Impact of relative intensity noise of vertical-cavitysurface-emitting lasers on optics-based micromachined

audio and seismic sensors

Robert Littrell,1,2,* Neal A. Hall,1 Murat Okandan,1 Roy Olsson,1 and Darwin Serkland1

1Sandia National Laboratories, P.O. Box 5800, Albuquerque, New Mexico 87185-1080, USA2Currently with the University of Michigan, 2250 G. G. Brown, 2350 Hayward Street, Ann Arbor, Michigan 48109-2125, USA

*Corresponding author: rlittrel@umich.edu

Received 26 March 2007; revised 25 July 2007; accepted 4 August 2007;posted 7 August 2007 (Doc. ID 81064); published 21 September 2007

The relative intensity noise of vertical-cavity surface-emitting lasers (VCSELs) in the 100 mHz to 50 kHzfrequency range is experimentally investigated using two representative single-mode VCSELs. Measure-ments in this frequency range are relevant to recently developed optical-based micromachined acousticand accelerometer sensing structures that utilize VCSELs as the light source to form nearly monolithic1 mm3 packages. Although this frequency regime is far lower than the gigahertz range relevant to opticalcommunication applications for which VCSELs are primarily designed, the intensity noise is found to below and well within the range of cancellation using basic reference detection principles. © 2007 OpticalSociety of America

OCIS codes: 130.6010, 250.7260.

1. Introduction

Ultraminiature micromachined optical interferome-ter array architectures that employ vertical-cavitysurface-emitting lasers (VCSELs) as light sourceshave recently been demonstrated. Since a wide vari-ety of micromachined sensors including microphonesand accelerometers ultimately rely on accurately de-tecting small displacements of moving structures,and since optical interferometry has the potential toprovide superior displacement detection resolutionover other popular microelectromechanical systems(MEMS) detection schemes (i.e., capacitive, piezore-sistive, etc.), these technologies have the ability toaddress applications where sensors and sensor ar-rays with extreme sensitivity to size ratio are desired.Optical detection has demonstrated the additionalbenefit of removing many of the mechanical designconstraints associated with capacitive detectionschemes—enabling the design of micromachinedstructures with low thermal mechanical noise levels

characteristic of larger, conventional high-fidelity in-struments [1].

A capacitive micromachined ultrasonic trans-ducer (cMUT) technology has been developed byDegertekin et al. that utilizes a monolithically in-tegrated diffraction-based optical detection archi-tecture to measure the vibration of individual160 �m diameter membranes with a displacementresolution of 20 fm��Hz, measured at 100 kHz [2]. Inthis case, the high detection sensitivity at ultrasonicfrequencies from individual array elements enabledthe demonstration of phased-array imaging in air us-ing only a 4 mm � 4 mm acoustic aperture—thesmallest demonstrated to date [3,4]. Ultrasound sen-sor arrays based on a multichip architecture, includingcustom designed complementary metal-oxide semicon-ductor (CMOS) photodiodes, have been demonstratedwith each sensor element occupying only 2 mm3

volume [4,5]. Hall et al. are also developing siliconmicrophone arrays based on the optical diffraction-based detection technology. Instrumentation-gradenoise performance has been demonstrated from1.5 mm diameter microphone elements [1]. Advancedbiomimetic microphone structures with directional

0003-6935/07/286907-05$15.00/0© 2007 Optical Society of America

1 October 2007 � Vol. 46, No. 28 � APPLIED OPTICS 6907

capabilities are also being developed for hearing aids[6,7]. For these applications, the detection of soundwhile pulsing the VCSEL has been demonstrated forlow-power operation [5].

In addition to these efforts, a research team atJohns Hopkins is pioneering the advanced packagingof MEMS interferometers based on the flip chip bond-ing of silicon on sapphire (SOS) substrates housingCMOS photodiodes and analog readout electronicsto VCSELs and silicon MEMS structures [8,9]. AFabry–Perot system has been developed, and the in-terference effect has been experimentally demon-strated. Additional MEMS-based optical transducersthat have been demonstrated include grating-basedaccelerometers [10] and Fabry–Perot-based micro-phones [11].

The fundamental displacement detection limit forthese detection schemes as well as any proposedoptical-based method is the quantum shot noise of thedetected laser light. In many cases, however, excessintensity or phase fluctuations of the light sourceimpose a substantially higher noise limit. The effectof phase noise on displacement resolving capabilitiesin a direct interferometric detection scheme dependson the optical path difference and the spectral line-width of the VCSEL. Displacement referred phasenoise levels are significantly below shot noise for com-mon VCSELs implemented in micromachined inter-ferometers, with optical path differences in the2–100 �m range [1,9]. Intensity noise of single-modeVCSELs has been experimentally and theoreticallyinvestigated previously in the megahertz and giga-hertz frequency range, resulting in an improved un-derstanding of how aspects of a VCSEL’s physicalstructure affect intensity noise [12,13]. Measure-ments on multimode structures below 100 kHz havealso been conducted for this purpose [14]. In whatfollows, we present measurements of the relativeintensity noise spectra of single-mode VCSELsthroughout the seismic and audio frequency regimes(in the range of 100 mHz to 20 kHz). These measure-ments determine the displacement detection reso-lution that an interferometer using such a VCSELand a direct detection scheme can achieve withoutnoise cancellation efforts. In addition, measure-ments utilizing laser intensity noise cancellationelectronics are presented, demonstrating the abilityto achieve shot-noise-limited displacement detec-tion with VCSELs down to millihertz frequencies.

2. Experimental Configuration and MeasurementResults

For the purpose of this paper, the audio and seismicfrequency ranges are considered to be 20 Hz to20 kHz, and 100 mHz to 10 kHz, respectively. Inten-sity noise measurements in the audio range areperformed using an 850 nm single-mode VCSEL fab-ricated at the Sandia National Laboratories (SNL)Compound Semiconductor Research Laboratory(CSRL). Figure 1 shows the experimental setup usedfor the measurement. The bare die VCSEL is packagedon a flat transistor outline (TO) header, and the injec-

tion current is controlled by a voltage divider circuitusing a 9 V battery as the source, a configuration thatproduces minimal noise in the injection current. Lightfrom the VCSEL is focused by a lens before passingthrough a variable beam splitter to create the signalbeam used for intensity noise measurements and areference beam that can be used for cancellation of thisnoise. The intensities of the beams are monitored withphotodiodes (PDs), with the resulting photocurrentsinput to an autobalanced noise canceling current-to-voltage amplifier circuit as described in detail byHobbs [15]. (Advanced Photonix SD 200-11-31-241type photodiodes were used for all measurements. Toeliminate the possibility of additional noise in the sys-tem, the windows were removed from the TO cans ofthese packaged components.) This circuit functions asa proportional integral controller, setting the dc level ofVout to zero by reducing the dc level of the referencephotocurrent to that of the signal photocurrent andthen subtracting one from the other to cancel the com-mon mode noise resulting from intensity fluctuationsin the laser. Because the circuit equalizes the dc pho-tocurrents continuously in time, a low-frequency sig-nal cutoff is inherent to this scheme as derived byHobbs. This cutoff frequency is a function of the designparameters of the circuit as well as the particular op-erating conditions (i.e., signal and reference photocur-rent levels). Circuits for both audio and seismicapplications were constructed. A schematic of the par-ticular circuit used for measurements in the audio

Fig. 1. (Color online) Noise measurement test setup.

6908 APPLIED OPTICS � Vol. 46, No. 28 � 1 October 2007

range, for example, is shown in Fig. 2. The transim-pedance gain in each case is 100 k�, as shown in Fig.2, and the voltage output is analyzed using a StanfordResearch Systems SR-785 Dynamic Signal Analyzer,as shown in Fig. 1.

For the VCSEL intensity noise measurement, thereference beam is covered and the circuit functions asa single transimpedance amplifier. The VCSEL out-put power in this particular experiment is 275 �W.(All optical power levels reported are measured usinga Newport power meter model 2930C with detectormodel 918-SL.) After passing through the lens andbeam splitter, 77 �W of optical power is incident onthe PD, resulting in 41.4 �A of dc photocurrent and a4.14 V dc output from the amplifier. The rms value ofthe current noise spectral density associated withshot noise equals �2qi�1�2, where q is the electroncharge and i is the dc photocurrent. This theoreticalquantum shot-noise limit for this photocurrent iscomputed as 3.64 pA��Hz, or �108.8 dB relative to1 �A��Hz. The shot-noise density level is lower thanthe measured intensity noise data as presented in thetop curve of Fig. 3, indicating a VCSEL intensitynoise level above the shot-noise limit in this fre-quency range.

To verify that the measured data is indeed VCSELintensity noise and to demonstrate how this noise canbe effectively canceled, the reference beam containing128 �W of optical power and 66.2 �A of dc photocur-rent is uncovered and used for noise cancellation.Because the circuit subtracts the two photocurrentsand because their shot noise is incoherent, the ex-pected shot-noise current density is �105.8 dB refer-enced to 1 �A��Hz. The measured noise data shownin Fig. 3 after cancellation is in close agreement withthe theoretical prediction in the 30 Hz to 50 kHz fre-quency range. There are a few frequencies at whichthere exists exceptionally high electromagnetic inter-ference, and these can be seen as narrowband peaksin the measurement. The roll-off below 30 Hz is ex-pected as this is the designed 3 dB cutoff frequency ofthe noise cancellation circuit under these conditions.Above 30 Hz, the subtraction of the comparison andsignal beams removes the common intensity noise

but not signal, as signal is not carried in the compar-ison beam. This cutoff frequency can be indepen-dently verified by directly measuring the frequencyresponse of the circuit. This measurement requiresthe use of a second VCSEL to provide an equivalentreference photocurrent level while modulating theVCSEL going into the signal photodiode. The result ofthis measurement can be seen in Fig. 4, which againshows the expected cutoff near 30 Hz.

It should be noted that the VCSEL intensity noiseas represented in the uncanceled measurement inFig. 3 was shown to scale approximately with the dcoutput intensity level of the VCSEL for the powerlevels used. For example, measurements using 2�the light power reported in this paper show an ap-proximate 2� (or 6 dB) increase in the measurednoise levels throughout the frequency range of the

Fig. 4. (Color online) Measured frequency response function ofthe laser noise cancellation circuitry, showing the anticipated low-frequency cutoff.

Fig. 2. Schematic of the intensity noise cancellation circuit usedin the measurements. This circuit is based on that initially dem-onstrated by Hobbs.

Fig. 3. (Color online) VCSEL intensity noise spectra in theaudio frequency regime with and without relative intensity noisecancellation.

1 October 2007 � Vol. 46, No. 28 � APPLIED OPTICS 6909

measurement. The measurement results presentedhere can be scaled and interpreted for systems em-ploying higher light power or PDs with different re-sponsivities. The ratio of the noise density to the dcphotocurrent level is expected to remain the same forthe same VCSEL.

Measurements in the seismic range were per-formed using a single-mode VCSEL from Avalon Pho-tonics with the same experimental setup and with thesame optical power, resulting in identical dc pho-tocurrent levels. The measured noise spectra are pre-sented in Fig. 5. The intensity noise of the AvalonVCSEL above 100 Hz is quite low, only 5 dB higherthan the shot-noise limit. The low-frequency inten-sity noise, however, increases substantially and isapproximately 35 dB above the shot-noise limit at100 mHz. In this frequency regime, the noise cancel-lation circuit, which was designed and verified tohave a cutoff frequency below 100 mHz for thesemeasurements, does not completely cancel the noise,suggesting the presence of an additional noise sourcein the measurement that is not common or correlatedamong the signal and comparison beams.

3. Discussion and Relevance of Results

The essence of optical-based microsensors including awide range of proposed microphone and accelerome-ter architectures is a scheme for modulating the in-tensity of an optical beam with respect to changes indisplacement of an integrated structure [16]. Conver-sion of the noise spectra presented in this study intounits of displacement depends solely on the efficiency(i.e., sensitivity) of the modulation technique used.For example, any scheme with Michelson-type inter-ference such as those highlighted in the introductionof this paper modulates the full optical power in thesignal beam across ��4 displacement, with a maxi-mum sensitivity equal to the dc optical power multi-plied by 2���. For the power levels used in thesemeasurements and expressed in terms of photocur-

rent instead of optical power, this sensitivity is com-puted as 0.306 �A�nm. The measured shot-noiselimit therefore corresponds to 16.8 fm��Hz displace-ment detection resolution for Michelson-type detec-tion. (Displacement resolution is used synonymouslywith minimum detectable displacement, both beingdefined as the signal level resulting in a signal-to-noise ratio equal to unity.) It is important to note thatthis noise level corresponds to the power levels usedin these measurements. The sensitivity increases lin-early with laser intensity while the shot noise is pro-portional to the square root of the laser intensity.Therefore, by increasing power, the displacementnoise limit corresponding to the shot noise is reduced.

To interpret the measurement results for a micro-machined microphone sensor application such asthose described in Section 1, noise contained in theaudio bandwidth is integrated using the commonlyused A-weighting function, which accounts for thefrequency dependant response of the human ear. Theactual A-weighted sound pressure resolution dependson the compliance of the particular microphone dia-phragm used. Polysilicon diaphragms with 1.5 mmdiameter and 20 nm�Pa compliance have recentlybeen demonstrated for optical microphone structures[1]. Calculations using this value and the measure-ment results in Fig. 3 are summarized in Table 1,demonstrating the potential to achieve below 15dB�A� noise performance from 1.5 mm diameterelements. This figure is comparable to the best in-strumentation and studio-application microphonesavailable that employ capacitive detection and 2.5 cmdiameter diaphragms.

4. Conclusion

The intensity noise of two representative single-modeVCSELs has been experimentally studied in the100 mHz to 50 kHz frequency range. Although thesefrequencies are far lower than those relevant to op-tical communication applications for which VCSELsare primarily designed, the intensity noise ofVCSELs in this low-frequency regime is quite lowand well within the range of common-mode cancella-tion using basic reference detection principles. Forclasses of optical-based sensors utilizing interferenceprinciples, intensity noise may be one of many im-portant noise sources to consider, with others includ-ing phase noise, wavelength drift, and noise resultingfrom mode and polarization flips. Acoustic and accel-

Fig. 5. (Color online) VCSEL intensity noise spectra in the seis-mic frequency regime with and without cancellation.

Table 1. Calculated A-Weighted Displacement and Sound PressureNoise Levels (SPLs)a

A-Weighted rmsDisplacement

(pm) SPL

Without Cancellation 13.2 30.4 dB(A)With Cancellation 2.07 14.3 dB(A)

aUsing the noise measurements presented in this paper andassuming Michelson type detection with a 20 nm�Pa compliantdiaphragm.

6910 APPLIED OPTICS � Vol. 46, No. 28 � 1 October 2007

eration sensor designs able to mitigate these effectshave the potential to achieve shot-noise-limited dis-placement detection, with corresponding sound pres-sure and acceleration resolution well beyond thecurrent state of the art. Furthermore, the VCSEL-based micromachined sensor designs discussed in theintroduction to this paper are well suited for the cre-ation of array architectures that leverage the precisepatterning inherent to micromachining processes.Such arrays may be ideally suited for integrationwith advanced source localizing audio signal process-ing algorithms, where extreme resolution from min-iature array elements is required [17].

The authors thank the Intelligence CommunityPostdoctoral Fellowship Program, the Department ofHomeland Security Scholarship and Fellowship Pro-gram, and Sandia’s Laboratory Directed Researchand Development Program. Sandia is a multipro-gram laboratory operated by Sandia Corporation, aLockheed Martin Company, for the U.S. Departmentof Energy under contract DE-AC04-94AL85000.

References1. N. A. Hall, R. Littrell, M. Okandan, B. Bicen, and F. L.

Degertekin, “Micromachined optical microphones with lowthermal-mechanical noise levels,” J. Acoust. Soc. Am. (to bepublished).

2. F. L. Degertekin, N. A. Hall, and W. Lee, “Capacitive micro-machined ultrasonic transducers (cMUTs) with integratedoptoelectronic readout,” in IEEE Ultrasonics Symposium(IEEE, 2001), pp. 875–881.

3. N. A. Hall, W. Lee, J. Dervan, and F. L. Degertekin, “Micro-machined capacitive transducers with improved optical detec-tion for ultrasound applications in air,” in IEEE UltrasonicsSymposium (IEEE, 2002), pp. 1027–1030.

4. W. Lee, N. A. Hall, and F. L. Degertekin, “Micromachinedacoustic sensor array with diffraction-based optical interfero-metric detection,” Proc. SPIE 4985, 140–151 (2003).

5. W. Lee, N. A. Hall, Z. Zhou, and F. L. Degertekin, “Fabricationand characterization of a micromachined acoustic sensor with

integrated optical readout,” IEEE J. Sel. Top. Quantum Elec-tron. 10, 643–651 (2004).

6. C. Weili, B. Bicen, N. A. Hall, S. A. Jones, F. L. Degertekin, andR. N. Miles, “Optical sensing in a directional MEMS micro-phone inspired by the ears of the parasitoid fly, Ormia ochra-cea,” in Micro Electro Mechanical Systems (IEEE, 2006), pp.614–617.

7. C. Gibbons and R. N. Miles, “Design of a biomimetic directionalmicrophone diaphragm,” ASME, Noise Control and AcousticsDivision (Publication) NCA (ASME, 2000), Vol. 27, pp. 173–179.

8. F. Tejada, A. G. Andreou, J. A. Miraqliotta, R. Osiander, and D.Wesolek, “Silicon on sapphire CMOS architectures for inter-ferometric array readout,” in IEEE International Symposiumon Circuits and Systems (IEEE, 2004), pp. 880–883.

9. F. Tejada, D. Wesolek, J. Lehtonen, J. A. Mirragliotta, A. G.Andreou, and R. Osiander, “An SOS MEMS interferometer,”Proc. SPIE 5346, 27–36 (2004).

10. E. B. Cooper, E. R. Post, S. Griffith, J. Levitan, S. R. Manalis,M. A. Schmidt, and C. F. Quate, “High-resolution micromach-ined interferometric accelerometer,” Appl. Phys. Lett. 76,3316–3318 (2000).

11. D. S. Greywall, “Micromachined optical-interference micro-phone,” Sens. Actuators A 75, 257–268 (1999).

12. D. Wiedenmann, M. Kicherer, C. Jung, M. Grabherr, M.Miller, R. Jager, and K. J. Ebeling, “Sub-Poissonian intensitynoise from vertical-cavity surface-emitting lasers,” Appl. Phys.Lett. 75, 3075–3077 (1999).

13. D. Wiedenmann, P. Schnitzer, C. Jung, M. Grabherr, R. Jager,R. Michalzik, and K. J. Ebeling, “Noise characteristics of 860nm single-mode vertical-cavity surface-emitting lasers,” Appl.Phys. Lett. 73, 717–719 (1998).

14. P. Signoret, G. Belleville, and B. Orsal, “Experimental inves-tigation of the 1�f amplitude noise of vertical-cavity surface-emitting lasers,” Fluct. Noise Lett. 1, L1–L5 (2001).

15. P. C. D. Hobbs, “Ultrasensitive laser measurements withouttears,” Appl. Opt. 36, 903–920 (1997).

16. N. Bilaniuk, “Optical microphone transduction techniques,Appl. Acoust. 50, 35–63 (1997).

17. M. Stanacevic, G. Cauwengberghs, and G. Zweig, “Gradientflow adaptive beamforming and signal separation in a minia-ture microphone array,” in Proceedings of the IEEE Interna-tional Conference on Acoustics, Speech, and Signal Processing(IEEE, 2002), pp. 416–419.

1 October 2007 � Vol. 46, No. 28 � APPLIED OPTICS 6911