Detection of lateral spontaneous emission for VCSEL monitoring Charlotte Bringera), Véronique Bardinal°a), Emmanuelle Darana),Thierry Campsa), Yann G. Boucheréb), Guilhem Almuneaua), Olivier Gauthier-Lafayea) ,Pascal Dubreuila), Jean-Baptiste Douceta) and Chantal Fontainea) a)

LAAS-CNRS, 7, Avenue du Colonel Roche, F-31077 Toulouse Cedex 4, France b) ENIB/RESO, CS 73862, F-29238 Brest Cedex 3, France

ABSTRACT VCSELs (Vertical Cavity Surface Emitting Lasers) are nowadays more and more exploited in optoelectronic applications, monitoring their lasing power in a compact and low cost manner becomes crucial. To collect and control the output light, an external photodetector associated with an optical microlens array can be used. Integrated solutions based on the use of a bulk or QW photodetection section added in single-or double-cavity structures have also been proposed. Here, we have investigated a simpler solution based on a standard VCSEL array. Light emitted by a VCSEL has been electrically detected by adjacent VCSELs located in the same array, using in plane optical waveguiding of spontaneous emission in the intrinsic central zone of the devices. We show that the detected photocurrent can be related to the power of the emitting VCSEL. Signal intensity has been studied as a function of VCSELs distance. This method could lead to a more efficient way to monitor VCSEL emission. Keywords: VCSEL, monitoring, spontaneous emission, photodetection, optical waveguide.

1. INTRODUCTION VCSELs are becoming largely involved in optoelectronic applications, and are now commonly used in sophisticated systems such as optical communications datalinks, optical sensors and MOEMs [1] [2] [3]. So, there is an increasing demand on monitoring the optical output power and the other characteristic parameters of the laser device in the system. In particular, the alignment to an optical fiber, and the forward adjustment of the optimized operating point can significantly improve the efficiency and the reliability of optical interconnects. Up to now, the optical power control in VCSEL packaged devices is achieved by a built-in monitor photodiode, practically done with angled lid TO package [4]. However, integrated solutions are often preferred for compactness and self-alignment requirements, cost reduction and signal to noise ratio improvement. A few solutions have been proposed, including a detecting section in the vertical structure, thus requiring an additional fabrication step (Fig.1a) [3] [4]. Moreover in this vertical scheme, the detected power hinders the total optical power emitted by the laser source. The mutual influence (electrical and optical) between the emission and the detection sections has also to be carefully considered [7]. In this paper, we investigate transverse geometries in standard epitaxial VCSEL structures (Fig.1b) to monitor the vertical emission characteristics. This type of lateral detection has been already exploited to measure the non-radiative coefficient of a VCSEL and can be used to detect the laser threshold [8]. However, this study has been carried out on devices dry etched through the cavity, down into the back-DBR, so that the light is scattered by the sidewalls, indeed reducing strongly the detection efficiency. In the geometry we present here, the mesa etch depth stops on the top of the cavity, in order to preserve the optical waveguiding between two adjacent devices in the layer plane. °

é

[email protected]; phone: 33/0 561 33 78 36; fax: 33/0 561 33 62 08 [email protected]; phone: 33/0 298 05 66 66; fax: 33/0 298 05 66 89

Emission

Emission Detection

Detection

(a)

(b) Fig.1: Principle of (a) vertical and (b) lateral detection integrated in a VCSEL.

By exploiting the waveguide formed by the VCSEL cavity section, we then enhance the efficiency of the photodetection, specially if the VCSEL-detector stands at a certain distance from the emitter. This principle is studied on standard VCSEL arrays, one VCSEL is emitting when the others are used as photodetectors. First, optical modellings of spontaneous emission and of waveguide using effective refractive index are presented. Secondly, optical measurements show that spontaneous emission is clearly obtained in the plane when the emitter is lasing. Finally, an electrical model is proposed to discriminate the parasitic current from photocurrent obtained during detection measurements.

2. DESCRIPTION OF THE STUDIED DEVICES The epitaxial structure of the 840nm-VCSELs array used in this study consisted of a lambda-cavity with three GaAsGa 0.7 Al0.3 As quantum wells (MQW zone) embedded by a 30x(AlAs-Ga 0.88Al0.12As) Silicon-doped bottom Bragg mirror and a 18x(AlAs-Ga 0.88Al0.12As) Be-doped top Bragg mirror (Fig.2). After molecular beam epitaxy, oxide-confined devices were fabricated by reactive ion etching of the top DBR, wet thermal lateral oxidation, polymer planarization and front and back metallic electrodes deposition steps. Anode Oxide aperture λλ-cavity

p-DBR 18 pairs 3 Quantum wells n-DBR 30 pairs Cathode

Fig. 2: Description of the VCSEL used in the study.

Before optical and detection measurements, we have checked that all the direct current-voltage I(V) and light-current L(I) characteristics of the studied devices were uniform and that their dark current without and with reverse bias was very low ( 0, n = 1, p = (jP – 1)/(1 – n trans).

(4.c)

As a matter of fact, threshold-crossing is actually continuous. If K2 ≠ 0, both p and n can be obtained by solving successively the following set of steady-state equations [13]: jP = r(n) + (n – n trans) (K2 /K1 ) n 2 /(1 – n),

(5.a)

p = (K2 /K1 ) n 2 /(1 – n).

(5.b)

We represent on Fig.3 the typical behavior of p(jP ) and n(jP ) across threshold. Total spontaneous emission, which varies as Bsp N2 , is also represented in a normalized way: as n 2 (jP ).

0.6

1

(a) p(jP)

0.5

(b) n(jP)

0.8

0.4 0.6 0.3

0.2

0.1

(a)

n 2(jP)

0.4

0.2

0

0.2

0.4

0.6

0.8

1

1.2

1.4

jP

(b)

0

0.2

0.4

0.6

0.8

1

1.2

1.4

jP

Fig.3: (a) Normalized optical power p vs. normalized pumping jP (in dotted line, classical approximation neglecting spontaneous emission); (b) normalized carrier density n, as well as its second power n 2, vs. normalized pumping jP . The curve n2(jP) can be thought of as the normalized overall spontaneous emission: note the plateau reached soon after threshold, quite typical of the carrier clamping caused by gain saturation. Numerical parameters in both cases are a nr = 0.49, b sp = 0.49, c au = 0.02, n trans = 0.5, K 2/K 1 = 0.002. However, the shape of the curves does not depend critically upon the choice of these parameters.

Some points should be emphasized: amplified spontaneous emission (ASE) is not only responsible for noise, it is the very driving source of the optical field. Material gain is therefore subject to saturation through ASE whatever the level of pumping with respect to threshold. Besides, optical gain is never able to reach its classical threshold value but asymptotically. It should also be noted that the precise knowledge of n and p is enough to derive the spectral linewidth of the laser field, as determined by the (generalized) transfer function of the microcavity [14][15]. As a result, we dispose of a simple but accurate description of the main processes at work around threshold for the devices studied here. 3.2. In-plane optical waveguiding In this study, we have only considered the guided modes in the intrinsic zone and neglected those guided in the bottom DBR [16]. The optical confinement in the intrinsic zone was calculated using a simple one-dimensional effective index model taking into account multilayer stacks [17], in three separate sections of the array: at the center of the emitting VCSEL, under the oxide aperture and in the etched region between the emitting and the detecting VCSELs (Fig. 4a). In each case, a single optical mode (TE0) at 840nm propagates in the waveguide limited by the intrinsic region. Corresponding profiles of this mode are shown in Fig. 4b. Effective index values for the 3 regions are respectively 3.297, 3.270 and 3.266. The good overlapping of these three modes profiles allows to assume that a part of the emission produced by the emitter can be guided in this TE0 mode in the plane of the cavity. 4

Refractive index

Oxidized Centre Etched

3

centre oxidized etched

2

1

Detector

0

(a)

Electric field

Emitter

(b)

0

1000

2000

3000

4000

5000

Thickness (nm)

Fig. 4: (a) Description of the 3 different sections used for the calculation of the optical waveguide between the two VCSELs: centre of t he emitter, oxidized zone and etched zone; (b) Refractive index and electric field profiles in the etched region between 2 VCSELs (solid lines) and electric field profiles in the centre (dotted line) and the oxidized region (dashed line).

In our devices, this mode will be no more guided if the layer thickness above the MQW zone is less than 80nm, so the etching step fabrication has to be carefully controlled. The optical confinement in the three quantum wells has been calculated and found to be equal to 5.5% in the etched zone compared to 60% in the whole cavity. Absorption in the unpumped lateral zone remains negligible due to the very low overlap factor between the guided mode and the MQWs. In-plane spontaneous emission being emitted in an "isotropic" way, its power decays as in a cylindrical wave: as the inverse of the distance. Therefore the losses are expected to be low for the distance considered between adjacent VCSELs (250µm). As a conclusion, we expect to obtain efficient in-plane guiding of the spontaneous emission of the quantum wells for the angles corresponding to the calculated numerical aperture of the guide and with no significant losses.

4. OPTICAL MEASUREMENTS 4.1 Description of set up The waveguided spontaneous emission and the stimulated emission were simultaneously measured using the set-up shown in Fig. 5. The sample was cleaved and the output was coupled to an optical fiber [18] which was carefully butted onto the facet, and signal was displayed by means of a spectrum analyzer. The laser output from the vertical face was detected owing to a x10 microscope objective focusing on a photodiode detector. Detector

Emitting VCSEL

Optical fiber

Lateral emission

Spectrum analyser

Cleaved Facet

Fig. 5: Experimental set up used for optical measurements of lateral emission guided in the plane of the cavity and output power of the VCSEL.

4.2 Experimental results The optical spectra obtained from the cleaved facet are found to be broad, whatever the current value is, and thus correspond to the electroluminescence of the VCSELs (Fig. 6a). The various peaks or shoulders observed on the curve could be explained by an in-plane cavity effect. Further measurements are in progress to check this effect. Both intensity emission from vertical face and emission spectrum from the lateral face were registered for currents varying from 0 to 10mA. The spontaneous emission intensity was calculated by integrating the spectra. The results are plotted in Fig. 6b. 1.0x10 -4

Intensity of lateral emission (u.a.)

-6

-6

3.0x10

2.5x10-6 -6

2.0x10

-6

1.5x10

1.0x10-6 -7

5.0x10

0.0

800

820

840

860

Wavelength (nm)

880

900

920

(b)

vertical emitted power lateral emission

8.0x10

-5

200 6.0x10 -5

4.0x10

100

-5

2.0x10 -5

Lateral emission (u.a.)

Applied current : 2mA 5mA 10mA

3.5x10

780

(a)

Vertical emitted power of the VCSEL (u. a.)

300 4.0x10-6

0 0.0 -1

0

1

2

3

4

5

6

7

8

9

10

11

Current applied on the VCSEL (mA)

Fig. 6: (a) Emission spectra measured at a cleaved facet situated 600µm far fro m an emitting VCSEL for 3 values of the applied current: 2, 5 and 10mA; (b) Integrated intensities for currents varying from 0 to 10mA compared to light-current characteristic measured simultaneously at the surface of the VCSEL.

Under laser threshold, the shape of the experimental curve is similar to the predicted n 2 (jP ) curve, which can be thought as the normalized overall spontaneous emission. Moreover, a slope variation is observed at laser threshold. However, no plateau occurs above threshold, which means that the model detailed in section 3.2 taking into account only spontaneous emission doesn’t describe the whole physical mechanisms implied in our devices. Nevertheless, a part of spontaneous emission is clearly guided and can be used for threshold-crossing detection.

5. PHOTODETECTION MEASUREMENTS 5.1. Electrical model

p

p

The basic principle of lateral photodetection has been studied on standard 250µm-pitch VCSEL arrays, one VCSEL (VCSEL1) was emitting when the adjacent VCSELs (VCSEL2, 3 and 4) were used as photodetectors. The equivalent electrical model shown in Fig. 7 allows to synthetically describe the whole conduction mechanisms in the devices. VA , VD and VE are the voltages between top and bottom electrodes of the emitter, the detector and supported by the cavity respectively. To ensure an optical guiding in the plane of the cavity, the mesa etch has been stopped just above the intrinsic layer. We measured a linear current-voltage I(V) characteristic between VCSEL1 and VCSEL2 and conclude that the parasitic conduction in this non intentionally doped layer is resistive, with a measured resistance RP equal to 120kÙ for a distance of 250µm. When the current IA is applied to VCSEL1, the collected current from VCSEL2, ID , is then given by: ID = IPH + IP

(6)

o

o

where IPH is the photocurrent created by the spontaneous emission of VCSEL1 and IP the parasitic current. To limit IP , measurements have been then performed without applied voltage (VD =0) on VCSELs 2, 3 and 4. VCSEL1

VCSEL2

Rc

A

RP VD

i

IE R Bottom

DBR n

IP

IPH R Bottom

VA V E

ID

RT

DBR p

Rc

RT

IA

Fig. 7: Equivalent electrical model of the performed measurements.

5.2 Experimental results The measured currents ID from VCSELs 2, 3 and 4 versus the direct current IA applied on VCSEL1 are plotted in Fig. 8a with a semi-log scale. Two different regimes can be distinguished: - for IA < 20µA (i.e. for VA < 1.2V), VCSEL1 is not conducting yet and the detected current is quite equal to the applied current IA . In this case: VE ~VA and ID is then given by: ID = IP = VE / RP

(7)

- for IA > 20µA (i.e. for VA > 1.2V), VCSEL1 is conducting: VE , and thus ID , evolves logarithmically with IA [19]. The extrapolation of ID is also plotted in Fig. 8a for VCSEL2.

Detected currents on VCSELs 2, 3 and 4

30µA

20µA

Fit IP 2 10µA

0A 1µA

(a)

10µA

100µA

1mA

Applied current I

A

15µA 1.5mW

10µA 1.0mW

5µA

IPH2

Output power

Detected currents on VCSELs 2, 3 and 4

2.0mW ID2 / VCSEL2 (250µm) ID3 / VCSEL3 (500µm) ID4 / VCSEL4 (750µm)

500.0µW

IPH3 IPH4 Output power

0A

10mA

0A

(b)

5mA

10mA 15mA 20mA Applied current IA

25mA

0.0W 30mA

Fig. 8: (a) Detected currents on VCSELs 2,3 and 4 versus the direct curre nt I A applied on VCSEL1 (semi-log scale); (b) Treated detected currents on VCSELs 2, 3 and 4 and light output power of VCSEL1 versus the applied current I A on VCSEL1.

The corresponding photocurrent IPH2 can then be derived by subtracting this extrapolation from the experimental curve. Final results after this treatment are shown in Fig. 8b for VCSELs 2, 3 and 4 as well as the optical output power of VCSEL1 as a function of its applied current IA . On the extracted photocurrents: a significant slope variation can be observed at the threshold current of VCSEL1, as already observed in optical measurements (see section 4-2). The extinction of the laser due to thermal effects is also clearly detected on the curve. This abrupt breakdown could be accounted for by the end of the gain clamping. Between these two working points, the lateral light detected does not present any saturation plateau, which is in agreement with optical measurements. For VCSELs far enough from the emitting one, the current variation is however monotonous. This behaviour could be easily exploited for monitoring applications. Moreover the reduction of the signal with the distance appears not to be due to absorption losses in the waveguide, as has been discussed in section 3.2; it is probably related to the reduction of the interception angle with the distance. Taking into account this angular modification, the level of the different photocurrents is actually the same whatever the distance.

6. CONCLUSIONS In conclusion we proposed a novel and simple way to monitor VCSELs characteristics by using in-plane waveguiding and lateral photodetection of spontaneous emission. Optical and electrical modeling and experimental measurements demonstrated that a part of spontaneous emission emitted by a VCSEL can be detected by an another standard VCSEL of the same array owing to optical waveguiding in the cavity if preserved from etching. After parasitic current elimination, the laser threshold as well as the laser extinction are clearly detected. The level of detected photocurrent is about 15µA (for an emitter-detector distance of 250µm) which appears to be enough for integrated monitoring of the device. Further work will consist in defining optimized devices to minimize the parasitic current and to increase the signal level.

ACKNOWLEDGEMENTS The authors wish to thank French RMNT (Réseau Micro et Nano Technologies) for supporting this study. Y.G. Boucher wishes to thank Pr. Jean Le Bihan, Head of RESO Laboratory at École Nationale d’Ingénieurs de Brest, for his support and encouragement, as well as Pr. Guy M. Stéphan, former Head of the Optronics Laboratory at ENSSAT, Lannion, for stimulating discussions.

REFERENCES 1. 2.

C. Wilmsen, H. Temkin, L.A. Coldren (Eds.), Vertical-Cavity Surface-Emitting Lasers, Cambridge University Press, 1999. H. Li, K. Iga (Eds.), Vertical-Cavity Surface-Emitting Laser Devices, Springer-Verlag, Berlin, 2002.

3. 4.

5. 6. 7. 8. 9.

10. 11. 12. 13. 14. 15. 16. 17. 18. 19.

M. Dragoman and D. Dragoman, "Micro/Nano-Optoelectromechanical systems", Progress in Quantum Electronics, 25, pp. 229-290, 2001. M.S. Lebby, P. Claisse, J. Wenbin, P. Kiley, M. Roll, L. Boughter, P. Sanchez, B. Lawrence, "VCSEL devices and packaging", Micro-Optics Integration and Assemblies, Proc. SPIE 3289, pp. 2-12, M.R. Feldman, Y.-C. Lee, 1998. G. Hasnain, K. Tai, Y.H. Wang, J.D. Wynn, K.D. Choquette, B.E. Weir, N.K. Dutta, A.Y. Cho, "Monolithic integration of photodetector with vertical cavity surface emitting laser", Electron. Lett., 27(18), pp. 1630-2, 1991. A.J. Fischer, W.W. Chow, K.D. Choquette, A.A. Allerman, and K.M. Geib, "Q-switched operation of a coupledresonator vertical-cavity laser diode", Appl. Phys. Lett., 76(15), pp. 1975-1977, 2000. A.J. Fischer, K.D. Choquette, W.W. Chow, H.Q. Hou, and K.M. Geib, "Coupled resonator vertical-cavity laser diode", Appl. Phys. Lett., 75(19), pp. 3020-3022, 1999. J.H. Lin and Y.H. Lee, "Determination of non radiative recombination coefficients of vertical-cavity surfaceemitting lasers from lateral spontaneous emission", Appl. Phys. Lett., 67(3), pp. 314-316, 1995. V. Bardinal, L. Averseng, C. Bringer, J. Polesel-Maris, T. Camps, P. Dubreuil, C. Fontaine, E. Bedel-Pereira, C. Vergnenegre, A. Muñoz-Yagüe, "Experimental demonstration of photodetection oxide modes presence in a dual-purpose single-cavity oxide-apertured VCSEL", Appl. Phys. Lett., 81(10), pp. 1771-1772, 2002. G.P. Agrawal, N.K. Dutta, Long-Wavelength Semiconductor lasers, Van Nostrand Rheinhold, New York, 1986. P.S. Zory, Jr. (Ed.), Quantum Well Lasers, Academic Press, Boston, 1993. A.E. Siegman, Lasers, University Science Books, Mill Valley, 1986. Y. Boucher, "Extended (3×3) transfer matrix formalism: From amplified spontaneous emission to thresholdcrossing", Research Signpost: Recent Res. Devel. Optics, 3, pp. 177-204, 2003. G. Stéphan, Semiclassical study of the laser transition, Phys. Rev. A, Vol. 55 (2), pp. 1371-1384, 1997. G. Stéphan, "Spectral properties in laser patterns", Quantum Semiclass. Opt., 10, pp. 1-12, 1998. A.Y. Cho, A. Yariv and P. Yeh, "Observation of confined propagation Bragg waveguides", Appl. Phys. Lett., 30(9), pp. 471-472, 1977. H. Kogelnik, "Theory of optical waveguides", Guided-wave optoelectronics, T. Tamir, 2nd edition, SpringerVerlag, Berlin, 1990. J.H. Shin, H.E. Shin, and Y.H. Lee, "Effects of carrier diffusion in oxidized vertical-cavity surface-emitting lasers determined from lateral spontaneous emission", Appl. Phys. Lett., 70(20), pp. 2652-2654, 1997. S.M Sze, Physics of semiconductors devices, 2nd edition, John Wiley & Sons, 1981.