In gravitational wave detection in space, spacecraft is exposed to the heliospheric environment and subjected to continuous bombardment by high energy particles (Jafry et al., 1996; Armano et al., 2023; Araújo et al., 2005). When the energies exceed a certain threshold, particles will penetrate the spacecraft and deposit on a TM, generating spurious acceleration noise and disturbing gravitational wave detection (Carbone et al., 2004; Sumner et al., 2020). Beginning from the Gravity Probe B (GP-B) mission and later on the LISA Pathfinder (LPF), photoelectric effect is employed as a non-contact way to neutralise charges where the mercury lamp is used as the ultraviolet (UV) light sourse (Buchman et al., 1995; Sumner et al., 2009; Wass et al., 2006; Armano et al., 2017). Instead of mercury lamp, UV-LED is considered as a better alternative for a number of technical reasons in the upcoming LISA and the prospective Chinese missions, such as Olatunde et al. (2015), Hu and Wu (2017), and Luo et al. (2016). UV-LED based charge management systems are now widely studied, and Stanford University has successfully demonstrated charge management system with UV-LEDs in space (Saraf et al., 2016; Hollington et al., 2015; Yang et al., 2020; Wang et al., 2022).

During a solar cycle, charge control strategies vary with particle flux intensity. Among them, continuous discharge is used to balance the charging-discharging rates, typically suited for the ascending and descending phase of a solar cycle in which particle flux is predominantly due to galactic cosmic rays and the solar activity is low. A continuous discharge method employing dual UV light sources was proposed to illuminate both the TM and the electrode housing, effectively maintaining the TM potential near zero (Pi et al., 2023; Wang et al., 2022; Armano et al., 2018). Previously, Wang and the Stanford group studied continuous charge management methods by employing UV light with a peak wavelength longer than the 255 nm commonly used, and they used a simplified parallel plate model derived from an inertial sensor (Wang et al., 2022).

The aim of the present work is to conduct this study by employing a more realistic model of an inertial sensor instead of a simple parallel plate model. A cubic TM is fixed within an electrode housing via insulating pillars and the surfaces are gold coated to mimic those of the LPF in terms of the work function and surface properties of the TM as well as the electrode housing. Continuous discharge requires sustained operation during solar minimum periods. However, constrained by the experimental conditions of this study, we use 4-h continuous discharge results to demonstrate in a preliminary way its performance, corresponding to the 0.1 mHz frequency band essential for space-based gravitational wave detection. A preliminary continuous charge management is then studied using multiple wavelengths of UV-LEDs. Experimental results indicate that, due to internal reflections of UV light within the electrical housing, the photoelectron emission is different from that of a parallel plate model. Further, our results show that 275 nm outperforms the standard 255 nm wavelength for both fast release of charged particles as well as for continuous charge management, it is capable of achieving smaller equilibrium potential with long time stability.

In the present work, we investigate the UV-LED light sources for charge management. Although these light sources have been extensively studied by a number of research groups with focus on the 255 nm wavelength. Our contribution lies in exploring longer wavelength light sources beyond conventional 255 nm, and we argue that during periods of low solar activity, the 275 nm light source seems to be a better option. This result is similar to our previous experiments based on UV micro-LED light sources (Jia et al., 2025), where we used four UV micro-LEDs of different wavelengths. Light sources with wavelengths longer than the 255 nm commonly used achieved a equilibrium potential closer to 0 mV, with 274 nm reaching −10 mV. However, in the paper (Jia et al., 2025), we focus more on the feasibility of applying this new light source to charge management in the detection of gravitational wave in space. We conducted a characterization analysis of the new light source and verified its potential application in the space environment.

The present work is structured as follows. In Section 2, charge management model based on a UV-LED with long wavelength is introduced and analyzed. In Section 3, based on the cubic charge management system, photoelectric effect for a UV-LED with peak wavelength of 275 nm is demonstrated. Continuous charge control using UV-LEDs with different peak wavelengths will be studied in Section 4, which will enable us to explore a method using light sources different from the standard 255 nm. The experimental results show that UV-LEDs with longer wavelengths achieve excellent long-term stability, maintaining equilibrium potentials within approximately 50 mV over a 4-h period, and achieving a charge noise level of 10 − 13 C / H z near 0.1 mHz. Moreover, 275 nm UV-LED achieves not only smaller equilibrium potential (−12 mV) but also faster discharge rate (6 s), making it a promising candidate as a better light source for charge management system.

In this section, we will present a charge management model for long-wavelength UV-LED illumination, on the basis of which we try to understand our experimental results and perform data analysis.

Figure 1 presents a schematic representation of the charge management mechanism employing long-wavelength (i.e., wavelength longer than 255 nm) UV-LED for direct illumination. The work function of Au exposed to air is approximately 4.2 eV (Schulte et al., 2009; Jiang et al., 1998). As a result, incident light with a photon energy greater than 4.2 eV (corresponding to a wavelength of less than 295 nm) is sufficient to induce photoemission for charge management. In this configuration, UV light illuminates the TM and reflects between the TM and the electrode housing (EH). Photoelectrons are subsequently emitted from both gold-coated surfaces.

Consider the case in which the potential of the TM is positive and greater than the initial kinetic energy of the photoelectrons, where the electrode housing (including its mounted electrodes) is grounded. The presence of the local electric field causes photoelectrons to migrate to the TM, thereby reducing its potential, as shown in Figure 1. When the potential approaches zero and becomes comparable to the initial kinetic energy of the photoelectrons, the photoelectron flow between the TM and the electrode housing reaches an equilibrium state. At this point, the TM attains a stable potential, as illustrated in Figure 2. A similar process occurs when the initial TM charge is negative. As the wavelength of the incident light increases, the photon energy decreases, which correspondingly reduces the kinetic energy of the generated photoelectrons. Thereby, a smaller equilibrium potential of the TM can be achieved.

The application of a bias voltage ( V B ) to the electrode will influence the migration of photoelectrons between the TM and the electrode housing. The potential variation of the TM under UV illumination follows an exponential function as shown in Equation 1, which has been verified through both theoretical modeling and experimental validation (Pi et al., 2023; Li et al., 2025): V TM t = A e − t τ + V eq ( V B ) (1) where τ is the time constant that reflects the potential variation rate of the TM, and V e q ( V B ) is the equilibrium potential to which the TM will settle under a given bias voltage after prolonged illumination.

In this experiment, long-wavelength UV-LEDs were used to evaluate the charge management performance. The performance was quantified by three metrics: the equilibrium potential of the TM, the stability of this potential, and the discharge rate.

Figure 7 illustrates the equilibrium potential V eq as a function of the bias electrode potential V B under illumination by UV-LEDs of five different wavelengths. The zero-bias regime ( V B = 0 V) represents the primary focus of this study, with its detailed characteristics highlighted in the enlarged inset in the upper-left corner.

The equilibrium potential of the TM under zero-bias conditions ( V B = 0 V) was used to quantify the charge management performance. Table 3 summarizes these measurements (corresponding to the shaded insert in Figure 7), demonstrating that direct TM illumination enables continuous charge control. The results show that all five UV-LEDs maintain the TM charge within ± 100 mV, the LISA requirement (equivalent to about 2 × 1 0 7 elementary charges (Antonucci et al., 2012). It should be noted that LED_255 and LED_265 achieve comparable equilibrium potentials around 50 mV, while the longer LED_275, LED_285, and LED_295) are moving the TM significantly closer to zero potential, indicating superior performance in precision charge management. It should be noted that the polarity of the equilibrium potential varies under different UV-LEDs. This is because the equilibrium potential is governed by the net photoelectron flux between the TM and the electrode housing, which in turn is influenced by factors such as the incident light wavelength, the geometry of the electrode housing, and the physical properties of the gold-coated surfaces. Therefore, further investigation through simulation is required to fully understand the behaviour of the equilibrium potential.

To evaluate the temporal stability of the long-wavelength UV-LED-based charge management system, 4-h continuous tests were conducted using LED_265, LED_275, LED_285 and LED_295. During these tests, the drive current was set to 10 mA, and the bias electrode was grounded. Figure 8a illustrates V TM over time for each wavelength, and Figure 8b shows the equivalent TM charge noise under the control of the LED_275.

The experimental results demonstrate stable maintenance of the TM potential within ± 100 mV of the target value over the 4-h duration, with all four wavelengths exhibiting peak-to-peak fluctuations below 2 mV. This confirms the viability of long-wavelength UV-LEDs for passive charge management in gravitational reference sensor applications. Under the control of the LED_275, the system achieved a charge noise level of 10 − 13 C / H z near 0.1 mHz, corresponding to an equivalent acceleration noise of 1 0 − 15 m ⋅ s − 2 / H z as derived from the literature (Sumner et al., 2020). Subsequently, we will conduct longer-duration continuous discharge experiments, including tests at lower frequency bands with both charge management on and charge management off, thereby validating its capability for sustained operation throughout the entire mission duration. And the 4-h test duration is relatively brief compared to a full space mission cycle, and additional risks may exist in long-term missions such as space gravitational wave detection (e.g., device aging, performance drift, etc.). Future ground-based experiments will be conducted over extended periods to simulate the complete mission cycle.

To evaluate the charging rate of V TM for long-wavelength UV-LED illumination, experiments were conducted using LED_265, LED_275, LED_285, and LED_295. The UV-LED drive current was set to 10 mA, and the bias electrode was grounded during measurements. Figure 9 shows the V TM variation. By fitting V TM with time using 2, time constants for each wavelength are summarized in Table 4. As shown in Figure 4, the four UV-LEDs exhibit distinct optical powers at a drive current of 10 mA. Given the inverse proportionality between time constants and optical power, Table 4 also lists the time constants normalized relative to a reference optical power of 1 mW for each UV-LED.

As shown in Table 4, the time constant after calibration increases with wavelength. The time constants of LED_285 and LED_295 are significantly larger than those of LED_265 and LED_275. Under conditions normalized to 1 mW output power and identical TM potentials, LED_265 and LED_275 stabilize the potential near equilibrium within 6 s. In contrast, LED_285 and LED_295 require approximately 30 s and over 600 s respectively, 5 times and 100 times longer than LED_265 and LED_275. This highlights a significant wavelength-dependent efficiency disparity in charge control.

While the preceding experiments utilized DC excitation currents for UV-LED operation, the methodology remains equally applicable in PWM mode. This is critical for space-based gravitational wave detectors, where DC charge control could introduce low-frequency noise into the mHz science band (Weber et al., 2007). Figure 10a illustrates the potential variation of the TM through three consecutive charge-discharge cycles, employing LED_275 in PWM mode. The LED was driven at a current of 20 mA and modulated at 1 kHz with 50% duty cycle. The TM was alternately biased with a positive or negative initial potential, followed by UV-LED illumination until equilibrium was achieved in each case. This alternating sequence was repeated for three cycles, and the shaded area indicates periods of UV-LED activation.

The experimental results demonstrate consistent convergence of the TM potential to equilibrium under UV-LED activation across all three cycles, regardless of initial potential polarity. Figure 10b details the transient behavior of the first cycle (0–400 s), revealing charge and discharge phases from the initial potentials to equilibrium. This bipolar charge control demonstrates the robustness of the methodology, as demonstrated in the previous research (Wang et al., 2022; Saraf et al., 2014).

The experimental results demonstrate that for gold-coated surfaces, the LED_265 showed performance comparable to the LED_255, and the LED_275, LED_285 and LED_295 exhibit a lower equilibrium potential (within 20 mV) compared to the LED_255, while wavelengths approaching 300 nm and beyond are ineffective due to insufficient photon energy. Moreover, LED_275 seems to be an optimal candidate for the source of charge management system, not only achieving a smaller equilibrium potential compared to LED_255 and LED_265 but also demonstrating a shorter time constant than long-wavelength UV-LEDs, such as LED_285 or LED_295. It should be clarified that our approach does not entirely replace the 255 nm UV-LED with the 275 nm variant. Instead, it offers more options for neutralizing cosmic ray charging, pending on the solar activity. On board, both the 255 nm and 275 nm should be placed at the four bottom corners of the caging. Depending on the intensity of the solar energetic particles and possibly contamination of the test mass, UV-LED of appropriate wavelengths will be activated for charge management.

Based on a model of inertial sensor that mimics the surface properties and work function of a TM for LISA, UV-LEDs with multiple wavelengths were tested to evaluate their performance in continuous charge management, in terms of both the equilibrium potential and long-term stability. Additionally, the discharge rates (time constant) of UV-LEDs with different wavelengths were measured. The results show that LED_275 appears to be the best choice if optimizing for both these parameters. These results demonstrate that the 275 nm UV-LED is a promising candidate for the continuous charge management and at the same time for regular rapid discharge operation.

It is our hope that our work will provide useful input when it comes to the system design and optimisation of the charge management system in the engineering phase of a mission to detect gravitational waves in space.