Sandia anchors a nanofiber between silicon pins to trap atoms with 5 milliwatts

The number people will repeat is 5 milliwatts, and anyone sizing a navigation unit for an airframe needs to know which line of the power budget it belongs on. In Sandia's work it is the light delivered to the atoms. "Building on our nanofiber results, we showed cesium atoms can be trapped with just five milliwatts of optical power and that atomic coherence can be measured using sub-microwatt fiber-coupled beams, all while minimizing in-vacuum heat loads," Lee said in a press release ^[5].

Heat is what made this hard. "This idea has not been fully realized by the community for decades, due to challenges in dissipating heat from photonic devices in vacuum and in efficiently loading atoms around them," said Jongmin Lee, a quantum sensing scientist at the lab ^[4][6]. Before the pinned design, the choice was between a fragile structure that loads atoms well and breaks under heat, and a sturdy one that handles heat and not the atoms ^[7].

Interesting Engineering describes the 5 milliwatts as about 2,000 times less than a standard LED bulb ^[3], which puts the implied bulb at 10 watts ^[15]. The coherence measurement is the smaller figure: a sub-microwatt beam sits at least 5,000 times below the 5 milliwatt trap ^[16]. The same account gives the fiber as 420 nanometers thick and 200 times thinner than a human hair, implying a hair about 84 micrometers across ^[2][17].

What the account does not contain is a navigation number. There is no interferometer sensitivity, no bias drift figure, no mass, and no total system draw ^[18]. The results published in AVS Quantum Science are trapping power, coherence measurement and in-vacuum heat load ^[11][5].

Milliwatt claims from this field sort onto three lines: optical power delivered to the atoms, electrical power to generate and control that light, and the power to hold vacuum and carry heat away. Sandia moved the first line and, on Lee's own account, cut the load the third has to remove ^[5]. A result that moves the first line is a component result, so a program office reading it is choosing whether to fund integration now or wait for the chip and the sensor array the group says it wants to build ^[10]. The price of waiting is that the integration knowledge sits with whoever gets the guide and its optics onto one circuit first.