1 \section{Cantilever Spring Constant Calibration}
2 \label{sec:cantilever-calib:intro}
4 In order to measure forces accurately with an AFM, it is important to
5 measure the cantilever spring constant. The force exerted on the
6 cantilever can then be deduced from it's deflection via Hooke's law
8 \nomenclature{$F$}{Force (newtons)}
9 \nomenclature{$k$}{Spring constant (newtons per meter)}
10 \nomenclature{$x$}{Displacement (meters)}
12 The basic idea is to use the equipartition theorem\citep{hutter93},
14 \frac{1}{2} k \avg{x^2} = \frac{1}{2} k_BT \;, \label{eq:equipart}
16 where $k_B$ is Boltzmann's constant, $T$ is the absolute temperature,
17 and $\avg{x^2}$ denotes the expectation value of $x^2$ as measured
18 over a very long interval $t_T$,
19 \nomenclature{$k_B$}{Boltzmann's constant,
20 $k_B = 1.380 65\E{-23}\U{J/K}$\citep{codata-boltzmann}}
21 \nomenclature{$T$}{Absolute temperature (Kelvin)}
22 \nomenclature{$\avg{s(t)}$}{Mean (expectation value) of a time-series $s(t)$
24 \avg{A} \equiv \iLimT{A} \;.
26 Solving the equipartition theorem for $k$ yields
28 k = \frac{k_BT}{\avg{x^2}} \;, \label{eq:equipart_k}
30 so we need to measure (or estimate) the temperature $T$ and variance
31 of the cantilever position $\avg{x^2}$ in order to estimate $k$.
33 To find $\avg{x^2}$, the raw photodiode voltages $V_p(t)$ are
34 converted to distances $x(t)$ using the photodiode sensitivity
35 $\sigma_p$ (the slope of the voltage vs.~distance curve of data taken
36 while the tip is in contact with the surface) via
38 x(t) = \frac{V_p(t)}{\sigma_p} \;. \label{eq:x-from-Vp}
40 By keeping $V_p$ and $\sigma_p$ separate in our calculation of $k$, we
41 can gauge the relative importance errors in each parameter and
42 calculate the uncertainty in our estimated $k$.
44 In order to filter out noise in the measured value of $\avg{V_p^2}$ we
45 fit the measured cantilever deflection to the expected theoretical
46 power spectral density ($\PSD_f$) of a damped harmonic oscillator
47 exposed to thermal noise
48 \nomenclature[PSD]{$\PSD_f$}{Power spectral density in
51 \PSD_f(g, f) \equiv \normLimT 2 \magSq{ \Fourf{g(t)}(f) }
53 \nomenclature{$f$}{Frequency (hertz)}
55 \PSD_f(V_p, f) = \frac{G_{1f}}{(f_0^2-f^2)^2 + \beta_f^2 f^2} \;.
57 In terms of the fit parameters $G_{1f}$, $f_0$, and $\beta_f$,
58 the expectation value for $V_p$ is given by
60 \avg{V_p(t)^2} = \frac{\pi G_{1f}}{2\beta_f f_0^2} \;.
61 \label{eq:Vp-from-freq-fit}
64 Combining \cref{eq:equipart_k,eq:x-from-Vp,eq:Vp-from-freq-fit}, we
67 k &= \frac{\sigma_p^2 k_BT}{\avg{V_p(t)^2}}
68 = \frac{2 \beta_f f_0^2 \sigma_p^2 k_BT}{\pi G_{1f}} \;.
71 For a complete derivation of the procedure presented in this section,
72 see \cref{sec:cantilever-calib}.