CD4051BE

125 Ω is the number that decides if you can use it

The switch resistance adds directly to the source impedance. Feeding a 12-bit ADC with a 10 kΩ source through a 125 Ω switch is fine, because the switch adds 1.2 % of error that a calibration removes. Feeding a 1 MΩ source through it is not: the switch forms a divider with the ADC's input impedance and the reading collapses. Keep source impedance below about 10 kΩ, or buffer each channel.

On-resistance varies with signal level

The switch is a CMOS transmission gate, and its resistance changes as the signal moves away from the rails — typically rising 30-50 % from mid-supply to near Vdd. That variation is why a multiplexed signal shows a small level-dependent gain error per channel. At 15 V supply and ±5 Ω channel matching, the variation between channels is small; at 5 V the matching is worse and the effect is larger.

Break-before-make and charge injection

The 4051 breaks the old channel before making the new one, so there is a brief open circuit during the transition — that is generally what you want, but with a high-impedance load the output floats during the interval. Charge injection of a few picocoulombs also puts a small step on the output at each switch, which a slow ADC will integrate into the reading. Allow 1-2 µs of settling after a channel change.

Sources

Texas Instruments — https://www.ti.com/lit/an/slva840/slva840.pdf · AEC — https://aecouncil.com/Documents/AEC_Q100_Rev_H_Base_Document.pdf · NIST — https://www.nist.gov/standardsgov/compliance-faqs-rohs