-
Development of VCSELs for Optical Nerve Stimulation
Matthew Dummera, Klein Johnsona, Mary Hibbs-Brennera, Matthew
Kellerb,Tim Gongb, Jonathon Wellsb, and Mark Bendettb
aVixar, 15350 25th Ave. N, Plymouth MN, USAbLockheed Martin
Aculight, 22121 20th Ave. SE, Bothell WA, USA
ABSTRACT
Neural stimulation using infrared optical pulses has numerous
potential advantages over traditionalelectrical stimulation,
including improved spatial precision and no stimulation artifact.
However,realization of optical stimulation in neural prostheses
will require a compact and efficient opticalsource. One attractive
candidate is the vertical cavity surface emitting laser. This paper
presentsthe first report of VCSELs developed specifically for
neurostimulation applications. The targetemission wavelength is
1860 nm, a favorable wavelength for stimulating neural tissues.
Continuouswave operation is achieved at room temperature, with
maximum output power of 2.9 mW. Themaximum lasing temperature
observed is 60 C. Further development is underway to achievepower
levels necessary to trigger activation thresholds.
Keywords: Nerve Stimulation, Infrared, VCSEL, Semiconductor
Laser
1. INTRODUCTION
Artificial stimulation of neural tissue has been an important
tool for identifying nerve connectivity and func-tionality, as well
as development of neural prostheses. Historically, the most widely
used methods for nervestimulation have been electrical. However the
electrode-tissue interface has many limitations including damageto
neural tissue by high current or mechanical contact, susceptibility
to environmental interference, and introduc-tion of high-frequency
artifacts to the stimulation signal.13 In addition, conductivity of
surrounding tissue leadsto undesired current spreading and poor
spatial specificity.1 These shortcomings have prompted exploration
intoalternative means of stimulation.46 Recently, it has been
discovered that relatively low levels of pulsed infraredlaser light
are capable of triggering neural activity in both motor and sensory
systems.7 This approach hasbeen determined to have many advantages
over direct electrical stimulation. For example, no contact is
requiredbetween the tissue and the source, and optical activation
appears not to produce any stimulation artifact.1 Fur-thermore, a
focused laser beam can be used to pinpoint small numbers of
neurons, thereby improving the spatialresolution.7
Optical neurostimulation could be instrumental in advancing
neurophysiological research and expanding itsuse in clinical
applications. Neural prostheses, devices which aim to restore
sensory or motor function by directlyinterfacing with the nervous
system, might benefit from this new technology. One such device is
the cochlearimplant, which restores hearing in deaf patients by
stimulating auditory nerves. These devices require
multiplestimulation sites that activate the spiral ganglion neurons
lining the cochlea, each site corresponding to a specificauditory
frequency. The spectral resolution of the implant depends on the
total number of stimulation channels aswell as the physical spacing
between them. Currently, electrical implants utilize up to 22
intra-cochlear electrodes.However, studies have shown that beyond
4-8 channels, speech comprehension does not improve due to
crosstalkbetween electrodes.8,9 (Some improvement in resolution has
been demonstrated by using specialized codingand multiple
electrodes simultaneously to induce current steering.10) On the
other hand, mid-infrared lighthas been shown to exhibit very
minimal scattering, and therefore does not spread laterally upon
incidence.11,12
Consequently the spatial resolution achievable by optical
stimulation could greatly improve the performance ofcochlear
implants beyond what is capable from even the best electrical
devices.13
Further author information: (Send correspondence to Matthew
Dummer)E-mail: [email protected], Telephone: (763) 746-8045
-
Figure 1. Reflectivity spectrum of the as-grown VCSEL wafer
compared with theoretical calculation
In vivo optical experiments thus far have been conducted in
laboratory animals using external infrared lasersto deliver the
stimulation signal through surgically implanted optical fibers.1,14
However for clinical applicationsin humans, such methods are not
practical. Therefore, implementing optical stimulation in neural
prostheses willrequire development of a miniaturized implantable
source. One especially suitable candidate is the vertical
cavitysurface emitting laser (VCSEL). VCSELs are type of diode
laser designed such that the optical beam is emittedorthogonal to
the wafer surface. These lasers offer small footprint, low power
consumption, high efficiency, andsimple packaging, all of which are
desirable for an implantable device. Also, their unique geometry
offers theability to be fabricated in 2-dimensional arrays for
increased output power, or addressing multiple
locationsindependently with a single chip.15
The goal of this work is to develop VCSELs specifically for
neurostimulation applications. The requiredspecifications for these
devices have been identified and a first round of VCSELs have been
fabricated andtested. Preliminary results demonstrate continuous
wave and pulsed lasing at the desired wavelength (1860 nm),with up
to 3 mW of output power. Power levels necessary for neural
activation threshold have not yet beendemonstrated, but the initial
measurements are an encouraging starting point for the future role
of VCSELsin neural prosthetics. This paper details the design and
characterization of these first VCSELs, and discusseschallenges
still ahead for this application.
2. REQUIREMENTS OF VCSELS FOR NEURAL PROSTHETICS
The physiologic mechanism responsible for optical activation of
the neuron is most likely attributed to directheating of the
tissue, rather than electric field interaction or photochemical
effects.14 That being the case,neural tissue activation does not
necessitate a precise wavelength. Rather, the initiation of the
action potentialrelies on the local temperature rise, and hence
total optical energy absorbed is the critical factor. Since
theabsorption coefficient of tissue varies as a function of
wavelength, selection of the wavelength can be used tospecify the
penetration depth of the optical signal. Wavelengths between 1840
nm and 1880 nm, correspondingto penetration depths from 1129 to 308
m, respectively, provide practical working distances for
stimulation.16
For the VCSEL development, we have chosen to target the center
of this range ( = 1860 nm, dp = 819 m).
The long wavelength of 1860 nm poses a significant challenge for
VCSELs due to fundamental materiallimitations. Furthermore, the
material composition of the VCSEL affects many other aspects of the
deviceperformance such as output power, temperature range, and
modulation rate. Besides wavelength, the greatestchallenge is
achieving high output power, since long-wavelength VCSELs have
traditionally been limited to afew milliwatts. Recently
large-aperture devices and multi-aperture arrays have been used to
achieve powers oftens to hundreds of milliwatts, respectively, at
1550 nm.17 Very little data has been reported on VCSELs near1860
nm,18 but similar results should be achievable.
-
Figure 2. Photograph of wafer with fabricated VCSEL die
3. VCSEL DESIGN AND FABRICATION
VCSELs are fabricated using wafer scale processes similar to
other optoelectronic devices. The material designis significantly
more complex than other diode lasers because all of the structures
comprising the laser mustbe integrated vertically. Though widely
commercially available at wavelengths between 800 and 1000 nm,
longwavelength VCSELs have required different materials platforms
that have taken much longer to develop. The keychallenge has been
finding semiconductor materials with the proper active-region
bandgap that are compatiblewith high index-contrast mirrors. For
our target wavelength of 1860nm, both indium phosphide and
galliumantimonide are possible substrate choices. Indium phosphide
is lattice-matched to various alloys emitting be-tween 1.0-1.7 m,
and strained materials can be incorporated to extend the emission
range beyond 2.0 m.19
Gallium antimonide has a wider range of alloy compositions,
although growth and fabrication techniques forthese materials are
not well established. Fabrication techniques for InP-based devices
are comparatively moremature, and commonly used for edge-emitting
lasers and high speed transistors. Significant efforts have
alsobeen made to develop InP-based VCSELs for telecommunications
and chemical sensing.20 Given the maturityof InP-based fabrication,
we have chosen to pursue this material platform for the 1860 nm
VCSEL development.However demonstrations of VCSELs emitting over
2.0 m on GaSb have also recently been reported, suggestingthat the
antimonide materials could be investigated for our application in
the future.21
Fabrication of the VCSELs begins with growth of the epitaxial
base structure by metal organic chemicalvapor deposition (MOCVD).
The layer stack consists of a gain region between two highly
reflective mirrorsto form a resonant optical cavity. Gain is
achieved at the target wavelength by incorporating
compressivelystrained InGaAs quantum wells in the active region to
lower the bandgap of the material. Tensile strainedInGaAsP barriers
compensate the wells to prevent relaxation of the lattice. Like
most VCSELs, the smalloverlap between the optical mode and the
active region necessitates very high mirror reflectivity (>99%)
toachieve lasing threshold. The lower mirror consists of an
epitaxially grown distributed Bragg reflector (DBR)with more than
30 periods of alternating high- and low-index InGaAsP. The upper
mirror is only partially grownepitaxially; an additional dielectric
coating is later deposited to increase the reflectivity. The layer
structure alsoincludes a tunnel junction above the active region to
improve lateral current spreading and reduce optical lossesdue to
p-doping. Reflectivity measurements of the wafers after growth
compare the theoretical design to theactual layer structure. Fig. 1
shows the wafer reflectivity compared with the simulated spectrum
obtained fromtransfer matrix calculations. The DBR stopband is
clearly shown from 1780nm to 1920nm with a reflectivity>99%. The
dip near the center of the stopband signifies the Fabry-Perot
resonance of the cavity, indicating thepreferential lasing
wavelength of the device (1842 nm). Photoluminescence experiments
were also performed toconfirm that the active region gain peak was
well aligned with the cavity resonance.
Post-growth device processing utilizes standard microelectronic
fabrication techniques. An optical photographof the fabricated
devices on the wafer is shown in Fig. 2. Annular metal contacts
surrounding each VCSELare defined to inject carriers into the laser
active region. The ring contacts are interconnected to
correspondingbondpads for flip-chip die bonding or wirebonding to
custom packages. Lateral current confinement within the
-
Figure 3. Schematic of VCSEL die containing a 2x2 VCSEL array.
Die dimensions are 350 x 350 x 250 m3 (LxWxH).
VCSEL is created by ion implantation, which reduces the
conductivity of the material outside the desired activeregion.
Deeply etched trenches are also used to electrically isolate
between adjacent VCSELs in an array. Thedielectric mirrors that
comprise part of the upper DBR are defined above the ring contacts,
and are terminatedwith a top metal reflector to boost reflectivity.
This necessitates that the VCSELs are bottom emitting, andtherefore
windows in bottom-side cathode metal are required for emission from
the subsrate. A schematic of thefinal die containing four
individually addressable VCSELs is depicted in Fig. 3. The die size
is 350 m per side.
Figure 4. VCSEL light output and operating voltage versus
applied current for implant aperture diameters between 10and 50 m.
Measurements were performed CW at 20 C
4. EXPERIMENTAL RESULTS
Initial device testing has been conducted at the wafer level.
Samples were placed on a copper stage to allowthermal and
electrical contact to the backside of the wafer. Anode contacts
were directly probed, and a hole in thestage allowed the emitted
light to be incident on a large area photodetector. Measurements of
light output versusapplied current and voltage (LIV) for VCSELs
with various active diameters are shown in Fig. 4. Measurementswere
taken continuous wave, with stage temperature controlled at 20 C.
Electrical characterization shows adiode turn-on voltage of 0.7 V
and the series resistance is inversely proportional to the current
aperture area.Continuous wave lasing was observed for all device
diameters between 10 and 100 m. Lasing threshold alsovaried as a
function of area, with the smallest devices exhibiting thresholds
as low as 1.4 mA. Figure 5(a)illustrates the trade-off between
series resistance and threshold current over the range of aperture
sizes. Thedifferential quantum efficiency at threshold was between
15-22% for all designs, although self heating resulted in
-
(a) (b)
Figure 5. (a) Measured series resistance and laser threshold
current versus aperture diameter. (b) Comparison of
LIVcharacteristic under CW and pulsed excitation for a 20 m
diameter VCSEL.
Figure 6. Continuous wave L-I measurements at various stage
temperatures for 12m and 20m diameter VCSELs.
decreased efficiency at higher currents. The 50 m aperture
devices exhibited the highest peak power, 2.9 mW at70 mA. Greater
than 1 mW of power was achievable from a 30 m device when biased at
15 mA. This operatingpoint corresponded to a peak wall plug
efficiency of 6%.
To isolate the effects of self-heating, pulsed measurements were
performed with 1 s pulses and 1% duty cycles,which is faster
modulation than necessary for neural stimulation. Figure 5(b) shows
pulsed versus continuouswave performance for a 20 m aperture VCSEL.
Under pulsed operation, peak power up to 4.0 mW at 50 mA
wasachieved (Fig. 4(b)). Slope efficiency is the same for CW and
pulsed operation. However the reduced thermalrollover increases the
maximum wall plug efficiency to 10% when operating in pulsed mode.
For longer pulselengths (>10 s), negligible increase in output
power was observed compared with CW due to the VCSELs shortthermal
time constant. The continuous wave output power is therefore a more
relevant measurement for ourtarget modulation rates. Effects of
external heating on the VCSEL output power have also been examined.
Fig.6 shows the LIV characteristic as a function of temperature for
12 and 20 m VCSEL designs. Both device sizesexhibit similar
temperture performance. Cooling the stage to 13 C results in a
significant increase in outputpower compared with room temperature
operation. Similarly, by heating the stage to 37 C, the peak
poweris reduced by about 50% compared with 20 C. An increase in
threshold current is also observed, owing to thereduction in gain
as the active region is heated. The maximum observed lasing
temperature was 60 C for bothdevices.
Wavelength measurements have been conducted by coupling the
output of the VCSEL into a multimode fiberand analyzing the signal
with a long-wavelength optical spectrum analyzer (OSA). The output
spectrum of theVCSEL is shown in Fig. 7(a). The device exhibits
single spectral mode operation with a peak wavelength of
-
(a) (b)
Figure 7. (a) Optical output spectrum of a 20 m VCSEL (b) Output
wavelength versus stage temperature for constantoperating
current
1859.6 nm. The spectral width is less than 0.1nm, limited by the
resolution of the OSA. Measurement of thewavelength at various
stage temperatures shows very stable operation over a wide
temperature range. The laserexhibits a linear red shift in
wavelength at a rate of 0.13 nm/C (Fig. 7(b)). The thermal tuning
rate is similarto the rate reported for other long wavelength
VCSELs.18
5. CONCLUSION
The recent discovery of optical neural stimulation could enable
new prosthetic devices for sensory impairedpatients. VCSELs look
especially promising as optical sources, and the long-wavelength
devices presented abovedemonstrate a first step toward implantable
devices. We have achieved CW and pulsed operation at the
desiredwavelength, and temperature operation up to 60 C. The
maximum CW power measured was 2.9 mW at 20 C.To our knowledge, this
is the highest continuous wave output power demonstrated for a
VCSEL at 1860 nm.A wall plug efficiency of 6% has also been
demonstrated at 1 mW of output power. Although power levels
toachieve neural activation have not yet been met, there do not
appear to be any fundamental roadblocks. Futurework will focus on
increased optical power, improvement in efficiency, and optimizing
the thermal performance.
ACKNOWLEDGMENTS
This material is based upon work supported by the Defense
Advanced Research Projects Agency (DARPA) underSPAWAR Systems
Center, Pacific (SSC PAC) Contract No. N66001-09-C-2008. The views,
opinions, and/orfindings contained in this article/presentation are
those of the author/presenter and should not be interpreted
asrepresenting the official views or policies, either expressed or
implied, of the Defense Advanced Research ProjectsAgency or the
Department of Defense
REFERENCES
1. A. Izzo, J. Walsh, E. Jansen, M. Bendett, J. Webb, H. Ralph,
and C. Richter, Optical parameter vari-ability in laser nerve
stimulation: a study of pulse duration, repetition rate, and
wavelength, BiomedicalEngineering, IEEE Transactions on 54(6), pp.
11081114, 2007.
2. S. L. Pinski and R. G. Trohman, Interference with cardiac
pacing, Cardiology Clinics 18(1), pp. 219 239, 2000.
3. K. McGill, K. Cummins, L. Dorfman, B. Berlizot, K.
Luetkemeyer, D. Nishimura, and B. Widrow, On thenature and
elimination of stimulus artifact in nerve signals evoked and
recorded using surface electrodes,Biomedical Engineering, IEEE
Transactions on (2), pp. 129137, 2007.
4. S. Norton, Can ultrasound be used to stimulate nerve tissue?,
BioMedical Engineering OnLine 2(1), p. 6,2003.
-
5. T. Wagner, M. Gangitano, R. Romero, H. Theoret, M. Kobayashi,
D. Anschel, J. Ives, N. Cuffin, D. Schomer,and A. Pascual-Leone,
Intracranial measurement of current densities induced by
transcranial magneticstimulation in the human brain, Neuroscience
letters 354(2), pp. 9194, 2004.
6. G. Allegre, S. Avrillier, and D. Albe-Fessard, Stimulation in
the rat of a nerve fiber bundle by a short UVpulse from an excimer
laser, Neuroscience letters 180(2), pp. 261264, 1994.
7. J. Wells, C. Kao, K. Mariappan, J. Albea, E. Jansen, P.
Konrad, and A. Mahadevan-Jansen, Opticalstimulation of neural
tissue in vivo, Opt. Lett 30(5), pp. 504507, 2005.
8. Q.-J. Fu, R. V. Shannon, and X. Wang, Effects of noise and
spectral resolution on vowel and consonantrecognition: Acoustic and
electric hearing, The Journal of the Acoustical Society of America
104(6),pp. 35863596, 1998.
9. Q. Fu and G. Nogaki, Noise susceptibility of cochlear implant
users: the role of spectral resolution andsmearing, JARO-Journal of
the Association for Research in Otolaryngology 6(1), pp. 1927,
2005.
10. D. Landsberger and A. Srinivasan, Virtual channel
discrimination is improved by current focusing incochlear implant
recipients, Hearing research 254(1-2), pp. 3441, 2009.
11. M. Niemz, Laser-tissue interactions: fundamentals and
applications, Springer Verlag, 2004.
12. A. Welch, Optical-thermal response of laser-irradiated
tissue, Plenum Press, 1995.
13. A. Izzo, E. Suh, J. Pathria, J. Walsh Jr, D. Whitlon, and C.
Richter, Selectivity of neural stimulation in theauditory system: a
comparison of optic and electric stimuli, Journal of Biomedical
Optics 12, p. 021008,2007.
14. J. Wells, C. Kao, P. Konrad, T. Milner, J. Kim, A.
Mahadevan-Jansen, and E. Jansen, Biophysical mecha-nisms of
transient optical stimulation of peripheral nerve, Biophysical
journal 93(7), pp. 25672580, 2007.
15. M. Hibbs-Brenner, K. Johnson, and M. Bendett, VCSEL
technology for medical diagnostics and therapeu-tics, in Proc.
SPIE, 7180, pp. 7180071810, 2009.
16. J. Walsh and J. Cummings, Effect of the dynamic optical
properties of water on midinfrared laser ablation,Lasers in surgery
and medicine 15(3), pp. 295305, 1994.
17. W. Hofmann, M. Gorblich, M. Ortsiefer, G. Bohm, and M.
Amann, Monolithic 2D high-power arrays oflong-wavelength VCSELs, in
Proc. of SPIE Vol, 6908, pp. 6908071, 2008.
18. M. Ortsiefer, R. Shau, G. Bohm, M. Zigldrum, J. Rosskopf,
and M. Amann, 90 C continuous-wave operationof 1.83-m
vertical-cavity surface-emitting lasers, Photonics Technology
Letters, IEEE 12(11), pp. 14351437, 2002.
19. H. Choi, Long-wavelength infrared semiconductor lasers,
Wiley-Blackwell, 2004.
20. M. Amann and M. Ortsiefer, Long-wavelength (? 1.3 m)
InGaAlAsInP vertical-cavity surface-emittinglasers for applications
in optical communication and sensing, physica status solidi (a)
203(14), pp. 35383544, 2006.
21. A. Bachmann, T. Lim, K. Kashani-Shirazi, O. Dier, C. Lauer,
and M. Amann, Continuous-wave op-eration of electrically pumped
GaSb-based vertical cavity surface emitting laser at 2.3 mm,
ElectronicsLetters 44(3), pp. 202203, 2008.