Electrochemical and Solid-State Letters, 10 共4兲 D35-D37 共2007兲
D35
1099-0062/2007/10共4兲/D35/3/$20.00 © The Electrochemical Society
Pore Formation on n-InP(100) in Acidic Liquid Ammonia
at 223 K
A True Water-Free Etching Process
A.-M. Gonçalves,z L. Santinacci,* A. Eb, I. Gerard,
C. Mathieu, and A. Etcheberry*
Lavoisier Institute-CNRS (UMR-8180), University of Versailles, 78035 Versailles cedex, France
For the first time, pore formation on n-InP共100兲 has been carried out by galvanostatic treatments in acidic liquid ammonia at
223 K. Voltage oscillations correlated to a specific current line oriented pore morphology have been evidenced by scanning
electron microscopy. Whatever the anodic charge, a constant pore depth was formed 共2–3 ␮m兲. Porous layers have been characterized by ex situ photoluminescence measurements that have revealed a dead layer behavior. This work demonstrates the crucial
role of interfacial phenomena illustrated by the use of this uncommon nonaqueous electrolyte.
© 2007 The Electrochemical Society. 关DOI: 10.1149/1.2434201兴 All rights reserved.
Manuscript submitted October 12, 2006; revised manuscript received November 21, 2006.
Available electronically January 24, 2007.
Due to their various morphologies, porous structures offer a wide
range of properties and applications 共see, for example, Ref. 1兲.
Shifted photoluminescence 共PL兲 responses have been observed2,3
according to the dissolution conditions and such phenomena are
essentially determined by the pore morphology.4,5 Wet porosification
of semiconductors depends on both nature, orientation, and doping
level as well as on the electrochemical parameters. The pore formation is an interfacial phenomenon controlled by the solubility and
the evacuation rate of the dissolution products. It is directly based on
physical 共surface tension, polarity, temperature, viscosity, and dielectric constant兲 and chemical properties of the electrolyte6 共pH,
protic nature兲. In aqueous media, only restricted ranges of pH can be
explored according to the oxide solubility. In the case of Si, the use
of HF mixing with usual organic solvents does not fully modify the
electrochemical conditions because some residual water from HF is
always present and active in those electrolytes.7 Although liquid
ammonia has still not been used for porous etching of semiconductors, specific cathodic processes have been previously evidenced
onto InP under true water free condition.8,9 In aqueous media, most
models10-12 describe the pore growth phenomenon according to a
two step mechanism—the anodic formation of an oxide layer, followed by its chemical dissolution. Obviously, in liquid ammonia the
porosification process does not start with the oxide layer formation.
Another etching mechanism and pore morphologies can be expected. Additionally, compared to water, liquid ammonia exhibits
low dielectric and viscosity constants 共four times lower兲, low polarity, and strong alkalinity 共1011 time higher兲.13 Under atmospheric
pressure, the use of liquid ammonia requires a low operating temperature 共223 K兲 inducing a decrease of kinetic constants. In this
paper, we report for the first time the galvanostatic porous etching of
n-InP共100兲 in acidic liquid ammonia. Particular electrochemical behaviors associated to singular pore morphologies are first described.
The structure of the porous layers is then characterized by scanning
electron microscopy 共SEM兲 and their optical properties are probed
by ex situ photoluminescence measurements.
n-InP共100兲 samples 共ND = 1018 cm−3兲 were chemomechanically
polished with Br2–methanol. Acidic liquid ammonia electrolyte was
obtained by condensation of gaseous ammonia 共electronic grade兲 at
atmospheric pressure and low temperature onto 1 M of anhydrous
NH4Br.14 The electrochemical setup was a classical three-electrode
cell. All potentials were measured vs a silver reference electrode
共SRE, E° = 0.79 V vs NHE兲. A 273 EG&G galvanostat was used to
apply, in the dark, constant anodic currents ranging from
2 to 50 mA cm−2 for various durations. After anodic treatments, the
ammonia was evaporated and 1 M H2SO4 was added to the residual
dissolved InP to perform quantitative chemical analyses by atomic
absorption spectroscopy.15 The morphology of the porous layers was
characterized by SEM using a JEOL JSM 5800 and the PL experiments were performed at room temperature using a He-Ne laser
共␭ = 633 nm兲 as excitation source.16
During galvanostatic treatments, specific oscillations of the interfacial potential are measured. Micro- 共Hz兲 and macro-oscillations
共mHz兲 features were observed in Fig. 1a and b. This oscillating
regime follows the characteristic and highly reproducible steep increase of the potential. The micro-oscillations could be ascribed to
the evacuation of dissolved materials out of the pores. After the
initial potential rising, the voltage oscillations follow a quasi-regular
periodic evolution. The oscillation frequencies have been measured
and averaged over more than three periods. Figure 1c shows therefore the linear increase of the macro-oscillation frequency with the
applied current. The slope of the straight line is equal to 1/Qp,a with
Qp defined as a coulometric charge contained in a single period
共Qp ⬇ 1.0 C cm−2兲. This is confirmed by the quasi-superimposition
of macro-oscillations when the potential is plotted against the anodic
charge 共Q兲 for different applied current 共Fig. 1a and b 兲. Depending
on the applied current density, it is noteworthy that the oscillation
amplitude is gradually damped until it reaches the average value of
the potential macro-oscillations 共Fig. 1a兲.
To determine the contribution of the semiconductor dissolution
in the whole measured current, the amount of dissolved InP 共nInP兲
has been measured by atomic adsorption for anodic treatments corresponding to dissolution charges ranging from 5 to 400 C cm−2.
The ratio Q/nInP is constant and equal to 6 F 共Faraday constant兲 over
the all range of current. The etching process occurs therefore according to the classical 6 holes dissolution mechanism 共Eq. 1兲17
InP + 6h+ → In+III + P+III
a
* Electrochemical Society Active Member.
z
E-mail: [email protected]
关1兴
It can be clearly deduced that no secondary reaction, such as
NH3 oxidation, occurs during the anodic process. Voltage macroand micro-oscillations are exclusively correlated to the semiconductor dissolution. The InP dissolution rate increases therefore directly
with the current density.
From every SEM characterizations, no matter the current density,
an unexpected constant morphology is observed with an estimated
porosity of 25%. A typical cross section of porous n-InP is shown on
Fig. 2. A tortuous current line oriented 共tCLO兲 pore morphology is
observed 共Fig. 2c兲. Because porous etching is performed onto n-type
Because f = 1/T = s ⫻ j then s = 1/共 j ⫻ T兲 = 1/Qp . f, T, s, j, and Qp, respectively, defined as the frequency, period, slope, current density, and periodic charge
of macro-oscillations.
Downloaded on 2016-05-18 to IP 130.203.136.75 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).
D36
Electrochemical and Solid-State Letters, 10 共4兲 D35-D37 共2007兲
Figure 1. 共Color online兲 Potential vs anodic charge of n-InP in acidic
liquid ammonia 共1 M NH4Br兲 at 223 K. 共a兲 j = 2 mA cm−2,
共b兲 j = 8 mA cm−2, 共c兲 inset corresponding to macro-oscillation frequency
plotted against the current density. Note that for Q = 1 C cm−2 a uniform
dissolution of 0.5 ␮m is calculated for a 6 holes mechanism.
material, the pore walls are expected to be in the range of the width
of the space charge layer 共WSCL ⬇ 100 nm兲.18 The observed wall
thickness of 50 to 100 nm is therefore in a good agreement. The
dissolution occurs at the pore tips inducing a pore growth perpendicular to the surface 共Fig. 2c兲. For Q ⬎ Qp, the applied current
density has no influence onto the morphology and the observed pore
depth 共dpore兲 remains constant 共dpore = 2 to 3 ␮m兲. The tCLO pores
are grown in hemispherical depressions with a diameter of ca.
Figure 2. Morphology of porous n-InP after galvanostatic treatments
in acidic liquid ammonia 共1 M NH4Br兲 at 223 K. 共a兲 SEM cross
section 共j = 2 mA cm−2, Q = 5 C cm−2兲. 共b兲 Focused view of the interface
between a detached porous layer and the semiconductor bulk.
共j = 4 mA cm−2, Q = 6 C cm−2兲. 共c兲 Details of the pore morphology
共j = 2 mA cm−2, Q = 3.6 C cm−2兲.
Figure 3. 共Color online兲 Normalized IPL /I0 共–䊏–兲 vs anodic charge compared to the corresponding macro-oscillation for j = 2 mA cm−2 共—兲. IPL /I0
共␭ = 920 nm兲 data were obtained from spectra performed between ␭ = 800
and 1000 nm.
10 ␮m. It appears clearly from SEM pictures 共Fig. 2a兲, that the
semiconductor losses its initial flatness leading to a strongly perturbed surface. Because no secondary reaction takes place, the film
thickness should be directly proportional to the applied anodic
charge. The constant pore depth is then unexpected. This invariable
thickness corresponds to the constant anodic charge 共Qp兲 observed
in each periodical macro-oscillation. It could be explained by the
synchronized layer lift off occurring on the entire surface. This simultaneous detachment is due to mechanical and electrical strengths
localized at the hemispherical depression boundaries. It is confirmed
by SEM observations 共see arrow on Fig. 2b兲. For each period, the
layer formation phase is followed by its lift off. These steps correspond, respectively, to the ascendant and descending parts of the
voltage peak 共Fig. 1a and b兲. As a consequence, the semiconductor
gets thinner. During the porosification a gradual increase of the disorder occurs. This induces the lost of the layer lift off synchronization that is correlated to the oscillation damping 共Fig. 1a兲. Such
porous morphology has never been observed in aqueous media
where the layer thickness is proportional to the anodic charge. It
confirms therefore the crucial contribution of liquid ammonia in the
interfacial processes.8,9
To probe the different stages of the porous layer growth, ex situ
photoluminescence was carried out. Only the intensity variation of
the band-to-band transition peak of InP 共␭ = 920 nm兲 was observed
as radiative recombination. Measurements were performed successively on different samples after several time of porous etching for
constant current density. Stable and reproducible results were obtained on a bunch of samples for various currents and durations.
Similar results were also observed with a single sample after the
repetition of immersion-treatment-emersion sequences. Figure 3
presents therefore the normalized photoluminescence intensity
共IPL /I0, where I0 is the IPL level of the initial surface兲 and the voltage
macro-oscillations plotted against the anodic charge.
Similar to the macro-oscillations of the interfacial potential, the
IPL /I0 evolution was again identical when it was plotted against the
anodic charge whatever the applied current. Three different zones
were clearly distinguished. The first 共I兲 corresponds to a constant
IPL /I0 plateau 共0 ⬍ Q ⬍ 0.1 C cm−2兲. It is associated to the very
steep rise of potential. In the second region 共II兲, IPL /I0 decays exponentially 共0.1 ⬍ Q ⬍ Qp兲. It is always related to the formation of
the first potential peak of macro-oscillation. Finally, an asymptotic
IPL /I0 decrease was observed in part III while the macro-oscillation
regime goes on 共Q ⬎ Qp兲. This is in line with literature because
IPL /I0 drop has been already reported for porous InP layers grown in
aqueous electrolytes.2,19
Downloaded on 2016-05-18 to IP 130.203.136.75 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).
Electrochemical and Solid-State Letters, 10 共4兲 D35-D37 共2007兲
In zone II, the IPL /I0 exponential decay is neither current nor
time dependent but only a function of the dissolution charge. The
quantitative chemical analyses of the amount of released InP demonstrate that exclusive semiconductor dissolution takes place during
the galvanostatic treatment without any secondary reaction. Because
a highly doped semiconductor is used, no significant IPL /I0 decrease
is expected if a flat InP surface is kept during the anodic treatment
like for a planar dissolution.20,21 While the porosification occurs, the
IPL /I0 decrease can be fully ascribed to the porous layer formation.
During the pore formation, the full-width at half maximum 共fwhm兲
values of the normalized PL spectra remain constant. This indicates
that neither additional convoluted contribution nor peak shift happen. The exponential decay in the region II suggests, therefore, that
the porous film exhibits a so-called dead layer behavior.20 Obviously, the apparent dead layer thickness increases with InP anodic
dissolution. With regard to the porous layer homogeneity it can be
considered that the pore depth is directly proportional to Q and the
IPL /I0 evolution can be related to the porous layer growth according
to
IPL
= exp共−aporedpore兲
I0
关2兴
where ␣pore the effective absorption coefficient of the porous film.
The porous homogeneity requires a constant ␣pore. For n-InP porous
etching, the pore wall thickness is indeed in the range of WSCL 共Fig.
2兲. This leads to the establishment of a majority carrier depletion in
the whole porous region. Photoexcited holes in this zone have a
reduced density of carriers for radiative recombination. Furthermore, minority carriers generated in this region will experience an
electric force toward the surface where nonradiative processes are
also likely to dominate. As a consequence, the photoluminescence of
bulk InP is progressively masked. At ␭ = 633 nm, ␣pore can be estimated from the exponential fit of the normalized IPL /I0 the
region II. The comparison to absorption coefficient of the bulk InP22
共␣bulk兲 has revealed a lower absorption of the porosified InP
共␣pore ⬇ 0.25 ⫻ ␣bulk兲. While step II is associated to the pore
growth, step I can be correlated to the nucleation phase. During this
stage, nanoscaled pits are mainly formed onto surface defects. The
charge distribution at InP surface is then not sufficiently perturbed to
alter the radiative recombination processes and IPL /I0 is constant as
for a bare sample one.
In part III, the asymptotic behavior reveals a IPL /I0 saturation
which agrees with the constant thickness of the porous layer. Because the porous etching restarts onto a compact InP surface, an
IPL /I0 recovery should be expected. However, the IPL /I0 decrease
goes on because the detached porous layer remains onto InP bulk
surface 共see Fig. 2b兲. Furthermore, no matter the applied current and
duration 共Q ⬎ Qp兲, a porous structure is always present onto InP
surface and multilayered porous films are sometimes observed.
D37
Conclusion
For the first time porous etching of InP has been carried out
under true water free conditions. The chemical analyses demonstrate
that dissolution occurs according to a six holes mechanism and secondary reaction can be neglected. New pore morphology is reported.
Due to the periodical lift off of the porous structure related to the
voltage oscillations, the pore depth remains constant. Photoluminescence experiments demonstrate that the specific geometry of the
porous film induces a “giant” dead layer behavior. A characteristic
periodical dissolution mechanism has been evidenced and is completely ascribed to the physicochemical properties of the acidic liquid ammonia.
Centre National de la Recherche Scientifique assisted in meeting the
publication costs of this article.
References
1. H. Föll, S. Langa, and J. Carstensen, Adv. Mater. (Weinheim, Ger.), 15, 183 共2003兲.
2. T. Takizawa, S. Arai, and M. Masafumi, Jpn. J. Appl. Phys., Part 1, 33, L643
共1994兲.
3. A. Hamamatsu, C. Kaneshiro, H. Fujikura, and H. Hasegawa, J. Electroanal.
Chem., 473, 223 共1999兲.
4. S. Langa, I. M. Tiginyany, J. Carstensen, M. Christophersen, and H. Föll, Appl.
Phys. Lett., 82, 278 共2003兲.
5. P. Schmuki, L. Santinacci, T. Djenizian, and D. J. Lockwood, Phys. Status Solidi A,
182, 51 共2000兲.
6. M. Christophersen, J. Carstensen, K. Voigt, and H. Föll, Phys. Status Solidi A, 197,
34 共2003兲.
7. H. Föll, M. Christophersen, J. Carstensen, and G. Hasse, Mater. Sci. Eng., R., 39,
93 共2002兲.
8. A.-M. Gonçalves, C. Mathieu, M. Herlem, and A. Etcheberry, J. Electroanal.
Chem., 462, 88 共1999兲.
9. A. Etcheberry, A.-M. Gonçalves, C. Mathieu, and M. Herlem, J. Electrochem. Soc.,
144, 928 共1997兲.
10. M. I. J. Beale, J. D. Benjamin, M. J. Uren, N. G. Chew, and A. G. Cullis, J. Cryst.
Growth, 73, 622 共1985兲.
11. R. L. Smith, S.-F. Chuang, and S. D. Collins, J. Electron. Mater., 17, 533 共1988兲.
12. J. Carstensen, M. Christophersen, and H. Föll, Mater. Sci. Eng., B, 69–70, 23
共2000兲.
13. See, for example, J. Jander, Anorganische und Allgemeine Chemie in Flüssigen
Ammoniak. Part I, Friedr. Vieweg & Sohn, Braunschweig 共1966兲.
14. D. Guyomard, C. Mathieu, and M. Herlem, J. Electroanal. Chem. Interfacial Electrochem., 246, 29 共1988兲.
15. O. Seitz, C. Mathieu, A.-M. Gonçalves, M. Herlem, and A. Etcheberry, J. Electrochem. Soc., 150, E461 共2003兲.
16. I. Gérard, F. Iranzo-Marín, J. Vigneron, and A. Etcheberry, J. Electroanal. Chem.,
401, 57 共1996兲.
17. P. H. L. Noten, J. E. A. M. Van den Meeraker, and J. J. Kelly, Etching of III-V
Semiconductors: An Electrochemical Approach, Elsevier, Amsterdam 共1991兲.
18. B. H. Erné, C. Mathieu, J. Vigneron, A. Millon, and A. Etcheberry, J. Electrochem.
Soc., 147, 3759 共2000兲.
19. A. Liu and C. Duan, Appl. Phys. Lett., 78, 43 共2001兲.
20. D. B. Wittry and D. F. Kyser, J. Appl. Phys., 38, 375 共1967兲; U. Langmann, Appl.
Phys., 1, 219 共1973兲; K. Mettler, Appl. Phys., 12, 75 共1977兲; K. Ando, A. Yamamoto, and M. Yamaguchi, J. Appl. Phys., 51, 6432 共1980兲; R. E. Hollingsworth
and J. R. Sites, J. Appl. Phys., 53, 5357 共1982兲.
21. A. Etcheberry, J. Gautron, and J. l. Sculfort, Appl. Phys. Lett., 46, 744 共1985兲; B.
Smandek, G. Chmiel, and H. Gerischer, Ber. Bunsenges. Phys. Chem., 93, 1094
共1989兲.
22. D. E. Aspnes and A. A. Studna, Phys. Rev. B, 27, 985 共1983兲.
Downloaded on 2016-05-18 to IP 130.203.136.75 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).