Evaluation of the frequency stability of a VCSEL locked to a micro- fabricated Rubidium vapour cell
J. Di Francesco, F. Gruet, C. Schori, C. Affolderbach, R. Matthey, G. Mileti Laboratoire Temps – Fréquence (LTF), Institute of Physics
University of Neuchatel, Avenue de Bellevaux 51, CH – 2009 Neuchatel, Switzerland Y. Salvadé
Laboratory of Metrology and Quality Engineering, University of Applied Sciences, Haute Ecole Arc Ingénierie, St-Imier, Switzerland
Y. Petremand, N. De Rooij
Ecole Polytechnique Fédérale de Lausanne (EPFL), Sensors, Actuators and Microsystems Laboratory, Rue Jaquet-Droz 1, 2000 Neuchâtel, Switzerland
ABSTRACT
We present our evaluation of a compact laser system made of a 795 nm VCSEL locked to the Rubidium absorption line of a micro-fabricated absorption cell. The spectrum of the VCSEL was characterised, including its RIN, FM noise and line-width. We optimised the signal-to-noise ratio and determined the frequency shifts versus the cell temperature and the incident optical power. The frequency stability of the laser (Allan deviation) was measured using a high-resolution wavemeter and an ECDL-based reference. Our results show that a fractional instability of ≤ 10-9 may be reached at any timescale between 1 and 100’000 s. The MEMS cell was realised by dispensing the Rubidium in a glass-Silicon preform which was then, sealed by anodic bonding. The overall thickness of the reference cell is 1.5 mm. No buffer gas was added. The potential applications of this compact and low-consumption system range from optical interferometers to basic laser spectroscopy. It is particularly attractive for mobile and space instruments where stable and accurate wavelength references are needed.
Keywords: Vertical cavity surface emitting laser (VCSEL), micro-fabricated absorption cell, Rubidium, laser frequency stabilization, Allan deviation, spectroscopy
1. INTRODUCTION
Stabilization of the output frequency of a semiconductor laser source [1] to reference etalons [2] or atomics reference lines [3, 4] has become a well-established tool in a wide field of research topics, and is also implemented in applications such as, e.g., atomic clocks and atomic magnetometers [5], interferometry [6], and wavelengths references [7].
In this paper, we investigate the frequency stability of a new frequency standard composed of a vertical-cavity surface- emitting laser (VCSEL) locked to a Doppler-broadened absorption resonance on the D1 line of atomic 85Rb, using either a micro-fabricated or a traditional cm-scale glass vapour cell. VCSEL diode lasers have the advantages of low power consumption, intrinsic mode-hop free single-mode operation, and a good potential for cost-effective mass production, making them ideal candidates for the realization of a compact, stabilized wavelength source. The Rb absorption line obtained from the vapour cell serves as a reference with good long-term stability, thanks to its small frequency shifts in response to external fields and operating conditions [8]. Recently, progress of silicon machining and anodic-bonding have resulted in the creation of micro-fabricated absorption cells filled with alkali vapour that allow a more radical miniaturization of the whole setup. Similar laser frequency stabilization techniques are employed in micro-fabricated atomic clocks and magnetometers [5], but not detailed studies of the obtained laser frequency stability have been published.
Published in
Semiconductor Lasers and Laser Dynamics IV (Proceedings of SPIE) 7720, 77201T, 1-9, 2010 which should be used for any reference to this work
1
2.1 Experim We use a VC 1.97 mA. Op coating for th the transmitte
Figure 1 –
Laser stabiliz traditional cy fabricated cel density at roo micro-fabrica by a feed-bac 1.4 mm.
The error sign lock-in detect WSU/2). In a measurement changes in ba
2.2 MEMS C The main step cavity etching
mental setup CSEL laser dio ptical feedback he lens which ed light power
– Experimental
zation to two c ylindrical glas
ll (see section om temperatur ated and cm-s ck regulated th
nal for laser fr tion. The VCS a typical mea t quality all th ackground ligh
Cell fabricati ps of the vapo g by photolith
ode emitting k on the VCS h collimated th r in the laser b
setup: frequenc
cells with diff s cell with 10 n 2.2). Both c re is low for b cale glass cel hermostat. Th frequency stab SEL frequenc asurement run he setup is pla ht in the test r
ion
our cells fabri hography. For
2. E
around 795 n SEL is limited he laser beam beam is detect
cy stability is m
= m
ferent sizes is 0 mm diamete cells contain a both cells resu ls are heated he typical opti bilization is ob cy measureme n the frequenc aced under an room.
ication techno this step, a si
EXPERIME
nm, operated a d thanks to the m (not show in ted on a Si pho
measured using mirror, BS = bea
s studied, usin er and 20 mm atomic Rubidi ulting in too w to 90 °C and ical power inc btained by mo
nt is measured cy is recorded opaque box w
ology is descr licon dioxide
ENTS
at a temperatu e use of an op n Figure 1). A
otodiode detec
a high precision m splitter.
ng the same se m length and t
ium in natural weak absorptio 55 °C, respec cident to the c odulation of th
d with a high d each 0.15 s which protect
ribed in Figur is grown on t
ure of 59.5°C ptical isolator After passing t
ctor.
n wavemeter. IS
etup and simil the second cel l isotope mixt on signals. To ctively. The ce cell is 250 µW he VCSEL cur
precision wav second during ts to against te
e 2. First, a si the surface. Th
and an inject r and an antire through the R
SO = optical iso
lar conditions ll is a 1.5 mm ture. The Rub o increase the ell temperatur W and the bea rrent at 50 kH vemeter (High g one day. To emperature flu
ilicon wafer i hen, after pho
tion current o eflection (AR) Rb vapour cell
olator, MR
s. One cell is a m thick micro
bidium vapour DC signal the re is stabilised am diameter is Hz and using 1
h-Finesse type o improve the uctuations and
s prepared for otolithography
f ) l,
a - r e d s f e e d
r y,
2
the wafer is dimension. T bonded wafe dimension tha The obtained and a high v dispensing po amount of Rb lid, mounted sealing of the includes the f heated up to that temperat bonding volta enabling a m diffusion, an
Figure 2 – Preparatio Photolitho bonding o preform.
cavity. f) C
etched by D The obtained h ers are then d
an the preform d chips (prefor
vacuum is ma osition, below b is deposited on a piston, i e chip by anod following proc about 300 °C ture, the sodiu age. A depleti echanical con anodic oxidat
– Process flow on of a silicon w ography and cav of the Si wafer
e) Silicon sur Cell bonding w
DRIE (Deep R hollowed waf diced to obtain ms. From this
rm and lid) ar ade inside the w a system bas into the prefo is finally mov dic bonding. T cess; once the C and a voltag um ions within
ion zone is cre ntact on the at tion forms a v
of the cell fabr wafer (500, 100
vity etching by with a glass wa face preparatio ith a glass lid b
Reactive Ion fer is then bon
n preforms. I point, the fabr re placed into e chamber. T sed on a com orm cavity, the
ved in contact The anodic bo e silicon prefo ge of 600 V or
n the borosilic eated in which tomic scale be ery strong bon
rication (cross s 00, 1500 or 200 y DRIE. c) Wa
afer. d) Dicing on and Rb de by anodic bondi
Etching) etch nded to a glas In parallel, a rication is con the deposition The preform,
mercially ava e preform is m t to the Rb fil onding techniq orm and the gl
r more is app cate glass mat h a voltage dro etween the tw
nd between th
section view). a 00 µm thick). b afer-level anodi of the obtained eposition in th
ing.
hing in order ss (borofloat) glass wafer i ntinued at chip n machine as placed on a ailable Rb disp moved to the b
lled preform t que is based o lass lid are in lied between trix become m op occurs. Thi wo surfaces to he glass and th
a) b) c d e
Figure 3 – closing ma is installed the perform horizontall
to obtain thr wafer by ano is diced to fo p level.
seen in Figur movable trol penser from S bonding positi
to close the ca on the work of contact at a g the glass lid mobile and can
is effect create be bonded. B he silicon surfa
– Picture of th achine. The gla d on a verticall
m (upper left y movable carr
rough holes o odic bonding.
orm glass lids re 3. The cham lley is displac SAES getters.
ion, below the avity and obta f Wallis and P given pressure
and the silico n be moved b es a large elec By field-assiste
faces.
he Rb depositio ss lid (upper ri ly movable sy corner) is mou rier
of the desired The obtained s of the same mber is closed ced to the Rb Once a smal e glass lid. The
ain a hermetic Pomerantz and , the system is on preform. A
y the negative ctrostatic force ed oxygen ion
on and cell ight corner) stem while unted on a
d d e d b ll e c d s At e e n
3
2.3 Laser pr The VCSEL Rubidium, th while the cur have measure density of fre 45 ± 3 MHz, This result wa
1 For this w this choice is n other manufact and found simi
roperties laser source he VCSEL is
rrent and temp ed the frequen equency fluctu , from a beat
as confirmed
ork we used a not critical for turers. In partic ilar laser charac
is from Ava operated at 59 perature tuning ncy noise spec uations at 1 kH
measurement in a second m
795nm single-m the results rep cular, we also e cteristics as the
alon1, emittin 9.5 °C and 1.
g coefficients ctrum shown Hz is approxi against an ex measurement u
Figure 4 – PI
mode VCSEL f ported here, and evaluated 795n
ones reported h
g around 795 97 mA. At th are -122.4 G in Figure 5, u imately2.5⋅10 xtended-cavity using a Fabry-
curve and tunin
from Avalon Ph d similar result nm single-mode here for the Ava
5 nm. To ach his operating p Hz/mA and -3 using a discrim
Hz Hz2
09 . Th
y diode laser Pérot interfero
ng coefficients.
hotonics (now B s can be expec e VCSEL from
alon device.
hieve laser em point the total 31.8 GHz/K, r minator slope he VCSEL lin (ECDL) havin ometer with a
Bookham Inc.), cted when using
ULM Photonic
mission on th l output powe respectively (
of the 85Rb li newidth was m
ng a linewidth a resolution of
, emitting at 79 g a single-mod cs (type 795-01
he D1 line o er is 0.25 mW (Figure 4). We ine (F=3). The measured to be h of 200 kHz f 1 MHz.
95nm. However de VCSEL from
1-TN-S46FOP) f W e e e z.
r, m ), 4
Figure 5 – F
2.4 Freque In a first step obtained with spectral densi
VCSEL
υ [10]:
D is obtained transmission VCSEL frequ
Frequency noi
ency stability p, we calculate
h our setup. Eq ity NPSD meas
d form a line signal through uency across t
ise spectrum o
y
ed the theoret quation (1) gi sured on the d
ear fit to the h the cell (bla the Rubidium
of the VCSEL
tical short-term ives the theore detector behin
) (τ σth error signal c ack trace) and
D1 absorption
L when operate
m frequency s etical limit wh
d the Rb cell,
2⋅ ⋅
= υ
N
VCSEL PSD
close to its ze the correspon n line.
ed at a temper
stability - in te hich is express the discrimin
2
−1
⋅τ D ero crossing.
nding error sig
rature of 59.5
erms of Allan sed in terms o nator slope D
Figure 6 sho gnal (grey trac
°C and a curr
n Deviation [9 of the detectio and the VCS
ows an examp ce) obtained by
rent of 2 mA.
] - that can be n noise power SEL frequency
(1 ple of the DC
y scanning the e r y
) C e
5
Figure 7 show function of av Table 1 and t a small differ term stability long-term dri (ECDL) [3], resolution bei
P D Detec Theoretica
T
ws the measu veraging time the experimen rence between y. From 10 sec ift of the wav achieving σE ing σWM(τ)≈2
Figure 6 –
Parameter Discriminator
ction noise PSD al frequency sta
Table 1 – Short-
ured fractiona e. First, we ca ntal values are n the stabilitie conds the mea vemeter used.
1012
) (τ ≤ −
ECDL 10 10
2⋅ − at τ = 1
Absorption and
U V D mVrm
ability
-term frequency
al frequency s an see that the perfectly mat es obtained w asured Allan D . This is conf
2 on all time 1 s, and its sta
d error signal fo
Unit cm
V/GHz ms/sqrt(Hz)
τ -1/2
y stability predi
tability of the theoretical es tched, up to in with two cells,
Deviation inc firmed by a s scales consid ability σWM(τ)
or the micro-fab
m-scale glass 3.26 0.768
10 10
41 .
4 ⋅ −
iction according
e stabilized V stimations of ntegration tim the cm-scale reases with th separate meas dered here. Th
10 9
1
)≈ ⋅ − at τ =
bricated cell.
cell µ-fabri 1 0 04 . 5
g to equation (1
VCSEL (in ter for the short-t mes of τ = 10 s
e glass cell giv he same tende surement of a his measureme
= 1000 s.
icated cell 1.83 0.431
10 10
4⋅ −
1).
rm of Allan d term laser stab s. Furthermore
ving a slightly ency for both a highly stable ent confirms t
deviation) as a bility given in e, there is only y better short cells, due to a e laser source the wavemeter a n y - a e r
6
Figure 7
2.5 Frequen Two main so Rb absorption observed freq cell temperatu µW around t results are sho
cm Micro
The temperat for the Rb re VCSEL. The frequency ins limitation to t at τ = 104 s, w
7 – Relative freq
ncy shift versu ources of syste n lines [8]: th quency stabilit
ure was varied the operating own in Table
Cell S
m-scale o-fabricated
Table 2 –
ture stability o ference line. T e typical fract
stability contr the laser stabi which is howe
quency stability
us cell tempe ematic freque e cell tempera ty and therefo d by ±1°C aro
point. In bot 2.
Shifts accordin 0.6 MH 1 MH
– Shift of the st
of the Rb cells This shift con tional in insta ribution of σV ility reported h ever not resolv
y of the VCSEL
rature and th ency shifts hav ature and the ore their impac
ound the oper th cases, the
ng to the cell t Hz/°K ± 1 MHz Hz/°K ± 1 MHz/
abilized laser fr
s at 104 secon ntributes an am
ability of the
VCSEL ≈ 8x10- here. In total, ved in our exp
L frequency stab Deviation
he incident op ve to be taken optical power ct on the stabi ating point. In resulting freq
temperature z/°K /°K
frequency in res
nds is approxim mount σVCSEL
optical powe
-11. The fluct we expect the periment due to
bilized to the to n.
ptical power n into account r incident on t ilized laser fre n a second ste quency shift w
Shifts accord
sponse to chang
mately 10 mK
≈ 3x10-11 to t er is <6x10-4 tuations in op e frequency in
o the insuffici
o the two Rb cel
t when stabili the cell. These equency was m ep, the optical
was measured
ding to the inc
< 200 kHz/
< 200 kHz/
ges in experimen
K, leading to a the fractional
at 104 s, wh ptical power p nstability to be
ient stability o
lls, in terms of
izing to Dopp e factors may measured. In a power was va d with the wa
cident optical /µW
/µW
ntal parameters
a frequency sh frequency ins hich correspon present the m e limited to σV
of the waveme
the Allan
pler-broadened well limit the a first step, the aried by a few avemeter. The
power
s.
hift of 10 kHz stability of the nds to a laser most importan
VCSEL ≤ 9x10-1 eter.
d e e w e
z e r nt 1
7
3. CONCLUSIONS
Our results show that Doppler-broadened Rb absorption lines obtained from micro-fabricated vapour cells of mm-scale dimensions can be used to stabilize the frequency of a VCSEL diode laser to the level of σVCSEL = 5x10-10 at τ=1 s. This stability is essentially the same as obtained with a classical cm-scale vapor cell and could be obtained in spite of the relatively large emission linewidth of the VCSEL. Our stability results compare well to the stabilities reported or predicted for similar schemes, but without the need for stringent requirements on cell thickness and diameter [11, 12] or additional pump laser beams [13, 14]. The simplicity of the used setup and small size of the few key components used opens the way for the realization of a low-power frequency-stabilized laser source with an overall physics package volume of only a few cm3.
We finally note that we observed similar frequency stability as reported above also for a VCSEL laser emitting at 780nm, stabilized to the Rb D2 absorption lines obtained from the same cells. This extends the wavelength range of the stabilized laser, without need to change the reference cell.
4. OUTLOOK AND APPLICATIONS POTENTIALS
Already with its present stability performance, the demonstrated frequency-stabilized VCSEL laser is appropriate for applications in miniaturized precision instrumentation requiring a stabilized frequency reference, such as, e.g., atomic clocks and atomic magnetometers [5]. Further improvement in short-term frequency stability may be expected when stabilizing the VCSEL to Doppler-free saturated-absorption lines that still can be resolved in spite of the relatively large VCSEL linewidth [12, 14]. Preliminary experiments on saturated-absorption lines observed with our setup show that the short-term frequency stability reaches σVCSEL = 2x10-10 at τ=1 s, which coincides with the resolution limit of the wavemeter used.
5. ACKNOWLEDGEMENTS
We acknowledge financial support from the Swiss National Science Foundation (Subsidy 200020-118162 and Sinergia program grant no. CRSI20_122693/1), the European Space Agency (ESTEC contracts 20794/07/NL/GLC and 19392/05/NL/CP), and the Swiss Space Office. We thank P. Scherler (LTF) for technical support.
REFERENCES
[1] Hall, J. L., Taubman, M. S., and Ye, J, “Laser stabilization”, in : [Handbook of optics], vol.4, 2nd edition, McGraw-Hill, New York (2000).
[2] Drever, R. W. P., Hall, J. L., Kowalski, F. V., Hough, J., Ford, G. M., Munley, A. J., and Ward, A., “Laser Phase and Frequency Stabilization Using an Optical Resonator”, Appl. Phys. B 31, 97 – 105, (1983).
[3] Affolderbach, C., and Mileti, G., “A compact laser head with high-frequency stability for Rb atomic clocks and optical instrumentation,” Rev. Sci. Instrum. 76, 073108, (2005).
[4] Ye, J., Swartz, S., Jungner, P., and Hall, J. L., “Hyperfine structure and absolute frequency of the 87Rb 5P3/2
state”, Opt. Lett. 21(16), 1280 – 1282, (1996).
[5] Knappe, S., Schwindt, P. D. D., Gerginov, V., Shah, V., Liew, L., Moreland, J., Robinson, H. G., Hollberg, L., and Kitching, J., “Microfabricated Atomic Clocks and Magnetometers”, J. Opt. A 8, S318 – S322, (2006).
[6] Salvadé, Y. and Dändliker, R., “Limitations of interferometry due to the flicker noise of laser diodes”, J. Opt.
Soc. Am. A 17(5), 927 - 932 (2000).
[7] Têtu, M., Cyr, N., Villeneuve, B., Thériault, S., Breton, M., and Tremblay, P., “Towards the Realization of a Wavelenght Standard at 780nm Based on a laser diode Frequency Locked to Rubidium Vapor“, IEEE Trans.
Instrum. Meas. 40(2), 191 - 195, (1991).
8
[8] Affolderbach, C., Mileti, G., Slavov, D., Andreeva, C., and Cartaleva, S., “Comparison of simple and compact Doppler and sub-Doppler laser frequency stabilization schemes”, Proceedings of the 18th European Frequency and Time Forum (EFTF), The Institution of Electrical Engineers, London, paper 084 (2004).
[9] Allan, D. W., “Time and Frequency (Time-Domain) Characterization, Estimation, and Prediction of Precision Clocks and Oscillators,” IEEE Trans. Ultrason., Ferroelec. Freq. Contr. 34(6), 647 - 654 (1987).
[10] Mileti, G. and Thomann, P., “Study of the S/N Performance of Passive Atomic Clocks using a Laser Pumped Vapor”, Proceedings of the 9th European Frequency and Time Forum, Société Française des Microtechniques et de Chronométrie, Besançon, 271 - 276, (1995)
[11] Zhao, Y. T., Zhao, J. M., Huang, T., Xiao, L. T., and Jia, S. T., “Frequency stabilization of an external-cavity diode laser with a thin Cs vapour cell”, J. Phys. D 37, 1316 – 1318 (2004).
[12] Fukuda, K.,Tachikawa, M., and Kinoshita, M., “Allan-variance measurements of diode laser frequency- stabilized with a thin vapor cell”, Appl. Phys. B 77, 823 – 827 (2003).
[13] Knappe, S. A., Robinson, H. G., and Hollberg, L., “Microfabricated saturated absorption laser Spectrometer”, Opt. Express 15(10), 6293 – 6299 (2007).
[14] Affolderbach, C., Nagel, A., Knappe, S., Jung, C., Wiedenmann, D., and Wynands, R., “Nonlinear
Spectroscopy with a Vertical-Cavity Surface-Emitting Laser (VCSEL)”, Appl. Phys. B 70, 407 – 413 (2000).
9