On the one hand, printing with reactive material is more complex, especially if several parts have to be mixed. On the other hand, this mixing allows for plenty of unique compositions which give the operators control over the desired characteristics.

The material I printed with during this project was a three-part polyurethane provided by the Canada based company, Amathane Inc. Each part of this set was liquid. One part was a polyurethane resin while the other two parts were hardener resins, respectively.

Polyurethane resin cures while exposed to the atmosphere, similar to silicone. However, by mixing the resin with a hardener resin, the curing time and mechanical properties of the finished product could be influenced. Along with the material itself, Amathane also provided sheets with various characteristics of different compositions which were necessary to predict the curing timelines I was working with.


The first step in this project was to determine the general characteristics of each part.

Poly printing 2.jpg

The most relevant property which I had to think about was the viscosity in order to choose a suitable nozzle size. To do this, I filled syringes with each part, mounted different nozzles onto the syringes and dispensed the material by hand.

The flow characteristic I was looking for was a continuous flow instead of individually dispensed drops. Ideally, the flow would have been initiated as soon as minimal pressure was applied to the syringe, and it would have stopped as soon as no additional pressure was applied.

However, the reality was that this behavior could be tested, but the flow characteristics were not consistent. The frequency of dispensed drops increased until a continuous extrusion was achieved. If no further pressure was applied, the continuous flow transitioned back to individual drops with a decrease in frequency.

Another important consideration pertains to the mixing of the individual parts. One option would be the curing rate if the resin and hardener mixture was stirred and distributed across the surface similar to a layer. Alternately, a few drops of hardener could be deposited onto a pool of the resin.

Three different approaches were taken for this project:

1.     Different parts of the material were extruded separately, on top of each other, thus starting the curing reaction after dispensing the material

2.     The different parts were mixed inside the syringe prior to the extrusion

3.     Different parts were mixed shortly before the extrusion by using an adapter that joins the two tubes before the nozzle

I will describe the details and outcome of each approach individually.


FIRST APPROACH: Separate extrusion

Choice of mixture composition and setup

For the first set of experiments, I extruded the polyurethane and the hardener separately. This meant I could only extrude two parts since I could only mount two nozzles to the 3D printer. This caused two restrictions for the composition. First, I had to choose the composition that allowed me to use only one hardener. Second, I had to choose the composition that cured the fastest. The fast curing was necessary to generate a three dimensional structure. Ideally, each deposited layer would cure shortly after the extrusion and thus provide a base for the next layer.

poly printing 3.jpg

The two nozzles were aligned on the y-axis, but featured an offset on the x-axis. As a result, the non-stop extrusion of at least droplets of material caused either an overlap or a deposit outside of the print layout. Since both nozzles were aligned on the y-axis, there were no additional lines to the print-layout-lines in this direction. Instead, these lines had excess material because both nozzles were constantly depositing. On the x-axis however, there were additional lines due to the offset. On the one side an additional resin-only line was generated, and an additional hardener-only line formed on the other side.

The deviation from the printing layout can be seen in the video below. The progression of the printing path was along a square and the deviation was a result of the nozzle offset. 

The nozzle mount I used was Structur3D’s custom nozzle mount for singe and dual extrusion on the FELIX 3.0 printer (only). A CAD file of the part can be downloaded here.

Printed geometry

Because of the low viscosity of each part, I expected a three dimensional structure with a large z-axis height was not achievable unless the blend would cure immediately after the extrusion.

Moreover, after slicing the model, the layers should have been printed with the two extruders in an alternating rhythm. This meant the model had to consist of layers made of two different materials alternating each layer. An alternative could have been to generate a support structure between every two layers, but either way, the complexity of the preparation would have been high.

For this reason, I wrote the G code for the printing outline myself, which consisted of only three layers. I arbitrarily chose a square as the shape to be printed. This allowed me to copy and paste the code if I wanted to print additional layers. The code for 21 layers can be seen below, in the code section.



I started printing with the hardener as the first layer in order to force the second layer of resin, and third layer of hardener to help cure the first layer faster in order to generate a solid base.

During the six prints I performed using this approach, the three factors I adjusted were the:

  • offset
  • nozzle size
  • number of layers

(All of which will be discussed more in detail in the following paragraphs.)

In addition, I printed the first print onto a building platform that was heated to 60°C. I assumed the higher temperature would reduce the curing time, but as it turned out, it just increased the liquids wetting behavior. (Wetting refers to the characteristic of a liquid to spread on a surface. For a non-wetting, a drop will just lay on the surface but not disperse. For a wetting and semi-wetting behavior, the drop will disperse and spread on the surface.) Even though the offset was the highest during the first print, the spreading of the liquid was significant.

To minimize this effect, I printed all following prints a room temperature, which was about 30°C, since it was summer, and the lab was upstairs. The difference was immediately apparent when comparing the print layout of the first and the second print.



As mentioned before, a continuous extrusion was only temporarily possible, and for that reason, the offset played a crucial role in depositing a consistent line. For example, if the offset was too high, the flow rate would eventually decrease, and individual drops of the liquids would be deposited. This could be prevented by reducing the offset to a distance smaller than the drop size. For this setup, there was no drop formation at the nozzle because the material was in constant contact with the building platform or the layer which enabled a constant deposition.

The difference between the deposition from an adequate offset and one that is too high can be seen in the video below.

Comparatively, the video below shows an adequate offset on both


Nozzle size

The first prints were printed with an internal nozzle diameter of 0.25 mm, while the last two were printed with an internal nozzle diameter of 0.61 mm. The difference being that only with the last two prints did the print layout feature a completely closed geometry. All prints before featured interrupted lines consisting of individual drops, except the first which was printed onto a heated platform. Obviously, the two other factors had big impact on the material being extruded as well, but the individual impact of nozzle size can be seen when comparing the fourth and fifth print where the offset and the number of layers were similar.

poly printing 6b.jpg


Number of layers

As shown in the pictures, a three dimensional structure was not possible with this approach.

This was expected, but the number of layers had another crucial impact nonetheless. During all prints, I varied the number of layers between 12 and 21. The modeled height could not be achieved, but the amount of extruded material increased with an increasing number of layers.

This effect can be seen especially in the last two prints, during which the number of layers was the highest.

This effect, called a ‘mountain effect’, appeared during printing a part with several layers. This means that the deposited material will not stay in the position it was deposited. Instead, it will not only spread onto a layer, but also flow down the sides of the printed geometry. The more layers that are printed, the stronger this effect appears, which can best be seen in the last approach.


SECOND APPROACH: Mixing in Syringe

Choice of mixture composition

For the second set of experiments, I used the second approach: Mixing the different materials inside the syringe, prior to the extrusion. By using this approach, I was actually able to use all three parts since I mixed and extruded them together.

The major difference to this approach was that the material would cured inside the syringe and the tube. In contrast to the previous approach, I needed a composition that would take more time to cure in order to keep the material in a printable state.


Printed Geometry

As I was mixing the resin and the hardener component prior to printing, I chose to print a different geometry. First of all, the curing reaction obviously started inside the syringe and the tube, which meant that the extruded material would increase in viscosity during the print. Eventually, a total curing inside the system was inevitable which meant I had a limited timeframe to perform the print.

The increase in viscosity promised that a three dimensional structure was achievable. For this reason, I designed a hollow thin-walled cube for the following prints.

poly printed geo.png



During this approach, the characteristics of the offset, nozzle size and number of layers was similar to the first approach. The crucial factors in this approach were the curing time and composition.

As stated before, the time during which the print was possible was dictated by the curing time of the composition. As soon as the material cured inside the syringe and the tube, no more extrusion was possible.

Interesting note about this was that this fully-cured point in time resulted in a clear cut-off line after. Obviously, the viscosity rose during the print and the flow rate dropped, but there was an instantaneous stoppage of extrusion which was notable.

Now that I was able to print from one syringe only, I could use all three parts of the material. During the first print, I used a composition which was supposed to reach a gel-like behavior after 20 minutes. This information provided by the material refiner was right, and after the gel-like state had been reached, the extrusion ceased.

For the second print, I used a composition which was supposed to have the highest curing time. During the first approach, I selected the other extreme using the lowest curing time which involved only one hardener. Since this was the other extreme, the composition consisted only of the resin and the other hardener.

One obvious advantage of printing with only one nozzle was that there is no additional material deposits next to the print layout.


THIRD APPROACH: Mixing in adapter, shortly before extrusion

For the third approach, an additional hardware component was necessary.

poly hardware.jpg

Choice of mixture composition

Similar to the first approach, ideally the curing would begin right after the extrusion. Since the material was mixed right before the extrusion, this was the expectation. For this reason, I used the same composition as in the first approach.


In order to maintain a constant mixing of the two parts, a simultaneous extrusion from both extruders was required. Since the software didn‘t allow for this, I had to connect both extruders to the same connection on the electronics board. This way, both extruders could be operated simultaneously.

I was expecting to achieve a three dimensional structure using this approach, which is why I chose to print the hollow thin-walled cube again.


Just like with my second approach, the characteristics of the offset, nozzle size and number of layers remained similar to the first. This time, the overall performance was better.

This was the first print during which a three dimensional structure seemed possible, and was in fact achieved. The geometry I was trying to print was the same cube from the previous approaches. As there was some time between mixing and extrusion, the deposited material likely featured a slightly higher viscosity which made it easier to achieve a three dimension structure.

Similar to the silicone printing experiments, the flow rate dropped throughout the print. This was counteracted by pausing the print and extruding the material manually until a sufficient flow rate was established again. During the last print, I performed this procedure several times. After about 13% of the print, the flow rate was too low to continue so I had to increase it. From this point onwards, I had to repeat this roughly after every 3% of the print.

For the majority of the print, the nozzle was constantly immersed in the deposited material, similar to the silicone printing experiments. As a result, most of the material was actually extruded into the existing structure. It appeared to cure faster on the surface and maintain a lower viscosity on the inside. This is supported by the observation that the nozzle visibly moved the cured skin on the surface when it contacted it. Polyurethane resin will cure without the hardener, but it takes longer. This means, there is an additional environmental reaction with the atmosphere. This supports the observation that the mixture cured faster on the surface because there was also the exposure to the atmosphere.

Mountain effect

As mentioned above, printing with a low viscosity material results in a mountain effect. In this case, most of the material was actually extruded into the already printed structure limiting the ability to flow down to the sides, and resulting in a broadening of the structure.

Considering the wall thickness of the cube was supposed to be 1mm, I succeeded with this print and this approach.

Tips and Tricks

Filling Syringes

Filling the syringes can actually be done without creating a mess. A little physical effort is required.

Basically, the liquid material can be sucked into the syringe by reversing the extrusion process. To start, the plunger needs to be inserted all the way into the syringe. Next, a connection adapter is used to mount a tube onto the syringe. The open end of this tube needs to be immersed into the liquid which allows the material to be sucked into the tube when the plunger is pulled out.

Material penetrates wax paper

As it turned out, wax paper seems to be permeable to the polyurethane resin and the hardeners. In many cases, I found the surface below the wax paper wet with some kind of liquid. When replicating, a second piece of wax paper should be used.

Mount a cup under home position

Pausing and manual purging

With this polyurethane printing project, the flow rate of the material dropped during extrusion.  To avoid big holes in the end product, the print can be paused, and a sufficient flow rate can be established by manually extruding the material.

Further factors to test

  • ·         Speed
  • ·         Temperature
  • ·         Compositions
  • ·         Flow rate at the start



Table of settings


Code for 21 layers

For printing under the first approach only because the code considers the nozzle offset. The outline of all other squares I printed are based on this code, only the layer size varied.


G28 ; home all axes

;G1 Z5 F5000 ; lift nozzle

M302 ; enable cold extrusion

M92 E5000 ; set axis steps per mm

G21 ; set units to millimeters

G90 ; use absolute coordinates

M82 ; use absolute distances for extrusion

G92 E0 ; Set current position to coordinates given


; G1- Coordinated Movement X Y Z E, S1 disables boundary check, S0 enables it


G92 E0

G1 E-1.00000 F5000.00000

G92 E0

;G1 Z0.2 F5000.000

; T1 = hardener

; T2 = resin



; first square

G1 X200 Y150 F5000;

G1 X200 Y100 F500 E2 ;

G1 X150 Y100 E4 ;

G1 X150 Y150 E6 ;

G1 X200 Y150 E8 ;



; second square

G1 Z 0.2 ;

G1 X200 Y165 ;

G1 X200 Y115 E1 ;

G1 X150 Y115 E2 ;

G1 X150 Y165 E3 ;

G1 X200 Y165 E4 ;



; third square

G1 Z 0.4

G1 X200 Y150 ;

G1 X200 Y100 E10 ;

G1 X150 Y100 E12 ;

G1 X150 Y150 E14 ;

G1 X200 Y150 E16 ; 



; fourth square

G1 Z 0.6 F5000.000

G1 X200 Y150 F5000;

G1 X200 Y100 F500 E5 ;

G1 X150 Y100 E6 ;

G1 X150 Y150 E7 ;

G1 X200 Y150 E8 ;



; fifth square

G1 Z 0.8 ;

G1 X200 Y165 ;

G1 X200 Y115 E18 ;

G1 X150 Y115 E20 ;

G1 X150 Y165 E22 ;

G1 X200 Y165 E24 ;



; sixth square

G1 Z 1.0

G1 X200 Y150 ;

G1 X200 Y100 E9 ;

G1 X150 Y100 E10 ;

G1 X150 Y150 E11 ;

G1 X200 Y150 E12 ; 



; seventh square

G1 Z 1.2 F5000.000

G1 X200 Y150 F5000;

G1 X200 Y100 F500 E26 ;

G1 X150 Y100 E28 ;

G1 X150 Y150 E30 ;

G1 X200 Y150 E32 ;



; eighth square

G1 Z 1.4 ;

G1 X200 Y165 ;

G1 X200 Y115 E13 ;

G1 X150 Y115 E14 ;

G1 X150 Y165 E15 ;

G1 X200 Y165 E16 ;



; ninth square

G1 Z 1.6

G1 X200 Y150 ;

G1 X200 Y100 E34 ;

G1 X150 Y100 E36 ;

G1 X150 Y150 E38 ;

G1 X200 Y150 E40 ; 



; tenth square

G1 Z 1.8 F5000.000

G1 X200 Y150 F5000;

G1 X200 Y100 F500 E17 ;

G1 X150 Y100 E18 ;

G1 X150 Y150 E19 ;

G1 X200 Y150 E20 ;



; eleventh square

G1 Z 2.0 ;

G1 X200 Y165 ;

G1 X200 Y115 E42 ;

G1 X150 Y115 E44 ;

G1 X150 Y165 E46 ;

G1 X200 Y165 E48 ;



; twelfth square

G1 Z 2.2

G1 X200 Y150 ;

G1 X200 Y100 E21 ;

G1 X150 Y100 E22 ;

G1 X150 Y150 E23 ;

G1 X200 Y150 E24 ; 



; thirteenth square

G1 Z 2.4 F5000.000

G1 X200 Y150 F5000;

G1 X200 Y100 F500 E50 ;

G1 X150 Y100 E52 ;

G1 X150 Y150 E54 ;

G1 X200 Y150 E56 ;



; fourteenth square

G1 Z 2.6 ;

G1 X200 Y165 ;

G1 X200 Y115 E25 ;

G1 X150 Y115 E26 ;

G1 X150 Y165 E27 ;

G1 X200 Y165 E28 ;



; fithteenth square

G1 Z 2.8

G1 X200 Y150 ;

G1 X200 Y100 E58 ;

G1 X150 Y100 E60 ;

G1 X150 Y150 E62 ;

G1 X200 Y150 E64 ; 



; sixteenth square

G1 Z 3.0 F5000.000

G1 X200 Y150 F5000;

G1 X200 Y100 F500 E29 ;

G1 X150 Y100 E30 ;

G1 X150 Y150 E31 ;

G1 X200 Y150 E32 ;



; seventeenth square

G1 Z 3.2 ;

G1 X200 Y165 ;

G1 X200 Y115 E66 ;

G1 X150 Y115 E68 ;

G1 X150 Y165 E70 ;

G1 X200 Y165 E72 ;



; eighteenth square

G1 Z 3.4

G1 X200 Y150 ;

G1 X200 Y100 E33 ;

G1 X150 Y100 E34 ;

G1 X150 Y150 E35 ;

G1 X200 Y150 E36 ; 



; nineteenth square

G1 Z 3.6 F5000.000

G1 X200 Y150 F5000;

G1 X200 Y100 F500 E74 ;

G1 X150 Y100 E76 ;

G1 X150 Y150 E78 ;

G1 X200 Y150 E80 ;



; twentieth square

G1 Z 3.8 ;

G1 X200 Y165 ;

G1 X200 Y115 E37 ;

G1 X150 Y115 E38 ;

G1 X150 Y165 E39 ;

G1 X200 Y165 E40 ;



; twenty-first square

G1 Z 4.0

G1 X200 Y150 ;

G1 X200 Y100 E82 ;

G1 X150 Y100 E84 ;

G1 X150 Y150 E86 ;

G1 X200 Y150 E88 ; 



G28 X0; home X axis

M84     ; disable motors

M84     ; disable motors