summaryrefslogtreecommitdiff
path: root/source/know/concept/salt-equation/index.md
blob: d47383f81e026c7d738123383a867e4c19586080 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
---
title: "SALT equation"
sort_title: "Salt equation" # sic
date: 2022-02-07
categories:
- Physics
- Optics
- Laser theory
layout: "concept"
---

The **steady-state *ab initio* laser theory** (SALT) is
a theoretical description of lasers, whose mode-centric approach
makes it especially appropriate for microscopically small lasers.

Consider the [Maxwell-Bloch equations](/know/concept/maxwell-bloch-equations/),
governing the complex polarization
vector $$\vb{P}^{+}$$ and the scalar population inversion $$D$$ of a set of
active atoms (or quantum dots) embedded in a passive linear background
material with refractive index $$c / v$$.
The system is affected by a driving [electric field](/know/concept/electric-field/)
$$\vb{E}^{+}(t) = \vb{E}_0^{+} e^{-i \omega t}$$,
such that the set of equations is:

$$\begin{aligned}
    - \mu_0 \pdvn{2}{\vb{P}^{+}}{t}
    &= \nabla \cross \nabla \cross \vb{E}^{+} + \frac{1}{v^2} \pdvn{2}{\vb{E}^{+}}{t}
    \\
    \pdv{\vb{P}^{+}}{t}
    &= - \Big( \gamma_\perp + i \omega_0 \Big) \vb{P}^{+}
    - \frac{i}{\hbar} \Big( \vb{p}_0^{-} \cdot \vb{E}^{+} \Big) \vb{p}_0^{+} D
    \\
    \pdv{D}{t}
    &= \gamma_\parallel (D_0 - D) + \frac{i 2}{\hbar} \Big( \vb{P}^{-} \cdot \vb{E}^{+} - \vb{P}^{+} \cdot \vb{E}^{-} \Big)
\end{aligned}$$

Where $$\hbar \omega_0$$ is the band gap of the active atoms,
and $$\gamma_\perp$$ and $$\gamma_\parallel$$ are relaxation rates
of the atoms' polarization and population inversion, respectively.
$$D_0$$ is the equilibrium inversion, i.e. the value of $$D$$ if there is no lasing.
Note that $$D_0$$ also represents the pump,
and both $$D_0$$ and $$v$$ depend on position $$\vb{x}$$.
Finally, the transition dipole matrix elements $$\vb{p}_0^{-}$$ and $$\vb{p}_0^{+}$$ are given by:

$$\begin{aligned}
    \vb{p}_0^{-}
    \equiv q \matrixel{e}{\vu{x}}{g}
    \qquad \qquad
    \vb{p}_0^{+}
    \equiv q \matrixel{g}{\vu{x}}{e}
    = (\vb{p}_0^{-})^*
\end{aligned}$$

With $$q < 0$$ the electron charge, $$\vu{x}$$ the quantum position operator,
and $$\ket{g}$$ and $$\ket{e}$$ respectively
the ground state and first excitation of the active atoms.

We start by assuming that the cavity has $$N$$ quasinormal modes $$\Psi_n$$,
each with a corresponding polarization $$\vb{p}_n$$ of the active matter.
Note that this ansatz already suggests
that the interactions between the modes are limited:

$$\begin{aligned}
    \vb{E}^{+}(\vb{x}, t)
    = \sum_{n = 1}^N \Psi_n(\vb{x}) \: e^{- i \omega_n t}
    \qquad \qquad
    \vb{P}^{+}(\vb{x}, t)
    = \sum_{n = 1}^N \vb{p}_n(\vb{x}) \: e^{- i \omega_n t}
\end{aligned}$$

Using the modes' linear independence to treat each term of the summation individually,
the first two Maxwell-Bloch equations turn into, respectively:

$$\begin{aligned}
    \mu_0 \omega_n^2 \vb{p}_n
    &= \nabla \cross \nabla \cross \Psi_n - \frac{1}{v^2} \omega_n^2 \Psi_n
    \\
    i \omega_n \vb{p}_n
    &= \big( i \omega_0 + \gamma_\perp \big) \vb{p}_n
    + \frac{i}{\hbar} \big(\vb{p}_0^{+} \vb{p}_0^{-}\big) \cdot \Psi_n \: D
\end{aligned}$$

With being $$\vb{p}_0^{+} \vb{p}_0^{-}$$ a dyadic product.
Isolating the latter equation for $$\vb{p}_n$$ gives us:

$$\begin{aligned}
    \vb{p}_n
    &= \frac{\big(\vb{p}_0^{+} \vb{p}_0^{-}\big) \cdot \Psi_n \: D}{\hbar \big((\omega_n - \omega_0) + i \gamma_\perp\big)}
    = \frac{\gamma(\omega_n) D}{\hbar \gamma_\perp} \big(\vb{p}_0^{+} \vb{p}_0^{-}\big) \cdot \Psi_n
\end{aligned}$$

Where we have defined the Lorentzian gain curve $$\gamma(\omega_n)$$ as follows,
which represents the laser's preferred frequencies for amplification:

$$\begin{aligned}
    \gamma(\omega_n)
    \equiv \frac{\gamma_\perp}{(\omega_n - \omega_0) + i \gamma_\perp}
\end{aligned}$$

Inserting this expression for $$\vb{p}_n$$
into the first Maxwell-Bloch equation yields
the prototypical form of the SALT equation,
where we still need to replace $$D$$ with known quantities:

$$\begin{aligned}
    0
    &= \bigg( \nabla \cross \nabla \cross - \, \omega_n^2 \frac{1}{v^2}
    - \omega_n^2 \frac{\mu_0 \gamma(\omega_n) D}{\hbar \gamma_\perp} (\vb{p}_0^{+} \vb{p}_0^{-}) \cdot \bigg) \Psi_n 
\end{aligned}$$

To rewrite $$D$$, we turn to its (Maxwell-Bloch) equation of motion,
making the crucial **stationary inversion approximation** $$\ipdv{D}{t} = 0$$:

$$\begin{aligned}
    D
    &= D_0 + \frac{i 2}{\hbar \gamma_\parallel} \Big( \vb{P}^{-} \cdot \vb{E}^{+} - \vb{P}^{+} \cdot \vb{E}^{-} \Big)
\end{aligned}$$

This is the most aggressive approximation we will make:
it removes all definite phase relations between modes,
and effectively eliminates time as a variable.
We insert our ansatz for $$\vb{E}^{+}$$ and $$\vb{P}^{+}$$,
and point out that only active lasing modes contribute to $$D$$:

$$\begin{aligned}
    D
    &= D_0 + \frac{i 2}{\hbar \gamma_\parallel} \sum_{\nu, \mu}^\mathrm{active}
    \bigg( \vb{p}_\nu^* \cdot \Psi_\mu e^{i (\omega_\nu - \omega_\mu) t}
    - \vb{p}_\nu \cdot \Psi_\mu^* e^{i (\omega_\mu - \omega_\nu) t} \bigg)
\end{aligned}$$

Here, we make the [rotating wave approximation](/know/concept/rotating-wave-approximation/)
to neglect all terms where $$\nu \neq \mu$$
on the basis that they oscillate too quickly,
leaving only $$\nu = \mu$$:

$$\begin{aligned}
    D
    &= D_0 + \frac{i 2}{\hbar \gamma_\parallel} \sum_{\nu}^\mathrm{act.}
    \bigg( \vb{p}_\nu^* \cdot \Psi_\nu - \vb{p}_\nu \cdot \Psi_\nu^* \bigg)
\end{aligned}$$

Inserting our earlier equation for $$\vb{p}_n$$
and using the fact that $$\vb{p}_0^{+} = (\vb{p}_0^{-})^*$$ leads us to:

$$\begin{aligned}
    D
    &= D_0 + \frac{i 2 D}{\hbar^2 \gamma_\parallel \gamma_\perp} \sum_{\nu}^\mathrm{act.}
    \bigg( \gamma^*(\omega_\nu) \big(\vb{p}_0^{+} \vb{p}_0^{-}\big)^* \!\cdot\! \Psi_\nu^* \cdot \Psi_\nu
    - \gamma(\omega_\nu) \big(\vb{p}_0^{+} \vb{p}_0^{-}\big) \!\cdot\! \Psi_\nu \cdot \Psi_\nu^* \bigg)
    \\
    &= D_0 + \frac{i 2 D}{\hbar^2 \gamma_\parallel \gamma_\perp} \sum_{\nu}^\mathrm{act.}
    \bigg( \gamma^*(\omega_\nu) \big(\vb{p}_0^{+} \cdot \Psi_\nu^*\big) \vb{p}_0^{-} \cdot \Psi_\nu
    - \gamma(\omega_\nu) \big(\vb{p}_0^{-} \cdot \Psi_\nu\big) \vb{p}_0^{+} \cdot \Psi_\nu^* \bigg)
    \\
    &= D_0 + \frac{i 2 D}{\hbar^2 \gamma_\parallel \gamma_\perp} \sum_{\nu}^\mathrm{act.}
    \Big( \gamma^*(\omega_\nu) - \gamma(\omega_\nu) \Big) \big|\vb{p}_0^{-} \cdot \Psi_\nu\big|^2
\end{aligned}$$

By putting the terms on a common denominator, it is easily shown that:

$$\begin{aligned}
    \gamma^*(\omega_\nu) - \gamma(\omega_\nu)
    &= \frac{\gamma_\perp ((\omega_\nu - \omega_0) + i \gamma_\perp)}{(\omega_\nu - \omega_0)^2 + \gamma_\perp^2}
    - \frac{\gamma_\perp ((\omega_\nu - \omega_0) - i \gamma_\perp)}{(\omega_\nu - \omega_0)^2 + \gamma_\perp^2}
    \\
    &= \frac{\gamma_\perp (i \gamma_\perp + i \gamma_\perp)}{(\omega_\nu - \omega_0)^2 + \gamma_\perp^2}
    = i 2 \big|\gamma(\omega_\nu)\big|^2
\end{aligned}$$

Inserting this into our equation for $$D$$ gives the following expression:

$$\begin{aligned}
    D
    &= D_0 - \frac{4 D}{\hbar^2 \gamma_\parallel \gamma_\perp} \sum_{\nu}^\mathrm{act.}
    \Big|\gamma(\omega_\nu) \vb{p}_0^{-} \cdot \Psi_\nu\Big|^2
\end{aligned}$$

We then properly isolate this for $$D$$ to get its final form, namely:

$$\begin{aligned}
    D
    &= D_0 \bigg( 1 + \frac{4}{\hbar^2 \gamma_\parallel \gamma_\perp} \sum_{\nu}^\mathrm{act.}
    \Big|\gamma(\omega_\nu) \vb{p}_0^{-} \cdot \Psi_\nu\Big|^2 \bigg)^{-1}
\end{aligned}$$

Substituting this into the prototypical SALT equation from earlier
yields the most general form of the **SALT equation**,
upon which the theory is built:

$$\begin{aligned}
    \boxed{
        0
        = \bigg( \nabla \cross \nabla \cross
        -\,\omega_n^2 \bigg[ \frac{1}{v^2(\vb{x})} + \frac{\mu_0 \gamma(\omega_n)}{\hbar \gamma_\perp}
        \frac{D_0(\vb{x})}{1 + h(\vb{x})} (\vb{p}_0^{+} \vb{p}_0^{-}) \cdot \bigg] \bigg) \Psi_n(\vb{x})
    }
\end{aligned}$$

Where we have defined **spatial hole burning** function $$h(\vb{x})$$ like so,
representing the depletion of the supply of charge
carriers as they are consumed by the active lasing modes:

$$\begin{aligned}
    \boxed{
        h(\vb{x})
        \equiv \frac{4}{\hbar^2 \gamma_\parallel \gamma_\perp} \sum_{\nu}^\mathrm{act.}
        \Big|\gamma(\omega_\nu) \vb{p}_0^{-} \cdot \Psi_\nu(\vb{x})\Big|^2
    }
\end{aligned}$$

Many authors assume that $$\vb{p}_0^- \parallel \Psi_n$$,
so that only its amplitude $$|g|^2 \equiv \vb{p}_0^{+} \cdot \vb{p}_0^{-}$$ matters.
In that case, they often non-dimensionalize $$D$$ and $$\Psi_n$$
by dividing out the units $$d_c$$ and $$e_c$$:

$$\begin{aligned}
    \tilde{\Psi}_n
    \equiv \frac{\Psi_n}{e_c}
    \qquad
    e_c
    \equiv \frac{\hbar \sqrt{\gamma_\parallel \gamma_\perp}}{2 |g|}
    \qquad \qquad
    \tilde{D}
    \equiv \frac{D}{d_c}
    \qquad
    d_c
    \equiv \frac{\varepsilon_0 \hbar \gamma_\perp}{|g|^2}
\end{aligned}$$

And then the SALT equation and hole burning function $$h$$ are reduced to the following,
where the vacuum wavenumber $$k_n = \omega_n / c$$:

$$\begin{aligned}
    0
    = \bigg( \nabla \cross \nabla \cross -\,k_n^2 \bigg[ \varepsilon_r
    + \gamma(c k_n) \frac{\tilde{D}_0}{1 + h} \bigg] \bigg) \tilde{\Psi}_n
    \qquad
    h(\vb{x})
    = \sum_{\nu}^\mathrm{act.} \Big|\gamma(c k_\nu) \tilde{\Psi}_\nu(\vb{x})\Big|^2
\end{aligned}$$


In addition, some papers only consider 1D or 2D *transverse magnetic* (TM) modes,
in which case the fields are scalars. Using the vector identity

$$\begin{aligned}
    \nabla \cross \nabla \cross \Psi
    = \nabla (\nabla \cdot \Psi) - \nabla^2 \Psi
\end{aligned}$$

Where $$\nabla \cdot \Psi = 0$$ thanks to [Gauss' law](/know/concept/maxwells-equations/),
so we get an even further simplified SALT equation:

$$\begin{aligned}
    0
    = \bigg( \nabla^2 +\,k_n^2 \bigg[ \varepsilon_r
    + \gamma(c k_n) \frac{\tilde{D}_0}{1 + h} \bigg] \bigg) \tilde{\Psi}_n
\end{aligned}$$

Given $$\varepsilon_r(\vb{x})$$ and $$D_0(\vb{x})$$,
the challenge is to solve this eigenvalue problem for $$k_n$$ and $$\Psi_n$$,
with the boundary condition that $$\Psi_n$$ is a plane wave at infinity,
i.e. light is leaving the cavity.

If $$\Imag(k_n) < 0$$, the $$n$$th mode's amplitude decays with time, so it acts as an LED:
it emits photons without any significant light amplification taking place.
Upon gradually increasing the pump $$D_0$$ in the active region,
all $$\Imag(k_n)$$ become less negative,
until one hits the real axis $$\Imag(k_n) = 0$$,
at which point that mode starts *lasing*:
its Light gets Amplified by [Stimulated Emission](/know/concept/einstein-coefficients/) of Radiation (LASER).
After that, $$D_0$$ can be increased even further until some other $$k_n$$ become real,
so there are multiple active modes competing for charge carriers.

Below threshold (i.e. before any mode is lasing), the problem is linear in $$\Psi_n$$,
but above threshold it is nonlinear via $$h(\vb{x})$$.
Then the amplitude of $$\Psi_n$$ adjusts itself
such that its respective $$k_n$$ never leaves the real axis.
Once a mode is lasing, hole burning makes it harder for any other modes to activate,
since they must compete for the carrier supply $$D_0$$.



## References
1.  L. Ge, Y.D. Chong, A.D. Stone,
    [Steady-state *ab initio* laser theory: generalizations and analytic results](http://dx.doi.org/10.1103/PhysRevA.82.063824),
    2010, American Physical Society.