On the expression of evolution operator for a time-independent Hamiltonian in quantum mechanics Given the TDSE for the evolution operator $U(t,t_0)$:
$$
{\displaystyle i\hbar {\frac {d}{dt}}U(t, t_0)={\hat {H}}U(t,t_0)}
$$
and assuming a time-independent Hamiltonian $\hat H$, and the boundary condition $U(t_0, t_0)=\mathbb{I}$, the general solution takes the form:
$$
U(t,t_0) = e^{-\frac i \hbar \hat H(t-t_0)}.
$$
In all the references I've come across so far, this solution is given without any proof, for it's probably seen as an obvious extension of a linear ordinary differential equation of order 1, which involves operators instead of functions.
I'm sorry, but I can't see all this resemblance.
For example, how should the following be interpreted if one blindly applies the separation of variables method?
$$
{\frac {d U(t,t_0)}{U(t,t_0)}}=-\frac{i}{\hbar}{\hat {H}}dt.
$$
You can't divide for a vector, so just imagine doing it with an operator...
 A: $\newcommand{\bl}[1]{\boldsymbol{#1}} 
\newcommand{\e}{\bl=}
\newcommand{\p}{\bl+}
\newcommand{\m}{\bl-}
\newcommand{\mb}[1]{\mathbf {#1}}
\newcommand{\mc}[1]{\mathcal {#1}}
\newcommand{\mr}[1]{\mathrm {#1}}
\newcommand{\mf}[1]{\mathfrak{#1}}
\newcommand{\gr}{\bl>}
\newcommand{\les}{\bl<}
\newcommand{\greq}{\bl\ge}
\newcommand{\leseq}{\bl\le}
\newcommand{\il}[1]{$\:#1\:$}
\newcommand{\plr}[1]{\left(#1\right)}
\newcommand{\blr}[1]{\left[#1\right]}
\newcommand{\vlr}[1]{\left\vert#1\right\vert}
\newcommand{\Vlr}[1]{\left\Vert#1\right\Vert}
\newcommand{\lara}[1]{\left\langle#1\right\rangle}
\newcommand{\lav}[1]{\left\langle#1\right|}
\newcommand{\vra}[1]{\left|#1\right\rangle}
\newcommand{\lavra}[2]{\left\langle#1|#2\right\rangle}
\newcommand{\lavvra}[3]{\left\langle#1\right|#2\left|#3\right\rangle}
\newcommand{\vp}{\vphantom{\dfrac{a}{b}}}
\newcommand{\Vp}[1]{\vphantom{#1}}
\newcommand{\hp}[1]{\hphantom{#1}} 
\newcommand{\x}{\bl\times}
\newcommand{\ox}{\bl\otimes}
\newcommand{\ol}[1]{\overline{#1}}
\newcommand{\qqlraqq}{\qquad\bl{-\!\!\!-\!\!\!-\!\!\!\longrightarrow}\qquad}
\newcommand{\qqLraqq}{\qquad\boldsymbol{\e\!\e\!\e\!\e\!\Longrightarrow}\qquad}
\newcommand{\tl}[1]{\tag{#1}\label{#1}}
\newcommand{\hebl}{\bl{=\!=\!=\!==\!=\!=\!==\!=\!=\!==\!=\!=\!==\!=\!=\!==\!=\!=\!==\!=\!=\!==\!=\!=\!==\!=\!=\!==\!=\!=\!==\!=\!=\!==\!=\!=\!=}}$
I think that for a time-independent Hamiltonian $\:\hat{\mr H}\plr{t}\e\hat{\mr H}\:$ we could derive the exponential expression of the time evolution operator
\begin{equation}
\mr U\plr{t,t_0}\e e^{\m i\,\hat{\mr H}\plr{t\m t_0}/\hbar}
\tl{01}
\end{equation}
using its property
\begin{equation}
\mr U\plr{t,t_0}\e \mr U\plr{t,t_k}\mr U\plr{t_k,t_0}
\tl{02}
\end{equation}
and the identity
\begin{equation}
\lim_{n\bl\rightarrow\infty}\plr{1\p\dfrac{z}{n}\,}^n \e e^z\qquad \plr{n\bl\in \mathbb N\,, z\bl\in \mathbb C}
\tl{03}
\end{equation}
So, suppose that we insert in the time interval $\:\plr{t_0,t}\:$ a number of $\:\plr{n\m 1}\:$ time moments  $\: \plr{n\bl\in\mathbb N}\:$ dividing it into $\:n\:$ sub-intervals
\begin{equation}
t_0\les t_1\les t_2\les\cdots\les t_k\les\cdots\les t_n\bl\equiv t  
\tl{04}
\end{equation}
Using the property \eqref{02} we can build the operator $\:\mr U\plr{t,t_0}\:$ from a product of $\:n\:$ operators $\:\mr U\plr{t_k,t_{k\m 1}}\:$

\begin{align}
\mr U\plr{t,t_0}&\e\mr U\plr{t_n,t_0} \e\mr U\plr{t_n,t_{n\m 1}}\mr U\plr{t_{n\m 1},t_0}\e\mr U\plr{t_n,t_{n\m 1}}\mr U\plr{t_{n\m 1},t_{n\m 2}}\mr U\plr{t_{n\m 2},t_0}
\nonumber\\
&\e\mr U\plr{t_n,t_{n\m 1}}\mr U\plr{t_{n\m 1},t_{n\m 2}}\cdots\mr U\plr{t_k,t_{k\m 1}}\cdots\mr U\plr{t_3,t_2}\mr U\plr{t_2,t_1}\mr U\plr{t_1,t_0}
\tl{05}
\end{align}

that is
\begin{equation}
\mr U\plr{t,t_0}\e \prod\limits_{k\e 1}^{k\e n}\mr U\plr{t_k,t_{k\m 1}}  
\tl{06}
\end{equation}
Now, make the $\:\plr{n\m 1}\:$ time moments equidistant
\begin{align}
t_k\m t_{k\m 1}&\e \Delta t \vp
\tl{07a}\\
\Delta t&\e\dfrac{t\m t_0}{n}
\tl{07b}
\end{align}
Taking $\:n\:$ extremely very large the time interval $\:\Delta t\:$ is infinitesimally very small so we could write
\begin{align}
t_k\m t_{k\m 1}&\e \mr dt \vp
\tl{08a}\\
\mr dt&\e\dfrac{t\m t_0}{n}\qquad \plr{n\gr\gr 1}
\tl{08b}
\end{align}
and  equation \eqref{06} yields
\begin{equation}
\mr U\plr{t,t_0}\e \prod\limits_{k\e 1}^{k\e n}\mr U\plr{t_{k\m 1}\p \mr dt ,t_{k\m 1}}  
\tl{09}
\end{equation}
But every factor in the rhs of \eqref{09} is an infinitesimal time evolution operator of the form $\:\mr U\plr{t\p \mr dt ,t}\:$ which satisfies the following equation
\begin{equation}
\mr U\plr{t\p \mr dt ,t}\bl\approx 1\!\!1\m\dfrac{i}{\hbar}\hat{\mr H}\,\mr dt
\tl{10}
\end{equation}
as proved and explained in the APPENDIX. Note that $\:1\!\!1\:$ is the identity operator.
So, we have
\begin{equation}
\mr U\plr{t_{k\m 1}\p \mr dt ,t_{k\m 1}}\bl\approx 1\!\!1\m\dfrac{i}{\hbar}\hat{\mr H}\,\mr dt\e 1\!\!1\m\dfrac{i\,\hat{\mr H}\plr{t\m t_0}/\hbar}{n}\qquad \plr{k\e1,2,\cdots,n}
\tl{11}
\end{equation}
and from \eqref{09} we obtain
\begin{equation}
\mr U\plr{t,t_0}\bl\approx \blr{1\!\!1\m\dfrac{i\,\hat{\mr H}\plr{t\m t_0}/\hbar}{n}}^n  
\tl{12}
\end{equation}
Using the identity \eqref{03} we prove the exponential expression \eqref{01} of the time evolution operator
\begin{equation}
\mr U\plr{t,t_0}\e \lim_{n\bl\rightarrow\infty}\blr{1\!\!1\m\dfrac{i\,\hat{\mr H}\plr{t\m t_0}/\hbar}{n}}^n \e e^{\m i\,\hat{\mr H}\plr{t\m t_0}/\hbar}
\tl{13}
\end{equation}
$\hebl$
APPENDIX : $\:\texttt{The infinitesimal time evolution operator }\mr U\plr{t\p \mr dt ,t}$
The Schrodinger equation
\begin{equation}
i\,\hbar\dfrac{\mr d\vra{\psi\plr{t}}}{\mr dt}\e \hat{\mr H}\plr{t}\vra{\psi\plr{t}}
\tl{A-01}
\end{equation}
is linear in $\:\vra{\psi\plr{t}}\:$ so we expect this state to be obtained from an initial state $\:\vra{\psi\plr{t_0}}\:$ at time  $\:t_0\:$  via a linear operator
\begin{equation}
\vra{\psi\plr{t}}\e \mr U\plr{t,t_0}\vra{\psi\plr{t_0}}
\tl{A-02}
\end{equation}
the time evolution operator $\:\mr U\plr{t,t_0}$. Since \eqref{A-02} would be  valid equally well for the evolution of $\:\vra{\psi\plr{t}}\:$ from time $\:t\:$  to time $\:t\p\mr dt\:$ we have
\begin{equation}
\boxed{\:\:\vra{\psi\plr{t\p\mr dt}}\e \mr U\plr{t\p\mr dt,t}\vra{\psi\plr{t}}\:\:\Vp{\tfrac{\dfrac{a}{b}}{\dfrac{a}{b}}}}
\tl{A-03}
\end{equation}
The operator $\:\mr U\plr{t\p\mr dt,t}\:$ is the infinitesimal time evolution operator.
From equation \eqref{A-01} we have
\begin{equation}
\!\!\!\!\!\!\!\!i\hbar\dfrac{\mr d\vra{\psi\plr{t}}}{\mr dt}\e \hat{\mr H}\plr{t}\vra{\psi\plr{t}}\bl\implies\vra{\psi\plr{t\p\mr dt}}\m\vra{\psi\plr{t}}\bl\approx  \m \dfrac{i}{\hbar}\,\hat{\mr H}\plr{t}\vra{\psi\plr{t}}\mr dt
\tl{A-04}
\end{equation}
so
\begin{equation}
\boxed{\:\:\vra{\psi\plr{t\p\mr dt}}\bl\approx  \blr{1\!\!1\m \dfrac{i}{\hbar}\,\hat{\mr H}\plr{t}\mr dt}\vra{\psi\plr{t}}\:\:\Vp{\tfrac{\dfrac{a}{b}}{\dfrac{a}{b}}}}
\tl{A-05}
\end{equation}
where $\:1\!\!1\:$ is the identity operator.
Comparing equations \eqref{A-03},\eqref{A-05} we obtain
\begin{equation}
\boxed{\:\:\mr U\plr{t\p\mr dt,t}\bl\approx 1\!\!1\m \dfrac{i}{\hbar}\,\hat{\mr H}\plr{t}\mr dt\:\:\Vp{\tfrac{\dfrac{a}{b}}{\dfrac{a}{b}}}}
\tl{A-06}
\end{equation}
We note that because of \eqref{A-06} equation \eqref{10} is valid for time-dependent Hamiltonian in general.
