Why this is hard
One micron is on the border of what is theoretically possible. You cannot create a parallel ray at this scale. One micron is almost as small as the wavelength of the light you are trying to focus, so the usual rules of optics don't apply.
Lasers normally produce what is known as a Gaussian beam, and they have a distinctive shape. You can focus them down to a small dot, but only if you are willing to let the beam get a lot wider before and after the spot. Looking side on at the beam, you'll get something like this:
(image from wikipedia)
The point where the beam is at it's narrowest is known as the "waist". At that point the beam has a diameter of $W_0$. At a distance of $Z_R$ from the waist, in either direction, the beam has grown to twice the area, or $\sqrt2$ times the diameter.
The equation which links $W_0$ and $Z_R$ is:
$$Z_R = \frac{\pi W_0^2}{\lambda} $$
Where $\lambda$ is the wavelength of the laser you are using. If we assume a 1000nm cheapish laser diode, and a 1um waist, then $Z_R$ is 3.1um. If you go to an expensive blue laser, you can get $Z_R$ up to 7.7um.
How to do it
You will need to bring the laser beam into a convergent lens, and place the sample at the focus. The convergent lens must be very high quality, and it must be held and oriented with micron accuracy. Realistically, you probably won't be able to obtain a good enough single lens, so you will need an assembly of lenses. A microscope objective would probably work, but you won't be able to get a second one close enough to the sample.
I expect the best option would be to buy a microscope with this feature built in. That also has the added advantage that it will have all the safeguards necessary to make sure the beam never makes it to the eyepieces.