A new quantum key distribution (QKD) scheme, twin-field QKD (TF-QKD), is introduced to overcome the fundamental rate-distance limit of point-to-point QKD without using quantum repeaters. The scheme involves generating phase-randomized optical fields at two distant locations and combining them at a central measuring station. These fields, called "twins," can be used to distill a quantum key. The key rate of TF-QKD scales with the square root of the channel transmittance, similar to a quantum repeater, but is feasible with current technology and can operate over 550 km of standard optical fibre with manageable noise levels. This is promising for overcoming the QKD rate-distance barrier and extending the range of secure quantum communications. The scheme is compared to conventional QKD in terms of key rate versus distance. Theoretical bounds for key rate versus distance are plotted, showing that TF-QKD can surpass the secret key capacity (SKC) bound after 340 km of optical fibre. The key rate of TF-QKD scales with the square root of the channel transmittance, unlike conventional QKD, which scales linearly. This results in a significant improvement in the key rate-distance figure. TF-QKD uses phase-randomized optical fields and decoy states to extend the distance of secure quantum communications. It retains the measurement-device-independent (MDI) characteristic of MDI-QKD while achieving the square-root dependence of the key rate on the channel transmittance. This provides an advantage over MDI-QKD even at short distances when Charlie's detectors have low efficiency. The scheme is implemented using components similar to decoy-state MDI-QKD. The users independently choose random phase values for their light sources, which are then revealed to identify the twin fields. This process results in an intrinsic quantum bit error rate (QBER) due to the close but not exact identical phases of the twin fields. The optimal number of phase slices, M, is determined to minimize the QBER and maximize the key rate. The main technical challenge in implementing TF-QKD is controlling the phase evolution of the twin fields over long distances. The phase drift between the two optical paths is measured and shown to be negligible for practical distances. The visibility of the interferometer remains high, contributing minimally to the QBER. The findings suggest that the point-to-point secret key capacity of a quantum channel can be overcome without using quantum repeaters. This is not at variance with existing results, as TF-QKD is not point-to-point. The security of TF-QKD does not depend on the measurement devices, and its single-photon nature entails count and error rates similar to standard QKD. Further work is needed to prove the unconditional security of TF-QKD, which is an important open question. The counter-intuitive features of the new scheme are expected to stimulate further research extending the limits of QKD.A new quantum key distribution (QKD) scheme, twin-field QKD (TF-QKD), is introduced to overcome the fundamental rate-distance limit of point-to-point QKD without using quantum repeaters. The scheme involves generating phase-randomized optical fields at two distant locations and combining them at a central measuring station. These fields, called "twins," can be used to distill a quantum key. The key rate of TF-QKD scales with the square root of the channel transmittance, similar to a quantum repeater, but is feasible with current technology and can operate over 550 km of standard optical fibre with manageable noise levels. This is promising for overcoming the QKD rate-distance barrier and extending the range of secure quantum communications. The scheme is compared to conventional QKD in terms of key rate versus distance. Theoretical bounds for key rate versus distance are plotted, showing that TF-QKD can surpass the secret key capacity (SKC) bound after 340 km of optical fibre. The key rate of TF-QKD scales with the square root of the channel transmittance, unlike conventional QKD, which scales linearly. This results in a significant improvement in the key rate-distance figure. TF-QKD uses phase-randomized optical fields and decoy states to extend the distance of secure quantum communications. It retains the measurement-device-independent (MDI) characteristic of MDI-QKD while achieving the square-root dependence of the key rate on the channel transmittance. This provides an advantage over MDI-QKD even at short distances when Charlie's detectors have low efficiency. The scheme is implemented using components similar to decoy-state MDI-QKD. The users independently choose random phase values for their light sources, which are then revealed to identify the twin fields. This process results in an intrinsic quantum bit error rate (QBER) due to the close but not exact identical phases of the twin fields. The optimal number of phase slices, M, is determined to minimize the QBER and maximize the key rate. The main technical challenge in implementing TF-QKD is controlling the phase evolution of the twin fields over long distances. The phase drift between the two optical paths is measured and shown to be negligible for practical distances. The visibility of the interferometer remains high, contributing minimally to the QBER. The findings suggest that the point-to-point secret key capacity of a quantum channel can be overcome without using quantum repeaters. This is not at variance with existing results, as TF-QKD is not point-to-point. The security of TF-QKD does not depend on the measurement devices, and its single-photon nature entails count and error rates similar to standard QKD. Further work is needed to prove the unconditional security of TF-QKD, which is an important open question. The counter-intuitive features of the new scheme are expected to stimulate further research extending the limits of QKD.