1. T-shaped quantum-wire laser
2004 Fall MRS meeting in Boston ( B3.1)
T-shaped quantum-wire laser
M. Yoshita, Y. Hayamizu, Y. Takahashi, H. Itoh, T. Ihara,
and H. Akiyama
Institute for Solid State Physics, Univ. of Tokyo and CREST, JST
L. N. Pfeiffer, K. W. West, and Ibo Matthews
Bell Laboratories, Lucent Technologies
1. Formation of high-quality GaAs T-shaped quantum wires
Cleaved-edge overgrowth with MBE, AFM, PL, PLE
2. Single-wire laser
PL scan, Lasing, PL, Absorption/Gain via Cassidy’s method, Transmission
3. Observation of RT 1D exciton absorption in 20-wire laser
4. Optical response of n-doped single-wire FET device

2. T-shaped Quantum Wire (T-wire)
T-shaped QWR, or T-wire, that I talk today, has a crosssectional structure shown here.
A QWR is formed at a T-intersection of two QWs, a stem QW and an arm QW with thickness of 14nm and 6nm.
Since quantum confinement is weaker on a low barrier than on a high barriers, 1-D state is formed quantum mechanically at the intersection.
In fact the red and blue contour curves show electron and hole wavefunctions in a T-wire exciton.

3. Cleave in situ 600oC 490oC (490oC)
Cleaved-edge overgrowth with MBE
by L. N. Pfeiffer et al., APL 56, 1679 (1990).
[110]
[001]
GaAs
substrate
T-wires are formed by cleaved-edge overgrowth method with MBE developed by Loren Pfeiffer in Bell Labs.
We first make a stem quantum well by conventional MBE growth on a (001) substrate at 600C.The wafer is once taken out from the MBE machine, thinned, cut into pieces, partly scribed, and then reloaded into the machine. After setting growth condition, we cleave the sample, and grow an arm well right away.
There are two major difficulties. The first is in-situ cleavage at growth temperature.
The second difficulty is the arm well growth. If you cleave you always have (110) surface, and the MBE growth on (110) surface is difficult. You need low temperature of about 500C, and high As pressure for (110) MBE.
(001) MBE Growth
Cleave
in situ
(110) MBE Growth
600oC
490oC
(490oC)

4. Nomarski Microscope Images of Cleaved-Edge- Good Overgrowth Surfaces
Poor
Bad
Nomarski
Microscope
Images of
Cleaved-Edge-
Overgrowth
Surfaces
“Hackling”

5. High Quality T-wire ??? 490C Growth Atomically flat interfaces
With this annealing technique, we are now able to fabricate very high quality T-wires.
All previous T-wires had a rough top interface on an arm well.
But we now have an atomically flat top interface by annealing, and also an flat bottom interface by cleavage, and remaining interfaces can be optimized in the conventional MBE growth.
Before we apply this technique to real T-wires, I have a comment.
If the amount of deposited film thickness is exactly integer monolayers, we get this flat surface.
But if we supply it a little too much or too less, we should not obtain this flat surface.
Atomically flat
interfaces
(By Yoshita et al. JJAP 2001)

6. High Quality T-wire !? 490C Growth 510-600C Anneal Atomically flat
With this annealing technique, we are now able to fabricate very high quality T-wires.
All previous T-wires had a rough top interface on an arm well.
But we now have an atomically flat top interface by annealing, and also an flat bottom interface by cleavage, and remaining interfaces can be optimized in the conventional MBE growth.
Before we apply this technique to real T-wires, I have a comment.
If the amount of deposited film thickness is exactly integer monolayers, we get this flat surface.
But if we supply it a little too much or too less, we should not obtain this flat surface.
Atomically flat
interfaces
(By Yoshita et al. JJAP 2001)

7. High Quality T-wire !!? 490C Growth 510-600C Anneal Atomically flat
With this annealing technique, we are now able to fabricate very high quality T-wires.
All previous T-wires had a rough top interface on an arm well.
But we now have an atomically flat top interface by annealing, and also an flat bottom interface by cleavage, and remaining interfaces can be optimized in the conventional MBE growth.
Before we apply this technique to real T-wires, I have a comment.
If the amount of deposited film thickness is exactly integer monolayers, we get this flat surface.
But if we supply it a little too much or too less, we should not obtain this flat surface.
Atomically flat
interfaces
(By Yoshita et al. JJAP 2001)

8. High Quality T-wire !!!!! 490C Growth 510-600C Anneal Atomically flat
With this annealing technique, we are now able to fabricate very high quality T-wires.
All previous T-wires had a rough top interface on an arm well.
But we now have an atomically flat top interface by annealing, and also an flat bottom interface by cleavage, and remaining interfaces can be optimized in the conventional MBE growth.
Before we apply this technique to real T-wires, I have a comment.
If the amount of deposited film thickness is exactly integer monolayers, we get this flat surface.
But if we supply it a little too much or too less, we should not obtain this flat surface.
Atomically flat
interfaces
(By Yoshita et al. JJAP 2001)

9. 20-wire laser sample 14nm x 6nm 1st growth 600C 2nd growth
490C (arm well)
600C 10min anneal
490C (cover layers)
This shows the schematic structure of our 20-wire laser.
20 wires with the size of 14nm by 6nm are formed in T-shaped optical waveguide.
After the growth of the arm well, we introduced the growth interrupt anneal at 600 degree C for 10 minutes.
We obtained 500micron-long laser bars by cleavage. The cleaved facet mirrors are left uncoated.
We cooled down a laser bar to 5K and measured PL from the top with a microscope or measured stimulated emission coming out from a cavity mirror facet.
Laser bars
500mm uncoated cavity

10. PL and PLE
spectra
Sharp PL width
Small Stokes shift
1D free exciton
1D continuum states
arm
well
T-wire
stem
well
(Akiyama et al. APL 2003)

11. E-field
E-field
// to wire
_ to wire
// to arm well I
点励起。

12. Single quantum wire laser
Probability of Photon
Cavity length　500 mm
G=5x10-4
In this work, we used the single quantum wire laser.
This is the schematic view.
Probability of Electron

13. Scanning micro-PL spectra
T-wire
stem well
T-wire
stem well
T=5K
ContinuousPL peak over 20 mm
PL width < 1.3 meV
We characterized our sample by scanning micro-PL spectra for 1 micro m spot along the quantum wire.
These were measured by 10 micro m steps for 500 micro m region.
And for this 25 micro m region we made scan by finer 0.5 micro m steps.
The high energy peaks are from stem well and the low energy peaks are from a quantum wire.
(Arm well peak was not observed at this low pumping level.)
These tiny low energy peaks of wire are attributed to localized excitons caused by monolayer fluctuation of the arm well thickness.
In this region, there is no tiny peak of localized excitons, and this shows that the region without monolayer fluctuation extends over 20 micro m.
This results prove the high quality of this quantum wire.

14. Lasing in a single quantum wire
500mm
gold-coated
cavity
Threshold 5mW
(Hayamizu et al, APL 2002)

16. Excitation power dependence of PL
M. Yoshita, et al.

17. Density Electron-hole Plasma Biexciton+Exciton Free Exciton
n1D = 1.2 x 106 cm-1
(rs = 0.65 aB)
aB ~13nm
Electron-hole Plasma
Density
n1D = 1.2 x 105 cm-1
(rs = 6.6 aB)
Biexciton+Exciton
EB =2.8meV
n1D = 3.6 x 103 cm-1
(rs = 220 aB)
Free Exciton
n1D ~ 102 cm-1

18. Absorption/gain measurement based on Hakki-Paoli-Cassidy’s analysis of Fabry-Perot-laser emission below threshold
：Absorption coeff.
The waveguide emission spectra are represented by this equation.
In the Cassidy’s method, we use the total intensity of the longitudinal modes peak “I sum” and minimal intensity “I min”.
By using the ratio of I sum and I min, p, the absorption coefficients of each longitudinal mode are derived analytically by this equation.
：Reflectivity
D. T. Cassidy JAP (1984)
Free Spectral Range
B. W. Hakki and T. L. Paoli JAP (1974)

19. Absorption Spectrum by Cassidy method
Single wire laser, uncoated cavity mirrors
Excitation Light ：cw TiS laser at 1.631eV
Spectrometer with spectral resolution of meV
Cassidy’s Method
Point
Waveguide Emission
By using the Cassidy’s method, we can derive the absorption spectrum from the waveguide emission spectrum.
This absorption spectrum shows the several peaks: exciton ground state, 1st excited state and continuum state.
These tiny peaks are attributed to the localized excitons caused by monolayer terrace of the arm well.
This spectra shows good agreement with the result of photo luminescence excitation spectrum.
Polarization parallel to Arm well

20. Measurement of absorption/gain spectrum
Excitation Light ：cw TiS laser at 1.631eV
Spontaneous emission
Spectrometer with spectral resolution of meV
Stripe shape
Cassidy’s Method
Waveguide Emission
This blue curve is the derived absorption spectrum of the single quantum wire laser by using Cassidy’s method at the high excitation power.
In this figure, the absorption is positive, the gain is negative.
The absorption spectrum has a continuous absorption band and a symmetric gain peak.
The energy of the gain peak corresponds to the shoulder of the broad PL peak, and also the lasing peak.
This absorption spectrum could be explain by the two band model.
The gain is yielded by the population inversion of the electron-hole plasma. We can attribute here to the band edge energy,
and here to the Fermi edge energy.
To investigate the gain peak in more detail, we measured the 20 wires laser. Each wire has same size as previous single quantum wire.
8.3mW
Polarization parallel to Arm well

21. Absorption/gain spectrum (High excitation power)
Electron-Hole Plasma
EFE
EBE
Gain
Absorption
This blue curve is the derived absorption spectrum of the single quantum wire laser by using Cassidy’s method at the high excitation power.
In this figure, the absorption is positive, the gain is negative.
The absorption spectrum has a continuous absorption band and a symmetric gain peak.
The energy of the gain peak corresponds to the shoulder of the broad PL peak, and also the lasing peak.
This absorption spectrum could be explain by the two band model.
The gain is yielded by the population inversion of the electron-hole plasma. We can attribute here to the band edge energy,
and here to the Fermi edge energy.
To investigate the gain peak in more detail, we measured the 20 wires laser. Each wire has same size as previous single quantum wire.
8.3mW
Hayamizu et al.

22. Electron-Hole Plasma Exciton
Exciton peak and continuum onset decay without shift.
Gap between exciton and continuum is gradually filled.
Exciton changes to Fermi edge
Exciton
Hayamizu et al.

23. Transmission measurement of a single quantum wire
~mm
~nm

24. Transmittance for single
Coupling
efficiency
= 20%
Takahashi et al. unpublished

25. Takahashi et al. unpublished
Absorption for single
Takahashi et al. unpublished

26. Absorption for 20
Y. Takahashi et al.

27. Room-Temperature 1D Exciton Absorption!
Absorption at 300 K
Room-Temperature 1D Exciton Absorption!
Y. Takahashi et al.

28. 14nmx6nm Doped Single Wire FET device with tunable 1D electron density
T. Ihara et al.

29. Quantum-Wire Devices

30. Summary
GaAs T-shaped quantum wires (T-wires) are formed by cleaved-edge overgrowth with MBE.
Growth-interrupt anneals dramatically improve T-wire quality.
AFM : No atomic steps over 100mm.
PL : Sharp PL width (~1meV) improved by a factor of 10.
PLE : Observation of 1D free exciton, & 1D continuum states
Single wire lasing : The world thinnest laser (14nm x 6nm),
5mW threshold optical pumping power at 5K.
Gain/absorption measurement by Hakki-Paoli-Cassidy’s method.
Strong photo-absorption by a single wire
84/cm (98.5% absorption / 500mm) at exciton peak at 5K
Room-temperature exciton absorption observed in 20-wire laser.
Single-wire FET: carrier-sensitive optical responses.

31. ここまで。25分のトーク。

32. (001) and (110) surfaces of GaAs
[001]
[110]
[110]
[001]

33. Growth rate of GaAs in MBE holder Substrate As Ga Substrate rotation
>>> uniform
holder
Substrate
Ga limited growth under
As4　overpressure As Ga

34. Interface control by growth-interruption annealing
490oC Growth
(by M. Yoshita et al.
JJAP 2001)
Atomically flat
interfaces
600oC Anneal
High Quality
T-wire
arm
well
6nm
stem
well
14nm

35. (By Yoshita et al. APL 2002)

36. Single wire laser with 500mm gold-coated cavity

37. Absorption at higher temperatures by Cassidy
前の図との共通点を述べる。
Hayamizu et al. unpublished

38. Absorption coefficients

39. Experiment for gain

40. Evolution of continuum
Takahashi et al. unpublished

41. Lasing & many-body effects in quantum wires
E. Kapon et al. (PRL’89) Lasing in excited-states of V-wires
W. Wegscheider et al. Lasing in the ground-state of T-wires, no energy shift,
(PRL’93) excitonic lasing
R. Ambigapathy et al. PL without BGR, strong excitonic effect in V-wires
(PRL’97)
L. Sirigu et al. (PRB’00) Lasing due to localized excitons in V-wires
J. Rubio et al. (SSC’01) Lasing observed with e–h plasma emission in T-wires
A. Crottini et al. (SSC’02) PL from exciton molecules (bi-excitons) in V-wires
T. Guillet et al. (PRB’03) PL, Mott transition form excitons to a plasma in V-wires
H. Akiyama et al. Lasing due to e–h plasma, no exciton lasing in T-wires
(PRB’03)
F. Rossi and E. Molinari
(PRL’96)
F. Tassone, C. Piermarocchi, et al.
(PRL’99,SSC’99)
S. Das Sarma and D. W. Wang
(PRL’00,PRB’01)
Theories
“1D exciton Mott transition”

42. Physical picture of 1D exciton–plasma transition
the exciton Mott transition
Our PL results show
Increase of e–h pair density causes
no energy shift of the exciton
band edge
・ reduction of exciton binding energy
・ red shift of the band edge
(band-gap renormalization (BGR))
plasma low-energy edges appear
at the bi-exciton energy positions,
and show BGR
no connection, but coexistence
of two band edges
eg. D. W. Wang and S. Das Sarma,
PRB 64, (2001).
band edge
most recent theory
this evolution is treated by the Mott transiion
no level-crossing between the band edges and the exciton level
exciton level

43. Exciton band edge & plasma band edge (T=30K)
(low energy edge of plasma PL)
starts at biexciton energy
and shows red shift.
▼ exciton band edge,
(onset of continuum states)
exciton ground and excited states
show no shift.

44. 75

45. Electron plasma
and minority hole e e e e h e e e e e e h e
X- Charged Exciton
X Exciton e h

46. 1D exciton and continuum states
Theory
1D exciton and continuum states