Construction techniques and commissioning of the Three-Backlink Experiment for the LISA mission

Designed to detect gravitational waves in the lower-frequency band, the space mission LISA will open a new window to astronomy after its launch in the 2030s. Each LISA spacecraft houses two optical benches that require the exchange of a phase reference between them via an optical connection, called a Backlink. Here we present the construction and commissioning of an ultra-stable quasi-monolithic optical testbed to investigate different Backlink implementations: a direct fiber, a frequency-separated fiber, and a free-beam link, compared in the Three-Backlink Experiment. Dedicated alignment techniques crucial for the construction of these optical benches are presented together with the development of a high-precision beam alignment and measurement tool - a Calibrated Quadrant Photodiode Singleton. An upper limit for the performance of all three investigated Backlink schemes, as determined by initial experiments, can be set at a $15\text{pm}/\sqrt{\text{Hz}}$-equivalent level within the LISA band, spanning 0.1mHz to 1Hz. Our measurements were able to verify the successful construction and commissioning of this very complex interferometer as an interferometric laboratory testbed for LISA. We find no limitations due to the construction on the here reported performance levels. Our results can support the construction of high-precision metrology testbeds for space-based laser interferometry for future gravitational wave or geodesy missions.

💡 Research Summary

The paper presents the design, construction, and initial commissioning of a laboratory test‑bed called the Three‑Backlink Experiment (3BL) that is intended to evaluate the performance of the LISA Backlink, the optical link that distributes the phase reference between the two optical benches on each spacecraft. LISA will operate in the 0.1 mHz–1 Hz band and requires the non‑reciprocal path‑length noise of the backlink to be below 1 pm/√Hz. To assess three candidate implementations under identical conditions, the authors built two ultra‑stable quasi‑monolithic optical benches from low‑CTE Clearceram, bonded with a UV‑curable epoxy (Opto‑cast 3553). The benches are mirror‑symmetric and host four interferometers each, allowing the subtraction of common laser phase noise.

The three backlink schemes tested are:

1. Direct Fiber Backlink (DFBL) – a single optical fiber directly connects the benches. This is the simplest approach but suffers from back‑scatter, which can couple into the measurement. In the experiment the fiber is used without additional attenuation to expose the worst‑case back‑scatter contribution.

2. Frequency‑Separated Fiber Backlink (FSFBL) – two extra lasers (Laser 3 and Laser 4) are introduced so that the light returning through the fiber is frequency‑shifted relative to the forward beam. This separates back‑scatter from the science band, at the cost of added lasers, interferometers, and control electronics.

3. Free‑Beam Backlink (FBBL) – a free‑space beam is steered between the benches by two piezo‑actuated mirrors. Differential Wavefront Sensing (DWS) loops keep the beam aligned, eliminating fiber back‑scatter entirely but introducing mirror‑induced phase dynamics and the need for a precise steering control system.

To extract the non‑reciprocal phase noise, the authors combine the phase readouts from photodiodes on both benches. By forming symmetric sums (Eq. 1) the common laser frequency noise cancels, leaving the backlink’s one‑way phase difference. Pairwise differences of the three backlink combinations (Eq. 2) then isolate the non‑reciprocal contribution of each scheme.

A major technical challenge was the sub‑microradian alignment of many optical components that are glued to the bench. The glue layer can introduce a wedge, leading to vertical tilts of up to 380 µrad. The team mitigated this by applying controlled pressure to create a deliberate wedge in the epoxy, iterating until both angular and translational degrees of freedom met the required tolerances.

Precise beam positioning was essential for both alignment and performance verification. The authors therefore developed the Calibrated Quadrant Photodiode Singleton (CQS), an evolution of the Calibrated Quadrant Photodiode Pair (CQP). The CQS consists of a large‑area quadrant photodiode mounted on a brass housing with Macor spacers, attached to two micrometer stages. By calibrating the distances between the housing surfaces and the QPD centre using a coordinate‑measurement machine, the system can determine beam centroid position with a standard uncertainty of ±3.5 µm and angular orientation with ±12 µrad. Long‑term stability tests showed drift below 5 µm over several days.

During the commissioning runs, all three backlink implementations demonstrated non‑reciprocal phase noise below 15 pm/√Hz across the full LISA measurement band (0.1 mHz–1 Hz). This level is comfortably above the mission requirement, indicating that the bench construction, alignment procedures, and the CQS met the stringent stability goals. The DFBL data revealed that back‑scatter is the dominant noise source and is sensitive to temperature fluctuations of the fiber. The FSFBL showed that the extra lasers and interferometers introduce electromagnetic coupling that must be carefully shielded. The FBBL performance was limited by the bandwidth and residual noise of the DWS steering loops, suggesting that higher‑speed, lower‑noise mirror actuators would be beneficial for future space implementations.

The authors conclude that the 3BL test‑bed successfully provides a platform for side‑by‑side comparison of backlink technologies under realistic laboratory conditions. The results confirm that none of the three concepts is fundamentally limited by construction tolerances; instead, the limiting factors are intrinsic to each implementation (back‑scatter, electronic complexity, or free‑beam steering). The CQS tool proved valuable for high‑precision alignment and could be adopted for other space‑based interferometric test‑beds.

Future work will focus on reproducing the space environment more closely—thermal cycling, vibration, and radiation exposure—to verify that the observed 15 pm/√Hz ceiling can be pushed down to the 1 pm/√Hz requirement. Additionally, the team plans to refine the free‑beam steering control algorithms, explore alternative low‑back‑scatter fiber designs, and integrate the backlink with a full Time‑Delay Interferometry (TDI) simulation to assess the impact on the overall LISA noise budget.