Quantum Dot Infrared Photodetectors

Quantum-dots (QDs) provide the ultimate quantum system with a

three-dimension carrier confinement resulting in discrete electronic energy state.

Bulk Well

Ec E

D(E)

Ec E

D(E)

Dot

Ec

E D(E)

Wire

Ec E

D(E)

3D 2D

1D 0D

Fig. 2.3 Density of states in bulk material (3D), quantum well (2D),

quantum wire (1D), and quantum dot (0D).

12

Great effort has been made for fabricating quantum dot structure by means of etching

or local diffusion of quantum wells [23], selective growth [24-25], self-organized

growth [26].

Self-organized growth of quantum dots has been demonstrated using both

molecular beam epitaxy (MBE) and metal-organic chemical vapor deposition

(MOCVD)[27]. The growth methods offer the distinct advantage of producing

defect-free QD structures. From the Stranski-Ќrastanov (SK) growth mode, the

mismatch between substrate and wetting layer and the resulting strain have been

considered to be responsible for the driving force for the formation of QDs.

QDIP is one of the applications of QD heterostructures. Compared to QWIP and

SLIP, QDIP’s advantages are high responsivity, low dark current,. polarization

independent, broad-band detection spectrum and high temperature operation. Due to

non-uniformity of QDs, the broad-band spectrum can be observed. Furtherrmore, zero

dimensional QDs results in δ-like densities of states and 3D confinement of electrons

on space geometry cause the possibility of room temperature operation.

Direct imaging methods, such as scanning tunneling microscopy (STM), atomic

force microscopy (AFM), and transmission electron microscopy (TEM) can be used

to study the surface morphology of QDs. Photoluminescence (PL) have been used to

describe optical properties of InAs/GaAs QDs [28].

2.2 Process Flow

A flow chart of the device fabrication and testing is shown in Fig. 2.4, and the

details will be described latter.

Fig. 2.4 The flow chart of device fabrication and testing.

2.2.1 Fabrication Processes

For better device performance, any defects caused in the process sequence

should be minimized. Etching depth, ohmic contact and optical coupling into the

devices are important issues that require much attention.

This section will describe the device fabrication processes in details. Fig.2.5

shows the fabrication process sequences of quantum dot infrared photodetectors.

(1) Surface cleaning

Cleaning refers to removing undesired materials from the wafer before

subsequent process steps. Cleaning operations are performed before all major steps

during device processing for better reliability and performance. These steps

14

Fig. 2.5 Device fabrication processes of infrared photodetector, (a) the

first photoresist coating, (b) the first developing, (c )the first wet etching,

(d) photoresist cleaning, (e) the second photoresist coating, (f) the second

developing, (g) the second wet etching, (h) photoresist cleaning, (i) the

third photoresist coating (j) the third developing (k) metals evaporation ,

and (j) lift-off.

may employ organic solvents, vapor degreasing, and acids. We use organic solvents to

remove oils, greases, particles and organic material such as photoresist and to keep

the surface clean. Table 2.1 lists the cleaning conditions and purposes of the solvent.

(2) Formation of mesa structure

The process of mesa structure formation includes two stages, which are

lithography and etching.

In lithography stage, SHIPLEY S1813 positive photoresist was spun and coated

on the sample surface first and soft baked it 5 minutes to evaporate solvent contained

in the photoresist. Soft baking before exposure is necessary because photoresist

freshly spun is sticky. SHIPLEY MF-319 development solution was used to remove

the portion of photoresist exposed. Finally, 20 minutes hard baking was used to

formalize the pattern. The conditions of photo-lithography are listed in Table 2.2.

In etching stage, wet etching by liquid chemical etchant is adopted to remove the

portion of epilayers without protection of photoresist. Such an etching procedure is an

important part of various processing steps.

For device without edge thinning as control group, the epilayer is etched down to

contact layer. For device with edge thinning as experimental group, it’s etched down

to the designated depth. The etching depth is measured by a surface profile. The etching solution will be discussed in section 2.2.3. After wet etching, 280ͪͪ180 μm²

16

mesa structure was formed for device isolation.

(3) Formation of edge thinning structure

After mesa structure has been formed, device with edge thinning structure requires a similar lithography process to form a larger pattern (300ͪͪ200 μm²) which

includes the width of edge . Then it’s etched down to the contact layer to reduce the

thickness of edge and the edge thinning structure is formed

(4) Formation of contact

The third photolithography mask was used to define the region of contact.

Finally, Au/Ge/Ni alloy and Au were evaporated under the pressure of 6ͪ10^-6 torr.

Table 2.3 lists the alloy type, thickness, and deposition rate. The sample was put in the

rapid thermal annealer (RTA) to anneal for ohmic contact formation. The anneal rate

is 420 ^o C/min up to the temperature of 420 ^o C for 120 seconds.

Table 2.1 Conditions and purposes of the cleaning solvent Chemical Solution Clean Time (min) purposes

Acetone (CH

3

COCH

3

) 5 Clean photoresist,

organism

Methanol (CH

₃

OH) 5 Clean Acetone

D.I water (H

₂

O) 5 Clean Methanol

Table 2.2 The photolithography conditions

Pattern Formation Conditions

Spinning and Coating 4000 rpm 40 sec Soft bake 90

^o

C 5 min Exposure 15 sec

Development 20 sec Hard bake 90

^o

C 20 min

Table 2.3 Evaporation condition

Source Thickness (nm) Evaporation rate (nm/sec)

Au/Ge/Ni ( 84 / 12 / 4 ) 70 0.05 ~ 0.1

Au 230 0.1 ~ 0.2

18

2.2.2 H

3

PO

4

-H

2

O

2

-H

2

O Etching Solution

Wet etching proceeds through chemical reactions at the surface of the material.

For chemical reactions to take place, the etchant species must reach the surface and

react with the material appropriately. The reaction products must be removed from the

surface. Almost all GaAs etchants operate by first oxidizing the surface and then

dissolving the oxide, thereby removing some of the Gallium and Arsenic atoms.

Generally, the etchant contains one component that acts as the oxidizer and the other

that acts as the dissolving agent [29]. The H2O2 is the oxidizing agent and H3PO4 is

the dissolving agent. For better control over the etching depth, the etching solution is

85% H3PO4

:

H2O2

:

H2O = 2 : 1 : 15 (8ml: 4ml: 60ml). Fig. 2.7 displays the linear fitting of etching rate trials with different etching time. The etching rate can be

reckoned as about 7 nm/sec and the deviation is about 0.5 nm/sec. Fig. 2.8 displays

the etching uniformity at different etching depths. The uniformity can be reckoned as

about 40 nm. Although it etches GaAs to some degree in almost all compositions,

GaAs will not be etched in either H2O2 or H3PO4 alone.

Freshly mixed etchants may be hot, because exothermic reaction occurs when

H3PO4 mixed with H2O. As all chemical reactions, etching rate is sensitive to the

temperature. So the etchant should be cooled to a steady temperature at least for 30

min after mixing. From our experiment, the etching rate for GaAs is about 6~8

nm/sec at 20 ^o C.

Fig. 2.6 The etching rates of different composition ratios in

H

₃

PO

₄

-H

₂

O

₂

-H

₂

O system.

20

Fig. 2.7 The linear fitting of etching rate trials of H

3

PO

4

/H

2

O

2

/H

2

O:

8ml/4ml/60ml etching solution.

Fig. 2.8 The uniformity at different etching depths.

0 30 60 90 120 150 180

200 400 600 800 1000 1200

Linear Fitting Slope: 6.98346

Standard Error: 0.5251

Et chi ng D e pt h ( n m )

Etching Time (sec)

0 200 400 600 800 1000

10 20 30 40

S. D of Et chi ng D e pt h ( n m )

Average Etching Depth (nm)

2.2.3 Lift-off Process

Lift-off is a very important procedure in the process flow. Device fabrication

may work in vain, if lift-off process fails. Two important issues should be noticed,

which can make lift-off process success and reduce the process time. (a) Photoresist

should be thicker and hard-baking is not needed at the third photolithography step. (b)

The sample could be adhered on a piece of glass before metalization. The sample

holder of the evaporation system is made of metal. When the gold or alloy evaporated

to the sample, it also evaporated to the sample holder. The holder was heated and

transferred heat to the sample. The photoresist would be heated to become harder

which was difficult to dissolve afterwards by acetone. The glass can retard the heat

transfer from the holder. Following these two procedures make the lift-off process

easier.

2.3 Measurement Systems

After devices are made, the performance of the device is tested, such as

current-voltage measurement at various temperatures, the relative spectral response,

responsivity and the calculation of the specific detectivity (D*). The various

measurement systems will be described later.

22

2.3.1 Current-Voltage Measurement

The devices were bonded on the ceramic plates. The devices with larger

resistance at 77 K are chosen to be measured. All the current-voltage (I-V)

measurements were done by the HP4145B semiconductor parameter analyzer at

various temperatures (10 ~ 200 K) in the cryogenic system. The dark current

measurements were measured by carefully shielding the device from the background

infrared radiation. For photocurrent measurement, the radiation shield is taken off.

The system for the current-voltage measurement is shown in Fig. 2.9.

2.3.2 Introduction of FTIR

The infrared spectroscopy is an efficient method to reveal the properties

of electrons in the minibands of superlattices and the performance of infrared

photodetectors. Unlike the traditional infrared spectrometers which use the grating to

detect each frequency component’s absorption to get the whole spectral response, the

Fourier Transform Infrared (FTIR) spectrometer uses the Michelson interferometer to

get the whole spectral response simultaneously. We choose the FTIR spectrometer to

measure our detector’s quality. So the detail knowledge of FTIR is necessary.

At the heart of an FTIR spectrometer is a Michelson interferometer, as shown in

Fig. 2.10. It consists of three active components: a moving mirror, a fixed mirror, and

a beamsplitter. The radiation from the broad-band IR source impinges on the

Compressor Gas

Lines

Displac er

Temperature Sensor Detector Under Test KRS-5

window

Radiation shield Vaccum

Pump Temperature

controller

Keithley 236

Computer

Fig. 2.9 The I-V measurement system.

24

Fixed mirror

IR detector

Sample

IR Beam

Beamsplitter

Moving Mirror

Moving Mirror Position

Intensity

D1 D2 D3

Fig. 2.10 The principle of Michelson interferometer.

beamsplitter, half the IR beam transmits to the fixed mirror and the remaining half

reflects to the moving mirror. Then, those divided beams reflect back to the

beamsplitter and recombine to generate the interference pattern. The resulting beam

passes through the sample and finally impinges upon the detector. We first consider a

frequency ƒ′ component of IR source. The intensity of the interfered beam depends on

the optical pass difference between two split beams. The inset in Fig. 2.8 is the

“interferogram”, which is the record of the interference signal. When the moving

mirror is moved with a constant velocity, the intensity of radiation reaching the

detector is a sinusoidal manner. The intensity of the sinusoidal wave will reduce if the

sample absorbs in this frequency ƒ′. Then, the FTIR spectrometer takes the

summation of superimposed sinusoidal waves, each wave corresponding to a signal

frequency, to get the whole interference patterns [30].

The interferogram is a time domain spectrum. By using the Fourier

transformation, we can convert the interferogram into a frequency domain spectrum to

show the intensity as a function of frequency.

26

2.3.3 Relative Spectral Response

Spectral response measurement is used to measure relative output electrical

signal as a function of wavelengths of incident infrared radiation. The entire system

for measuring the spectral response is shown in Fig. 2.11.

Our system adopted PERKIN ELMER Fourier Transform Infrared Spectrometer

(FTIR). The FTIR spectrometer has several basic advantages over a classical

dispersive instrument, such as simultaneous measurement of the source wavelength,

higher energy throughput, negligible stray light, constant resolution and no

discontinuities. The detector was plastered on the closed-cycle cryostat system by the

copper adhesive tape. The infrared radiation is incident on the detector by passing

through the KRS-5 window to filter out the visible light and incident on the detector

to excite photo-electrons. The current was transferred into voltage signal by

STANFORD RESEARCH SYSTEM model SR570 transconductance preamplifier

and then the preamplified signal is converted into spectral by Fourier Transform.

Since the measured spectrums include background spectral response, we recorded this

background by the pyroelectric detector of FTIR first and then ratio the measured

spectrums against it to get the relative spectral response.

ˣ

ˣ˘˥˞˜ˡʳ˘˟ˠ˘˥ʳ˙̂̈̅˼˸̅ʳ

˧̅˴́̆˹̂̅̀ʳ˜́˹̅˴̅˸˷ʳ

˦̃˸˶̇̅̂̀˸̇˸̅

˜́˹̅˴̅˸˷ʳ˥˴˷˼˴̇˼̂́

˜́˹̅˴̅˸˷ʳ˥˴˷˼˴̇˼̂́

˗˸̇˸˶̇̂̅ʳ˼́ʳ˶̅̌̂̆̇˴̇

˖̈̅̅˸́̇ʳ˔̀̃˿˼˹˼˸̅ʳ

˦˥ˈˊ˃

˖̂̀̃̈̇˸̅

˼́̃̈̇

˼́̃̈̇

̂̈̇̃̈̇

̂̈̇̃̈̇

Fig. 2.11 The setup to measure relative spectral response.

28

2.3.4 Absolute Responsivity

After the relative response is taken, the blackbody radiation at high temperature

is used to get the absolute responsivity. Responsivity is the ratio of the output S

(usually in amperes or volts) to the radiant input

I

e⁽O) (in watts). Considered blackbody source at temperature T modulated at a frequency f that produces the

observed output

radiation power. T and f express blackbody source temperature and chopper

frequency.

Figure 2.12 displays the schematic diagram showing the setup for measuring the

responsivity. The blackbody source is 800 K for a thermal infrared detector test. The

variable-speed chopper modulates the signal at a frequency f by rotating a notched

wheel in front of the source. The notches alternately cover and uncover the source,

producing a nearly square-wave signal if the source aperture is small compared to the

notch width. The detector is located at a known distance from the source so that the

signal on the detector can be calculated. The photocurrent with applied bias was

amplified by SR570 current preamplifier and transferred into the voltage signal. Both

the chopper frequency controller and the preamplifier signal were connected with

SR830 DSP lock-in-amplifier to demodulate the voltage signal.

The absolute peak responsivity (R_peak) formula can be obtained by measuring

both the responsivity and the relative spectral response

]

Where S is the photo-voltage measured by lock-in-amplifier, ABB, A_d are areas of

blackbody source and detector, respectively, TKRS-5, TGaAs are transmission coefficients

of KRS-5 windows (0.7) and GaAs (0.8) respectively, FF is chopper modulation factor

and our system is 0.45, R is the distance between blackbody source and detector, TBB,

Trm are absolute temperature of blackbody source (800 K) and background (300 K),

R’(O) is normalized relative spectral response, M(O^,TBB) is blackbody radiation energy.

The Eq. (2-2) indicates the radiance of difference of blackbody light source and

background light source. The π ʳ factor is taking into account that blackbody light

source and background light source are Lambertian Radiators. The absolute spectral

response can be obtained by absolute peak responsivity multiplying by normalized

relative spectral response.

30

˕˿˴˶˾˵̂˷̌ʳ̅˴˷˼˴̇˼̂́ʻˋ˃˃˾ʼ

˗˸̇˸˶̇̂̅ʳ˼́ʳ˶̅̌̂̆̇˴̇

˗˦ˣʳ˿̂˶˾ˀ˼́ˀ˴̀̃˿˼˹˼˸̅

ʻ˦˥ˋˆ˃ʼ ˖̂̀̃̈̇˸̅

˖̈̅̅˸́̇ʳ˴̀̃˿˼˹˼˸̅

ʻ˦˥ˈˊ˃ʼ

˼́̃̈̇

̂̈̇̃̈̇

˥˸˹˸̅˸́˶˸

˶˻̂̃̃˸̅

ˠ̂˷̈˿˴̇˸˷ʳ

˜˥

Fig. 2.12 The setup to measure absolute spectral response.

2.3.5 Specific Detectivity

Detectivity is the signal to noise ratio (SNR). The larger SNR, the better device

is. The more useful figure of merit is the normalized detectivity D*, which normalizes

the detector area and bandwidth

D* =

n p d

i f A

R '

(2-3)

Where R_p is the responsivity of the detector, A_d is the area of the detector, 'f ^{is the}

bandwidth of the measurement and

i

n is the noise current, which is attributed to the

shot noise. It can be expressed as

in = 4eI^dgⁿ'f (2-4)

The advantage of D* as a figure of merit is that it is normalized to an active

detector area of 1 cm² and noise bandwidth of 1 Hz. Therefore, D* may be used to

compared directly the merit of detectors of different size whose performance was

measured using different bandwidths.

32

2.4 Edge Thinning Introduction

Edge thinning is the structure which reduces the surface leakage current by

pinching off surface depletion region and junction depletion region in the edge. It was

facilitated on heterojunction bipolar transistor to reduce the surface recombination and

consequently enhance the current gain. In this work, edge thinning structure is

adopted to reduce the surface leakage current of quantum dot infrared photodetector

(QDIP) and consequently enhance its operation temperature.

Fig. 2.13 The pinching off mechanism of edge thinning structure.

As Fig. 2.13 shows, the profile of surface depletion region is different between

devices with edge thinning structure and those without because of the profile of

surface edge. In device with edge thinning structure, the junction depletion region and

the surface depletion region may pinch off in the edge and consequently block the

path of surface leakage current. As a result, the surface leakage current may be

reduced and the operation temperature could be enhanced.

n

Junction depletion region Surface depletion region

No edge thinning Edge thinning

pinch off

The thickness of junction depletion region can be

Where ߝ_௦ is the dielectric constant; V_bi , the built-in potential ; e, the electronic charge; Nd , the n-type doping concentration; ni , the intrinsic carrier concentration.

The thickness of surface depletion region can be evaluated by the minority

carrier hole concentration as a function of distance from the surface.

¸ ¸

The parameter s is called the surface recombination velocity ; g’, the generation

rate of the excess carrier; ߬_௣௢, the excess minority carrier lifetime; Dp ,the diffusion

coefficient of the excess minority carrier; L_P , the diffusion length of the excess

minority carrier. The thickness of surface depletion region can be evaluated as the

distance from the transition point of the steady state of

G

p(x) to the surface.

The appropriate thickness of edge thinning structure is supposed to be close to

the sum of the thickness of junction depletion region and the surface depletion region

so that they can pinch off and consequently block the surface leakage current.

34

2.5 Atomic Layer Deposition Mechanism and the transmission of Al 2 O 3 in the infrared

spectrum

In air H2O vapor is adsorbed on most surfaces, forming a hydroxyl group. After

placing the substrate in the reactor, Trimethyl Aluminum (TMA) is pulsed into the

reaction chamber as Fig. 2.14 (a). Then, Trimethyl Aluminum (TMA) reacts with the

adsorbed hydroxyl groups, producing methane (CH4) as the reaction product as shown

in Fig. 2.14 (b). Trimethyl Aluminum (TMA) will react with the adsorbed hydroxyl

groups, until the surface is passivated. However, TMA does not react with itself,

terminating the reaction to one layer. This causes the perfect uniformity of ALD. Then,

the excess TMA is pumped away with methane reaction product as Fig. 2.14 (c). After

the TMA and methane reaction product is pumped away, water vapor (H2O) is pulsed

into the reaction chamber again as Fig. 2.14 (d). H2O will react with the dangling

methyl groups on the new surface forming aluminum-oxygen (Al-O) bridges and

hydroxyl surface groups, and waiting for a new TMA pulse. The reaction product

methane is pumped away again as Fig. 2.14 (e). However, excess H2O vapor does not

react with the hydroxyl surface groups, again causing perfect passivation to one

atomic layer as Fig. 2.14 (f). One TMA and one H2O vapor pulse form one cycle, with

approximately 1 Angstrom per cycle. Each cycle including pulsing and pumping takes

3 seconds as Fig. 2.14 (g). Since each pair of gas pulses (one cycle) produces exactly

one monolayer of film, the thickness of the resulting film may be precisely controlled

by the number of deposition cycles. Figs. 2.14 (a) to (g) show the Al2O3 forming

process sequences.

Fig. 2.15 shows the transmission of GaAs substrate (S.I) and GaAs with

10-nm-thick Al2O3 layer deposited by ALD in the infrared spectrum. Obviously, the

transmission of GaAs substrate and GaAs with Al2O3 layer in the infrared spectrum

are similar. It means that the Al2O3 layer does not absorb infrared light. Hence, the

Al2O3 layer deposited by ALD doesn’t block the QDIPs’ absorbsion of infrared light.

36

Fig. 2.14 The Al

2

O

3

layer forming process sequence.

Fig. 2.15 The transmission of GaAs and GaAs with Al

2

O

3

layer in the infrared spectrum

.

4 8 12 16 20 24

0.00 0.05 0.10 0.15 0.20 0.25 0.30

T ran sm ittan c e

Wavelength

⁽

_P O

⁾

GaAs

GaAs/10nm Al₂O₃

38

Chapter 3 Edge Thinning Structure with Different Depths on QDIPs

Self-assembled InAs quantum dots (QDs) on GaAs substrate using Stranski-

Ќrastanov (SK) growth mode by molecular beam epitaxy (MBE) have attracted much

attention in recent years [38-39]. Due to its long capture and relaxation times, the QD

structure is suitable for the optoelectronic applications of the infrared photodetectors

[40-42], lasers [43], and optical memories [44]. The fabrication of QDIPs has

attracted growing interest recently [45-47]. Although the advantages such as

high-temperature operation and insensitivity to incident light polarization have made

QDIPs superior than quantum-well infrared photodetectors (QWIPs) for applications.

It still needs lots of efforts to achieve room temperature operation.

In this work, the edge thinning structure is adopted to reduce the dark current and

enhance the operation temperature. In this chapter, the edge thinning structure with

different depths on QDIPs will be investigated.

3.1 The Effect of Edge Thinning Structure with Different Depths on IV Characteristics of QDIPs

3.1.1 Sample Preparation

The samples investigated in this article named QDIP75 were grown on

(100)-oriented semi-insulating GaAs substrates using a Riber Compact 21

solid-source molecular-beam epitaxy (MBE) system. The device structure is shown in

Fig. 3.1 The active region consists of ten periods of 2.4 ML (monolayer) InAs QD

layers separated by 30-nm-thick undoped GaAs barrier layers. 600 and 300 nm n⁺

GaAs layers doped with n=2×10¹⁸ cm⁻³ were grown to sandwich the active region as

the bottom and top contact layers.

Three devices are fabricated as shown in Fig. 3.2 (a), (b), (c), respectively.

Device A is the standard device without edge thinning structure, which was etched

down directly to bottom contact layer in a mesa pattern (ʹͺͲ ൈ ͳͺͲρ^ଶ) to form

bottom ohmic contact. Device B is the device with edge thinning structure at top

contact layer, which was first etched down to top contact layer in a mesa pattern

(ʹͺͲ ൈ ͳͺͲρ^ଶ) and then further etched down to bottom contact layer in a larger

mesa pattern (͵ͲͲ ൈ ʹͲͲρ^ଶ) to form bottom ohmic contact. Device C is the device

with edge thinning structure at quantum dot layer, which was first etched down to

40

the bottom contact layer in a larger mesa pattern (͵ͲͲ ൈ ʹͲͲρ^ଶ) to form bottom

ohmic contact. The thicknesses of the edge structure are between 100nm~200nm.

Fig. 3.1 The schematic diagram of device structure (QDIP75).

Fig. 3.2 The schematic diagram of devices (a) A, the standard device without edge thinning structure, (b) B, the device with edge

thinning structure at top contact layer, (c) C, the device with edge thinning structure at quantum dot layer.

(a)

Device A

(b)

Device B

(c)

Device C

Edge Thinning Edge Thinning

3.1.2 Results and Discussion

Figs. 3.3 to 3.5 show the I-V characteristics of devices A (without edge thinning

structure), B (with edge thinning structure at top contact) and C (with edge thinning

structure at quantum dot layer), respectively. The background limited performance

(BLIP) of all devices can be reckoned as 70 K. Fig 3.6 (a) and (b) show the

comparison of dark I-V characteristics of devices A, B, and C at 20 K and 90 K

respectively. At 20 K, the dark current of devices B and C are very similar. The dark

current of both devices B and C are about two orders of magnitude lower than that of

device A at bias voltage from about -0.8V to -1.3V. It’s because of the natural

device A at bias voltage from about -0.8V to -1.3V. It’s because of the natural

在文檔中 利用邊緣減薄結構提升量子點紅外光偵測器操作溫度之研究 (頁 21-0)