Turn a smartphone from portrait to landscape and the display follows within a fraction of a second. The motion feels almost effortless, but the measurement begins with a physical structure inside the phone moving through a distance measured in nanometres.
That structure is part of a MEMS accelerometer, short for micro-electro-mechanical system. Inside it, a silicon proof mass hangs from flexible beams while rows of microscopic conductive fingers sit beside matching fixed electrodes, forming a mechanical machine small enough to disappear inside a chip package.
A mechanical machine built in silicon
A MEMS accelerometer works around the same basic arrangement as a weight suspended on springs. When the sensor package accelerates, or when its orientation changes relative to gravity, the proof mass shifts slightly against the beams holding it in place.
The movement is tiny, but the underlying principle is ordinary mechanics. As NIST explains in its overview of accelerometers, the device tracks the position of a proof mass relative to a fixed reference inside the sensor.
Manufacturers can create these structures through several forms of micromachining. In surface micromachining, engineers deposit and pattern thin layers on a wafer before removing a sacrificial layer to release freestanding springs, beams and other movable components.
A University of Maryland guide to surface micromachining describes how spacer layers are removed to free the final structure. Other accelerometers use bulk micromachining or specialised silicon processes, so not every phone sensor is manufactured in exactly the same way.
Why capacitance reveals the movement
The phone cannot inspect the moving mass with a camera or a miniature ruler. Instead, most consumer accelerometers convert its movement into a change in capacitance, the ability of two conductors to store electric charge.
The amount of capacitance depends partly on the area of the conductors and the distance between them. As this overview of capacitors explains, changing the geometry between conductive surfaces changes the amount of electrical charge they can store.
A typical MEMS design places movable fingers between fixed electrodes. When the proof mass shifts toward one set of electrodes and away from another, one capacitance increases while the opposing capacitance decreases.
This creates a differential signal that is easier for the electronics to detect than either measurement alone. Connecting many finger pairs in parallel also strengthens the combined change produced by the movement.
Peer-reviewed research into multi-axis capacitive MEMS accelerometers describes how fixed and movable comb electrodes translate displacement into changing capacitance. Depending on the sensor’s mass, spring stiffness and measurement range, an acceleration of one g may move the structure by a few nanometres or several tens of nanometres.
When a phone is resting on a table, its accelerometer reports roughly one g along the axis opposing gravity. Tilt the phone and that reading is redistributed across the X, Y and Z axes, producing a three-dimensional vector that the software can use to determine which direction is down.
Three axes turn movement into orientation
A smartphone needs acceleration readings in three dimensions. Its sensor therefore measures motion along three perpendicular axes, although the exact internal arrangement varies between manufacturers and chip designs.
The X and Y measurements commonly use comb-shaped electrodes that sense movement within the plane of the silicon die. The Z-axis structure must detect movement perpendicular to the die and may use a tilting proof mass, vertical electrodes, plates beneath the mass or specially arranged comb fingers.
The same review of three-axis accelerometer designs documents several of these arrangements, including in-plane combs and out-of-plane capacitor plates. There is no single microscopic layout shared by every accelerometer used in a smartphone.
The sensor can produce readings tens, hundreds or thousands of times each second, depending on its configuration. The operating system smooths those samples and waits until the change remains stable before rotating the interface.
That pause is deliberate. Without filtering and a minimum duration threshold, the display could rotate whenever the phone briefly wobbled in the user’s hand.
Applications can use the same readings for far more than screen rotation. A phone can become a digital level or angle gauge, as shown in this guide to using an Android phone as a measuring tool, while games can translate tilting and shaking into controls.
The accelerometer can also help diagnose a screen that refuses to rotate. Problems with the sensor, calibration or orientation settings are among the possible causes covered in this guide to fixing Android screen rotation.
From airbag sensors to billions of pockets
MEMS accelerometers entered a major commercial market through automotive safety systems. A University of Maryland history of surface micromachining records that the first commercial application of the process was announced in 1991 for an automotive accelerometer.
One of the important early devices was the ADXL50, a single-chip accelerometer developed for detecting the sudden deceleration associated with a collision. A 2025 peer-reviewed history of inertial measurement technology identifies it as the first manufactured MEMS accelerometer in 1993 and describes its integrated signal-conditioning electronics.
Consumer electronics soon gave the technology a more visible role. Nintendo released the Wii in 2006, and the company later explained that the original Wii Remote measured acceleration along three axes.
Apple followed with the original iPhone in 2007. Its launch announcement specifically described an accelerometer that detected changes between portrait and landscape orientation and adjusted the display automatically.
Accelerometers now count steps in fitness trackers, help drones remain level and detect impacts in vehicles and wearable devices. They also protected the spinning disks in older laptops by helping park their read heads when a fall was detected, although solid-state drives no longer need that particular safeguard.
The limits of measuring nanometres
A structure sensitive enough to detect nanometre-scale movement is also sensitive to disturbances unrelated to orientation. Temperature changes can alter the stiffness of its suspension, mechanical stress can shift its resting position and electrical noise can compete with the capacitance signal.
Manufacturers compensate by calibrating sensors during production, measuring offsets and applying corrections to the output. Phone software may then combine accelerometer readings with information from the gyroscope and magnetometer in a process known as sensor fusion.
There is no single filtering algorithm used by every device. Depending on the hardware and required accuracy, software may use low-pass filtering, complementary filters or Kalman-based estimators, all approaches represented in inertial sensor-fusion systems.
The result still begins with something strikingly physical. Each time the screen turns, gravity changes the load on a silicon mass, microscopic electrodes move through an invisible distance and an electrical circuit notices that the capacitance between them is no longer quite the same.
That movement continues whenever the phone is picked up, tilted, dropped onto a sofa or carried through a room. Long before the interface reacts, a machine built into silicon has already flexed, settled and translated the direction of gravity into numbers.
The mechanism belongs to the same hidden world of physical processes found throughout everyday electronics. Just as a microwave oven can be understood through water molecules responding to an alternating electric field, a rotating phone screen begins with a mechanical structure moving by nanometres, and with the remarkable fact that this is enough.
