Depends on your task and your SEM.

In many (most?) cases researchers use coating and are happy with results.

However, as with any additional step in specimen preparation, coating may induce artifacts. If you have field emission SEM, you can work at low voltages (0.5-1.5 kV) without any coating. The same true if you have environmental SEM, but with ESEM you will work with higher voltages - can be useful for BSE and EDS. FESEM can reach pretty high magnifications even in low voltage and environmental mode.

Limited or non-conductive material samples require carbon and/or metal coating. Generally sputtering is used , 2-20 nm of electrically-conducting metal such as gold (Au), gold/palladium (Au/Pd), platinum (Pt), silver (Ag), chromium (Cr) or iridium (Ir) prevents charging.

Depends on your task and your SEM.

In many (most?) cases researchers use coating and are happy with results.

However, as with any additional step in specimen preparation, coating may induce artifacts. If you have field emission SEM, you can work at low voltages (0.5-1.5 kV) without any coating. The same true if you have environmental SEM, but with ESEM you will work with higher voltages - can be useful for BSE and EDS. FESEM can reach pretty high magnifications even in low voltage and environmental mode.

The answer of Prof.Vladimir Dusevich is right and proper. The low or high votage is the key factor for high magnification imag. However the imag quality may be not good, and for a long time e beam exposure, your sample may be destucted more or less specially for pmma.So AFM is proper tool to scan your nonconductive sample. Also if your sem does not work good at the low voltage, I still suggest you coat conduct atop your sample, the lowcost process is to spincoat condutive resin which can be washed out by water if your sample is not affected. Or you can choose to coat other mental, as long as it would be removed by other material and this material will not affect your sample, also it is high cost.