8.7 Experimental Walk-Through: Temperature Measurement Using the Detuned Probe
9.1.3 Experimental Parameters: What to Expect
vibrational mode, and is given by the ratio of the antinode displacement and the displace- ment of the membrane surface S averaged over the intensity profile of the cavity mode:
ηij≡R
S|ψmn(x, y, z)|2φ(x, y, z)~ ·zdσ/ˆ R
S|ψmn(x, y, z)|2dσ (see Section 7.1.3).
In the MIM system, the generalized radiation force experienced by coordinatebij is given by:
Fradij (t) =−gijUc(t)/ωc =ηijgm(zm)Uc(t)/ωc. (9.16)
By direct analogy to the two mirror system, the thermal noise spectrum for a single, underdamped vibration of the membrane in the presence of a weak optical spring is given by (assuming the membrane mode in question is well isolated in frequency from other modes):
Sbij(Ω)≈4kBTbΓij(Ω) mij
1
((Ωij+ ∆Ωijopt)2−Ω2)2+ Ω2(Γij(Ω) + Γijopt)2 (9.17a) Γijopt= 2hnci(gijbijzp)2
κ
1
1 + (∆ + Ωij)2/κ2 − 1
1 + (∆−Ωij)2/κ2
(9.17b)
∆Ωijopt≡=hnci(gijbijzp)2 κ
(∆ + Ωij)/κ
1 + (∆ + Ωij)2/κ2+ (∆−Ωij)/κ 1 + (∆−Ωij)2/κ2
(9.17c) Tef fij ≈mij
kB
Z ∞ 0
Ω2Sbijbij(Ω)dΩ/2π≈ Γij
Γij+ ΓijoptTbath, (9.17d) where hnci ≡ h|a|2i/~ωc is the generalized intracavity photon number and bijzp = p
~/2mijΩij is the zero-point displacement. To deal with multimode thermal noise, we here retain the frequency dependence of the Γij(Ω) associated with structural damping: Γij(Ω) = Γij(Ωij)Ωij/Ω = Ω2ij/QijΩ (Section 7.3.1.2). The integration R
Ωij is here understood to exclude off-resonant displacement of other vibrational modes for the multimode resonator.
values very with the exact position of λ on the mirror coating curve). Variables {Ωm, Qm} are the typical frequency and mechanical quality of a higher-order mode of the membrane, which has material properties{dm, wm,T, ρ}, corresponding to thickness, square width, tension, and density.
9.1.3.1 Intracavity Photon Number
We first compute the intracavity photon number, which depends on the steady-state input and output power,{hPini,hPouti={h|Ein|2i,h|Eout|2i}(we here assume that the input field is monochromatic), and the magnitude of cavity decay rates{κ, κ1, κ2}. With the membrane removed, our Fabry-Perot cavity is nearly symmetric and lossless: κ1≈κ2≈κ/2 (Section 6.2). However, with the membrane in the cavity, decay rates{κ, κ1, κ2} are all functions of the membrane position. The steady-state intracavity photon number in this more generic case is given by solving Eq. 9.15 with ˙a = 0 and bij = 0 (Eq. 8.5):
hnci=h|a|2i
~ωc =hPouti
~ωcκ κ
2κ2 = hPini
~ωcκ 2κ1
κ 1
1 + ∆2/κ2 (9.19a)
hPouti=hPout(∆ = 0)i
1 + ∆2/κ2 . (9.19b)
The ratio 2κ1/κcan be determined from the resonant transmission/reflection of the MIM cavity (Eq. 8.5). Assuming negligible internal loss, so thatκ1+κ2=κ:
hPref(∆ = 0)i
hPini = hPini − hPout(∆ = 0)i hPini =
2κ1
κ −1 2
(9.20a)
→ 2κ1
κ = 2−2κ2 κ =
s
1−hPout(∆ = 0)i
hPini + 1. (9.20b)
The magnitude ofhPout(∆ = 0)i/hPinidepends on the position of the membrane in the cavity. It can be computed numerically by the method discussed in Section 3.4, and has been measured for our
“science” cavity as discussed in Section 6.5. For our low-loss, 50-nm-thick membrane with reflectivity rm= 0.42 at 935 nm, the magnitude ofhPout(∆ = 0)i/hPinivaries between 0.8 and 1.0 and has a magnitude of ≈0.9 when the membrane is positioned halfway between a node and an antinode of the intracavity field. At this position, for which gm= 2rmg0, we find 2κ1/κ≈(gm/g0)2≈0.7 and 2κ2/κ≈2−(gm/g0)2≈1.3.
For relevant cavity parameters, we have:
hnci= 0.750×106×
hPini 10µW
λ 935 nm
10 MHz κ/2π
2κ1 κ
1 1 + ∆2/κ2
(9.21a)
= 0.750×106×
hPout(∆ = 0)i 10µW
λ 935 nm
10 MHz κ/2π
κ 2κ2
1 1 + ∆2/κ2
. (9.21b)
It’s useful to use the quantity hPout(∆ = 0)i because in the lab we can measure this number
without worrying about input mode-matching. Also, for a perfectly mode-matched, symmetric cavity,hPout(∆ = 0)i=hPini.
9.1.3.2 Effective Mass and Zero-Point Amplitude
The vibrational modes of the membrane were discussed in detail in Sections 4.2 and 7.1.1. The effective mass of generalized displacement amplitude bij coinciding with the displacement of an antinode on the membrane surface is given bymij =mphys/4. Thus we have, for the effective mass and the zero-point amplitude:
mij = 8.44 ng× ρ 2.7 g/cm2
! dm
50 nm
wm
500µm 2
(9.22a)
bijzp = s
~
2mijΩij = 4.45×10−16 m×
8.44 ng mij
1/25 MHz Ωij/2π
1/2
. (9.22b)
9.1.3.3 Optomechanical Coupling and Spatial Overlap
For a nearly lossless 50-nm-thick film at 935 nm, the reflection coefficient isrm= 0.42. Using Eq.
3.22 derived forgm(zm) and defining the vibrational amplitudeb as the amplitude of an antinode, we have for a single membrane mode (dropping the indicesiandj):
gijm≡ δωc bij
=ηijgm(zm) =ηij× −2|rm|sin(2kz)
p1− |rm|2cos2(2kz)×g0≤ηij×2rm×ωc
L (9.23a)
= 2π×0.363MHz pm ×ηij
1
rm
0.42
743µm L
935 nm λ
. (9.23b)
9.1.3.4 Optical Damping and Spring Shift
The full expressions for “weak” optical damping and spring rates given in 9.17 can be expressed:
Γijopt= 2π×3.92 kHz×
hPout(∆ = 0)i 10µW
10 MHz κ/2π
2743µm L
2 rm
0.42 2ηij
1 2
(9.24a)
×
935 nm λ
5 MHz Ωij/2π
8.44 ng mij
κ 2κ2
1 1 + ∆2/κ2
κ2
(Ωij+ ∆)2+κ2 − κ2 (Ωij−∆)2+κ2
(9.24b) and
∆Ωijopt= 2π×1.96 kHz×
hPout(∆ = 0)i 10µW
10 MHz κ/2π
2743µm L
2 rm
0.42 2ηij
1 2
(9.25a)
×
935 nm λ
5 MHz Ωij/2π
8.44 ng mij
κ 2κ2
1 1 + ∆2/κ2
κ(Ωij+ ∆)
(Ωij+ ∆)2+κ2 − κ(Ωij−∆) (Ωij−∆)2+ (κ)2
.
(9.25b)
Two important limits exist for the unitless terms at the far right, corresponding to the difference of the strength of the blue sideband and the red sideband generated on the intracavity field by the vibrating membrane. In the sideband unresolved or “bad” cavity limit (Ωij << κ), they reduce to
κ2
(Ωij+ ∆)2+κ2 − κ2 (Ωij−∆)2+κ2
Ωij<<κ
−−−−−→ 4Ωij κ
∆/κ
(1 + (∆/κ)2)2 =±Ωij
κ for ∆ =±κ (9.26a) κ(Ωij+ ∆)
(Ωij+ ∆)2+κ2 − κ(Ωij−∆) (Ωij−∆)2+ (κ)2
Ωij<<κ
−−−−−→ 2∆/κ
1 + (∆/κ)2 ≤ ±1 for ∆ =±κ. (9.26b) In the sideband resolved or “good cavity” limit, Ωij >> κ, with ∆ =±Ωij (in which case only one sideband is resonant), the prefactors reduce to
κ2
(Ωij+ ∆)2+κ2 − κ2 (Ωij−∆)2+κ2
Ωij>>κ
−−−−−→ ±1 for ∆ =±Ωij (9.27a) κ(Ωij+ ∆)
(Ωij+ ∆)2+κ2 − κ(Ωij−∆) (Ωij−∆)2+κ2
Ωij>>κ
−−−−−→ ± κ 2Ωij
for ∆ =±Ωij. (9.27b) In practice, we currently operate the cavity in between the “good” and “bad” cavity limit, using the (6,6) mode with Ω66/2π≈5 MHz and the cavity at a wavelength for whichκ/2π≈10 MHz. In this case the unitless factors (Eqs. 9.27a and 9.27b) are roughly 0.5 for both the spring and damping rates.
9.1.3.5 Effective Temperature and Thermal Occupation Number
For parameters used in the previous section, we expect substantial radiation pressure damping rate for ∼ 10 microwatts of power coupled into the cavity. For damping rates much smaller than the mechanical frequency or the cavity linewidth, we can naively applyTef fij =Tbath×Γij/(Γij+ Γijopt) (Eq. 9.17d) to predict the effective temperature and occupation number ¯nij =kBTef fij /~Ωij of the damped mode. In the limit Γijopt>>Γij:
Tef fij = 0.38 K·
10µW hPout(∆ = 0)i
κ/2π 10MHz
2 L 743µm
20.42 rm
2 1 ηij
2 λ 935 nm
(9.28a)
· 106
Qij
Ωij/2π 5 MHz
2 mij
8.44 ng 2κ2
κ
1 1 + ∆2/κ2
−1
κ2
(Ωij+ ∆)2+κ2 − κ2 (Ωij−∆)2+κ2
−1
(9.28b)
¯
nij = 1590·
10µW hPout(∆ = 0)i
κ/2π 10MHz
2 L 743µm
20.42 rm
2 1 ηij
2 λ 935 nm
106 Qij
(9.28c)
·
Ωij/2π 5 MHz
Tbath 300 K
mij 8.44 ng
2κ2 κ
1 1 + ∆2/κ2
−1 κ2
(Ωij+ ∆)2+κ2 − κ2 (Ωij−∆)2+κ2
−1
.
(9.28d)
For cavity and membrane parameters used in the measurements below, {κ/2π, L,∆, λ,Ω66/2π, Q66, m66, rm, η} ≈ {10 MHz, 742 µm, 5 MHz, 935 nm, 5 MHz, 1.5×106, 8.44 ng, 0.42, 0.64}, the full expression for weak optical damping (Eq. 9.24) gives Γ66opt/2π≈ 4.0(κ/2κ2) kHz for hPouti= 100µW. The resulting effective temperature is and Tef f66 ≈ 0.25(2κ2/κ) K, corresponding to a thermal occupation number ¯n66 ≈ 1.0×103(2κ2/κ), which represents significant optomechanical cooling for moderate circulating powers of∼1 W. We now test these predictions.