Using cotton thread and a modified 3D printer nozzle of my own design I created high-strength composite parts with a flexural modulus (stiffness) over 14 times greater than unreinforced nylon with my low cost Prusa I3 printer.
By coaxially extruding a continuous thread of cotton through a 0.7mm nozzle with nylon, much stronger parts can be produced - similar to the method used by Markforged composite 3D printers. However extensive cooling and far slower print speeds (≈ 6mm/s) have to be used to prevent the cotton being pulled out of the composite matrix. Printing with fine copper wire was also tried with possible future uses for 3D printing electronics e.g. NFC / RFID antenna, electromagnets and motors.

Not only is this technology able to produce stronger parts but ones that have the potential to use less fossil fuels and energy during their production - less plastic for the same stiffness is required and cotton is a renewable bio-degradable resource. Although I demonstrate this technology with natural cotton fibres, there are no reasons why carbon and Kevlar fibres could not be used instead.
In this instructable I will provide information on how to convert your 3D printer for composite 3D printing as long as you have a lathe. At the end of this instructable I explain how I gathered data on the flexural modulus of some 3D printed test specimens. Also I include my design process from an initial prototype to a working final design of the nozzle.
If you like my instructable please vote for me in the Invention and Explore Science contest.

Tools and materials :

Lathe (with vertical slide)
1mm OD, 0.3mm ID Teflon/PTFE tube
A 3D printer
Piercing saw and junior hacksaw
Brass rod
Stainless steel rod

Step 1: Final Nozzle Design

Guiding the thread into the nozzle so that both plastic and cotton can be extruded simultaneously with as little friction as possible is not a simple task- a few iterations were required (initial versions documented at the end of this instructable).

Note that I used a lathe to produce this nozzle.

I have included a sketch of my final iteration which used a pre-manufactured Teflon tube with a threaded "collet" tube clamp - no metal outside of nozzle to conduct heat - high temperature gradient means the plastic is either molten or not - no oozing back up fibre inlet.
I found that using a long thin Teflon tube to guide the cotton into the nozzle worked best - low friction, low thermal conductivity (to prevent the flow of molten plastic back up guide tube instead of out of printing orifice) and small diameters are available so that a good seal between fibre and tube can be made.

For the fibre inlet hole an angle of 40° to the nozzle axis was found to work well although there are quite a variety of angles which would work well, generally a shallow angle works better.

My final iteration used a small custom machined titanium Teflon tube clamp which threads into the side of the nozzle at an angle. This provided a good seal between tube and nozzle to prevent plastic leakage. The original version of this was made of brass, but re-machined out of titanium due to its flexibility - the prongs of the clamp must be able to spring to hold the tube. Post-processing of the brass nozzle was required to remove sharp edges which could increase the chance of cutting of the fibre or wire and increase friction.

On this iteration of the nozzle you may have noticed that the method of attachment of the nozzle to the heater block is not a standard M6 thread as in most printers but rather a quick-change grub-screw locking system of my own design which allows the nozzle to be rotated into any desired orientation for convenient positioning of the fibre inlet. However this is not essential and was part of an upgrade to an all-metal hotend to allow me to print high temperature thermoplastics such as nylon. Earlier iterations of the nozzle did use the standard M6 thread attachment.

Step 2: Making the Nozzle

The nozzle is without doubt the most defining part of this fibre printing system. I used a lathe with a vertical slide to machine the nozzle so the production of these parts may be hard to access for some people.

Machining main nozzle body:
1. Machine outer diameter.
2. Machine M6 thread or attachment system onto end of nozzle.
3. Drill hole down middle of nozzle to required depth for filament entry 3.1mm
4. Drill a 0.7mm hole for the printing orifice
5. Part off with some excess
6. Hold in Chuck and face off until the 0.7mm hole appears. Face off until the nozzle is an exact length - you want to be able to swap nozzles with no change of z stop.
7. Machine angle on nozzle with topside set to 50° - Machine until the tip of the nozzle is the desired diameter e.g. 1.2mm

8. Drilling of the angled hole, I used a vertical slide and held the nozzle in the vice on the vertical slide. Use a slot drill held in the lathe headstock to produce a flat spot for drilling on the side of the nozzle at the desired angle (40°). Use a centre drill to reduce drill wander.
9. Drill a 0.9mm hole until the drillbit breaks through into the 3.1mm hole just before the tip of the nozzle.

10. Drill a 2.5mm thread (tapping drill size for M3) a few mm deep, ensure some of the 0.9mm hole remains for sealing onto the Teflon tube.

11. Carefully tap the M3 angled thread using a tap holder and a centre in the headstock - tapping an angled hole by hand is not easy.

The nozzle is machined.

Machining Teflon tube sealing collet clamp: I recommend a springy material for this such as Titanium or stainless steel - when made of brass the prongs have a tendency to snap. Be carefull when drilling the 0.9mm hole down the centre of this - the hole should be a snug fit onto the Teflon tube.
Use a fine piercing saw to cut the slit so as to serve as a springy collet. Bend the prongs of the collet to help make a good seal but not add friction to the cotton.

And use a junior hacksaw to cut the slot for a Screw-driver bit for tightening of the Teflon tube clamp in the nozzle.

Polishing of sharp edges:
I used a tooth pick with grinding paste to smooth the sharp edges of the nozzle printing orifice and the entrance to the angled fibre inlet hole.
Also I used a piece of cotton coated with grinding paste pulled through the nozzle. I pulled the cotton through at the same angle as it would be pulled through in use. This reduces the chance of a cut fibre mid-print.

Step 3: Setting Up and Essential Equipment

Ensure you have a very good cooling system - I 3D printed this excellent all-round cooling vent (DiiCooler by Pawpawpaw85 on Thingiverse) to cool the part being printed - https://www.thingiverse.com/thing:1025471 .

To clear out the nozzle so that you can thread the cotton through I found that the most effective method is to burn out the plastic using a cooker hob.

To make threading the cotton thread through the fibre inlet tube easier I soaked a section of the thread in superglue. Keep the thread taught as the glue dries. After the glue dries run your finger nails along the thread to remove any larger lumps of dried glue.

You will need to make a spool holder that clips onto the print-head. I used a standard sowing machine spool as a sowing machine can be used to wind the cotton onto it from a larger spool and the spools are light-weight. However for larger high-cotton-density prints a larger spool size will be needed.

Step 4: Recommended Printing Settings

Slow printing speed ≈ 6mm/s
Slow travel speed ≈ 20mm/s
Flow of 80% of normal extrusion to account for volume used by cotton to prevent oversized parts.
100% cooling after 1st layer - ensures plastic cools so that fibre follows the correct path and doesn't cut corners.
0.25mm layer height gave the highest strength parts
Decreasing infill width and infill line separation below the nozzle width allowed the fibres to be placed closer together and vastly increased the flexural modulus. Tried an infill line separation of 0.50mm despite using a 0.7mm nozzle.

I used Coats cotton thread (nr. 50) and also tried 80um copper wire and found that 115um wire did not work as well as it was less flexible.

Step 5: Strength Testing of Printed Specimens

I used a DIY 3 point bend testing apparatus to find the flexural modulus of printed specimens in comparison to pure nylon. A DTI is used to measure the deflection of the beam specimen as it is loaded with different weights. I recorded the deflection and mass into a spreadsheet (attached). Samples are 80 x 10 x 5 mm, however the exact dimensions varied slightly so I used verniers and a micrometer to measure exact dimensions which are required as part of the flexural modulus calculations. Once the gradient of the force deflection graph has been plotted, you just need to copy and paste the gradient into the appropriate spreadsheet box as this is aso required as part of the calculation.

I have included a downloadable spreadsheet as well as a Google Sheets link here of 3D printed specimens and the results of the tests: Density and flexural modulus. This DIY 3 point bend tester obtained correct modulus values for aluminium strips so it can be assumed the modulus values for the 3D printed parts are accurate. The spreadsheet calculates these metrics using the dimensions of the specimen and deflection of the specimen under several loads. Sample 5 gave the highest flexural modulus of 4.6GPa compared to 0.3GPa of nylon.

Specimen data V2.xlsx
Download

Step 6: Failed Nozzle Designs

Here I have documented my errors and how I finally obtained a working design.

First I tried just an angled hole in the side of the nozzle - molten plastic would ooze out of the cotton inlet rather than out through the printer hole.

Then I tried using a machined 10mm Teflon guide tube - although otherwise performed well, plastic pressure would strip the weak threads of the attached tube. Therefore I made an identical tube from brass- however brass has too high a thermal conductivity and plastic would still ooze back out of fibre inlet.

Therefore I machined a thin titanium case and inset a machined Telfon tube - far better but after long prints plastic would still ooze due to conduction. Plastic would then solidify once far enough from nozzle and cause the cotton thread to snap.

As a result I concluded that in much the same way as a 3D printer extruder, there must be a short and sharp change from aprox. room temperature to the temperature of molten plastic.

To view the final design which used a pre-manufactured small diameter Teflon tube to make a very sharp thermal gradient to prevent oozing go to step 2 where I have a design and some photos.