High-precision laser-based measurements, such as gravitational wave interferometry (The LIGO Scientific Collaboration, 2015), precision metrology (Giovannetti et al., 2004), and quantum information (Wang et al., 2020), rely heavily on the quantum properties of light. These measurements, while crucial for scientific advancements, are often limited by quantum noise, specifically shot noise and radiation pressure noise. The challenge lies in finding ways to improve the sensitivity of these measurements beyond the standard noise limits imposed by quantum mechanics with systems that are robust and high performing. To address this, researchers have turned to quantum engineering, particularly the use of squeezed vacuum states (Takeno et al., 2007), which offers a promising route to overcoming these limitations. At present the highest level of squeezing has been generated from a bulk optics apparatus (Vahlbruch et al., 2016) that used a semi-monolithic cavity configuration. The development and application of an integrated squeezed vacuum source (Kashiwazaki et al., 2023) have great potential for many fields including gravitational wave detection, quantum sensing (Casacio et al., 2021), and quantum computing (Eli Bourassa et al., 2021; Larsen et al., 2025) since it can greatly simplify the experimental set-ups currently built with hundreds of bulk optics components (Takeno et al., 2007).

Here we will focus on the potential of the use of lithium niobate periodically poled waveguides (Lenzini et al., 2018) and their suitability for the generation of highly squeezed states that can be used for GW detection when the squeezing is measured in the acoustic band between 10 Hz and 10 kHz, which is the most sensitive frequency region of the interferometric GW detectors, and quantum computation for the high frequency region. This material is characterized by large second order nonlinearity and the compatibility with several fabrication techniques for making optical waveguides. Proton exchanged (Lenzini et al., 2018) and titanium in-diffused (Stefszky et al., 2017) waveguides have been used in the past for the generation of squeezed vacuum, but more recently other approaches based on thin film lithium niobate (Zhu et al., 2021) (TFLN) geometries have been used. These devices include ridge TFLN waveguide (Nehra et al., 2022), strip-loaded waveguide (Peace et al., 2022), and bonding-dicing waveguides (Kashiwazaki et al., 2023).

In interferometric measurements, quantum noise manifests in two primary forms: shot noise and radiation pressure noise. Shot noise arises from the statistical fluctuation in the number of photons detected, which results in a variation of the photocurrent, while radiation pressure noise is due to the fluctuation of the momentum transferred between photons and mechanical objects. In most cases, shot noise dominates; however, radiation pressure noise becomes significant when very light mechanical objects or very high laser powers are involved, such as in interferometric GW detectors.

A common approach to mitigate shot noise is by increasing the laser power. Since shot noise scales with the square root of the circulating power while the signal scales linearly, increasing the power improves the signal-to-noise ratio (SNR). However, this strategy is not without its challenges. The power required to achieve the desired SNR might exceed the capabilities of the laser systems available, or the optical components might not be able to handle the increased power without introducing additional noise. Furthermore, increasing power beyond a certain threshold can also introduce additional radiation pressure noise, which may negate the benefits of reducing shot noise. In these situations, quantum engineering offers a viable alternative.

The fundamental limit to reducing both shot noise and radiation pressure noise simultaneously arises from the Heisenberg uncertainty principle, which dictates that the amplitude and phase fluctuations of the coherent vacuum state cannot be minimized at the same time. However, it is possible to engineer specific quantum states of light, such as squeezed states, in which one type of noise is reduced at the expense of increasing the other. Squeezing a quantum state involves reducing the uncertainty in one quadrature (such as amplitude) while increasing the uncertainty in the conjugate quadrature (such as phase).

By applying squeezed vacuum states, the sensitivity of high-precision instruments can be enhanced. For instance, in gravitational wave detectors like Advanced LIGO and Advanced Virgo during the observational run O3, squeezed light has been used to reduce shot noise (Virgo Collaboration Aga et al., 2019; Tse et al., 2019), resulting in an increase of the Universe explorable volume by approximately 50%. This improvement is particularly crucial: to achieve the same sensitivity by merely increasing the laser power in the interferometer, it will require almost doubling the amount currently present in the interferometric GW detector, something that is technically challenging with current technology. In a more recent observational run (O4) the observable universe has been further increased up to 65% using well engineered squeezed vacuum states (Ganapathy et al., 2023) (see Figure 1). These are generated by exploiting the dispersion in reflection of a detuned optical cavity to which a frequency-independent squeezing field is injected. This technology is the baseline design of all current and future detectors such as the Einstein Telescope (Einstein Telescope, 2020) and Cosmic Explorer (Cosmic Explorer, 2019), providing them with the sensitivity to detect signals that are nowadays impossible to measure.

While squeezed vacuum sources have demonstrated their potential in improving the sensitivity of gravitational wave detectors, their implementation remains complex. These systems often require extensive infrastructure, expert knowledge, and dedicated space, which make them difficult to integrate into existing experimental setups. Additionally, traditional squeezed vacuum sources are typically bulky and require delicate alignment and tuning.

To overcome these challenges, researchers have focused on the development of integrated squeezed vacuum sources that are compact, stable, and easily adaptable to existing experimental frameworks. One promising approach is the use of integrated optical devices, such as lithium niobate (LiNbO3) waveguides, which offer a range of advantages over bulk optical sources. These integrated devices can provide the necessary nonlinear response to generate squeezed states through processes like parametric down-conversion, with the added benefits of stability and scalability.

Lithium niobate has become a leading material for integrated quantum technologies due to its excellent nonlinear optical properties, low propagation loss, and electro-optical reconfigurability. Moreover, the creation of periodically-poled nonlinear waveguides and its transparency window that goes from UV to mid infrared allows the generation of squeezed light over a broad spectral range. Among the different fabrication techniques mentioned in the introduction, devices based on thin film lithium niobate (TFLN) (Saravi et al., 2021) offer greater nonlinear conversion efficiency because of the strongest confinement when compared to proton exchanged or titanium in-diffused waveguides which are generally long several centimeters. The bonding-dicing (Kashiwazaki et al., 2023) approach has shown great potential for the generation of highly squeezed states. These devices are fabricated by bonding a TFLN layer over a carrier substrate and then carving ridges a few microns wide, depending on the working wavelength, with a high precision, self-polishing dicing saw (Berry et al., 2019) (see Figure 2C). This saw leaves the sidewall of the device extremely smooth achieving propagation losses as low as 0.1 dB/cm (Kashiwazaki et al., 2021) compared to the 0.3 dB/cm of etched ridge waveguides (Luke et al., 2020). This type of waveguide has demonstrated the generation of more than 8 dB squeezing (Kashiwazaki et al., 2023), resulting in the highest value reported from an integrated device. They also have a lower numerical aperture which makes the collection of the generated squeezed light easier than from ridge waveguides. Figures 2A,B shows some results of parametric gain and squeezing measured at 1,550 nm in our laboratory via homodyne detection at the 8 MHz sideband. While the set-up still requires optimization in terms of losses of the optical components to improve the level of squeezing, it is quite simple to build and operate compared to bulk optical systems. Furthermore, to measure squeezing in the frequency region of interest for GW detection (from 10 Hz to 10 kHz), schemes for the reduction of low frequency noise are required such as the coherent control described in the next section, and low noise electronics for the circuitry of the homodyne detector.

The main limitation of this type of waveguides is that because of the dicing step, only straight waveguides can be fabricated, while the other techniques can integrate on a single chip optical circuits with multiple functionalities such as nonlinear waveguides, directional couplers, modulators, MMIs and more.

The development of integrated quantum sources for squeezed vacuum has the potential to revolutionize high-precision measurements in interferometric GW detectors, quantum metrology, and quantum technologies. By addressing the limitations imposed by quantum noise, such as shot noise and radiation pressure noise, these devices enable measurements beyond the standard quantum limit, enhancing the sensitivity and performance of scientific instruments. The integration of squeezed light sources into experimental setups will significantly reduce costs, improve ease of implementation, and open the door to new discoveries in the field of gravitational wave detection and quantum technologies.