Influence of limiting the actuation strain on the functional fatigue behavior of Ni 50.3 Ti 29.7 Hf 20 high temperature shape memory alloy

Abstract

The cyclic stability and the fatigue life of NiTiHf alloys are very important for their functionality at high temperatures. The previous studies have shown that the increase in Upper Cycle Temperature (UCT) and the magnitude of applied stress decreased the cyclic stability and fatigue life of NiTiHf high temperature shape memory alloys due to plasticity with dislocation formation at high temperature. On the other hand, this study was dedicated to the effect of limited actuation strain against the applied constant stress to the functional fatigue life and the possible damage mechanism such as dislocation and crack formation. Strain limitation led to observe insignificant amount of accumulated irrecoverable strain but significant amount of crack formation such that the reason of the failure of Ni_50.3Ti_29.7Hf₂₀ (at%) alloy was found to be the crack formation instead of dislocation accumulation.

Keywords

Shape memory alloys functional fatigue high temperature shape memory alloy

1. Introduction

Shape Memory Alloys (SMAs) are favorable materials in aerospace applications for their shape memory behavior which can be utilized in actuation purposes (Benafan et al., 2014, 2016; Godard et al., 2003; Hartl and Lagoudas, 2007; Jani et al., 2014; Singh et al., 2003). Although vast amount of knowledge on shape memory behavior has been gained through NiTi alloy, this binary alloy can be only applicable up to 100°C. For an actuator application in extreme temperature applications, it is necessary to use shape memory alloys with higher transformation temperatures. For this purpose, Au, Pd, Pt, Zr and Hf elements are used as ternary elements for NiTi alloys (Ma et al., 2010; Santamarta et al., 2013). In the last two decades NiTiHf alloys are found to be the most promising ones with their higher strength, high sensitivity to Hf percentage for manipulating the transformation temperatures and cost effectiveness (Karaca et al., 2014). Among all Hf alloys, nickel rich Ni_50.3Ti_29.7Hf₂₀ (at%) alloy has been studied extensively since high strength and easy tailoring of transformation temperatures can be achieved by nano-precipitation via aging (Coughlin et al., 2012; Karaca et al., 2013, 2014; Saghaian et al., 2016).

Most of the studies on NiTiHf alloy have been about the characterization of the transformation temperatures, microstructures and the evolution of actuation and plastic strain via heating-cooling under constant stress conditions (Karaca et al., 2013, 2014; Kockar et al., 2006; Saghaian et al., 2016). In structural fatigue experiments, high magnitudes of loads are applied cyclically till the failure of metals or metal alloys including SMAs. On the other hand, in the functional fatigue experiments SMAs are cycled via cooling-heating under constant stress for repeated martensite-austenite transformation and the actuation or recoverable and irrecoverable strain evolution together with the change in the transformation temperatures and thermal hysteresis are evaluated throughout the repeated phase transformation. Additionally, the fatigue behavior of NiTiHf alloys highly depends on the thermal and/or thermomechanical heat treatments (Ma et al., 2010).

Karaca et al. (2013, 2014) have stated that the shape recovery properties of nickel rich Ni_50.3Ti_29.7Hf₂₀ (at%) alloy can be enhanced by aging the material via producing nano-precipitates in the matrix. The optimized aging properties were attained as 550°C for 3 h to achieve highest strength magnitude with shape recoverability property (Karaca et al., 2013, 2014; Saghaian et al., 2016). However, the number of thermal cycles in these studies was kept very low while fewer functional fatigue studies were conducted to show the shape memory behavior of Ni_50.3Ti_29.7Hf₂₀ (at%) alloy via running higher number of thermal cycles. Besides, there have been also studies on the cyclic stability of equiatomic NiTi based alloys in which the severe plastic deformation techniques were used to improve their shape recovery properties via increasing the critical stress for slip (Frick et al., 2005; Gall et al., 2008; Kockar et al., 2006; Treppmann and Hornbogen, 1997; Tugrul et al., 2019). However, up to now, the functional fatigue experiments with high cycle numbers were run on Ni-rich Ni_50.3Ti_29.7Hf₂₀ (at%) alloy. For instance, effect of Upper Cycle Temperature (UCT) and applied stress magnitudes on the actuation and accumulated irrecoverable strain magnitudes with the evolution of transformation temperatures and fatigue life were investigated by Karakoc et al. (2017, 2018). The study on the effect of UCT temperature to the functional fatigue properties of Ni-rich Ni_50.3Ti_29.7Hf₂₀ (at%) alloy has shown that the increase in UCT led to an increase in accumulated irrecoverable strain magnitudes with the number of cycles such that the fatigue life decreased. On the other hand, the actuation strain magnitudes also increased with the increase in UCT due to the partial recovery of the retained martensite (Karakoc et al., 2017). It is important to note that, the applied stress magnitudes were 300 MPa and 400 MPa and the UCTs were kept as 300°C and 350°C such that 50°C of increase in UCT temperature was investigated in the aforementioned study (Karakoc et al., 2017). In another study, it was found that the increase in applied stress magnitude showed a very strong influence on the actuation and irrecoverable strain magnitudes and the fatigue life of the aged Ni_50.3Ti_29.7Hf₂₀ (at%) alloy as expected (Karakoc et al., 2018). The actuation strain together with the irrecoverable strain increased but the functional fatigue life decreased with the increased stress level (Karakoc et al., 2018). Besides, the actuation strain decreased with the number of thermal cycles throughout the experiments (Karakoc et al., 2018). More intriguing finding in this study was about the relation between the applied stress magnitude and the fracture surface appearance (Karakoc et al., 2018). The fracture mode type was determined to be transgranular under 200 MPa while the fracture modes were observed to be intergranular with rock-candy appearance under 500 MPa (Karakoc et al., 2018). Saygili et al. (2019) investigated the effect of nano-precipitate formation to the functional fatigue life of Ni_50.3Ti_29.7Hf₂₀ (at%) alloy without considering the UCTs during the experiments and it was found that aging at 550°C for 3 h increased the fatigue life of the alloy 3 times with respect to the hot extruded sample.

In the aforementioned studies, UCT has proven to be important since the increase in UCT led to a decrease in remnant/retained martensite phase and an increase in the actuation strain value. On the other hand, in this study, actuation strain was limited to a constant value, instead of limiting the UCT, and in every cycle, UCT was increased to recover approximately 2% of actuation strain. The main aim of keeping the actuation strain as 2% was to increase the fatigue life of the Ni-rich Ni_50.3Ti_29.7Hf₂₀ (at%) alloy via maintaining the transforming volume constant. However, UCT increase led to a decrease in the dislocation density as stated in the literature for the binary NiTi alloys (Atli et al., 2013). Additionally, it was previously shown that, the crack propagation proceeds by zigzagging easily between different twin plates to avoid the nano-precipitates in Ni-rich NiTiHf alloy (Amin-Ahmadi et al., 2019). Therefore, the crack formation and propagation due to the incompatibility between the martensite-austenite boundary and the easier propagation of the cracks between the different twin plates in martensite became the main damage mechanism, which was responsible for the failure of the strain limited sample. The fatigue life of the strain limited sample stayed the same as the one that was achieved via keeping UCT constant. The rationale behind the evolution of the damage mechanism throughout the thermal cycles under constant stress was also investigated for strain limited and UCT limited experiments via analyzing the evolution of actuation and the accumulated irrecoverable strain together with the transformation temperatures.

2. Experimental method

50.3at%Ni-29.7at%Ti-20at%Hf alloy was produced with vacuum induction melting process using high purity Ni, Ti and Hf. As cast billet was sealed into a mild steel can and hot extruded at 900°C with area reduction ratio of 4:1 for the homogenization of the casting structure. Tensile test samples used in this study were cut by wire electro discharge machining in the direction of extrusion.

Aging heat treatment was applied at 550°C for 3 h to enhance the strength of the alloy via precipitation hardening which was shown to be the most promising aging parameters in the literature as stated before.

Details of the custom made functional fatigue testing machine which was built by our group were explained in the aforementioned study (Saygili et al., 2019). However, it is worth to mention that the measurement and the control systems were designed and controlled with NI LabVIEW program since the cooling-heating cycles under 200 MPa constant stress magnitude were run via limiting the actuation strain magnitude to 2%. Limited strain value of 2% which was lower than the ultimate actuation strain magnitude for 200 MPa loading condition was set for each cycle such that the samples were cooled and heated to a temperature at which the limited strain value was achieved at each cycle. Therefore, UCTs were not kept constant in strain limited experiments. These experiments will be called as Strain Limited (SL) and the other experiments at which the UCTs were kept constant as 300°C will be called as UCT Limited (UCTL) through the rest of the paper.

High cycle functional fatigue experiments were repeated two times under constant stress magnitude of 200 MPa for SL and UCTL experiments. Two samples were thermally cycled with constant UCT of 300°C. Loading was done at 300°C when the samples were in austenite phase, cooled down to complete the martensitic transformation and reheated again to 300°C. Heating was performed with joule heating using direct current which was controlled by a PID script in NI Labview. These samples will be expressed as UCTL1 and UCTL2 throughout the text and in the Figures.

Strain Limited samples were heated to 300°C into fully austenite phase and then loaded to 200 MPa. After loading, cooling was performed for martensitic transformation. Then, heating was conducted up to achieving 2% of actuation strain and ended by the interruption of austenitic transformation. Strain Limited experiments which were conducted on two samples will be mentioned in the rest of the text as SL1 and SL2.

Postmortem thermal analyzes using Differential Scanning Calorimeter (DSC) equipment were also done to compare the change in the transformation temperatures and enthalpy values after SL and UCTL experiments. Additionally, Optic Microscope (OM) images were taken from the surface of the failed SL and UCTL samples after the fatigue experiments to compare the crack formations. The surfaces of the samples were grinded without applying force using 1200 and 2500 grinding papers to maintain the formed cracks during the fatigue experiments and then polished with 1 µm diamond suspension to get clear images of the cracks.

3. Results

3.1. Functional fatigue experiments

The strain versus temperature curves for UCTL functional fatigue experiments with high cycle numbers of 16,500–20,000 cycles were presented in our previous study (Saygili et al., 2019). The strain versus temperature curves which were obtained from the first and the second strain limited fatigue experiments were shown in Figures 1 and 2. UCTL and SL experiments were run twice to show the consistency of the results. It is important to note that, some of the cycles from the functional fatigue experiments were plotted to clearly show the evolution of the strain and temperature values and these cycles were separated for better visibility. The procedure which was followed to determine martensite strain (ε_M), austenite strain (ε_A) (or total irrecoverable strain) and the actuation strain (ε_act) values together with the transformation temperatures was shown in Figure 3. Actuation strain values were calculated by taking the difference of martensite and austenite strain values. The comparison of the martensite and austenite strain was presented in Figure 4 and the calculated actuation strain values were shown in Figure 5.

The ε_A values obtained from UCTL experiments increased with the number of cycles as shown in Figure 4. On the other hand, it can be clearly seen that there was a little increase in the ε_A for the first 300 cycles and then the ε_A became almost constant till the end of 13,000th cycle. Specifically, the increase in the ε_A values with the number of cycles in the UCTL experiments was considerably higher than that of the increase with the number of cycles in the SL experiments. Although it was very difficult, the ε_act with the number of cycles in the SL experiments were tried to keep constant at 2%. Therefore, a small shift from 2% strain value was observed throughout the SL experiments as demonstrated in Figure 5. On the other hand, the ε_act values obtained from UCTL experiments continuously decreased with the number of cycles.

Since the first and second SL and UCTL experiments showed almost the same results, the transformation temperatures obtained from the strain versus temperature curves of the SL1 and UCTL1 were compared for simplicity in Figure 6. Additionally, SL1 and UCTL1 samples showed almost the same fatigue life. It can be seen from this Figure that there was a similar increasing trend in the Martensite Finish (M_f) and Austenite Start (A_s) temperatures. However, the tendency in the evolution of the Martensite Start (M_s) and Austenite Finish (A_f) temperatures was different. M_s and A_f temperatures increased with the number of cycles in UCTL experiment. However, strain limited experiments have shown that there was a considerable increase in the M_s and A_f temperatures up to around 6000 cycles and then stabilized. All the possible reasons behind these results will be explained in the discussion section.

3.2. Postmortem analysis with DSC

Figure 7 represents the DSC results which were obtained from the pre-fatigue, SL1 and UCTL1 samples to compare the transformation temperatures and the enthalpy values. These values together with the calculated thermal hysteresis (ΔT) and the difference between the M_s temperatures before and after fatigue experiments were shown in Table 1.

Table 1.

Transformation temperatures and transformation enthalpy values which were drawn from DSC curves and thermal hysteresis and the difference between the Ms temperatures after and before fatigue experiments which were calculated from the transformation temperatures.

Samples	M_f(°C)	M_s(°C)	A_s(°C)	A_f(°C)	ΔH(J/g)	ΔT (°C)(A_f–M_s)	M_s (after fatigue)-M_s (before fatigue)ΔM_s (°C)
Pre-Fatigue	101	142	138	171	15.9	29	–
SL1	128	180	161	224	7.3	44	38
UCTL1	109	192	147	240	6.5	48	50

As can be realized from Table 1 that the transformation enthalpy value decreased down to half values after the fatigue tests. DSC curves in Figure 7 also showed that the fatigue experiments led to observe wider transformation peaks and the peaks obtained after UCTL1 experiment were relatively wider than that of the one obtained after SL1 experiment. Moreover, ΔT and ΔM_s values also increased after the fatigue experiments.

3.3. Postmortem analysis with DSC

The amount and the forms of the cracks on the surfaces of the failed SL1 and UCTL1 samples were presented in Figure 8(a) and (b), respectively to reveal the damage accumulation after running the experiments. It is important to note that Optical Microscope (OM) images were taken from the same locations and almost 1 mm away from the fracture surfaces of SL and UCTL samples. It can be easily seen from the OM images that the crack density of SL1 sample was noticeably higher than that of the UCTL1 sample. The surfaces on which the OM studies were done are referred as length × width surfaces.

4. Discussion

The main aim to conduct SL experiments was to investigate the possibility to increase the functional fatigue life of NiTiHf HTSMA with stable shape memory properties. The results of cooling-heating experiments under 200 MPa following the SL and UCTL paths have shown that the number of cycles which was achieved with the strain limited experiments has not been increased but ε_A values, in other terms the accumulated irrecoverable strain values have become considerably stable throughout the number of cycles. As it was shown in Figure 4, the accumulated irrecoverable strain value (austenite strain) initially increased to 0.3% over the first 300 cycles, decreased down to around 0.2% and then stabilized for more than 10.000 cycles. On the other hand, the samples tested by following the UCTL path, exhibited a continuous increase in the ε_A values. Additionally, the actuation strain values obtained in UCTL experiments decreased from 2.5% down to 1.5% with the increase in ε_A. The continuous decrease in ε_act and the increase in ε_A with the number of cycles were due to the accumulation of dislocation substructures and internal stress which led to retained martensite formation (Karakoc et al., 2017). The transforming volume and thus the actuation strain values decreased with the number of cycles since UCT was kept constant in UCTL experiments. On the other hand, the actuation strain values in the SL experiments were kept constant since UCT increased throughout the cycles to maintain 2% constant actuation strain as shown in Figures 1 and 2, such that the increase in UCT may lead to hinder or mitigate the local retained martensite formation. The accumulated irrecoverable/austenite strain was still noticeable due to the plastic deformation in SL experiments at the beginning as it can be seen in Figure 4. However, dislocation annihilation due to the increase in UCT overcame the dislocation formation due to cyclic deformation such that there was a small decrease in ε_A value of SL1 and SL2 samples after 300 cycles and then ε_A values were stabilized until the samples got close to the fracture.

In addition to the evolution of ε_A and ε_act values in SL and UCTL experiments, the transformation temperatures also showed an increasing trend with the number of cycles. The evolution in A_f and M_s temperatures was noteworthy in Figure 6. There was a rapid increase in both of these temperatures through the cycles of the SL experiments starting from the beginning up to 5000th cycles, then the increase in these temperatures slowed down and they became stabilized while both of these temperatures continuously increased in UCTL specimens. The reason in the rapid increase of A_f and M_s during the first 5000 cycles of SL experiments is correspondent with the increase in the rate of UCT such that higher UCT can lead to a decrease in the strength of the alloy which might be the possible reason of more dislocation formation and it has been already known from the literature that the dislocation density increase is the main reason in the increase of the transformation temperatures (Karakoc et al., 2017, 2018).

Transformation temperatures and the enthalpy values before and following the fatigue experiments were drawn from DSC curves in Figure 7 and were presented in Table 1. The transformation enthalpy values (ΔH) of the failed SL and UCTL samples reduced to the half values of the pre-fatigue condition. The reduction in ΔH values is the sign in the decrease of the transforming volume after fatigue thermal cycle. This might be due to the pinned martensite plates with the dislocations formed during cycles. Although the number of cycles up to failure for SL1 and UCTL1 samples is similar, ΔH value obtained after UCTL fatigue experiment is lower than that of SL experiment. There might be two possible reasons for this. SL1 sample showed highest actuation strain due to increasing UCT throughout the experiment such that there might be less amount of retained martensite. The second reason is the dislocation annihilation and thus smaller dislocation density due to heating the sample to higher UCTs. Additionally, the transformation temperatures after fatigue experiments are determined to be higher than that of the pre-fatigue sample. The difference between M_s temperatures after SL, UCTL and the prior to fatigue experiments were found as 38°C and 50°C, respectively as shown in Table 1. This observation is also consistent with the relative increase of the transformation enthalpy values after SL and UCTL experiments. However, there is still a remarkable increase in M_s temperature after SL experiment since most of the damage and dislocation formation occur during the first 5000 cycles.

Although the accumulated irrecoverable strain values, in other terms, ε_A values were determined to be very much lower for SL1 sample than that of the UCTL1 sample, the functional fatigue life times which were achieved via following SL and UCTL experiments were similar. This observation might be not only due to the fact that UCT temperature was increased but also the actuator strain was kept constant. The higher and constant transforming volume, in other words, the constant actuation strain values in SL experiment throughout the cycles may lead to higher amount of crack formation with respect to the crack formation during UCTL experiment owing to the incompatibility between martensite and austenite phases. Additionally, cracks may easily propagate between the different plates of twins in NiTiHf alloys (Amin-Ahmadi et al., 2019). The OM images, which were taken from the surfaces of SL and FT samples in Figure 8 showed the higher amount of crack formations after strain limited experiments. Therefore, one can conclude that the damage in UCTL experiments proceeded with dislocation formation while the damage occurred with crack formation in SL experiments. Although the failure of SL samples was expected at later cycles due to less retained martensite and thus the dislocation formation, more crack formation leads SL samples to show approximately the same number of cycles till failure in functional fatigue experiments.

5. Conclusion

The main aim of this study was to evaluate the effect of strain limitation to the performance of the fatigue properties and the life of Ni_50.3Ti_29.7Hf₂₀. Here are the main conclusions which can be drawn from this study:

(1) Strain limitation has a strong effect on the accumulated irrecoverable strain values in fatigue experiments. ε_A values are higher in SL experiments at the beginning since the dislocation formation was more significant. However, a small decrease in ε_A values was observed since the dislocation annihilation due to the increase in UCT overcame the dislocation formation due to cyclic deformation. When the samples got close to the failure, accumulated irrecoverable strain showed a slight increase again.

(2) Although SL1 and UCTL1 samples showed similar fatigue life cycles, less shift of transformation temperatures was observed after strain limited experiments in DSC experiments, suggesting less dislocation formation during the SL thermal cycles. On the other hand, the number of cycles that SL1 sample showed in fatigue experiments was similar to the number of cycles that UCTL1 sample showed; more crack formation was found in SL1 sample due to the possible incompatibility between martensite and austenite phases since more volume was transformed at each cycles. Otherwise, one could achieve higher fatigue life via running SL experiments. Therefore, if the crack formation and propagation can be hindered, functional fatigue life can be enhanced by following SL path.

(3) The damage was observed to be higher during the first 300 cycles due to higher rate of dislocation formation for both of SL and UCTL experiments on the basis of the increase in ε_A values.

Actuator which can be produced by aged Ni_50.3Ti_29.7Hf₂₀ alloy can be used by keeping the actuation strain constant via increasing the UCT temperature to keep the accumulated irrecoverable strain low with a close control of the hysteresis loop.

Footnotes

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study was supported by the Turkish Aerospace Industries under Grant no. DKTM/2015/10.

ORCID iD

Benat Kockar

References

Amin-Ahmadi

Noebe

Stebner

(2019) Crack propagation mechanisms of an aged nickel-titanium-hafnium shape memory alloy. Scripta Materialia 159: 85–88.

Atli

Karaman

Noebe

, et al. (2013) The effect of training on two-way shape memory effect of binary NiTi and NiTi based ternary high temperature shape memory alloys. Materials Science and Engineering: A 560: 653–666.

Benafan

Garg

Noebe

, et al. (2014) Mechanical and functional behavior of a Ni-rich Ni50.3Ti 29.7Hf20 high temperature shape memory alloy. Intermetallics 50: 94–107.

Benafan

Noebe

Halsmer

(2016) Static rock splitters based on high temperature shape memory alloys for planetary explorations. Acta Astronautica 118: 137–157.

Coughlin

Phillips

Bigelow

, et al. (2012) Characterization of the microstructure and mechanical properties of a 50.3Ni–29.7Ti–20Hf shape memory alloy. Scripta Materialia 67(1): 112–115.

Frick

Ortega

Tyber

, et al. (2005) Thermal processing of polycrystalline NiTi shape memory alloys. Materials Science and Engineering: A 405(1–2): 34–49.

Gall

Tyber

Wilkesanders

, et al. (2008) Effect of microstructure on the fatigue of hot-rolled and cold-drawn NiTi shape memory alloys. Materials Science and Engineering: A 486(1–2): 389–403.

Godard

Lagoudas

(2003) Design of space systems using shape memory alloys. Proceedings of SPIE, Smart Structures and Materials 5056: 545–558.

Hartl

Lagoudas

(2007) Aerospace applications of shape memory alloys. Proceedings of the Institution of Mechanical Engineers, Part G: Journal of Aerospace Engineering 221(4): 535–552.

10.

Jani

Leary

Subic

, et al. (2014) A review of shape memory alloy research, applications and opportunities. Materials and Design 56: 1078–1113.

11.

Karaca

Saghaian

Ded

, et al. (2013) Effects of nanoprecipitation on the shape memory and material properties of an Ni-rich NiTiHf high temperature shape memory alloy. Acta Materialia 61(19): 7422–7431.

12.

Karaca

Acar

Tobe

, et al. (2014) NiTiHf-based shape memory alloys. Materials Science and Technology 30(13): 1530–1544.

13.

Karakoc

Hayrettin

Bass

, et al. (2017) Effects of upper cycle temperature on the actuation fatigue response of NiTiHf high temperature shape memory alloys. Acta Materialia 138: 185–197.

14.

Karakoc

Hayrettin

Canadinc

, et al. (2018) Role of applied stress level on the actuation fatigue behavior of NiTiHf high temperature shape memory alloys. Acta Materialia 153: 156–168.

15.

Kockar

Karaman

Kim

, et al. (2006) A method to enhance cyclic reversibility of NiTiHf high temperature shape memory alloys. Scripta Materialia 54(12): 2203–2208.

16.

Karaman

Noebe

(2010) High temperature shape memory alloys. International Materials Reviews 55 (5): 257–315.

17.

Saghaian

Karaca

Souri

, et al. (2016) Tensile shape memory behavior of Ni50.3Ti29.7Hf20 high temperature shape memory alloys. Materials and Design 101: 340–345.

18.

Santamarta

Arroyave

Pons

, et al. (2013) TEM study of structural and microstructural characteristics of a precipitate phase in Ni-rich Ni–Ti–Hf and Ni–Ti–Zr shape memory alloys. Acta Materialia 61(16):6191–6206.

19.

Saygili

Tugrul

Kockar

(2019) Effect of aging heat treatment on the high cycle fatigue life of Ni50.3Ti29.7Hf20 high-temperature shape memory alloy. Shape Memory Superelasticity 5(1): 32–41.

20.

Singh

Sirohi

Chopra

(2003) An improved shape memory alloy actuator for rotor blade tracking. Journal of Intelligent Material Systems and Structures 14(12): 767–786.

21.

Treppmann

Hornbogen

(1997) On the influence of thermomechanical treatments on shape memory alloys. Journal de Physique IV 7(C5): 211–220.

22.

Tugrul

Saygili

Velipasaoglu

, et al. (2019) Comparison of the transformation behavior of cold rolling with aging and hot extrusion with aging processed Ni50.3Ti29.7Hf20 high temperature shape memory alloy. Smart Materials Structures 28: 105029.