P44
ECS Solid State Letters, 2 (5) P44-P46 (2013)
2162-8742/2013/2(5)/P44/3/$31.00 © The Electrochemical Society
Precise Formation of Dovetail Structures for InP-Based Devices
Iman Hassani Nia and Hooman Mohseniz
Bio-Inspired Sensors and Optoelectronics Laboratory (BISOL), EECS, Northwestern University,
Evanston, Illinois, USA
Anisotropic etching of InP along specific crystallographic directions leads to negative sidewall angles along dovetail direction. This
is an important process for self-alignment of electrical contacts where sub-micron alignment is needed. However, existing etching
methods are only suitable for shallow etching, and for stress-free layers. Here we demonstrate a new etching method that is capable
of producing dovetail patterns with 10 times larger etch depth, and where interface stress exists. We believe this new etching method
is useful for many electronic and optoelectronic devices that require precise negative angle sidewalls in their fabrication scheme.
© 2013 The Electrochemical Society. [DOI: 10.1149/2.007305ssl] All rights reserved.
Manuscript submitted January 25, 2013; revised manuscript received February 21, 2013. Published March 8, 2013.
The vast majority of electronic devices require complicated fabrication processes. In recent years, the introduction of many new
nano-devices has motivated the development of robust self-alignment
methods, especially when a high throughput is required. Etching is
one of the most widely used and important steps in fabrication; it
has been investigated extensively during the last few decades for
semiconductors1–3 and it has been shown that well-controlled etching
recipes can lead to very novel structures such as air bridges, suspended
structures, and micro and nano electromechanical (MEMs and NEMs)
devices. Furthermore, reliable etching increases the yield of fabrication and reduces the cost of final products. Precise etching of InP
is particularly important, because the epitaxial growth of other III-V
compounds lattice matched to InP wafers is used in many high-speed
electronic and optoelectronic devices operating at the telecommunication wavelength.4
In our previous work, we demonstrated for the first time, the operation of an electrically tunable nano-wire structure that was developed
using negative angle sidewalls for the self-alignment of gate metal.5
The lateral transport of electrons in this device was confined by gate
contacts that require nanoscale spacing. The fabrication of closely
spaced gate metals without shorting the top contact metal was impossible without the self-alignment property of the negative sidewall of
the ridge. However, all existing wet etching methods fail at producing a deeper etch in quantum well structures with significant lattice
mismatch.6
In this letter we describe a new wet etching method that is capable
of achieving the above goal. Systematic change of the composition and
temperature of the previously reported solutions, as described in the
next section, led to slight improvements but not satisfactory results.
For this reason we changed our strategy and tried to implement the
preferential protection of polar crystallographic plane (111A) using a
non-ionic surfactant.
Experimental
In this section the materials and methods of etching are explained.
The multiple quantum well (MQW) region consists of five periods.
Etch period contains the following layers in order: 17 nm of InAlAs,
18 nm of Q1.43 InGaAsP, 1 nm of InAlAs, 3 nm of InGaAs and
17 nm of InP. The layers were grown on (100) oriented n-doped InP
substrate using MOVCD. All of the grown layers are nominally lattice
matched to the InP substrate. The MQW region is capped with 1.5 μm
of p-doped InP. X-ray and PL data confirmed the excellent crystalline
quality of the grown epitaxial layers, but with some small strains.
A thin layer of low-stress silicon nitride was deposited on top
of the p-doped InP layer and was patterned using the photoresist AZ
5214. Then the silicon nitride was etched using CF4 based reactive ion
etching. The photoresist was removed subsequently by acetone and
ethanol and opened windows were used for etching the semiconductor.
After each wet etching, the samples were cleaved perpendicular to
z
E-mail: [email protected]
dovetail direction in order to reveal the cross section as shown in
Fig. 1a.
The first solution was HCl:H3 PO4 :H2 0 with ratio 3:1:1 at
T = 0◦ C. This solution has been used before by Inamura et al.,7
and was shown to have accurate negative sidewall angle for shallow
etching (approximately 100 nm). The etch profile of this solution after
50 μm etching of InP along [100] direction is shown in Fig. 1b. Clearly
the quantum wells have been attacked from the side walls along [11̄0]
direction with a [11̄0]/[100] etch ratio of about 0.1. Etching with HCl
at elevated temperatures resulted in a more unfavorable etch, with an
even higher etch ratio (see Fig. 1c).
Next we suspected that the strained InAlAs layers in the MQW are
responsible for an HCl based etchant to penetrate into the quantum
well and then subsequently etch the rest of the layers. To test this, an
etching solution which is extremely selective to InP as compared to
the InAlAs was used. The solution consists of HCl:H3 PO4 :CH3 COOH
with the ratio of 1:1:2 and was previously introduced by He et al.8
This solution has a selectivity of more than 85 for InP versus InAlAs.
As shown in Fig. 1d, thin layers of InP in the MQW region show
substantially higher etch rate compared with the thick InP cladding
layers, most probably due to a small epitaxial strain within the MQW
layers.
Finally, we changed our strategy and used a surfactant to protect
the InP polar sidewalls preferentially. We added 1% non ionic Triton
X-100 surfactant by volume to HCl:H3 PO4 :H2 O 3:1:1 at zeros degrees. This surfactant has been previously used to improve the surface
roughness of silicon etching with KOH-IPA.9 It also prevents the formation of hydrogen bubbles for the etching of silicon10 and increases
the anisotropic etching of silicon by TMAH solution.9 However to the
best of our knowledge it has never been used for wet etching of InP.
The solution was stirred every 20 minutes manually for 5 minutes.
Real time monitoring of the surface of the semiconductor by an optical
microscope showed that the use of the surfactant and the stirring helped
to prevent the formation of hydrogen bubbles on the surface. The final
result is shown in Fig. 1e.
The anisotropic etching of InP originates from the different etching rate for different crystallographic directions. It has been proposed
that different crystallographic planes of InP have different chemical
reactivity.11 Without the presence of strain the etching continues until the acid molecules reach the slow etching plain (111A).12 (111A)
contains Indium atoms, each of which have three bonds to phosphate
atoms on the underlying (111B) plane as shown in Fig. 2. The Indium
atoms on this plane don’t have any free electrons and therefore
they are not chemically active. As the etching goes deeper into the
material along [100] direction the negative sidewall angle is formed.
However the presence of strain in the material promotes the chemical
reaction between the acid and the material in the (111A) plane and
facilitates the etching and penetration of the acid laterally into MQWs
as depicted in Fig. 1. One hypothesis for the significant improvement
achieved by adding surfactant is that Triton X-100 molecules protect the atomic planes of the InP lattice perpendicular to the [111]
direction, as depicted in Fig. 2. Triton X-100 molecules have both
hydrophilic and hydrophobic ends. They are non-ionic and are not
ECS Solid State Letters, 2 (5) P44-P46 (2013)
P45
Figure 1. Side wall etching profile after 50 μm etching of InP substrate toward [100] direction for (a) HCl:H3 PO4 :H2 O 3:1:1 at zero degrees (b) only HCL at
room temperature (c) HCl:H3 PO4 :CH3 COOH at 0◦ C and (d) HCl:H3 PO4 :H2 O 3:1:1 mixed with 1% vol of Triton X-100. The depths of the etching along the
lateral direction [11̄0] was also measured and indicated. The layer compositions are as follows 1) p-doped InP 2) Intrinsic MQW 3) n-doped InP 4) InGaAs 5) InP
substrate.
Figure 2. Illustration of the adsorption of Triton X-100 molecules to a polar
(111A) plane because of electrostatic attraction. The indium atoms have partial
positive charges as opposed to phosphate atoms.
ionized in the acidic solution; the hydrophilic side of the molecule
is polar and is attracted to the polar InP planes as shown in Fig. 2.
Therefore these molecules can act as a protection layer on a (111A)
plane, and slow down the lateral penetration of the acid into the material. This is also true for {100} planes of the crystal because they are
also polar, however the distance of a pair of (111A) and (111B) planes
near the surface to the underlying pair is larger than the distance of
{100} planes which are spaced equally. Therefore the screening of
the surface electrostatic field after removal of a (111B) plane by the
underlying planes will be much lower than that of a (100) plane. As
a result the surface density of adsorbed polar Triton X-100 molecules
will be higher near a (111A) plane as compared to a (100) plane. This
explains the lower lateral to vertical etching ratio after addition of
Triton X-100.
Table I includes etching characteristics of the all of the solutions
and shows that the addition of the Triton X-100 surfactant suppresses
the strain meditated etching significantly. By carefully looking at the
trend of the [11̄0]/[100] etch ratio in table 1 one can notice that the
acetic acid solution is not the best choice however it improves the
etch ratio. It has been shown that carboxylic acids can act a surface
selective surfactant and can alter the anisotropic chemical reactivity13
and growth of TiO2 nanocrystals.14 Acetic acid molecules have a polar
hydroxyl bond similar to Triton X-100 molecules and therefore we
think that they can also preferentially protect polar crystallographic
planes.
For the special solution under consideration the etch rate versus
time for different Triton X-100 percentage volumes has been measured and plotted in Fig. 3. As time goes on, the etching rate reduces
significantly for the solution with Triton X-100. This suggests that the
diffusion-limited deposition of Triton X-100 might be responsible for
the reduction of the etch rate. For a solid liquid interface the following
formula can be used to calculate the kinetics of the diffusion process
for nonionic surfactants:15
kt ka,θ (1 − θ)
m dθ
=
(cb − ce,θ )
[1]
dt
ka,θ (1 − θ) + kt
In the above equation m is the maximum surface density of surfactant molecules on the solid interface, θ is the normalized density of
surfactant on the solid surface with respect to m , kt is the transport
coefficient, ka,θ is the coefficent of adsorption and cb is the bulk density
of the surfactant molecules, ce,θ is calculated as follows:
ce,θ =
kd,θ θ
ka,θ 1 − θ
[2]
Table I. Etching properties of different solutions used in this article.
Etching solution
HCl:H3PO4:H2O 3:1:1
HCl
HCl:H3PO4:H2O 3:1:1
HCl:H3PO4:CH3COOH 1:1:2
HCl:H3PO4:H2O 3:1:1 1% vol Triton X-100
Temperature
(C)
Duration
(hours)
Etching
depth (μm)
Average etch rate
of [100] during the
etching (nm/sec)
[11̄0]/[100]
etch ratio
0◦
20◦
20◦
0◦
0◦
3
0.33
0.5
5
3.5
∼50
∼50
∼50
∼50
∼50
4.6
41.55
30
2.7
3.2
∼0.1
∼0.1
∼0.1
∼0.05
∼0.01
P46
ECS Solid State Letters, 2 (5) P44-P46 (2013)
Conclusions
7
measured 0.5 % vol Triton X
6.5
measured 1 % vol Triton X
theoretical 0.5 % vol Triton X
Etch rate (nm/sec)
6
theoretical for 1% vol Triton X
5.5
5
4.5
4
3.5
3
10
20
30
40
50
60
Time(min)
70
80
90
100
Figure 3. The measured etch rate for 0.5 and 1 volumetric percentage of
Triton X-100 in the solution at zero degrees.
The effect of different etching solutions on the properties of InP,
InGaAs and InAlAs has been explored. It is concluded that adding a
non-ionic surfactant such as Triton X-100 reduces the unwanted side
etching caused by the presence of strain in the epitaxial layers. This
is an important step in the fabrication of InP based devices where
the negative sidewall angle is crucial. We believe that the significant
improvement, demonstrated by SEM images, is due to polarity of
the Triton X-100 molecules that increases their adhesion to the polar
(111A) plane, and leads to formation of a protective layer. Although
similar phenomenon applies to all crystallographic planes of InP, the
relatively higher surface electrostatic field of {111A} planes leads
to a significant enhancement in the (100)/(111A) etching ratio. The
etching rate versus time and volume of Triton X-100 was measured
and was shown to decrease over time because of the diffusion limited
adsorption of Triton X-100. Modeling results show a similar trend
compared to the measured etch rate data.
References
Where kd,θ is the coeffiecnt of desorption. For the diffusion limited
process i.e kt ka,θ , Eq. 1 is simplified to :
m dθ
= ka,θ (1 − θ) (cb − ce,θ )
[3]
dt
The above formula predicts an exponetial reduction in the absorption rate of the surfactant molecules to the InP surface. As a first
order approximation we assume that the etching rate linearly decreses with the increase in coverage of Triton X-100 molecules on the
surface, therfore the etching rate decays exponentionaly over time
because of the deposition of Triton X-100 molecules. The theoretical
curves follows a similar trend to the measured data using ka = 1.25
× 10−9 cm/s kd = 1.34 × 10−12 mole/cm2 s and m = 2.9
× 10−9 mole/cm2 for two different molar density Triton X-100 as
indicated by Fig. 3. The value of the maximum surface density, m ,
has been chosed from Ref. 16. The values of kd and ka are within the
acceptable range when compared to reported values of different nonionic surfactants.15,16 After 1.5 hours, θ increases from 0 to 0.43 for
the solution with 1% vol of Triton X-100 and to 0.33 for the solution
with 0.5% vol of Triton X-100.
1. P. Mounaix, P. Delobelle, X. Mélique, L. Bornier, and D. Lippens, Materials Science
and Engineering: B, 51, 258 (1998).
2. A. R. Clawson, Materials Science and Engineering: R: Reports, 31, 1 (2001).
3. D. J. Stirland and B. W. Straughan, Thin Solid Films, 31, 139 (1976).
4. R. Schwertberger, D. Gold, J. P. Reithmaier, and A. Forchel, Journal of Crystal
Growth, 251, 248 (2003).
5. W. Wu, I. Hassani, and H. Mohseni, ACS nano, 5, 7488 (2011).
6. E. Fitzgerald, Journal of Vacuum Science & Technology B: Microelectronics and
Nanometer Structures, 7, 782 (1989).
7. E. Inamura, Y. Miyamoto, S. Tamura, T. Takasugi, and K. Furuya, Japanese Journal
of Applied Physics, 28, 2193 (1989).
8. Y. He, B. Liang, N. Tien, and C. Tu, Journal of The Electrochemical Society, 139,
2046 (1992).
9. D. Resnik, D. Vrtacnik, U. Aljancic, M. Mozek, and S. Amon, Journal of Micromechanics and Microengineering, 15, 1174 (2005).
10. C. Ashruf, P. French, P. Bressers, and J. Kelly, Sensors and Actuators A: Physical,
74, 118 (1999).
11. P. Eliáš, I. Kostič, J. Šoltýs, and S. Hasenöhrl, Journal of Micromechanics and
Microengineering, 14, 1205 (2004).
12. S. Phatak and G. Kelner, Journal of The Electrochemical Society, 126, 287 (1979).
13. R. Tanner, Y. Liang, and E. Altman, Surface science, 506, 251 (2002).
14. Y. Jun, M. F. Casula, J. H. Sim, S. Y. Kim, J. Cheon, and A. P. Alivisatos, Journal of
the American Chemical Society, 125, 15981 (2003).
15. L. H. Torn, L. K. Koopal, A. de Keizer, and J. Lyklema, Langmuir, 21, 7768 (2005).
16. C. H. Chang and E. I. Franses, Colloids and Surfaces A: Physicochemical and Engineering Aspects, 100, 1 (1995).