rf losses in a high gradient cryogenic copper cavity

The development of high brightness electron sources can enable an increase in performance and reduction in size of extreme X-ray sources such as free electron lasers (FELs). A promising path to high brightness is through larger electric fields in radio-frequency (rf) photoinjectors. Recent experiments with 11.4 GHz copper accelerating cavities at cryogenic temperatures have demonstrated $500\mathrm{MV}/\mathrm{m}$ surface electric fields with low rf breakdown rates. However, when the surface electric fields are larger than $300\mathrm{MV}/\mathrm{m}$, the measured cavity quality factor, $Q_0$, decreases during the input rf pulse by up to 30%, recovering before the next rf pulse. In this paper, we present an experimental study of the rf losses, manifested as degradation of $Q_0$, in a copper cavity operated at cryogenic temperatures and high gradients. The experimental conditions range from temperatures of 10–77 K and rf pulse lengths of 100–800 ns, using surface electric fields up to $400\mathrm{MV}/\mathrm{m}$. We developed a model for the change in $Q_0$ using measured field emission currents and rf signals. We find that the $Q_0$ degradation is consistent with the rf power being absorbed by strong field emission currents accelerated inside the cavity.

Figure 1

Surface electric and magnetic fields for the cryogenic cavity. The fields are scaled to 2.5 MW dissipated in the cavity, $T=45\mathrm{K}$ and $Q_0=30$, 263. (a) shows the surface electric fields, with a maximum $400\mathrm{MV}/\mathrm{m}$. (b) shows the surface magnetic fields, with a maximum $580\mathrm{kA}/\mathrm{m}$.

Figure 2

Measurement (red) and linear model (blue) for two pulses with flat gradient section of 400 ns of low (left) and high (right) power at 45 K and 10 Hz. The input rf pulse (green), begins at 50 ns and ends at 600 ns. The model agrees with the measured signal for $Q_0=30$, 263 at low power ((a)–(c)). For high power ((d)–(f)), the linear model does not match the measured rf and dark current signals.

Figure 3

Measured data (red) for 45 K, also shown in Fig. 2, and nonlinear model (blue) are plotted in (a)–(c). (d) shows the calculated gradient for linear and nonlinear models. (e) and (f) show the predicted change in $Q_0(t)$ and $f_0(t)={\omega }_0(t)/2\pi$ respectively. Comparing the linear model in Fig. 2 and the nonlinear model in Fig. 3, the nonlinear model better predicts the data.

Figure 4

$Q_0$ at the end of the rf pulse, when it reaches a minimum, versus calculated gradient for different rf pulse lengths. The plot shows no obvious dependence on pulse length, but a strong dependence on peak surface electric fields.

Figure 5

Measured (red) and simulated (blue) missing power in the accelerating cavity versus the measured dark current magnitude for rf pulse lengths 100–800 ns. The missing power is the difference between what we measured as the power lost to the cavity walls, and what would have been lost to the walls in the linear model. The simulation of power being lost to dark currents is consistent with the presented data.