delivery structure which can be an inductor has to generate a strong homogeneous
microwave field to increase the contrast of the ODMR curve (i.e. signal) and drive all
the NV centers in the ensemble with the same strength.
In order to detect the spin dependant red fluorescence a photodiode on the chip is
integrated below the microwave delivery structure. However, the absorption probability
21
of the NV to green light is very low, which means the majority of the green light
transmits into the on-chip photodiode. This hurts the overall noise of the experiment,
hence the sensitivity. This is due to two factors: (i) the intensity fluctuation of the
green light. (ii) the shot noise due to the large output DC current which can be given
by Eq. 3.1:
i2 = 2qID/f - 2q(Ig + Ir)Af. (3.1)
where, Af is the noise bandwidth, and ID is the DC current detected by the photodiode.
I., and I, are the green and red current portions detected by the photodiodes. This
green background also hurts the contrast of the ODMR curve, which consequently
hurts the sensitivity. Therefore an optical filter is needed to suppress the green
excitation light and pass the red fluorescence. This filter should be added between
the diamond and the photodiode. We choose to integrate this filter on the same chip;
pushing towards a monolithic solution using standard CMOS process.
3.2 CMOS-Diamond Hybrid Quantum Magnetome-
ter
Figure 3-2 shows the system schematic of the first prototype of the hybrid CMOS-
diamond quantum magnetometer. This chip, using TSMC 65 nm CMOS technology,
integrates most of the critical components for the ODMR operation. Shown in Fig. 3-2,
an on-chip ~ 2.87 GHz voltage-controlled oscillator (VCO) is phase-locked to an
external reference with a tunable frequency. Through a differential current driver
and a loop inductor, the circuit applies a vertical AC magnetic field to excite the
m = 1 ground levels of NV centers inside a thin layer of nanodiamond particles.
22
Filter+ Green...- :GreenMicrowave Coupling Photodiode ..- ' i ...--- :ExcitationStructure ..----------------. ---Diamond
-*with NVsMicrowaveSignal DMicrowave
Generation Coupling Structure
Optical Filter.................... .............. P od o e~~Ph odiode
Readout ''sCircuit '-
Figure 3-1: The basic schematic of the proposed CMOS-Diamond sensing architecture.
On Chip
Ring VCO Green U 1 0 ' Nano DiamondExcitation L with Ns
Magnetic FieldGenerator on Metal 9
cuus
Swich 5 Opt Cal FilterSon Metal 8Ih 0
5506gags, 0 Red
Y V~VV Fluorescence
Vem Current Driver PatternedPhotodiode
fm-1KHz
Loop Freq. |Oz!Filter Divider Fluorescence
Readout
Lock-in amplifier |Phase/
Charge - -uoncPump Frqenyfo0Mr iDetcto - -0MN
Off ChipSi nal Generator
Figure 3-2: The schematic of the first CMOS-diamond quantum magnetometer with alayer of nanodiamond on the top of the chip.
The nanodiamond coating is formed by appling the nanodiamond solution on the
chip surface and then evaporated the liquid (extra details are given in Section 3.6.1).
The nanodiamond is excited from above with a green light, and the red fluorescence
is detected by a p+/n-well/p-sub photodiode placed under the loop inductor. The
23
.......................................................................................
On Chipis@~ Green
Ring VCO ExcitationBulk Diamond
Magnetic Field I with NVs
Generator on Meta 9Ctd
VDD
3 3 S 3 U Optical FilterVI IV4*I on Metal 8
egging RedFluorescence
V rrent Driver PatternedPhotodlode
....................................
LoopFreq Ogg FluorescenceFilter Divider Readout
f% - 1 KHz oki mlfe
Chre Phase/aFrequency fa-10MzO
Off ChipI Signal Generator|
Figure 3-3: The schematic of the first CMOS-diamond quantum magnetometer with a
single crystalline slab on the top of the chip.
diode is partitioned into smaller portions in shunt, in order to prevent the induction
of large eddy current from the inductor (hence RF loss). As described in Section 3.1
the majority of the green light is not absorbed by the diamond, but transmits into
the chip. A plasmonic nano-photonic filter, using a grating of the CMOS interconnect
metal (Metal 8), is implemented above the photodiode (Fig. 3-2). The detailed design
of the filter is provided in Section 3.4. In addition, techniques for improving the overall
filtering performance are discussed in Chapter 5.
The lattice orientations in the nanodiamond particles are random. This means the
amount of frequency splitting in each NV is also random. As a result, the resulting
ODMR can be only used for scalar-field sensing. Another prototype with a slab of
single crystalline diamond attached to the same CMOS chip as shown in Fig. 3-3 is
also reported. This allows for vector-field sensing capabilities. The diamond is also cut
24
and attached in a way that direct the green light horizontally to enhance the filtering
as explained in Section 3.6.2.
3.3 On-chip Microwave generation
The ground-state spin transitions are driven by the on-chip generated microwave fields.
Figure 3-4 shows the circuitry for on-chip microwave generation and delivery. This
circuitry is composed of a phase-locked loop (PLL), a current driver, and a resonant
loop inductor. The PLL generates the microwave sweep signal from 2.6 GHz to
3.1 GHz required for the ODMR experiment. The main component of this loop is an
on-chip voltage-controlled ring oscillator (VCO) with 3 differential inverter stages. The
usage of a ring VCO [15] avoids any large-size inductor and minimizes the cross-talk
between the oscillator and the microwave inductor which drives the NV ensemble.
The mutual-locking inverter pair (e.g., INV 2 in Fig. 3-4 forms a latch and ensures the
differential phases between the left and right branches of the VCO. Inside each inverter
stage of the on-chip ring VCO in Fig. 3-4 (e.g., INV 1), the sizes of the NMOS and
INV1,INV2
Loop Filter Off Chip On ChipC 4
Charge Pump-
PhaselFrequency VDetector er c
[Freq. Divider
f,o«=~120 MHz
Ring VCOVoD
Qx lo(t) Vtune Current/ Driver
lo(t) C1 C2
SM 1 M2
lamp 'bias
M 3 M4
Figure 3-4: Schematic of the microwave generation circuitry.
25
PMOS (gate width/gate length) are 24 pm/280 nm and 54 pm/280 nm, respectively.
Inside each latch inverter (e.g., INV 2 in Fig. 3-4), the NMOS and PMOS sizes are
5.2 pm/280 nm and 12 pm/280 nm, respectively. The frequency tunability of the
oscillator is realized via 3 pairs of MOS variable capacitors (e.g., Cvco in Fig. 3-4,
of which the capacitance changes from 22 fF to 75 fF when the PLL control voltage
Vtri varies from 0 to 5 V. The simulated and the measured VCO tuning curves are
shown in Fig. 3-5. The curves show very good agreement between them. The VCO
gain is ~180 MHz/V. The entire phase-locked loop is closed with off-chip components
to enhance the stability and decrease the phase noise of the signal. The loop filter of
the PLL is a typical second-order low-pass filter and the values of the components
shown in Fig. 3-4 are R1=0.4 kQ, C3 =4.5 nF and C4 =150 pF.
The microwave fields are delivered to the NV ensemble through the loop inductor
(Fig. 3-4) implemented on the top-most copper layer (Metal 9). To efficiently deliver
the microwave field, the loop inductor and a pair of shunt capacitors (C1 and C2 in
Fig. 3-4) form a resonating load for the current driver. C1 and C2 are MOS variable
3
2.95-N
2.9
0 2.85
L..
LL 2.8
2.750.3
SimulationMeasurements
0.5 0.7 0.9 1.1 1.3 1.5 1.7
Tuning Voltage (V)
Figure 3-5: The simulated and measured tuning curve of the on-chip ring VCO.
26
capacitors with capacitance ranging from 312 fF to 1.4 pF. By electrically tuning them
via Vtue, the load resonates near Dgs as shown on Fig. 3-6. This current driver fed by
the output of the ring VCO produces oscillating current in the inductor at the VCO
microwave frequency. In Fig. 3-4, the sizes (gate width/gate length) of transistors
M1~M4 are 80 im/280 nm, 80 pm/280 nm, 72 pm/500 nm and 720 pm/500 nm,
respectively. To improve the performance of this inductor for advanced NV sensing
protocols [16, 17], we need to increase the applied microwave field amplitude. The
amplitude is enhanced by a factor Q compared to the driver DC bias current (Ibias
~ 5 mA from a 2.5 V power supply), where Q (~ 15) is the quality factor of the
inductor. In addition, a three-turn loop is used to multiply the microwave field
strength. Overall, we have 25x enhanced microwave field strength compared to a
non-resonant single turn loop. The High-frequency electromagnetic fields simulations
(HFSS) result is plotted in Fig. 3-7. The magnetic field amplitude is plotted as a
function of distance from the inductor center
The advanced NV sensing protocols mentioned above also require highly uniform
microwave fields over the excitation volume. To achieve this, three capacitive parasitic
loops are inserted. The radius of the these loops is tailored, so that their opposite
induced field homogenize the overall generated field. Another degree of freedom
is the capacitive gaps in the parasitic loops. This controls the amount of current
flowing in these loops. A comparison between the field distribution of a simple one
turn loop and another one with the capacitively loaded parasitic loop as a function
of the distance from the center of the inductor is shown in Fig. 3-8. This loading
enhances the homogeneity of the field without sacrificing the field magnitude. The
two important parameters (i.e., the parasitic loop radius and the capacitive gap) for
the three parasitic loops are optimized to achieve > 95% uniformity. The detailed
dimensions of the final loop inductor implemented in the chip is shown in Fig. 3-9.
The loop outer diameter is 236 pm, and exhibits an inductance of -3 nH.
27
1.8 2 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8Frequency (GHz)
Figure 3-6: The amplitude of the AC current flowing in the inductor as a function offrequency. The plot shows a resonance behavior at 2.87GHz.
1 4
E
CDC3CM
Res. 3-turn -on-res. 1-turn
25X
1.4
1.2-N
0.8-
0.6-
0.4-
0.2-
-45 -30 -15 0 15Distance from antenna center (pm)
Figure 3-7: The simulated magnetic field profiles as a function of distance from thecenter of the proposed resonant inductor and a non-resonant one.
28
14-
12 -
U..
=O
0
LL
aO
10 -
8
6
4
2
0
K
30 45
1.4
1.2
E
%N,
ca
1=
0.8-
0.6-
0.4-
0.2-
T I V
0 -R 1-R/4 0 R/4 R/2
Distance from inductor center
Figure 3-8: The simulated magnetic field profiles as a function of distance from theinductor center with and without capacitive parasitic loop.
l0p
Metal 9
Via 8-9Metal 8
Via 7-9 11 8p Sp-i -
Metal 7
2 36p
Figure 3-9: The layout of the 3-turn on-chip inductor with parasitic capacitive loops.
29
9 4
-- ----------4%
................................. f.
................................. I -
-S
2.6p
-H
3.4 On-chip Optical Detection
The NV spin transitions are detected using an on-chip photodetector. A CMOS-
compatible periodic metal-dielectric structure in the Metal 8 interconnect layer filters
green pump light (Fig. 3-10). The incident light couples to the surface plasmon
polariton (SPP) at the metal-dielectric interface, where it undergo frequency-dependent
Ohmic loss [18, 19]. Therefore green light is attenuated more than red light, achieving
a finite green to red rejection. As shown in Fig. 3-10 each slit is considered as a parallel
plate waveguide transmitting light inside a dielectric (relative permitivity ed) in the z-
direction. The incoming light is modeled as a plane wave with a transverse electrical
field E., and a propagation constant ko=owVQ/c (c is the speed of light in vacuum).
Then, it is coupled to the TEM mode of the parallel-plate waveguides, which has
identical propagation constant ko. As the propagating wave interacts with the metal,
the surface plasmon polariton (SPP) mode at the metal-dielectric interface is excited.
Note that the dispersion relation of SPP mode is:
kzSPP =OSPP + jcespp = ko E (3.2)Em + Ed
900nm
Figure 3-10: The layout of the single-layer plasmonic grating filter impleniented on
Metal 8.
30
0
0
c,-10 0
.wI w 4--j -15- A >
OC ~E-20- A 2-
0
-25-
-30- 0 ,500 600 700 800 500 600 700 800
Wavelength (nm) Wavelength (nm)(a) (b)
Figure 3-11: The (a) real and (b) imaginary parts of the relative dielectric constant ofcopper. The calculated plots (solid lines) are based on the Drude-Brendel-Bormannmodel. The measured data (triangles and squares) are from 120, 21].
To quantify the loss of the SPP mode using the real and imaginary parts of the
dielectric constant of copper shown in Fig. 3-11, the attenuation factor aspp is derived
in (3.2):
caSPP = Im(ko + ~ k0 emi (3.3)
where em=Emr + jcmi is the permittivity of the metal, which is copper for M8 in
the technology we use. This exhibits large difference between red (aSpp~O.O1ko)
and green (asppe0.26ko). That is due to not only the decreased Emr of copper in
green, which is normal plasmonic property described by the Drude model, but also an
abnormal dispersion with increased emi, which is caused by interband transitions of
bound electrons excited by the photons. The relative permitivity of the dielectric (Ed)
is 1.5 in the above calculations.
The simulated Lumerical FDTD transmission through the filter at the wavelengths
of 532 nm (green) and 700 nm (red) are shown in Fig. 3-12. The measured green-to-red
31
0.5 0.5
(a) (b)
Figure 3-12: FDTD simulated light transmission through the optical filter at (a) green
light (A=532 nm) (b) red light (A=700 nm).
suppression ratio due to this filter (d=900 nm) is 10 dB. Increasing the slit thickness
by stacking more grating layers (in M7, M6...) is expected to further increase the
suppression of the green light [19]. However, that causes degradation of red-light
transmission due to the scattering at the side walls formed by the sparse inter-layer
via pillars.
To detect the NV-emitted red fluorescence, a P+/N-well/P-sub photodiode as
shown in Fig. 3-13a is used. This specific configuration is preferable for long wavelength
detection [22]. Since we place the photodiode with its conductive layers below the
inductor (Fig. 3-2), large eddy currents near 2.87 GHz can be induced if we use a
conventional unpatterned physical layout as shown in Fig. 3-13b. This reduces the
quality factor of the inductor, resulting in microwave amplitude reduction. By dividing
the photodiode area into four subareas (2x2 array) as shown in Fig. 3-14a, this eddy
current is reduced by half as shown in Eq. 3.4, and Eq. 3.5:
Peddy X (d@(t)/dt)2 L4 ( )2 (3) dB 234
R L o dt
PeddyProp 3( =(.)3 5)
PeddyConv La 2
where, Peddy is the eddy current loss and L is the side length of the photodiode. 0(t)
32
Cathode P+ Layer
Anode N-Well CathodeConnection
N-Well EddyLCurrents
AnodeP Substrate Connection
___________________________________P Substrate
(a) (b)
Figure 3-13: (a) The cross section of the P+/N-well/P-sub (b) unpatterned photodiodelayout with eddy current loops in the active area and possibly the anode and cathodeconnection.
and B are the magnetic flux and the magnetic field generated by the loop inductor
in Metal 9, respectively. t is the time and R is the active area resistance. Similarly,
by dividing the photodiode active area into N x N subareas, the eddy current loss is
reduced by 1/N. Furthermore, the anode/cathode connectors are arranged in a radial
way that is similar way to patterned ground shielding used in CMOS inductors [23].
This arrangement (Shown in Fig. 3-14b, and Fig. 3-14c) avoids any closed loops,
which helps to cut the eddy current that may flow in the metallic connections. As
shown in Fig. 3-14d the eddy current loops flow only in the patterned active area
only. The photodiode has a measured responsivity of 0.23 A/W at the wavelength of
532 nm, which corresponds to a quantum efficiency of 0.54.
3.5 Passivation Layer Etching
The chip is fabricated with a standard 65nm low-power CMOS technology from TSMC.
This CMOS technology provides a top nitride layer for surface passivation. This
passivation layer emits unwanted background red fluorescence under green excitation.
Fig. 3-15 shows an optical image for the red fluorescence of the chip with and without
33
P+ LayerN-Well
L
Shallow-TnchIsolation (STI)
(a)
N-Well P+ Layer CathodeConnection
AnodeConnection
Shallow-TrenchIsolation (STI)
(c)
P+ LayerN-Well
AnodeConnection
Shallow-TrenchIsolation (STI)
(b)
N-Weil P+ Layer CathodeConnection
AnodeConnection
EddyCurrents
Shallow-TrenchIsolation (STI)
(d)
Figure 3-14: The proposed patterned photodiode layout (a) 2x2 active area layout (b)the radially connected anode connection that prevents any closed loops implementedin M1. (c) the full layout with cathode connection in M2 added. (d) the full layoutwith eddy current loops in the patterned active area only.
passivation etching. As the figure clearly shows fluorescence intensity drops significantly
after the passivation removal. CF4 plasma dry reactive ion etching is used in this
process. The etching is done at the clean rooms at MIT. This is the only post processing
step required after fabrication. However, this step can be done later at the CMOS
foundry for product level fabrication in the future. In the measurements discussed in
this thesis, we focus on ensemble measurements (more details in Chapter 4) . The
background noise that may come from the passivation layer doesn't affect much the
SNR of the experiments discussed in this thesis. However, In future experiments,
where we address single or small number of NV centers (Chapter 5), this background
34
15000 E 15000
50 50'A n
100 10000 100 10000
150 150
5000 5000 4.200 200
250 25050 100 150 200 250 50 100 150 200 250
(a) (b)
Figure 3-15: Optical image for the red fluorescence emitted by the chip surface undergreen excitation (a) without passivation layer removed (b) with passivation layerremoved. Scale bar is 100pm.
noise can be very critical.
3.6 Diamond Sample Preparation
As mentioned in Section 3.2, two prototypes in this thesis are discussed. The NV
centers is embedded in nanodiamond layer that covers the chip in the first one, and
in a slab of single crystalline diamond in the second one. The diamond sample in
both cases are placed on the top of the chip on the sensing area (See Fig. 3-16a)
including the on-chip inductor, the optical filter and the photodiode. In this section,
the preparation of both samples and how they are deposited on the top of the chip
are discussed.
3.6.1 Nanodiamond Placement
A solution of nanodiamond particle is deposited on top of the chip by adding a droplet
using pipette. The solution then evaporates leaving the nanodiamond particles on the
chip surface after drying. Figure 3-16b shows the sensing area of the chip with the
nanodiamond particle covering the chip.
35
3.6.2 Bulk Diamond Placement
A 500 pm x 500 pm x 500 pm diamond slab is picked and placed on top of the chip
under a microscope. The diamond is a single crystalline CVD-grown diamond from
Element 6. It is electronically irradiated with a dosage of 1018 e- /cm 2 at 1 MeV, and
then annealed for 2 hours at 8500C. This produces an estimated NV centers density
of - 0.01 ppm. Immersion oil is used to adhere the diamond slab to the chip. By
bridging the difference of the refractive index, the oil also minimizes the fluorescence
loss. A 450 cut is introduced in the diamond's corner as shown in Fig. 3-3 to direct the
vertical incident green laser horizontally to further enhance the overall green rejection
ratio. Fig. 3-16c shows an optical image of the diamond slab on top of the chip.
36
On-chipInductor
*N Sensing Area(Optical Filter &
Photodiode)
(a)
Nanodiamoid Particles Single Cry al Diamond
(b) (c)
Figure 3-16: Top-view micrograph of (a) the fabricated CMOS chip sensing areaincluding the inductor, optical filter, and photodiode. (b) the same area on thechip covered with nanodimaond particles. (c) the sensing area covered by the singlecrystalline diamond slab. Scale bar is 100pm.
37
38
Chapter 4
Experimental Results
In this section the measurement results of the two prototypes discussed in Chapter 3
are introduced. The CMOS chip is wire-bonded on a PCB board. The PCB photo
with the chip micrograph are shown in Fig. 4-1.
4.1 Nanodiamonds Results
We start with measuring the ODMR of the nanodiamonds attached on top of the chip
using external optical detection. This step is important to verify the operation of the
microwave source. Figure 4-2a shows the test setup of the experiment. A linearly
polarized DPSS green laser beam (2 W, A = 532 nm, Verdi G2, Coherent) is delivered
to the diamond through a telescope of fi = 35 mm and f2 = 150 mm. The beam
diameter incident on the diamond is -500 pm. The lens system given is used as
our laser is placed far from the sample for experimental convenience, though it is
important to note that the lens system could be avoided if the laser was positioned
nearby the CMOS chip or even replaced with other options as discussed in Chapter 5.
The microwave frequency is swept to address the NV center and excite the magnetic
resonance. This is done by changing the reference frequency of the PLL loop through
external signal generator (~ 120 MHz). The VCO output signal is frequency-divided
39
Microwave 0.8mmGeneration Circuitry
0s3"
Sensing Area
Figure 4-1: Optical micrograph of the CMOS chip (bottom) and photo of the printedcircuit board for testing (top).
40
A
F
EM-CCD 00-, Bext, = 0 .-
S0.98Bex = 1.72 mT--%4 0.96
0.96-Dichroic U)
532 nm 0.94Laser
Microscope 0.92Objective
E 0.9-Diamond 0LayersZ 0.88.
2.8 2.82 2.84 2.86 2.88 2.9 2.92 2.94. Frequency (GHz)
(a) (b)
Figure 4-2: (a) The experimental setup of the ODMR experiment using off-chip camera.(b) The corresponding measured and fitted ODMR spectrum at no external magneticfield and at 1.72 mT permanent magnet.
by 24 and then compared to the refrence signal provided by an external signal source
(HP ESG-D4000A). Then the spin dependant red fluorescence is collected using a
camera. As shown in Fig. 4-2b, an ODMR spectrum is measured under no external
biasing magnetic field. Due to the strong microwave signal genertated by the chip
a contrast of 10% is successfully obtained. An external magnetic field through a
permanent magnet is then applied. Figure 4-2b shows an extra broadening in the
ODMR curve in this case, which corresponds to the Zeeman splitting of the randomly
oriented NV centers. Spectral broadening, rather than splitting, is observed because
the effective Bz (hence the amount of splitting) for each orientation varies. The
magnetic field of the permanent magnet is estimated through curve fitting to be
1.72 mT.
The next step is to test the whole on-chip system by detecting the NV red
fluorescence using the on-chip photodiode. Similar laser setup is used as the one
shown in Fig. 4-2a, However, lock-in amplifier is used to detect the photodiode signal
as shown in Fig. 3-2, and Fig. 3-3. In this experiment A half-wave plate rotates the
41
14 i
12- Bext2 =2.2 mT -
C 8-
r 6-
2.
02.8 2.82 2.84 2.86 2.88 2.9 2.92 2.934
Frequency (GHz)
Figure 4-3: The measured and fitted ODMR spectrum using on-chip photodiode at
no external magnetic field and at 2.2 mT permanent magnet.
polarization of the laser beam to maximize the laser absorption through the periodic
mietal/dielectric structure in the Metal 8 layer. Since the filter rejection is still only
10 dB, there is a huge background green signal that complicates the measurements.
A lock-in detection technique is used to detect the NV red fluorescence. The green
laser beam continuously excites the NV ensemble, and the amplitude-modulated
(AM) microwave fields (fm = 1 kHz) drive the NV electron spin transition. This
is done by switching the microwave signal on and off with fm, rate (Fig. 3-2). The
spin-dependent fluorescence produces photo-current within the on-chip photodiode.
Then, the modulated photo-current is detected through the voltage drop across a 50 Q
resistor at fm, with a Stanford Research Systems lock-in amplifier (SR865A). The use
of the lock-in amplifier rejects the DC current offset of the photodiode, which is caused
by the unmodulated green laser, and avoids the low-frequency flicker noise accordingly.
The measured ODMR is shown in Fig. 4-3. Similar to the off-chip detection experiment,
42
Bexti =0 -
The ODMR with and without external magnetic field is recorded. A clear Zeeman
splitting is observed when 2.2 mT from permanent magnet is applied. Replacement
of nanodiamond with single-crystalline diamond addresses spectral broadening issue,
which is due to the random orientation of NV centers, and helps in demonstrating
vector field sensing as discussed in Section 4.2. In this experiment the sensitivity of
the sensor is estimated as given in Eq. 4.1:
S = - v t. (4.1)7e C
Here, u is the noise over the measured data, Av is the linewidth of the ODMR curve
in Fig. 4-3. -ye is the gyromagnetic ratio (28 GHz/T), C is the contrast, and t is the
integration time. The calculated DC magnetic field sensitivity of 74 pT/v Hz.
4.2 bulk diamond Results
To detect the ODMR of an ensemble of NV centers in a single crystalline diamond
slab, a similar setup as discussed in Section 4.1 is used, in addition to the same laser
setup as before, but with laser power of 500 mW . Although the overall filtering for
the green light is enhanced due to the cut introduced in the diamond, there is still
huge green light background detected by the photodiode. A lock-in technique is also
used in this experiment. The green laser beam continuously excites the NV ensemble,
and the frequency-modulated (FM) microwave fields (fm = 1.5 kHz and modulation
depth of 6 MHz) drive the NV electron spin transition (Fig. 3-3). A 50 Q resistor
is also used here to read the photocurrent at fm with one second integration time,
which corresponds to the equivalent noise bandwidth of 0.078 Hz (considering the
filter roll-off of 24 dB/oct) of the lock-in amplifier.
Figure 4-4 shows the lock-in signal for the ODMR experiment under zero external
magnetic field applied. This spectrum corresponds to the derivative of the ODMR
43
I I I I I I I I
> 600* 400C 200
S01
-200 --
o -400--600-
2.7 2.75 2.8 2.85 2.9 2.95 3 3.05
Frequency (GHz)
Figure 4-4: Frequency-modulated (FM) lock-in signal of NV spin-dependent fluores-
cence at zero external magnetic field.
spectrum shown in Fig. 2-4. Next, a permanent magnet (6.27 mT) is aligned to
split the spin transitions of the four NV orientations. Figure 4-5 plots the ODMR
spectrum, which exhibits the expected eight spin transitions (Fig. 2-4). The use of the
corresponding four NV ensembles enables vector magnetometry. In particular, it is
important to note that the spin transitions at v_ = 2.8303 GHz and v+ = 2.9330 GHz
of the NV ensemble.
Monitoring the lock-in signal V at v_. and v+ enables independent measurements
of magnetic field and temperature. Specifically, the sum of the lock-in signal change
AV at vi± is proportional to AT, while the difference provides ABz:
11 AV AXVA T = 1 V + AV(4.2)
20T dV/df + dVdf (4_
and1 (Av AV
ABZ = .VA (4.3)2-ye dV/df + dVdfv) (
Figure 4-6 plots the detected AB, induced by an electromagnet (blue) and measured
44
150-
C 100-
e 50-
0--50--
-100
-1502.7 2.75 2.8 2.85 2.9 2.95 3 3.05
Frequency (GHz)
Figure 4-5: FM lock-in signal with a permanent magnet (B = 6.27 mT). The linewidthof the ODMR is 7 MHz. Slopes dV/df at v- = 2.8303 GHz and v+ = 2.9330 GHz are42.969 nV/MHz and 42.450 nV/MHz, respectively.
center frequency shift (red). The square-wave magnetic field applied in Fig. 4-6
is generated by an electromagnet. Alternating electrical current is used to avoid
magnetization. lock-in signals at both v± are observed while switching the polarity of
external electromagnet with a period of 26 min. The center frequency shift represents
the temperature shift in the experiment which is calibrated to determine the magnetic
sensitivity. The magnetic field sensitivity is given by the following relation:
S = UB, (4.4)VENBW
Here, B,B is the noise in AB, measurement, and ENBW is the equivalent noise
bandwidth of the lock-in detector. In the measurements, ENBW = 5/(64T) with a
time constant T of 1 second, accounting for the 24 dB/oct of the lock-in amplifier filter
roll-off. By measuring UB, of 6.3 [pT from the modulated spin-dependent fluorescence
(Inset in Fig. 4-6), a DC magnetic field sensitivity of 32.1 pT/Vii is determined. The
45
-- 1
2053
5
25tL " -. -d f --------SU -
32 -250-50
0 20 40 60 80 100 120Time (min)
Figure 4-6: On-chip magnetometry (Blue) and temperature effect (Red) separation by
detecting the effect of switching electromagnet on v± of the ODMR curve of Fig. 4-5.
sensitivity includes additional V/2 factor of v+ and v- signal average. This DC magnetic
field sensitivity is limited by the noise detected in the ENBW at fm = 1.5 kHz.
In order to understand the contribution of each component on the overall noise
performance of the system, the noise is measured using the lock-in amplifier. The
measured overall noise in the experiment is 38 nV/v/III at fm = 1.5 kHz. This is
primarily because of from the green laser intensity noise, due to the limited performance
of the optical filters. This laser intensity noise is orders of magnitude larger than other
noise sources:
1. The thermal noise of the 50 Q resistance (R), which is used to convert the photo-
current to voltage. This noise is 0.9 nV/vlz¶ as calculated from Eq. 4.5:
V2= 4KTRAf. (4.5)
where K is the boltzman constant, T is the temperature, and A f is the band-
width.
46
2. The NV red fluorescence shot noise is - 9 pV/vHz at fm 1.5 kHz as given in
Eq. 4.6:
V-2= R2 (2qiD)Af ~ R2(2q DaAf.CR
(4.6)
where Vmax is the maximum voltage in the ODMR curve in Fig. 4-5, which is
100 nV, C is the ODMR contrast - 0.02, and q is the electronic charge.
3. The amplitude noise converted from the microwave generator spectral purity
is 1 fV/V Hz (see Eq. 4.7). The measured phase noise (#p) of the PLL is -90
dBc/Hz at an offset frequency of 1.5 kHz (FM modulation frequency fmn).
V2 c (Vq5fm/) 2Af. (4.7)
where V, is the signal voltage amplitude, and # is the slope of the FM-ODMR
curve. We assume that p < 1.
47
48
Chapter 5
Conclusion and Future Work
In this thesis, the first implementation for a chip-scale quantum magnetometer by
integrating diamonds with CMOS technology is reported. An architecture in which
the essential components to detect NV-ODMR - a microwave generator, an inductor,
an optical pump beam filter, and a photodetector - are fabricated throughout the
CMOS multi-layers is created. Two prototypes to address NV spin ensembles in a layer
of nanodiamond particles and single crystalline diamond slab. In the nanodiamond
prototype, the measured sensitivity is 74 pT/v/Hz for scalar field magnetic field
sensing. In the second experiment with single crystal diamond, vector-field magnetic
field measurements are realized. The sensor magnetic sensitivity is 32.1 pT//l Hz.
In the prototypes reported in this thesis, the achieved magnetic field sensitivity
is orders of magnitude worse compared to the best DC sensitivities reported: 290
pT/v/fzl and 28 pT/v/III for vector [101 and scalar [11] magnetometry, respectively, to
our best knowledge. The sensitivity is mainly limited by the green laser intensity noise
as described in Chapter 4. However, this performance can be improved by including
(i) metal gratings in multiple CMOS metal layers based on the wavelength-dependent
Talbot effect [24] and (ii) fabricating a resonant grating [25] in diamond. These
additionally attenuate the green laser and consequently reduce the laser intensity noise
49
by several orders of magnitude. In addition, using a diamond waveguide geometry [10],
possibly with a higher NV density [26] (0.01 ppm to 10 ppm), should increase the
signal-to-noise ratio (SNR) by orders of magnitudes. Moreover, dynamical decoupling
sequences [16, 17, 27] can improve the sensitivity by a few orders of magnitude for
measuring magnetic fields at frequencies above the NV decoherence rate.
For practical sensor realization in the future, one critical component not presently
integrated into the presented CMOS-diamond platform which is the pump laser for
NV optical excitation. This optical pump can be integrated into the platform through
using a chip-scale laser diode [28], green Vertical-Cavity Surface-Emitting Lasers [29],
or CMOS-compatible waveguided delivery of the optical pump beam [30]. It is also
important to note that the CMOS-integration of all currently off-chip electronic
components, such as the phase-locked loop with frequency modulation and the lock-in
amplifier, has been demonstrated in prior research [31]. These indicate the feasibility
of millimeter-scale form factor for future quantum-sensing systems.
This hybrid CMOS-diamond platform can be extended toward on-chip sensing of
other quantities such as electric fields. The architecture can also be extended towards
an imaging platform (see Fig. 5-1). An array of photodiodes can be implemented
below the microwave delivery structure to create the image by measuring the NV-
ODMR on each pixel. In addition to chip-scale quantum sensing capability, the
CMOS-based spin control and readout scheme presented in this thesis can uniquely
provide a scalable solution for implementing spin quantum-bit (Qubit) controls. This
is, in particular, essential to develop a large-scale quantum system [1, 2, 32, 33]. A
conceptual scheme is shown in Fig. 5-2. Individual on-chip control and readout of
Qubits can be implemented. Since the required components are integrated in the same
platform, feedback control can be implemented through on-chip logic. This would
enable on-chip quantum information processing [34-36].
50
Microwave DeliveryStructureI g
Photodio A rray
Microwave SignalGeneration and
Switching
Electric Field .. T.
Tuning
ntroiu Ciogic
Figure 5-1: NV-based magnetic imager with an array photodiodes below a singlemicrowave delivery structure.
Qubit Node---........... ..----------..
ElectricField Tuning
MicrowaveControl
ReadoutCircuit
ElectricField Tuning
MicrowaveControl
ReadoutCircuit
Microwave Delivery PhotodiodeStructure
Qubit
ElectricField Tuning
MicrowaveControl
I ReadoutCircuit
51
Control logic
Figure 5-2: NV-based individual Qubit control for scalable quantum informationprocessing applications.
I Control logic
52
Bibliography
[1] Edoardo Charbon, Fabio Sebastiano, Masoud Babaie, Andrei Vladimirescu, MinaShahmohammadi, Robert Bogdan Staszewski, Harald A.R. Homulle, BishnuPatra, Jeroen P.G. van Dijk, Rosario M. Incandela, Lin Song, and BahadorValizadehpasha. Cryo-CMOS Circuits and Systems for Scalable Quantum Com-puting. In IEEE International Solid-State Circuits Conference, pages 264-265, SanFrancisco, CA, 2017. ISBN 9781509037582. doi: 10.1109/ISSCC.2017.7870359.
[2] Joseph C Bardin, Evan Jeffrey, Erik Lucero, Trent Huang, Ofer Naaman, RamiBarends, Ted White, Marissa Giustina, Daniel Sank, Pedram Roushan, KunalArya, Benjamin Chiaro, Julian Kelly, Jimmy Chen, Brian Burkett, Yu Chen,Andrew Dunsworth, Austin Fowler, Brooks Foxen, Craig Gidney, Rob Graff, PaulKlimov, Josh Mutus, Matthew Mcewen, Anthony Megrant, Matthew Neeley,Charles Neill, Chris Quintana, Amit Vainsencher, Hartmut Neven, and JohnMartinis. A 28nm Bulk-CMOS 4-to-8GHz <2mW Cryogenic Pulse Modulatorfor Scalable Quantum Computing. In IEEE International Solid-State CircuitsConference, pages 456-457, San Francisco, CA, 2019. ISBN 9781538685310.
[31 Georg Kucsko, PC Maurer, Norman Ying Yao, Michael Kubo, HJ Noh, PK Lo,Hongkun Park, and Mikhail D Lukin. Nanometre-scale thermometry in a livingcell. Nature, 500(7460):54, 2013.
[4] Florian Dolde, Helmut Fedder, Marcus W Doherty, Tobias N6bauer, FlorianRempp, Gopalakrishnan Balasubramanian, Thomas Wolf, Friedemann Reinhard,Lloyd CL Hollenberg, Fedor Jelezko, et al. Electric-field sensing using singlediamond spins. Nature Physics, 7(6):459, 2011.
[5] JR Maze, PL Stanwix, JS Hodges, S Hong, JM Taylor, P Cappellaro, L Jiang,MV Gurudev Dutt, E Togan, AS Zibrov, et al. Nanoscale magnetic sensing withan individual electronic spin in diamond. Nature, 455(7213):644, 2008.
[6] Gopalakrishnan Balasubramanian, IY Chan, Roman Kolesov, Mohannad Al-Hmoud, Julia Tisler, Chang Shin, Changdong Kim, Aleksander Wojcik, Philip RHemmer, Anke Krueger, et al. Nanoscale imaging magnetometry with diamondspins under ambient conditions. Nature, 455(7213):648, 2008.
[7] D. Le Sage, K. Arai, D. R. Glenn, S. J. Devience, L. M. Pham, L. Rahn-Lee,M. D. Lukin, A. Yacoby, A. Komeili, and R. L. Walsworth. Optical Magnetic
53
Imaging of Living Cells. Nature, 496(7446):486-489, 2013. ISSN 00280836. doi:10.1038/nature12072.
[8] Gopalakrishnan Balasubramanian, Philipp Neumann, Daniel Twitchen, MatthewMarkham, Roman Kolesov, Norikazu Mizuochi, Junichi Isoya, Jocelyn Achard,Johannes Beck, Julia Tissler, et al. Ultralong spin coherence time in isotopicallyengineered diamond. Nature materials, 8(5):383, 2009.
[9] Thomas Wolf, Philipp Neumann, Kazuo Nakamura, Hitoshi Sumiya, TakeshiOhshima, Junichi Isoya, and J6rg Wrachtrup. Subpicotesla diamond magnetome-try. Physical Review X, 5(4):041001, 2015.
[10] Hannah Clevenson, Matthew E Trusheim, Carson Teale, Tim Schr6der, DanielleBraje, and Dirk Englund. Broadband magnetometry and temperature sensingwith a light-trapping diamond waveguide. Nature Physics, 11(5):393, 2015.
[11] Georgios Chatzidrosos, Arne Wickenbrock, Lykourgos Bougas, Nathan Leefer,Teng Wu, Kasper Jensen, Yannick Dumeige, and Dmitry Budker. Miniaturecavity-enhanced diamond magnetometer. Physical Review Applied, 8(4):044019,2017.
[12] A. Gruber, A. Drabenstedt, C. Tietz, L. Fleury, J. Wrachtrup, and C. VonBorczyskowski. Scanning Confocal Optical Microscopy and Magnetic Resonanceon Single Defect Centers. Science, 276:2012-2014, 1997.
[13] S. Gambini, K. Skucha, P. P. Liu, J. Kim, and R. Krigel. A 10 kpixel cmos hallsensor array with baseline suppression and parallel readout for immunoassays.IEEE Journal of Solid-State Circuits, 48(1):302-317, Jan 2013. ISSN 1558-173X.doi: 10.1109/JSSC.2012.2224531.
[14] M. F. Snoeij, V. Schaffer, S. Udayashankar, and M. V. Ivanov. Integratedfluxgate magnetometer for use in isolated current sensing. IEEE Journalof Solid-State Circuits, 51(7):1684-1694, July 2016. ISSN 1558-173X. doi:10.1109/JSSC.2016.2554147.
[15] Behzad Razavi. RF microelectronics. Prentice Hall Press, USA, 2nd edition, 2011.ISBN 0137134738.
[16] Christian L Degen, F Reinhard, and P Cappellaro. Quantum sensing. Reviews ofmodern physics, 89(3):035002, 2017.
[17] John F Barry, Jennifer M Schloss, Erik Bauch, Matthew J Turner, Connor AHart, Linh M Pham, and Ronald L Walsworth. Sensitivity optimization fornv-diamond magnetometry. arXiv preprint arXiv:1903.08176, 2019.
[18] Anatoly V Zayats, Igor I Smolyaninov, and Alexei A Maradudin. Nano-optics ofsurface plasmon polaritons. Physics reports, 408(3-4):131-314, 2005.
54
119] Lingyu Hong, Hao Li, Haw Yang, and Kaushik Sengupta. Fully integratedfluorescence biosensors on-chip employing multi-functional nanoplasmonic opticalstructures in cmos. IEEE Journal of Solid-State Circuits, 52(9):2388-2406, 2017.
[20] Aleksandar D. Raki6, Aleksandra B. Djurisid, Jovan M. Elazar, and Marian L.Majewski. Optical Properties of Metallic Films for Vertical-Cavity Optoelec-tronic Devices. Applied Optics, 37(22):5271, 1998. ISSN 0003-6935. doi:10.1364/ao.37.005271.
[21] H.-J. Hagemann, W. Gudat, and C. Kunz. Optical Constants from the Far Infraredto the x-Ray Region: Mg, Al, Cu, Ag, Au, Bi, C, and Al 2 03. Journal of the OpticalSociety of America, 65(6):742, 1975. ISSN 0030-3941. doi: 10.1364/josa.65.000742.
[22] Kartikeya Murari, Ralph Etienne-Cummings, Nitish Thakor, and Gert Cauwen-berghs. Which photodiode to use: a comparison of cmos-compatible structures.IEEE sensors journal, 9(7):752-760, 2009.
[23] C Patrick Yue and S Simon Wong. On-chip spiral inductors with patternedground shields for si-based rf ics. IEEE Journal of solid-state circuits, 33(5):743-752, 1998.
[24] Jianming Wen, Yong Zhang, and Min Xiao. The talbot effect: recent advances inclassical optics, nonlinear optics, and quantum optics. Advances in optics andphotonics, 5(1):83-130, 2013.
[25] Song Peng and G Michael Morris. Resonant scattering from two-dimensionalgratings. JOSA A, 13(5):993-1005, 1996.
[26] VM Acosta, E Bauch, MP Ledbetter, C Santori, K-MC Fu, PE Barclay, RG Beau-soleil, H Linget, JF Roch, F Treussart, et al. Diamonds with a high density ofnitrogen-vacancy centers for magnetometry applications. Physical Review B, 80(11):115202, 2009.
[27] JM Taylor, P Cappellaro, L Childress, L Jiang, D Budker, PR Hemmer, A Yacoby,R Walsworth, and MD Lukin. High-sensitivity diamond magnetometer withnanoscale resolution. Nature Physics, 4(10):810, 2008.
[28] Thomas Oeckinghaus, Rainer St6hr, Roman Kolesov, Julia Tisler, FriedemannReinhard, and J6rg Wrachtrup. A compact, diode laser based excitation systemfor microscopy of nv centers. Review of Scientific Instruments, 85(7):073101,2014.
[29] Daiji Kasahara, Daisuke Morita, Takao Kosugi, Kyosuke Nakagawa, Jun Kawa-mata, Yu Higuchi, Hiroaki Matsumura, and Takashi Mukai. Demonstrationof blue and green gan-based vertical-cavity surface-emitting lasers by currentinjection at room temperature. Applied Physics Express, 4(7):072103, 2011.
55
[30] David J Moss, Roberto Morandotti, Alexander L Gaeta, and Michal Lipson. Newcmos-compatible platforms based on silicon nitride and hydex for nonlinear optics.Nature photonics, 7(8):597, 2013.
[31] Cheng Wang, Xiang Yi, James Mawdsley, Mina Kim, Zihan Wang, and RuonanHan. An On-Chip Fully-Electronic Molecular Clock Based on sub-THz RotationalSpectroscopy. Nature Electronics, 1(7):421-427, 2018.
[32] Norman Y Yao, Liang Jiang, Alexey V Gorshkov, Peter C Maurer, Geza Giedke,J Ignacio Cirac, and Mikhail D Lukin. Scalable architecture for a room tempera-ture solid-state quantum information processor. Nature communications, 3:800,2012.
[33] M Veldhorst, HGJ Eenink, CH Yang, and AS Dzurak. Silicon cmos architecturefor a spin-based quantum computer. Nature communications, 8(1):1766, 2017.
[34] Hannes Bernien, Bas Hensen, Wolfgang Pfaff, Gerwin Koolstra, MS Blok, LucioRobledo, TH Taminiau, Matthew Markham, DJ Twitchen, Lilian Childress, et al.Heralded entanglement between solid-state qubits separated by three metres.Nature, 497(7447):86, 2013.
135] Wolfgang Pfaff, BJ Hensen, Hannes Bernien, Suzanne B van Dam, Machiel S Blok,Tim H Taminiau, Marijn J Tiggelman, Raymond N Schouten, Matthew Markham,Daniel J Twitchen, et al. Unconditional quantum teleportation between distantsolid-state quantum bits. Science, 345(6196):532-535, 2014.
[36] Peter C Humphreys, Norbert Kalb, Jaco PJ Morits, Raymond N Schouten,Raymond FL Vermeulen, Daniel J Twitchen, Matthew Markham, and RonaldHanson. Deterministic delivery of remote entanglement on a quantum network.Nature, 558(7709):268, 2018.
56