ALD-Coated Hollow Borosilicate Glass Fibers

The influence of ALD-coating on the resistance during internal pressure of thin hollow glass fibers were tested with an aluminum oxide coating of four different coating thicknesses of s_c = (50, 100, 200, 500) nm on the surface of hollow borosilicate C5 fibers. In previous tests, borosilicate glass proved to have the highest resistance when loaded with internal pressure.

The fibers were packed after being produced in special transportation brackets made of stainless steel and transported in a stainless steel box, as shown in Figure 64. The brackets were secured against shifting by screws.

Figure 64: Transport system for coated single fibers consists of stainless steel box (left picture) and the stainless steel brackets (right picture)

The production of the transportation system from stainless steel was necessary because of its inert behavior during the ALD-process. The different coloration of brackets in Figure 64 was related to the different coating thicknesses.

The brackets prevented any contact between the single fibers. The fibers did not have to be removed from the bracket for coating process but the bracket with the located fibers was put in the ALD-chamber and coated as well. The contact surface was kept as small as possible to ensure a homogeneous and large-area application of the coating.

Additional possible injuries caused by the steel bracket itself should be hindered.

The ALD-procedure processes stepwise and at each step a thin layer of s_c = 0.1 nm is applied on the surface [137]. The number of steps is repeated until the desired thickness is reached. Consequently thicker coatings require longer process duration. The necessity of heat during ALD-process brought along an influence of temperature wherefore the different coating thicknesses were applied not only at T = 300 °C but also at T = 500 °C. The chosen temperatures are the most used values when ALD coating is applied on glass surface by the enforcing company [136], [137]. The handling and gluing together into the stainless steel pipes was not changed to the untreated single fibers.

In Table 53 the characteristic results of fibers coated at T = 300 °C are listed. As reference value a series of uncoated fibers was tested and listed whereby all fibers were produced from glass tube taken from the same batch. Hence, differences due to deviations in chemical composition can be excluded.

Table 53: Characteristic test results of fibers covered with different coating thickness at a temperature of T = 300 °C, all fibers exhibit dimensions of d_o ≈ 500 µm, di ≈ 465 µm and s ≈ 17.5 µm

Serial number

Coating thickness sc

[nm]

Min. burst pressure

pmin

[MPa]

Max. burst pressure pmax [MPa]

Form parameter

b

Characteristic pressure P

[MPa]

ALD 0

(reference) uncoated 9.8 19.1 9.3 16.1

ALD 1 50.0 10.7 23.5 6.2 17.9

ALD 2 100.0 11.1 19.0 8.1 15.3

ALD 3 200.0 7.1 21.6 5.6 17.5

ALD 4 500.0 3.9 19.2 2.6 14.2

Comparing the reference series to series ALD 1 a slight increase of all pressure values is noticeable. The minimum burst pressure and the characteristic pressure features an improvement of about 10% with 50 nm coating thickness. However, the form parameter decreases due to a wider spread which is caused by an improvement of about 20% for the maximum reached burst pressure.

The minimum pressure value of series ALD 2 increases again slightly but pressure remains characteristically below the according values of the reference series. The maximum burst pressures of both series are comparable. Indeed, the pressure range between minimum and maximum reached burst pressure is smaller than of the reference series. The accumulation of measured pressure values of ALD 2 leads to a lower form parameter than ALD 0.

Series ALD 3 exhibits a minimum burst pressure by the factor 1.4, which is lower than the reference series. However, the maximum burst pressure and as well the characteristic pressure feature higher values and an improvement of 10%. Therefore an outlier can lead to the lower minimum burst pressure. The resulting wider spread of measured test pressures consequently leads to a decrease of the form parameter b.

The series ALD 4 with a coating thickness of s_c = 500 nm features a maximum burst pressure comparable to the reference series but the minimum burst pressure remains distinctively below the according value of the reference by having only 40% of the reference. A significantly lower form parameter points out not only a wide spread but together with the decreased characteristic pressure, it indicates a distinct degradation of the pressure resistance.

Theoretically and bottom-up on the literature reference the resistance against inner pressure of the fibers should be increased by filling up small flaws and increasing the radius of larger cracks. The unsteady behavior of test results can be shown more clearly in Figure 65 by plotting the failure probabilities against the burst pressure.

The diagram shows a narrow development of the graphs of the tested series. Series ALD 0 to ALD 2 proceed similar in lower pressure region and the initial points of these curves are close together. The slight improvement of the minimum burst pressures of ALD 1 and ALD 2 constitute that close physical proximity. The graph of ALD 1 then proceeds with a lower slope than the reference series graph but at higher pressure level.

The similarity of test results of reference series and series ALD 2 is shown by comparable developments.

The S-curve of ALD 3 with s_c = 200 nm exhibits a comparable development to reference series in the low pressure range whereby the initial point slides to the left. In further development the graph converges to the S-curve of series ALD 1 with sc = 50 nm. That behavior can be explained by the similar form parameter and characteristic pressure.

A considerable different development shows the S-curve of series ALD 4 with a coating thickness of s_c = 500 nm. The significantly lower minimum burst pressure and the form parameter create the low slope of the graph. Therefore the calculated failure probabilities are much higher for series ALD 4 than for the other series in comparison vis-a-vis up to pressures of p = 15 MPa. It can be seen that no outlier leads to the described development of the graph but a straight and consistent distribution is noticeable.

Figure 65: Failure probability of hollow borosilicate C5 fibers covered with ALD-coating of different thicknesses at T = 300 °C, the reference series was carried out without coating and temperature treatment

Furthermore, single hollow borosilicate fibers were coated with the same four thicknesses at a process temperature of T = 500 °C. An elevated process temperature is necessary to deliver the essential energy for the chemical reaction. The increased temperature should accelerate the reaction again and support the formation of the coating layer.

The characteristic results of test series with fibers coated at increased process temperature is listed in Table 54.

Table 54: Characteristic test results of fibers covered with different coating thickness at a temperature of T = 500 °C, all fibers exhibit dimensions of d_o ≈ 500 µm, d_i ≈ 465 µm and s ≈ 17.5 µm

Serial number

Coating thickness sc

[nm]

Min. burst pressure

pmin

[MPa]

Max. burst pressure pmax [MPa]

Form parameter

b

Characteristic pressure P

[MPa]

ALD 0

(reference) uncoated 9.8 19.1 9.3 16.1

ALD 5 50.0 12.1 20.0 9.8 17.2

ALD 6 100.0 2.3 16.1 2.1 13.1

ALD 7 200.0 8.6 14.3 10.3 11.9

ALD 8 500.0 2.0 21.6 2.6 16.9

The comparison of the reference and series ALD 5 with sc = 50 nm shows a slight improvement of pressure data in the range of 20% at minimum and about 5% at characteristic as well as at maximum pressure. The slight increase of all pressure values leads to comparable form parameters of both series.

Series ALD 6 features a characteristic test result which remains below the data of the reference series. Series ALD 6 has a minimum burst pressure that is decreased by the factor 5 in comparison to the reference series. The maximum and the characteristic pressure values remain at 80% of the according reference value. Due to the large deviation at minimum burst pressure and the approach at maximum burst pressure the spread of the measured values is significantly wider. Consequently series ALD 6 has a small form parameter.

Fibers coated with sc = 200 nm of ALD 7 have pressure values below the reference series. The minimum burst pressure is similar to the value of the reference series by having about 90% of reference value. The deviation gets higher by approaching the maximum burst pressure. Therefore, the characteristic and the maximum burst pressure exhibit only 75% of series ALD 0 values and decreased by the factor 1.3 to 1.4. The

narrow values of minimum and maximum burst pressure are indicative of a consistent distribution, which is shown by the high form parameter of b = 10.3.

Series ALD 8 with sc = 500 nm of coating thickness shows a similar behavior to ALD 6.

The minimum pressure remains at only 20% of the reference but the maximum burst pressure shows an improvement of about 15%. The characteristic pressure of series ALD 8 also has a higher value compared to ALD 0 and increases about 5%, which indicates a small number of outliers at minimum burst pressure. Nevertheless, the wide spread between minimum and maximum burst pressure originates the small form factor of b = 2.6 which is similar to that of series ALD 6.

Calculating the failure probabilities and plotting it in a diagram against the single burst pressure gives the opportunity to compare the series not only by the characteristic results but on their developments and distributions as well. The diagram is given in Figure 66.

Figure 66: Failure probability of hollow borosilicate C5 fibers with the dimensions of d_o≈ 500 µm, di≈ 465 µm and s ≈ 17.5 µm covered with an aluminum ALD-coating of different thicknesses at T = 500 °C, the reference series was carried out with samples without coating and temperature treatment

The diagram clearly shows the differences between the single tested series. While series ALD 5 and ALD 7 show a comparable development to the reference series ALD 0, they show different locations. The graph of ALD 5 slides to the right which is indicated by the higher pressure values, as seen in Table 54 and, therefore, exhibits an improved pressure resistance. The S-curve of ALD 7 slides to the left wherefore a decreased pressure resistance is shown. Nevertheless, it shows the same narrow and consistent distribution than the reference and ALD 5.

A completely different development regarding the graphs of ALD 6 and ALD 8 can be observed. The graph conforms more of a straight line than an S-curve, resulting from Weibull distribution. It can be seen that the initial points are in the range of p = 2 MPa.

However, both graphs exhibit only one and three points, respectfully in that range of initial points wherefore the assumption of outliers is confirmed. Those outliers lead to the small form parameter of both series which is responsible for the low slope and development of the graphs. Nevertheless, the multitude of measured pressure values is below those of the reference series and consequently no improvement can be observed.

The temperature treatments during ALD coating process average a duration of four hours wherefore an annealing effect could be ensued. The used process temperatures of T = 300 °C and T = 500 °C remain distinctively below the annealing point, which is at a temperature of T = 570 °C [151]. However, an annealing effect can be reached at a lower temperature, at strain point of T = 530 °C, which is still above the process temperatures. An investigation of a possible annealing effect of uncoated hollow fibers was carried out. Test samples were temperature treated for four hours at given process temperatures of T = 300 °C and T = 500 °C before being prepared and tested. Again the test series consist of a minimum of 30 samples.

The test results are summarized and listed in Table 55. The comparison of the pressure values of the three series shows only slight deviations. Series ALD 9 treated with T = 300 °C exhibits a slight decreased pressure resistance of the test samples. All pressure results are below the reference values whereby the degradation of pressure resistance is only in the range of 5% to 10%.

ALD 10 treated with T = 500 °C exhibits a degradation of about 5% at minimum burst pressure. However, at maximum burst pressure an improvement of about 15% can be detected. The characteristic pressure of reference series and ALD 10 feature similar values. The wider spread due to decreased minimum but increased maximum burst pressure leads to a smaller form parameter of ALD 10.

Table 55: Characteristic test results of uncoated fibers with dimensions d_o ≈ 500 µm, d_i ≈ 465 µm and s ≈ 17.5 µm annealed at different temperatures

Serial number

Annealing temperature TA

[°C]

Min. burst pressure

pmin

[MPa]

Max. burst pressure pmax [MPa]

Form parameter

b

Characteristic pressure P

[MPa]

ALD 0

(reference) untreated 9.8 19.1 9.3 16.1

ALD 9 300 8.6 18.6 7.1 15.1

ALD 10 500 9.3 21.7 6.5 16.5

Again the failure probabilities were calculated and for better visualization plotted in a diagram against the burst pressure values, which is given in Figure 67.

Figure 67: Comparison failure probability of uncoated hollow fibers but different thermal history

In the diagram only slight differences between the single test series can be seen.

Nevertheless, it is detectable that both temperature treated series feature initial points

Im Dokument The pressure resistance of hollow glass fibers at internal pressure load (Seite 192-200)