Tunable nanophotonic systems that can be integrated with silicon and manipulate light at deep-subwavelength scales is of paramount importance for photonic integrated circuits, [1] enhanced light-matter interaction, [2] sub-diffraction imaging, [3] metamaterials, [4] and negative refraction [5] among other feats. In particular, polaritons in 2D materials exhibit the highest degree of mode confinement. [6] A variety of highly confined polaritons with unique properties [6,7] have been demonstrated in 2D materials, for example, electrically tunable plasmons in graphene, [7a,8] ultralow-loss phonon polaritons in hexagonal boron nitride (h-BN) [9] and α-MoO 3 , [7c,7d] room-temperature exciton polaritons in MoSe 2 , [10] WSe 2 [11] and CsPbBr 3 microsheets, [12] and Cooper-pair polaritons. [13] These surface modes have prompted the exploration of multiple potential applications, such as surface-enhanced infrared spectroscopy, [8a,14] optical sensors, [15] nanolasers, [16] light modulators, [17] and nanophotonic circuits. [18] However, in most practical applications, important challenges, such as relatively large optical losses, reduced operation frequency range, and slow modulation speed, are remaining.The hybrid light-matter nature of polaritons (i.e., quasiparticles made up of a photons coupled to polarization charges in a material) can be passively tuned by designing the composition and geometry of the hosting materials, [9,19] offering important advantages for nanophotonic systems. [6a,20] Recently, van der Waals heterostructures (vdWHs), which can be fabricated by exfoliating 2D materials down to isolated monolayers and subsequently reassembling them in a layer by layer fashion in precisely chosen compositions, stacking sequence, and twist angles, provide a versatile way to customize polaritons. [21] For example, an ultralong lifetime graphene plasmon in graphene/h-BN vdWHs, [7a,22] an efficient all-optical modulation of graphene plasmons in the graphene/ monolayer MoS 2 vdWHs, [23] a strong plasmonic nonlinear response in graphene/metal heterostructures [24] have been demonstrated. Moreover, the behavior of polaritons can also be modulated by creating moiré patterns in the vdWHs, such that one can obtain, for example, an extremely long plasmon lifetime (≈3 ps) in stacked bilayer graphene, [25] photonic crystals of twisted bilayer graphene (TBLG), [26] and tunable topological transitions of phonon polaritons in twisted α-MoO 3 bilayers. [27] 2D monolayers supporting a wide variety of highly confined plasmons, phonon polaritons, and exciton polaritons can be vertically stacked in van der Waals heterostructures (vdWHs) with controlled constituent layers, stacking sequence, and even twist angles. vdWHs combine advantages of 2D material polaritons, rich optical structure design, and atomic scale integration, which have greatly extended the performance and functions of polaritons, such as wide frequency range, long lifetime, ultrafast all-optical modulation, and photonic crystals for nanoscale light. Here, th...