This is a method that, based on my experiments, I would describe as “not there yet,” at least for the home tinkerer. I’ve tried a few different approaches without much success, due to some minor problems with the design of the clip and ultimately a problem with focusing through the glass beads. So as of now I cannot recommend this method, and I suggest you go with the laser pointer method or a clip-on magnifier product. However, I am posting about this method here because I think it does have potential, and anyone wishing to experiment may benefit from hearing about my progress so far. I invite my fellow tinkerers to leave comments about your own experiments.
The basic method has been described by Pacific Northwest National Laboratories (PNNL). It is a 3D printed clip that holds a small glass bead against the lens of your smartphone. The beads, which are already available cheaply for other industrial uses, act as a magnifying lens. PNNL says they developed the technology as a cheap way for law enforcement to detect biohazards. A field investigator may be able to snap a microscopic photo of a suspicious material using a smartphone, which can instantly send the photo to a lab for analysis.

PNNL has kindly made the STL files freely available here. There are magnifications of 100x, 350x and 1000x, and clips designed for iPhones or Android phones. To create the clip, you’ll need a 3D printer and software, or you can use a service like 3D Hubs to connect you to someone who will print the object for you. That’s what I did, and I found a price $10 per order plus $2-3 per object. I tried all 5 different styles of clip. For the material, I used nylon (FDM), which was a good choice, because it is very soft and flexible, and it was necessary to enlarge the hole where the glass bead goes.
One problem with the design of the Android clips is that they did not quite fit my Moto X. The hole for the glass bead needs to be located a little farther away from the bend in the clip. It may fit other phones as is. I also tried jeff1958’s design at Shapeways, and it solved that problem, at least for my phone. The Shapeways product is described as nylon, but it was more brittle than the nylon used by 3D Hubs. I had to enlarge the hole to fit the 3.5mm glass bead, and it was difficult to do so without chipping too much away.
A larger problem is focus. For the glass beads, I ordered the sample pack of different sizes from Biospec Products. They seemed to be as advertised, but ultimately I was never able to capture an in-focus image. There are other sources for glass beads listed on PNNL’s page, which may work better. If you have achieved better results, or have tips on focus using this method, please leave a comment below.