In the first article, I made a connection between the resolution of the human eye and the resolution in a CT scan. A quick internet search revealed that “the optical resolution of the eye is, at best, about 15 lp/mm”. I would state 7 lp/mm.

The 2D resolution test of the imaging chain with an ISO 19232-5 (EN 462-5 or ASTM E2002) Duplex IQI showed more than 20% modulation for the D15 (32 µm) which is equal to 16 lp/mm. In such a high magnification, it almost only depends on the focal spot size.

Using ASTM E2698 equations, the optimum magnification for the nominal 60 µm focal spot and a detector SRb of 150 µm would be 12. Even with magnification 9, one may reach a 30 µm (17 lp/mm) resolution. A CT test according to ASTM E1695 at a magnification of 5.1 showed 10% MTF at 14 lp/mm. Due to the tube case of the COMET 450 compact version with a bulky exit window-frame and a little lack of magnification in the first setup, we still have room for improvement to show 5 µm more resolution.

Fig. 1: Additively manufactured copper cooling device

In this last blog, I like to show some results of a 3D printed part made of copper. Without being an expert in such manufacturing and testing applications, I can see how the quality check reveals more relevant information by higher resolution.

The part displayed in Fig. 1 is supposed to be a cooling device, where water flows around inside the flat and inside the cone-shaped part. Therefore, it must be watertight, which can be difficult with 3D printing if the walls are too thin, and something went wrong during the print.

The most outer diameter of the copper part is 140 mm, and the top disk has a diameter of 100 mm. According to table 1 of the ISO 15708 2, by the use of 450 kV and a lot of filtering, 90% of the Bremsstrahlung gets absorbed by just 20 to 25 mm of solid copper. Between this and double thickness, CT with a flat-panel detector becomes problematic, and beam hardening artifacts, as well as noise, begin to dominate the images.

We don’t have an exception here, as the radiation is blocked by 100 mm of copper when the disk plane is aligned to the beam. Using the highest possible acceleration voltage of 450 kV delivered by the MesoFocus tube turns out to be a great advantage. But as it is flat, the material thickness quickly decreases.

The scan parameters of the later shown results of a simple continuous 360° CT are 450 kV, 1.5mm Sn filter and the 100 W, nominal 100 µm big focal spot, a focal detector distance of 900 mm, and a magnification of 3.6. Just one half of the sample has been imaged by 1800 projections with 0.8Hz.

For comparison, the part has also been scanned with a 450 kV mini-focus tube, also at 450 kV and 2 mm Sn filter, 675 W.

Fig. 2: nominal/actual comparison

The lower resolution mini-focus result is sufficient to perform a nominal/actual comparison or variance analysis with the corresponding stl-file. There we see that sections near the inlet and outlet are very far out of tolerance (pink in Fig.2 ). A cross-sectional look inside the part (see Fig. 3) shows that a finishing machining of the material to the nominal dimension would open it.

Fig. 3: Cross-sectional view of the mini-focus CT scan

For the water tightness, the recommendation was to use a construction being at least 1.2 mm thick. From several additively manufactured steel parts, I know that even thinner walls can be flawless. That could possibly be more difficult with copper. With a look at the scan with mini-focus CT, however, I still don't understand why this wall thickness is necessary. Although, when I look to Fig. 4 with a higher magnified crop on the left side, I can see an even distribution of points with slightly different gray values. Due to the resolution, I cannot see the actual extent and the depth of these flaws. At least I know that mini-focus CT can detect cracks in additively manufactured parts.

Fig. 4: Mini-focus CT result with a cropped magnification on the left side

It looks quite different when the part is scanned with a higher resolution which is now also available at 450 kV with MesoFocus.

Fig. 5: MesoFocus CT result with a cropped magnification on the left side

Most of the small indications are small in all three dimensions, and maybe the tiniest flaws have arisen due to non-melted metal. But some of them are channels through half of the wall thickness (see Fig. 6), and therefore, may be shrinking artifacts. Most likely, the deeper channels can be starting points of cracks when the part gets mechanically stressed.

Fig. 6: 450 kV MesoFocus depth cross-sectional view with channels through the material

Something we saw was that the resolution now dives deeper than the surface roughness of such part. Roughness is not easy to quantify. Sometimes it is categorized by a finger testing of the part and a reference surface. This way of testing is impossible at inner surfaces which are quite common with additive manufacturing.

The surface determination of a CT volume scanned with MesoFocus now reveals micro-spikes and grooves, which have been formed by the laser firing into loose powder to build the first layer after a hollow. With a low-resolution CT using a mini-focus (nominal 400 µm focal spot), these details are smoothed out by the unsharpness of the imaging chain (see left side of Fig. 7 for a 3D look into the sample and 1:1 comparison).

Fig. 7: Comparison of the two different resolutions and their effect on the surface roughness of inner surfaces (inset height distribution of a smaller surface region)

With the resolution of the MesoFocus, the roughness can now be sampled with the voxel resolution of the volume (see inset in Fig. 7). One can probe a smaller area, and the standard deviation of the distribution (in this case 54 µm) or the commutation of a certain percentage (90% in 62 µm) is then a measure of the roughness.

I hereby close the blog series. Over the last few weeks, I wanted to show you what will soon be possible with this great new solution in CT. The 450 kV MesoFocus tube opens up a previously completely unreachable area of very high energy and at the same time very high resolution for the closed tube x-ray sources. The smallest details inside and enclosed in gold have been made visible and with additive manufacturing, even the smallest defects which quickly become safety-relevant, can now be made visible in steel, nickel, and copper parts.

I wish you a Merry Christmas and stay healthy.
See you in 2022 for MesoFocus CT.
Daniel Stickler