Air-gap heterostructures

APPLIED PHYSICS LETTERS 98, 033105 共2011兲

Air-gap heterostructures Ch. Heyn,a兲 M. Schmidt, S. Schwaiger, A. Stemmann, S. Mendach, and W. Hansen Institut für Angewandte Physik und Zentrum für Mikrostrukturforschung, Jungiusstraße 11, D-20355 Hamburg, Germany

共Received 1 November 2010; accepted 30 December 2010; published online 19 January 2011兲 We demonstrate the fabrication of thin GaAs layers which quasi hover above the underlying GaAs substrate. The hovering layers have a perfect epitaxial relationship to the substrate crystal lattice and are connected to the substrate surface only by lattice matched nanopillars of low density. These air-gap heterostructures are created by combining in situ molecular beam epitaxy compatible self-assembled droplet-etching and ex situ selective wet-chemical etching. © 2011 American Institute of Physics. 关doi:10.1063/1.3544047兴 Usually, molecular beam epitaxy 共MBE兲 grown heterostructures are built of alternating semiconductor or metallic layers with varied material composition. This letter demonstrates the realization of a thin epitaxial semiconductorlayers that are separated by a nanometer-thin air-gap. Local droplet-etching 共LDE兲 is the central technology for the fabrication of such air-gap heterostructures. LDE is related to the droplet epitaxy which functionalizes metallic droplets for instance for self-assembled generation of III/Vsemiconductor nanostructures.1–7 At high process temperatures and minimized As pressure, the droplets locally remove the substrate material which results in the self-assembled formation of nanoholes on the substrate surface.8–13 Recently, the fabrication of strain-free GaAs quantum dots has been demonstrated by filling of such nanoholes in AlGaAs and AlAs surfaces.14–16 In the present letter, in situ LDE and nanohole filling was combined with ex situ selective etching to form air-gap heterostructures. A scheme of the different steps during fabrication of the air-gap heterostructures is shown in Fig. 1. The initial GaAs/ AlAs-heterostructures were grown using solid-source MBE. Starting on a 共001兲 GaAs substrate, 300 nm GaAs and subsequently an AlAs layer with thickness dA = 2 . . . 8 nm were grown. Afterwards, the temperature was adjusted to T = 600 ° C, the As shutter and valve were closed, and Ga droplets were deposited for a time t = 4 s at a growth speed of F = 0.8 ML/ s. After droplet formation, a thermal annealing step of t = 180 s was applied. During this annealing time, the nanohole etching took place.11 Figure 2 shows a typical atomic force microscopy 共AFM兲 image of a GaAs surface with Ga-droplet etched nanoholes. With the above process conditions, the average hole depth is of about 10 nm, the diameter of 50 nm, and the density of around 3 ⫻ 108 cm−2. The holes were filled with GaAs in a pulsed mode using 10 pulses of 0.5 s deposition and 10 s growth interruption, respectively. This filling procedure is identical to the fabrication of GaAs QDs by nanohole filling.14–16 This is followed by continuous growth of a GaAs layer with thickness of dG = 50 or 100 nm. In the subsequent ex situ lithography processing, a mask was defined by electron beam lithography. After removal of the exposed photoresist, the mask structure was transferred into the sample with depth of 350 nm by etching with a兲

Electronic mail: [email protected].

0003-6951/2011/98共3兲/033105/3/$30.00

H2O : H2O2 : H3PO4 共500:10:1兲. Finally, after removal of the photoresist, the AlAs layer was completely removed by highly selective etching with 5% HF for 300 s. After this selective etching, the GaAs filled nanoholes remain as nanopillars. This final step yields the formation of an air-gap stabilized by the GaAs nanopillars and with thickness given by the initial AlAs layer thickness. The above procedure is able to provide thin epitaxial semiconductor-layers that nearly freely hover above the substrates. The only connections to the underlying substrates are low-density nanopillars given by the filled nanoholes. To prove the stabilization of the air-gaps by the GaAs nanopillars, we have fabricated a reference sample with layer sequence identical to that of the air-gap heterostructures except the droplet-etching and refilling step. After ex situ selective etching of the reference sample, we observe the detachment of the top GaAs layer. This result demonstrates that the stabilization of the air-gaps is provided by the GaAs nanopillars and not by etching residues or by incomplete removal of the AlAs layers. Since the initial heterostructures are fabricated under optimal MBE growth conditions, these air-gap heterostructures exhibit a perfect epitaxial relationship between the hovering layers and the substrate crystal lattice. Optical microscopy and reflectivity measurements were applied in order to prove the formation of the air-gap between the GaAs layers and its stabilization by the nanopilGa

Ga

AlAs

AlAs

AlAs

GaAs

GaAs

GaAs

1) Growth of AlAs/GaAs heterostructure

2) Ga droplet deposition

3) In situ local droplet etching

GaAs

Photoresist GaAs

GaAs

AlAs

AlAs

Air

GaAs

GaAs

GaAs

4) Overgrowth with GaAs

5) Ex situ lithography

6) Wet-chemical selective etching

FIG. 1. 共Color online兲 Scheme of the steps for air-gap heterostructure fabrication. Steps 1–4 are performed in situ inside the MBE growth chamber and steps 5 and 6 ex situ.

98, 033105-1

© 2011 American Institute of Physics

Author complimentary copy. Redistribution subject to AIP license or copyright, see http://apl.aip.org/apl/copyright.jsp

033105-2

Appl. Phys. Lett. 98, 033105 共2011兲

Heyn et al.

c2

c1 Deep etched window

50 µm

[110]

[110]

500 nm

[110]

(a) (b)

FIG. 2. 共Color online兲 AFM image of a GaAs surface with self-assembled nanoholes created by local etching with Ga droplets.

[110]

c2 GaAs AlAs

c1 Air

GaAs lars. Figures 3共a兲 and 3共c兲 show top-view micrographs on a sample after processing according to the above procedure. Two typical mask features are shown: square deep-etched windows with respective size of 100⫻ 100 ␮m2 关Fig. 3共a兲兴 and a deep-etched letter string “100⫻ 20 ␮m” 关Fig. 3共c兲兴. The arrows c1 in Fig. 3共a兲 mark the contrasts related to the lithographically defined windows and the arrows c2 additional a weaker contrast that we attribute to the borderline up to which AlAs has been removed during the HF underetching step. The origin of the different contrasts is illustrated in Fig. 3共b兲. The additional contrast c2 was observed for all samples with dA = 4 . . . 8 nm. In contrast to this, the dA = 2 nm sample shows no such additional contrast. This observation indicates that a minimum AlAs layer thickness of dA = 4 nm is required for the underetching process. Furthermore, Fig. 3共c兲 shows a clear contrast around the deep-etched letter string “100⫻ 20 ␮m,” as well. This indicates that also here a large area surrounding the deep-etched letter string has been underetched. The most interesting features are the “0”s from the letter string where the inner parts are laterally not connected to the surrounding area. It demonstrates that the GaAs films remain attached to the substrate by the nanometer sized GaAs pillars. From the distance between the contrasts c1 and c2, we have determined values for the etched length of 20. . . 30 ␮m. The data reveal that the HF etching is anisotropic with an etched length along 关⫺110兴 direction being slightly reduced compared to 关110兴: l关−110兴 ⯝ 0.85l关110兴. Furthermore, we find an influence of the AlAs layer thickness d. For dA = 4 nm, the values of l关110兴 and l关−110兴 are reduced by approximately 10% compared to dA = 8 nm. The measurements yield a minimum etched length of about 20 ␮m for etching along 关⫺110兴 direction and dA = 4 nm. Since this value exceeds the half width of 13 ␮m of the “0” in Fig. 3共c兲, the complete underetching and, thus, the realization of quasi hovering GaAs layers inside these structures is confirmed. The air-gap thickness of a completely underetched “0” was studied with spatially resolved reflectivity. A monochromatized and focused white light source was used for irradiation. The diameter of the spot is of about 10 ␮m. This yields an irradiated area containing approximately 235 nanopillars.

(c)

100 µm

FIG. 3. 共Color online兲 Optical micrographs from different mask features on a sample with dA = 8 nm and dG = 100 nm. 共a兲 Deep-etched windows with size 100⫻ 100 ␮m2. The arrows c1 mark contrasts caused by the lithographically defined windows and the arrows c2 the borderline up to which AlAs has been removed during the HF underetching step. 共b兲 Crosssectional scheme along the dashed line in b illustrating the contrasts c1 and c2. Gap and nanopillar size are strongly magnified with respect to the lateral distance. 共c兲 Deep-etched letter string “100⫻ 20 ␮m” with “0”s being laterally not connected to the surrounding area.

The intensity of the reflected radiation was measured at normal incidence and room temperature by an InGaAs diode. Figure 4 shows a normalized reflectivity spectrum I0 / ISub, with the reflectivity spectra I0 from the center of the “0” and ISub from the substrate far away from the underetched regions. The data reveal a clear dependence of I0 / ISub on the wavelength. At ␭ = 800 nm, there is no effect of the air-gap on the reflectivity in comparison to the original AlAs/GaAs heterostructure. On the other hand, at shorter ␭ in the visible range, the reflectivity signal with air-gap is increased. This interesting result explains the contrast c2 in the microscopy images 关Figs. 3共a兲 and 3共c兲兴. The higher reflectivity of the underetched air-gap area in the visible spectral range yields a contrast at the border to the AlAs heterostructure area with lower reflectivity. For a quantitative analysis of the air-gap thickness, the measured reflectivity data are compared with theoretical values. The theoretical reflectivity I0 is calculated using the transfer-matrix method assuming a stack build of a layer with thickness dA and refractive index one 共air兲 and a dG = 50 nm thick GaAs layer. The reflectivity ISub is calculated assuming a stack of a film of AlAs with thickness dA nm and 50 nm GaAs. The wavelength dependent refractive index of


033105-3

Appl. Phys. Lett. 98, 033105 共2011兲

Heyn et al.

defined nanometer air-gaps in single crystalline heterostructures. As an example, the present air-gap heterostructures are suggested for applications in the field of thermoelectricity, where thin semiconductor-layers thermally isolated from the substrate are highly interesting.18 The authors would like to thank the “Deutsche Forschungsgemeinschaft” for financial support via Grant Nos. HA 2042/6-1, GrK 1286, and SSP 1386. T. Chikyow and N. Koguchi, Jpn. J. Appl. Phys., Part 2 29, L2093 共1990兲. T. Mano, K. Watanabe, S. Tsukamoto, N. Koguchi, H. Fujioka, M. Oshima, C. D. Lee, J. Y. Leem, H. J. Lee, and S. K. Noh, Appl. Phys. Lett. 76, 3543 共2000兲. 3 J. S. Kim and N. Koguchi, Appl. Phys. Lett. 85, 5893 共2004兲. 4 Ch. Heyn, A. Stemmann, A. Schramm, H. Welsch, W. Hansen, and Á. Nemcsics, Phys. Rev. B 76, 075317 共2007兲. 5 M. Abbarchi, C. A. Mastrandrea, T. Kuroda, T. Mano, K. Sakoda, N. Koguchi, S. Sanguinetti, A. Vinattieri, and M. Gurioli, Phys. Rev. B 78, 125321 共2008兲. 6 E. Stock, T. Warming, I. Ostapenko, S. Rodt, A. Schliwa, J. A. Töfflinger, A. Lochmann, A. I. Toropov, S. A. Moshchenko, D. V. Dmitriev, V. A. Haisler, and D. Bimberg, Appl. Phys. Lett. 96, 093112 共2010兲. 7 J. H. Lee, Zh. M. Wang, E. S. Kim, N. Y. Kim, S. H. Park, and G. J. Salamo, Nanoscale Res. Lett. 5, 308 共2010兲. 8 Zh. M. Wang, B. L. Liang, K. A. Sablon, and G. J. Salamo, Appl. Phys. Lett. 90, 113120 共2007兲. 9 A. Stemmann, Ch. Heyn, T. Köppen, T. Kipp, and W. Hansen, Appl. Phys. Lett. 93, 123108 共2008兲. 10 P. Alonso-González, D. Fuster, L. González, J. Martín-Sánchez, and Y. González, Appl. Phys. Lett. 93, 183106 共2008兲. 11 Ch. Heyn, A. Stemmann, and W. Hansen, Appl. Phys. Lett. 95, 173110 共2009兲. 12 A. Stemmann, Ch. Heyn, and W. Hansen, J. Appl. Phys. 106, 064315 共2009兲. 13 Ch. Heyn, A. Stemmann, R. Eiselt, and W. Hansen, J. Appl. Phys. 105, 054316 共2009兲. 14 Ch. Heyn, A. Stemmann, T. Köppen, Ch. Strelow, T. Kipp, S. Mendach, and W. Hansen, Appl. Phys. Lett. 94, 183113 共2009兲. 15 Ch. Heyn, A. Stemmann, T. Köppen, Ch. Strelow, T. Kipp, M. Grave, S. Mendach, and W. Hansen, Nanoscale Res. Lett. 5, 576 共2010兲. 16 Ch. Heyn, M. Klingbeil, Ch. Strelow, A. Stemmann, S. Mendach, and W. Hansen, Nanoscale Res. Lett. 5, 1633 共2010兲. 17 E. D. Palik, Handbook of Optical Constants of Solids 共Academic Press, San Diego, 1985兲. 18 T. Zeng, Appl. Phys. Lett. 88, 153104 共2006兲. 1 2

FIG. 4. 共Dashed line兲 normalized reflectivity I0 / ISub from the center of a completely underetched “0” at normal incidence. The nominal AlAs layer thickness was dA = 6 nm and dG = 50 nm. 共lines兲 Normalized reflectivity calculated as is described in the text for dA = 6 and 8.5 nm.

GaAs and AlAs was taken from Ref. 17. For both, the air-gap as well as the AlAs layer, the effect of the GaAs inclusions caused by the low-density nanopillars is considered. From the nanohole density and diameter, we estimate that the nanopillars cover an areal density of about 0.6%. Calculated values of the reflectivity are shown in Fig. 4. Best agreement with the experimental spectrum is found for dA = 8.5 nm. For a sample with dG = 50 nm and an AlAs layer thickness of 4 nm, the calculations yield best agreement assuming dA = 3 nm. As a key result, these reflectivity studies demonstrate that the nanopillars stabilize the air-gaps between the epitaxial layers. In conclusion, we have demonstrated the fabrication of heterostructures providing epitaxial semiconductor-layers being separated by a thin air-gap. The air-gap is stabilized by low-density nanopillars which are generated by combining self-assembled in situ droplet-etching and ex situ selective etching. We expect a number of applications of such well
