Optics Communications 273 (2007) 256–259

www.elsevier.com/locate/optcom

Temperature effects on the emission properties of Yb-doped

optical fibers
T.C. Newell a, P. Peterson a,*
, A. Gavrielides a, M.P. Sharma b

a
Directed Energy Directorate, 3550 Aberdeen Avenue, Kirtland AFB, NM 87117, United States
b
Physics Department, 800 Yale Blvd NE, University of New Mexico, Albuquerque, NM 87131, United States

Received 29 September 2006; received in revised form 9 January 2007; accepted 10 January 2007

Abstract

Ytterbium-doped fiber lasers are making impressive leaps in power production. Yet in spite of fiber’s large surface area to volume
ratio which is beneficial for cooling, such power inevitably leads to high core temperatures that in turn affect the laser performance.
In this paper, the temperature effects on the emission and fluorescence lifetime of ytterbium-doped optical fibers are investigated. From
these the temperature dependent emission and absorption cross-sections are calculated.
 2007 Elsevier B.V. All rights reserved.

Keywords: Spectroscopy; Lasers; Emission

1. Introduction the most ideal rare-earth element for lasing. Weak multi-
phonon decay is practically the only non-radiative channel
When ytterbium (Yb) is seeded in the amorphous silica that exists. Fig. 1 shows our energy level diagram for Yb
glass core of an optical fiber, the emission and absorption with a silica core. This diagram is inferred from our low
spectrum due to the subbands of the two manifolds, 2F5/2 temperature measurements which we discuss in the follow-
and 2F7/2, broaden beyond that which would be found in ing sections. The three levels of the upper 2F5/2 manifold
a crystalline structure like YAG [1]. While there is a large are denoted a, b, and c. The four levels of the lower 2F7/2
interest in cryogenic Yb:YAG lasers [2], with the exception manifold are labeled i, j, k, and l. Included in this diagram
to fiber temperature sensors [3,4] less research exists for are arrows indicating absorption lines at 977 nm, 960 nm,
Yb:silica fiber temperature effects. Such temperature effects 920 nm, and emission lines at 977 nm, 1020 nm, 1032 nm,
have generally not been of much concern although Grukh 1069 nm. The latter four lines originate from level (a) in
et al. [5] observed that improved long wavelength lasing the upper manifold as can be seen from the low tempera-
can be obtained at high temperatures. Furthermore, now ture emission spectra shown in Fig. 3. Note that this places
that demonstrated CW laser powers have exceeded 2 kW the four lower levels (i, j, k, l) at 0, 425 cm1, 530 cm1,
very high core temperatures may be reached. Here heating 865 cm1, respectively. Additionally, the absorption
inevitably occurs in spite of the fiber’s large surface area to cross-section derived via the McCumber formula shows
volume ratio. an absorption line at 960 nm. This then verifies the exis-
Compared to other rare-earth ions, Yb possesses a tence of the (j) level at 425 cm1. The other lower levels
remarkably simple atomic structure. With only two princi- are in the neighborhood of previous measurements [6–8].
ple manifolds that are separated by some 8750 cm1, it is The apparent discrepancy between these references is
resolved by realizing that the energy spacing depends on
*
Corresponding author. Tel.: +1 505 846 9301; fax: +1 505 846 1191. the constituent elements in the fiber core. Also broadening
E-mail address: [email protected] (P. Peterson). of the levels renders it difficult to establish their true values.

0030-4018/$ - see front matter  2007 Elsevier B.V. All rights reserved.
doi:10.1016/j.optcom.2007.01.027
 T.C. Newell et al. / Optics Communications 273 (2007) 256–259 257

reflections, and the fiber is short enough to minimize any

amplified spontaneous emission and radiative trapping.
Should radiative trapping occur, one observes a decrease
in the ratio of the spontaneous emission peak at 977 nm
to the peak at 1050 nm and an increase in the fluorescence
lifetime. For a cryogenic measurement, the doped fiber sec-
tion is bathed in liquid nitrogen cryostat. For high temper-
ature measurements, it is placed within a custom built
oven. The doped fiber is pumped with a low frequency
sinusoidally modulated 910 nm diode laser. The spontane-
ous emission propagating counter to the direction of the
pump is input into a Spectral Products DK480 monochro-
Fig. 1. Energy levels of Ytterbium:silica. mator and detected by a New Focus femtowatt photodetec-
tor. Prior to the experiments the detector was spectrally
calibrated using a white light source and calibrated inte-
Emission data obtained at cryogenic temperatures reveals grating sphere. The resulting voltage signal is measured
the dominant energy transitions, and in particular, an with a Stanford Research Systems lock-in amplifier that
energy level at 425 cm1. Additionally, the energy spacing is frequency locked to the laser diode modulation fre-
between levels a and b can be verified by curve-fitting to quency. Stray pump light is blocked using a 950 nm long
fluorescence lifetime data, as we show in Section 2.2. wave pass filter. A computer controls the acquisition
Lasing typically occurs from 1030 nm to beyond process.
1150 nm. Although the laser is considered as being quasi The setup for a fluorescence lifetime measurement uses
three-level, it is more of a four-level system at cryogenic the same coupler and fiber, but the fiber is pumped with a
temperatures when the thermally induced broadening of pulsed source at a low duty cycle. Backward propagating
the levels is minimized and when the thermal equilibrium light from the Yb-fiber was filtered with a 950 nm long wave
distribution of electrons render them frozen in the ground pass filter to remove the pump and coupled to a high gain
state. It also should behave more like a four-level system high bandwidth detector. The ensuing signal was captured
for very long lasing wavelengths when the state into which by an oscilloscope operating in a signal averaging mode.
the electron falls is well above the ground state.
This energy diagram is also seen in Yb heavy-metal fluo- 2.2. Fluorescence lifetime
ride glasses [10]. Experiments done by Lei et al. show a sim-
ilar diagram [10], except that long wavelength emission is Measurements were made between temperatures from
shifted to lower wavelengths by about 30 nm. Their motiva- 77 K to 443 K. Near 77 K, the observed lifetime is
tion was to determine the temperature dependence of both 0.885 ms. At 443 K, the lifetime decreases by 59 ls to
the homogeneous and inhomogeneous linewidths, and in 0.826 ms. The fluorescence lifetime as a function of temper-
particular, to resolve the low temperature linewidths. ature is shown in Fig. 2. The error bars are two standard
In this work a 910 nm diode laser pumps the doped deviations above and below the average as determined by
fiber. Consequently level c is not pumped. This is more rep- a linear curve fit to the log of the decay data. The dotted
resentative of typical fiber laser pumping schemes, and fur- line through the data points is a curve fit from a two-level
thermore the system can be modeled using only the two model [9]. By considering the decay from the lower two
upper levels [9]. In Section 2 the experiments are described thermalized levels of the 2F5/2 manifold to the ground state
from which the fluorescence lifetime and spontaneous emis-
sion are obtained. The emission cross-section is then com-
puted from the results. Conclusions are summarized in
Section 3.

2. Experimental investigations

2.1. Experimental arrangement

The fiber used is commercially available dual-clad Yb-

doped fiber manufactured by OFS (part #Comcode 107
986 820). The 6 lm diameter silica core is Yb-doped. A
210 lm diameter silica cladding follows, and the outer
polymer jacket is 300 lm. A 0.65 mm section of the doped
fiber is fusion spliced to one arm of a single mode 2 · 2 Fig. 2. The fluorescence lifetime as a function of temperature. The dotted
coupler. The end is angle cleaved to suppress back line traces a curve fit using a two-level model.
258 T.C. Newell et al. / Optics Communications 273 (2007) 256–259

of the 2F7/2 manifold, the lifetime can be modeled by

s = (1 + eDE/kT)/(1/sa + 1/sbeDE/kT). Here sa is the life-
time of the lowest level, a, and sb is that of level b. DE is
the energy separation between the two states. When this
function is fit to our data the energy separation between
the lowest two 2F7/2 levels is 630 cm1. This value matches
that which is obtained from observing the emission peaks
seen in Fig. 3. We find that sa = 0.884 ms and sb = 0.547
ms. In fibers that have been subjected to very high temper-
atures, the lifetime has been observed to decrease very rap-
idly beyond activation energy [3]. We did not operate the
fiber at high enough temperatures to observe this quench- Fig. 4. Gain cross-section computed using the measured fluorescence
ing effect. lifetime and spontaneous emission data.

2.3. Spontaneous emission

Spontaneous emission was measured at temperatures of

77 K, 293 K (20 C), 323 K (50 C) 383 K (110 C), and
443 K (140 C). The emission data is shown in Fig. 3 The
77 K spectra shows emission at 1020 nm, 1032 nm, and
1069 nm. These lines show that there are levels at
425 cm1, 530 cm1, and 865 cm1 above the ground state
(i). This is the first time, as far as we know, that a level near
425 cm1 has been identified in Yb silica, although mea-
surements in low temperature Yb phosphates indicate this
same level [11]. At high temperatures the 1019 nm plateau Fig. 5. The absorption cross-section computed using experimentally
and 1032 nm peak coalesce into a smooth broad rise, which measured emission data along with the McCumber transform.
is due to the broadening of the states, particularly level a.
The 977 nm peak decreases substantially due in part to 77 K, 293 K, 323 K, 383 K, and 443 K. Again, in Fig. 4
broadening around the shoulders. And as the thermaliza- we see the four dominate emission lines at 978 nm,
tion of electrons populates the 2F5/2 manifold’s higher 1020 nm, 1032 nm, and 1069 nm at the low temperature
energy levels, absorption of the 910 nm photon decreases. of 77 K. While it is clear that an increase in emission at
low temperatures should lead to a greater emission cross-
2.4. Gain cross-section section, the expected benefit is not realized in its entirety
since the emission lifetime is also increasing. In fact,
Using the work of McCumber [12], Moulton [13] derives beyond 1035 nm, the data shows very little change in
the relationship between the measured emission spectra cross-section with temperature. As the absorption
s(k), seen in Fig. 3, and the emission cross-section. They decreases with increasing wavelength, the net gain portends
find that r(k) = k5(8pcn2s)1 s(k)hki1 where hki is the inte- excellent high temperature performance at long wave-
gral of the product ks(k), and r(k) is the emission cross-sec- lengths. Indeed, this has been observed in Ref. [5].
tion. This equation applied to the data in Fig. 3 yields McCumber formula [12] applied to the data in Fig. 4
Fig. 4 for the emission cross-section at temperatures of generates the absorption cross-section shown in Fig. 5.
The low temperature cross-section shows a weak but dis-
cernable absorption peak near 960 nm which grows as
the temperature increases. This corresponds to the (j–b)
transition and again establishes the DEj,i = 425 cm1 and
the DEb,a = 630 cm1 levels. We note that in generating
the absorption spectra the zero photon line is shifted by
about 1% between 77 K and 443 K.

3. Summary

In this paper we use temperature as an investigative tool

to determine spectroscopic characteristics of moderately
doped Yb:silica fiber. Spontaneous emission and fluores-
Fig. 3. Spontaneous emission spectra for temperatures ranging from 77 K cence lifetime measurements are used to calculate the emis-
to 443 K. sion cross-section as a function of temperature and to
 T.C. Newell et al. / Optics Communications 273 (2007) 256–259 259

provide insight into the evolution of the emission lineshape. References

Cryogenic emission measurements yield insights into the
lower 2F7/2 manifold. While fluorescence lifetime studies [1] S.Z. Zhao, A. Rapaport, J. Dong, B. Chen, P.Z. Deng, M. Bass, Opt.
indicate the energy split between the lower two levels of Mater. 27 (2005) 1329.
[2] S. Sumida, T.Y. Fan, OSA Proc. Adv. Solid State Lasers 20 (1994)
the upper 2F5/2 manifold. 100.
A substantial amount of homogenous and inhomoge- [3] T. Sun, Z.Y. Zhang, K.T.V. Grattan, A.W. Palmer, Rev. Sci.
neous broadening mechanisms remain at the cryogenic Instrum. 69 (1998) 4179.
temperatures we were ale to reach. Still observed peaks in [4] E. Maurice, S.A. Wade, S.F. Collins, G. Monnom, G.W. Baxter,
the spontaneous emission spectrum at 1020 nm, 1032 nm Appl. Opt. 36 (1997) 8264.
[5] D.A. Grukh, A.S. Kurkov, V.M. Paramonov, E.M. Dianov, Quan-
and 1069 nm in the emission indicate the energy levels of tum Electron. 34 (2004) 579.
the lower 2F7/2 manifold. As for the fluorescence lifetime, [6] H.M. Pask, R.J. Carman, D.C. Hanna, A.C. Tropper, C.J. Mack-
when the temperature increases from 77 K to 443 K a small echnie, P.R. Barber, J.M. Dawes, IEEE Select. Topics Quantum.
decrease of 8%, or roughly, 0.00013 ms/K occurs. This Electron. 1 (1995) 2.
[7] J.Y. Allain, M. Monerie, H. Poignant, Electron. Lett. 28 (1992)
decrease can be successfully explained by emission from a
988.
thermalized two upper level model with levels separated [8] D.C. Hanna, R.M. Percival, I.R. Perry, R.G. Smart, P.J. Suni, A.C.
by 630 cm1. Absorption spectra not presented here also Tropper, J. Mod. Opt. 37 (1990) 517.
confirmed this separation. [9] Z. Zhang, K.T.V. Grattan, A. W Palmer, J. Appl. Phys. 73 (1993)
At high temperatures, the evolution of the cross-section 3493.
shows that temperature insensitive lasing should occur for [10] G. Lei, J.E. Anderson, M.I. Buchwald, B.C. Edwards, R.I. Epstein,
Phys. Rev. B 57 (1998) 7673.
longer wavelengths than the typical 1050–1080 nm wave- [11] S. Dai, A. Sugiyama, L. Hu, S. Liu, G. Huand, Z. Jiang, J. Non-
lengths demonstrated. There are a number of applications Crystalline Solids 311 (2002) 138.
where such qualities are highly desirable, and such a laser [12] E. McCumber, Phys. Rev. 134 (1964) A299.
is the focus of future work. [13] F. Moulton, J. Opt. Soc. Am. B 3 (1986) 125.