UPDATE : I found some precise references that answer the OP's question and fill in some details in my original answer. See the end for them.

I don't have a precise reference for the cases of $n=2$ or $n=3$, though I suspect that they could be found in Moise's "Geometric Topology in Dimensions 2 and 3".

However, I do have a nice reference for the amazing theorem (due to Stallings) that $\mathbb{R}^n$ has a unique $C^{\infty}$ structure for $n \geq 5$. It is written up very nicely in Steve Ferry's Geometric Topology Notes. See Chapter 10 starting on page 56. What he proves is that $\mathbb{R}^n$ has a unique PL structure, but that is the key to the result. The rest of the notes are also wonderful.

This theorem is surprising for a number of reasons. For instance, it says that if $\Sigma$ is an exotic $n$-sphere for $n \geq 5$, then $\Sigma \setminus \{p\}$ is diffeomorphic to $\mathbb{R}^n$ for any point $p \in \Sigma$. In other words, the "exoticness" is "concentrated at a point". Of course, this also follows from the usual proof of the high dimensional Poincare conjecture using the $h$-cobordism theorem, which constructs a homeomorphism between a homotopy sphere and the usual sphere which is differentiable except at one point (that one point giving trouble due to the "Alexander trick").

Another remark that should be made is that Freedman and Donaldson proved that $\mathbb{R}^4$ has uncountably many $C^{\infty}$ structures.

UPDATE : OK, here are some precise references. In

MR0121804 (22 #12534)
Munkres, James
Obstructions to the smoothing of piecewise-differentiable homeomorphisms.
Ann. of Math. (2) 72 1960 521--554.

it is proven in Corollary 6.6 that two PL-isomorphic differentiable manifolds that are homeomorphic to $\mathbb{R}^n$ are actually diffeomorphic. To make sense of this, recall that Cairns proved that differentiable manifolds have canonical PL structures; the original reference for this is his paper

MR0017531 (8,166d)
Cairns, Stewart S.
The triangulation problem and its role in analysis.
Bull. Amer. Math. Soc. 52, (1946). 545--571.

I think JHC Whitehead might have also proven this, but I don't have a reference for that.

This reduces us to showing that $\mathbb{R}^n$ has a unique PL structure. The cases $n=2$ and $n=3$ can be found in Moise's book "Geometric Topology in Dimensions 2 and 3". As far as the original results go, the 2-dimensional case is classical, while the 3-dimensional case was originally proven by Moise in his paper

MR0048805 (14,72d)
Moise, Edwin E.
Affine structures in $3$-manifolds. V. The triangulation theorem and Hauptvermutung.
Ann. of Math. (2) 56, (1952). 96--114.

For dimensions at least $5$, the original reference is the following

MR0149457 (26 #6945)
Stallings, John
The piecewise-linear structure of Euclidean space.
Proc. Cambridge Philos. Soc. 58 1962 481--488.

I hope this is helpful!