Riiven Threads

Barometric Pressure Sensor

Atmosphere, Carved in Silicon

Four trades, one grain of silicon

Step out of an elevator and your phone already knows you climbed three floors. It did not ask. Your phone knew you took the elevator, not the stairs, without you telling it anything, because a sensor smaller than a grain of rice counted the air. Air thins as you rise, and a single floor shifts pressure by roughly 12 pascals out of about 101,000. Reading a change that small means carving a silicon membrane thin enough to flex under the weight of the sky, then trusting the number it gives. The question is how four separate trades, none of which set out to track footsteps, ended up doing it together.

50x
Doped silicon reads strain about 50 times more strongly than a metal gauge.
300micrometers
Depth-to-width ratio DRIE carves into silicon to shrink sensor cavities.
5x
Naive membrane scaling raises central deflection fivefold, breaking the linear range.
1hPa
Uncompensated drift across the ambient range, faking about 8 meters of altitude.

How each idea was handed down

Barometric Pressure Sensor inherited its parts in sequence. Each field handed its result to the next.

Follow the inheritance, link by link.

Ordered from the earliest ancestor to the most recent heir.

01

Keystone

The silicon trick that made stress speak

Piezoresistive Effect in Semiconductors materials science matured 1962 C. S. Smith, T. Manku

Squeeze a doped silicon bar and its resistance jumps far more than the squeeze alone explains. C. S. Smith measured that in 1962.

A bathroom scale turns your weight into a number by stretching a metal strip until it resists current a little more. Silicon does the same job, but wildly better. The piezoresistive effect, the way a material's electrical resistance changes when you bend it, runs about 50 times stronger in doped silicon than in metal. Without semiconductor piezoresistance, a barometric sensor would be about 50 times less sensitive, dropping its output from roughly 10 mV/V to around 0.2 mV/V for the same strain. That gap decides everything downstream. A weak signal needs a huge membrane and hungry electronics to be readable; a strong one fits inside a phone running on a coin cell. Four resistors wired into a Wheatstone bridge, a simple balance that flags tiny imbalances, turn membrane stress into a clean voltage.

Without this link

Without the piezoresistive effect in doped silicon, MEMS barometric sensors could not translate diaphragm stress into a sizeable resistance change, leaving outputs near metal-gauge levels and forcing much larger diaphragms and higher excitation power to reach usable resolution.

Without semiconductor piezoresistance, barometric sensors are about 50 times less sensitive than doped silicon designs, dropping output from 10 to 0.2 mV/V.

How we know

Smith's 1954 measurements, formalized for microsystems in Manku's 2010 review, showed silicon's gauge factor reaching the low hundreds versus roughly 2 for metal foil, the headroom that makes compact piezoresistive bridges practical.

Source: NIH piezoresistance review (2010) · tier1

A strong signal still needs something to flex. Next came a way to carve the silicon membrane thin enough to bend.

02

Etching walls straight down into silicon

Deep Reactive Ion Etching Microfabrication chemistry matured 1995 Klaus Bosch, Yu-Chong Tai

Wet acid eats silicon sideways as fast as it eats down, rounding every wall. In 1995 the Bosch process etched straight trenches instead.

Older chemical etching dissolves silicon in all directions at once, so it cannot cut a narrow deep slot without widening the top. Deep reactive ion etching, a method that alternates a cutting gas with a protective coating, carves nearly vertical walls one thin layer at a time. Without DRIE grooves reaching a 300 micrometer depth-to-width ratio, sensor cavities cannot shrink, forcing larger membrane openings and sharply weaker response to small pressure changes. That control is what lets a diaphragm be both small and thin at once.

Without this link

Without deep reactive ion etching, silicon diaphragms could not be thinned and structured with high aspect ratio cavities, leaving membranes too thick and stiff. Wet etching alone yields limited aspect ratios, larger openings, and poorer vertical-wall control, reducing full-scale output and accuracy.

Without DRIE grooves at a 300 micrometer depth-to-width ratio, cavities cannot miniaturize, forcing larger openings and weaker barometric sensitivity.

How we know

A thin membrane you can carve is only useful if you know how thin to make it before it bursts.

03

Knowing how far a membrane can bend

Thin Diaphragm Mechanics and Plate Theory engineering matured 1999 N. D. N. T. Sirimanna, F. Pashmforoush

A drum skin pulled too tight tears; pushed too gently it barely moves. A sensor membrane lives in the same narrow band.

Push on a thin plate and it bows in the middle. Push too hard and pressure and stress stop tracking in a straight line, which wrecks accuracy. Thin plate theory, the math Kirchhoff and Love worked out for how flat sheets bend, predicts deflection and peak stress from radius and thickness. Size the membrane by naive scaling instead, and central deflection jumps 5 times for the same pressure, pushing it past the linear range toward burst failure. The theory tells engineers exactly how thin is too thin.

Without this link

Without thin diaphragm mechanics, engineers could not hold membrane deflection inside the small-deflection regime needed for linear stress-pressure behavior, causing unpredictable sensitivity and nonlinearity, nor could they predict deflection and peak stress from geometry to avoid burst or touch-mode failure.

Without Kirchhoff-Love plate theory sizing the membrane, naive scaling raises central deflection 5 times, pushing the sensor out of its linear range.

How we know

The 1999 TechConnect analytical solution tied central deflection directly to geometry, and naive scaling drives metrological sensitivity to 9.1e-10 m/Pa, outside the valid small-deflection regime where stress stays proportional to pressure.

Source: TechConnect Briefs 1999 diaphragm deflection (1999) · tier2

A perfectly sized membrane still lies if heat alone makes it swell. The last fix separated temperature from true pressure.

04

Telling real pressure from a warm day

Thermodynamics of Gas Laws and Temperature Compensation physics matured 2024 Guo Zhang, MDPI Sensors team

Leave a sealed sensor in a hot car and the trapped air expands, faking a pressure drop. Heat and altitude look identical.

Air pressure and air temperature are bound together by the same gas law, so a warming sensor reports altitude change that never happened. Self-temperature compensation models that thermal expansion and subtracts it before the number leaves the chip. Without compensation drawn from the gas laws, a barometer drifts up to 1 hPa across a minus 20 to plus 60 degrees C range, which fakes roughly 8 meters of altitude. Eight meters is more than two floors, enough to ruin the elevator trick the whole package exists to do. Zhang's 2024 quartz-resonant design folds the correction into the sensor itself.

Without this link

Without thermodynamic gas law modeling and temperature compensation, sensors interpret thermal expansion of the element and surrounding air as real pressure change. Across the ambient range this drives full-scale errors from below 0.1% up to several percent, corrupting altitude, weather, and gas readings by tens of pascals.

Without gas-law temperature compensation, a barometer drifts up to 1 hPa across the ambient range, faking about 8 meters of altitude.

How we know

Watch

A visual companion to the fields above.

The history of the barometer (and how it works) - Asaf Bar-Yosef ยท TED-Ed

Takeaway

None of these four trades was chasing your footsteps. Smith was probing how strain moves electrons in silicon in 1962. The Bosch process in 1995 wanted vertical trenches for entirely different chips. Plate theory predates the transistor by decades. Temperature compensation only matured in 2024. Forced together inside a cavity smaller than a grain of rice, they produced a sensor that resolves the 12 pascals between one floor and the next. The piezoresistive effect supplies a signal worth reading. DRIE carves a membrane thin enough to flex. Plate theory keeps that flex honest and linear. Gas-law compensation strips out the heat that would otherwise masquerade as height. Pull any one link and the chain drops below the resolution a phone needs to tell stairs from an elevator. The atmosphere has not changed in millions of years; what changed is that we learned to carve something small enough to feel it shift by a single floor.

References

  1. NIH piezoresistance review (2010) tier1

    T. Manku, Semiconductor Piezoresistance for Microsystems, Micromachines, 2010

  2. Micromachines 2020 DRIE pressure sensors (2020) tier1

    Zhang et al, Recent Progress of Miniature MEMS Pressure Sensors, Micromachines, 2020

  3. TechConnect Briefs 1999 diaphragm deflection (1999) tier2

    TechConnect: A New Analytical Solution for Diaphragm Deflection and its Application to a Surface Micromachined Pressure Sensor, 1999

  4. Self temperature compensation barometer (MDPI Sensors 2024) (2024) tier1

    Zhang et al, A Self-Temperature Compensation Barometer Based on All-Quartz Resonant Pressure Sensor, Sensors, 2024

Enjoyed this?

New Thread every week. Each one pulls a technology apart and traces it back to the fields of science that made it possible.

By subscribing, you agree to our Privacy Policy . Unsubscribe anytime.