Intraband stark shift in twocolour quantum dot infrared photodetectors

1
Intraband stark shift in two-colour quantum dot
infrared photodetectors
P. Aivaliotis1, E.A. Zibik1, L.R.Wilson1, J.P.R.
David2, M. Hopkinson2, C. Groves2 and J. W.
Cockburn1
1 Department of Physics and Astronomy, Sheffield
University, Sheffield, S3 7RH, UK 2 Department of
Electronic and Electrical Engineering, Sheffield
University, Sheffield, S1 3JD, UK e-mail
luke.wilson_at_shef.ac.uk, e-mail
j.p.david_at_shef.ac.uk
2
Talk Outline

• Introduction
• Current Mid-Infrared Detector technology
• The benefits of Quantum dot Infrared
  photodetectors
• Device design
• Dots-in-a-well (DWELL) based device design
• Results
• Two colour operation Dot size and well width
  dependence
• Intraband Stark shift - Theoretical calculations
  and experimental results
• Well-within-a-well model
• QDIP Performance - Dark current, Responsivity,
  Detectivity

3
Introduction

• Current Mid-Infrared Detector
  technology
• Quantum Well Infrared Photodetectors (QWIPs)
• Cannot detect at normal incidence (Intraband
  selection rules)
• Low temperature operation
• Short excited state lifetimes (1ps)
• CdHgTe (CMT)
• Low temperature operation to avoid Auger
  recombination
• Fabrication is complicated and expensive

4
QDIP Advantages

• Intraband absorption allows operation in the
  mid-IR region, making mid-IR detectors suitable
  for applications requiring day and night
  operation or detection of camouflaged objects

Quantum Dot Infrared Photodetectors (QDIPs) have
a number of intrinsic advantages

• Normal incidence operation
• Low dark current
• Long excited state lifetimes (reduced
  electron-phonon scattering-phonon bottleneck
  effect)
• Bias tunable operation

5
Device Fabrication design

• Dots-in-a-well structure (DWELL)

• InAs dots grown by MBE, using Stranski-Krastanow
  growth in an InxGa1-xAs well with GaAs spacer
  layer
• Active region consists of 5 periods of InAs dots
  in an InxGa1-xAs well (x0.15) capped by n GaAs
  barrier layers
• 2 sets of samples with 2.2, 2.55, 2.9 monolayers
  (ML) of dots with doping densities corresponding
  to 1 and 2 electrons per dot
• 3d set of samples with varying well width (10-70,
  10-90, 10-120 (x0.1))

Cross-sectional view
Conduction Band diagram for active region
6
Two-Colour Operation

• Two-colour behaviour at 130meV (9µm) and
  230meV (5 µm) corresponding to transitions from
  the QD to QW ground state and QD ground state to
  continuum respectively.
• Transition energy increases with number of
  monolayers due to deeper confinement for larger
  dots
• ML dependence enables the pre-growth wavelength
  tunability via dot design
• Also possible to tune the spectral range by
  varying the QW width

??/?20
1e- per dot
7
Well Width Variation

• The E1?EQW transition energy decreases with
  increasing well width
• Smaller effect than varying the number of ML
• From previous growth 90A is maximum width before
  strain becomes too large - introduction of
  dislocations for x0.15
• Decrease of In concentration to reduce
  strain-Increase of E1?EQW transition energy
• 3 potential variables for pre-growth Wavelength
  tuning

8
Bias Dependence - Intraband Stark shift

• The E1?EQW transition is not observed at zero
  bias due to low probability of carriers
  tunnelling from QW state
• Furthermore, the E1?EQW transition is bias
  tuneable due to the intraband Stark effect
• Photocurrent spectrum at 1V is red-shifted with
  respect to that at -1V, indicating an asymmetric
  dependence on applied bias due to off-centre
  position of the QD layer in the well

9
Well within a well model
positive
negative

• Well within a well approximation for DWELL
• off-centre position of the QD layer in the well
  giving rise to a smaller red-shift for reverse
  bias compared with forward bias - relative
  position of the centroid of the QD ground state
  and the QW ground state wavefunctions

10
Bias Dependence
Bias Dependence - Intraband Stark shift

• Theoretical prediction indicates larger dots
  less asymmetry -smaller shift
• Opposite observed experimentally
• The effect that the increase of number of ML has
  on the shape, size and composition of the dot
  cannot be determined by spectral measurements-or
  calculated with simple model used

• However if the well-within a well simulation
  assumes the dots are becoming smaller the
  experimental data behaviour is fitted
• Therefore the model assumes that the DWELL
  asymmetry increases for larger number of ML

11
Bias Dependence
Bias Dependence - Intraband Stark shift

• The agreement of this model with experimental
  data indicates that the electron wavefunction
  becomes more localised at the base of dot for
  increasing number of ML

12
Dark current

• Dark current decreases with increasing number of
  MLs
• After a certain number of ML, it appears that
  strain introduced in the DWELL results in
  structural defects
• 2.55MLs optimum (red line)

13
Responsivity

• Responsivity increases with number of ML and was
  found to be in the order of
• 1A/W _at_ 1V _at_10K
• for all the samples
• No significant reduction up to 77K

?8-9?m
?10-11?m
14
Detectivity

• The Id limited Detectivity corresponding to the
  responsivity values shown previously were
  estimated to range between the order of
• 1010-1012 cmHz1/2W-1 _at_ 10K (2.55ML optimum)
• D at 77K of the order of 1010 cmHz1/2W-1

15
Summary

• Demonstrated the two colour operation of DWELL
  QDIPs
• Pre-growth wavelength tunability and Dark
  current control via
• Control over the number of MLs
• The variation of the well width
• Post-growth applied bias voltage tuneable
  wavelength response
• Provided detailed information regarding the
  conduction band structure of the DWELLs, using a
  well-within a-well model to simulate the bias
  dependence of the transitions
• Reported peak responsivities of 1A/W at 10K and
  detectivities of 1012cmHz1/2W-1_at_10K and
  1010cmHz1/2W-1_at_77K