Jpn. J. Appl. Phys. Vol. 42 (2003) pp. L 1469–L 1471 Part 2, No. 12A, 1 December 2003 #2003 The Japan Society of Applied Physics

Miniaturized Scanning Near-Field Microscope Sensor Based on Optical Feedback Inside a Single-Mode Oxide-Confined Vertical-Cavity Surface-Emitting Laser Dominique HEINIS, Christophe G ORECKI, Charlotte B RINGER1 , Ve´ ronique B ARDINAL1 , Thierry C AMPS1 , Jean-Baptiste D OUCET1 , Pascal DUBREUIL1 and Chantal F ONTAINE1 LOPMD - IMFC (UMR CNRS 6603), Universite´ de Franche-Comte´, 16 Route de Gray 25030 Besanc¸on Cedex, France 1 Groupe Photonique, LAAS-CNRS, 7 Avenue du Colonel Roche 31077 Toulouse Cedex 4, France (Received June 20, 2003; accepted September 10, 2003; published November 18, 2003)

We propose and demonstrate an original concept of a near-field sensor using the optical feedback properties of a vertical cavity surface-emitting laser (VCSEL) as the detection principle. This is based on the monitoring perturbation induced in the laser cavity by the backscattered light coming from the specimen. Test images confirm the efficiency of the proposed scanning near-field optical microscopy (SNOM) and a solution for the integration of the proposed architecture is given. [DOI: 10.1143/JJAP.42.L1469] KEYWORDS: VCSEL, SNOM microscopy, optical feedback, microsystems

Optical feedback influence on conventional edge-emitting lasers is a well-known problem, particularly studied in single-mode lasers.1) The dependence of the laser power on the feedback conditions has been successfully used as the detection principle in Doppler velocimeters, displacement interferometric sensors, and confocal microscopy.2–4) The feedback of a semiconductor laser has already been used in the atomic force microscope (AFM) to measure the displacement of a cantilever5) Betzig et al.6) demonstrated the applicability of optical feedback as a read-out system in a near-field fiber laser probe. More recently, a cavity scanning near-field optical microscopy (SNOM) head using a laser diode has been proposed.7,8) Since the mid-1980s verticalcavity surface-emitting lasers (VCSELs) have received considerable attention due to their high speed, low threshold current, low divergence and high integration capacity. On one hand, the very short cavity length makes VCSEL operation inherently in the single longitudinal mode, avoiding mode hops typical of conventional diode lasers. On the other hand, VCSELs potential in integrating with microelectromechanical system (MEMS) elements was demonstrated.9) Optical near-field technology has attracted much interest since the demonstration of subwavelength resolution unlimited by optical diffraction. In particular, recent advances in batch fabrication of probes with subwavelength aperture and the progress in laser techniques have pushed conventional SNOM to be applied routinely to a wide range of applications such as high-density optical recording.10) New integrated technologies exploiting the parallel access capabilities of optical storage have emerged to satisfy these requirements. In Japan, Hashizume et al.11) proposed a technique of VCSEL-based probing for optical storage. Our goal is to use the feedback signal from the object to be imaged for perturbing the VCSEL cavity, with the VCSEL operating both as an illumination source and as a detector.12) This idea is demonstrated here and an integrated architecture for the SNOM sensor is proposed. Figure 1 shows the architecture of an ideal SNOM head composed of a VCSEL cavity with the microtip mounted on the top facet, and a PIN photodetector integrated into the back of the VCSEL structure. The microtip is positioned at a nanometer distance z (20-40 nm) from the sample to be measured, the couple extremity of the microtip-back facet of VCSEL playing the role of a compound cavity. When an

SNOM cavity R1

PD

R3

R

2

sample

VCSEL tip R3

R3 + r

Fig. 1. Structure of a VCSEL-based compound cavity.

evanescent wave is emitted from the microtip aperture, it is reflected by the sample, transformed into a propagating wave by the microtip, and then propagated into the VCSEL cavity. The evanescent interaction leads to a change r in the reflectivity R3 of the compound cavity. As the light beam is first emitted by the microtip, then reflected by the sample and finally backscattered by the same microtip, the SNOM head both plays the role of the light emitter and the nearfield detector. In the presence of feedback, the threshold current of VCSEL is a function of compound cavity length z12)   pffiffiffiffiffi  1 Ith ðzÞ ¼ C
th 2i d
ln ðR1 R2 Þ
 R3 þ r cos ð2k0 zÞ ; 2 ð1Þ where C is the coupling efficiency of the back-reflected light into the VCSEL cavity based on the coupling geometry, R1 and R2 are the reflectivities of the top and bottom Bragg reflectors, respectively, and
th is the quantum efficiency based on the structure of the active medium. Using this model, we demonstrated (see ref. 12) that the threshold current decreases with an increasing R3 , while the output power increases with a decrease in R3 . The shift of the threshold current provides a convenient evaluation of the depth of power modulation, providing a SNOM detection based on the power modulation measured by the photodetector. We carried out the experiment with an oxide-apertured GaAs-based VCSEL grown by molecular beam epitaxy under in situ optical control. The VCSEL contains 3

L 1469

Jpn. J. Appl. Phys. Vol. 42 (2003) Pt. 2, No. 12A

Fig. 2. Single-mode emission spectrum of VCSEL for a 400 mA current.

V

in

VCSEL

Tapered Optical Fiber

95% Y Junction

5% z

y x

Photodetector

Piezo for Z distance control

Tunning Fork

quantum wells (QWs) embedded between two high-reflectivity distributed bragg reflectors (DBRs). The top (n-type doped) and bottom (p-type doped) DBRs consist of stacks of 20 and 30.5, respectively, alternating quarter-wave GaAlAs and AlAs layer periods. To obtain low-resistivity mirrors, heterojunction band discontinuity effects are minimized by inserting a thin gradual composition layer at each heterointerface. Furthermore, p-type doped DBRs are grown at a low temperature (750 K) to limit Beryllium diffusion and the QWs’ region at high temperature (970 K) to achieve a maximum optical gain. In VCSEL structures, the spectral alignment between the gain of the QWs and the single longitudinal Fabry-Prot resonance are very sensitive to layer thickness. For this reason, an optical real-time thickness control technique has been used together with the VCSEL growth : tunable dynamic optical reflectometry (TDOR).13) The influences of both the QWs in the cavity and the numerous graded compositions inserted at each DBR interface have been considered in the TDOR modelling, as well as the different substrate temperatures applied during growth. After growth, the structures are processed using reactive ion etching (RIE) to define mesas, benzocyclobutene (BCB) for planarization, and finally, selective lateral oxidation of an AlAs layer near the cavity to form a buried oxide aperture. These devices exhibit a lasing wavelength within 0.2% of the aimed value. Moreover, its variation along a 2 cm radius on the wafer has been found to be less than 0.3%. Finally, significantly reduced threshold currents have been measured to as low as 200 mA as well as a singletransverse-mode operation, as shown in Fig. 2. The beam divergence has been found to be less than 10 . The experimental setup is shown in Fig. 3. The key element is the single-mode VCSEL described previously. The output light from VCSEL is couple-injected into a single-mode optic fiber and then divided as 5%:95% by the arms of a Y-junction. One arm of the Y-junction is tapered using a commercial puller. Its apex size is about 50-100 nm (allowing the SNOM subwavelength resolution) and is not coated. The other arm is connected to a photodetector. Light transmitted through the tapered arm is diffracted by the tip aperture. The evanescent wave is reflected by the sample, transformed into a propagating one by the tip and then recollected by the tapered arm of the Y-junction. The feedback light is propagated into the VCSEL cavity,

D. HEINIS et al.

20–40nm

L 1470

Shear Force Detection

Scan X–Y Sample HV High Voltage

Fig. 3. Experiment setup.

modulating the VCSEL power, and detected by the detector. To insure optical near-field detection at a constant distance between the end of the near-field probe and the sample, a well-known technique of shear force detection is used. The end of the tapered optical fiber interacting with the surface is glued onto the leg of a quartz tuning fork, which is excited in transverse vibrations around a resonance frequency of 32 kHz. The oscillation of the tip creates shear forces between the tip and the sample upon interaction. The magnitude of the shear force signal depends on the tip-tosample distance, and is maintained constant by the feedback system. The amplitude signal is fed into the control electronics which generates a signal to the vertical piezo actuator. An additional x-y axis piezodriver stage is used to scan the sample. The first analyzed sample is a 3-mm-period grating etched into a silica layer deposited on a silicon substrate. The peakto-valley size of the grating is 20 nm. The sample is imaged simultaneously via SNOM and shear force microscopy, as shown in Figs. 4(a) and 4(b). The scanning area is 55 mm2 . The sampling period is 80 nm. On the topographical image, black represents silicon, while white represents silica. The optical image (Fig. 4(b)) is obtained by measuring the power emitted by the photodetector. We note strong similarities between the two images. Because of the constant distance between tip and sample, SNOM detection is only possible because of the optical index difference between silica and silicon (n ’ 2). In this case, the modulation of the total VCSEL-emitted power is nearly 5%. The cross section of this sample (see Fig. 5) allows us to estimate the optical resolution to be 200 nm. Then, we investigate a second specimen with a very weak index modulation. It is the 7-mm wide rib region of a Ti:LiNbO3 waveguide. It is fabricated

Fig. 4. Topographical (a) and SNOM (b) images of a silica diffraction grating.

Jpn. J. Appl. Phys. Vol. 42 (2003) Pt. 2, No. 12A

D. HEINIS et al.

Side view

L 1471

VCSEL Wafer

bump cantilever SNOM tip

Optical signal

Arbitrary Units

sample

Front view

VCSEL Wafer

bump cantilever SNOM tip

sample

Shear force

Fig. 7. Architecture of the integrated SNOM sensor.

0

1

2

3

Length (µ µm)

4

5

Fig. 5. Cross-sectional profile of the shear force and optical image.

Fig. 6. Topographical (a) and SNOM (b) images of a Ti:LiNbO3 waveguide.

by the standard diffusion of titanium on LiNbO3 , modifying locally the refractive index. In this case, the optical index contrast is less than 10
3 . Rib heigh is approximately 20 nm. The scanned area is now 3030 mm2 and the sampling period is 500 nm. Figure 6(b) shows the optical image, while the Fig. 6(a) corresponds to the topographical image. All the conclusions deduced previously concerning the difference between topographical and optical images remain true. Beneath the optical image, we have displayed a color bar to show the entire optical signal modulation. The extreme values are obtained at the edges of the rib. This is caused by their soft slope compared with the very sharp edges of the first sample. We also note an opposite behavior for each edge (white on left side, black on right side). This is certainly due to the use of a nonsymetrical tip.

The presented optical images are of excellent quality. The sensitivity of SNOM detection based on monitoring the VCSEL power is limited to a sample with an index contrast of less than 10
3 . To improve the resolution a heterodyne detection as presented by ref.14) will be used. We demonstrate the performance of the SNOM sensor and justify the interest of the integrated architecture. The basic design (Fig. 7) will consist of a transparent metalized tip mounted on a cantilever with a VCSEL bonded by flip-chip above the tip. An aperture in the metallization layer at the apex is illuminated by the VCSEL and the light beam that is backreflected is detected using a detector integrated in the structure of the VCSEL. Because of the easy fabrication of the VCSEL matrix, we can imagine a matrix of such a SNOM head working in parallel for high-density optical storage applications or for biochip analyses. 1) W. M. Wang, K. T. V. Grattan, W. J. O. Boyle and A. W. Palmer: Appl. Opt. 33 (1994) 1795. 2) P. J. de Groot, M. Gallatin and S. H. Macomber: Appl. Opt. 27 (1988) 4475. 3) A. Dandrige, R. O. Milles and T. G. Giallorenzi: Diode Laser Sensor Lett. 16 (1980) 948. 4) R. Juskaitis, N. Rea and T. Wilson: Opt. Commun. 99 (1993) 105. 5) D. Sarid: Scanning Force Microscopy (Oxford University Press, New York, 1991) Chap. 8, pp.101-109. 6) E. Betzig, S. G. Grubb, R. J. Clichester, D. J. DiGiovanni and J. S. Weiner: Appl. Phys. Lett. 63 (1993) 3550. 7) K. Ito, T. Shintani, S. Hosaka and M. Muranishi: Jpn. J. Appl. Phys. 37 (1998) 3759. 8) U. Schwartz, M. L. Berthie´, D. Courjon and H. Bielefeldt: Opt. Commun. 134 (1997) 301. 9) S. Heisig, H. U. Danzebrink, A. Leyk, W. Mertin, S. Mu¨ster and E. Oesterschultze: Ultrasmicroscopy 71 (1998) 99. 10) Y. Martin, S. Rishton and H. K. Wickramasinghe: Appl. Phys. Lett. 71 (1997) 1. 11) J. Hashizume, S. Shinada, F. Koyama and K. Iga: Opt. Rev. 9 (2002) 186. 12) C. Gorecki, S. Khalfallah, H. Kawakatsu and Y. Arakawa: Sens. & Actuat. 2799 (2000) 113. 13) F. Van Dijk, V. Bardinal, C. Fontaine, E. Benel-Pereira and A. Mu~nozYague: J. Cryst. Growth 201/202 (1999) 1028. 14) I. Koltchanov, K. Petermann and J. Roths: Proc SPIE 3098 (1999) 325.