This inequality appeared in a print publication:
$$0 \le e^{-nx} - (1-x)^n \le nx^2 e^{-nx} \qquad (0 \le x \le 1)$$
No conditions on $n$ were stated but it's probably safe to take $n \in \mathbb{N}$. I'm specifically interested in proving the right-hand portion of the result. Differentiation produced inconvenient expressions; however, maybe I made bad choices of functional form. Would series expansion be the preferred approach here?
First, note that $1-x \le e^{-x}$ for all $x \in [0,1]$. Hence, $(1-x)^n \le e^{-nx}$, i.e. $0 \le e^{-nx}-(1-x)^n$ for all $x \in [0,1]$.
Also, since the map $y \to ny^{n-1}$ is non-decreasing on $[0,1] \supseteq [1-x,e^{-x}]$, we have $$e^{-nx}-(1-x)^n = \int_{1-x}^{e^{-x}}ny^{n-1}dy \le \int_{1-x}^{e^{-x}}n(e^{-x})^{n-1}dy = ne^{-(n-1)x}[e^{-x}-(1-x)] = ne^{-nx}[1-(1-x)e^x].$$
For $x \in [0,1]$, we have $e^x \ge 1+x$ and $1-x \ge 0$. Hence, $1-(1-x)e^x \le 1-(1-x)(1+x) = x^2$.
Therefore, $e^{-nx}-(1-x)^n \le ne^{-nx}[1-(1-x)e^x] \le nx^2e^{-nx}$, which is the desired upper bound.