Bond Strength and Corrosion Protection Properties of Hot-Dip Galvanized Prestressing Reinforcement in Normal-Strength Concrete

Abstract

Several prestressing reinforced structures have recently collapsed due to chloride-induced steel corrosion. This study investigates the effect of the corrosion of hot-dip galvanized conventional prestressing steel reinforcement under hydrogen evolution on bond strength in normal-strength concrete. The impact of hydrogen evolution on the porosity of cement paste at the interfacial transition zone (ITZ) is verified through image analysis. The whole surface of prestressing strands is hot-dip galvanized, and their corrosion behavior when embedded in the cement paste is investigated by measuring the time dependence of the open-circuit potential. Concerning the uniformity of the hot-dip galvanized coating and its composition, it is advisable to coat the individual wires of the prestressing reinforcement and subsequently form a strand. It is demonstrated that the corrosion of the coating under the evolution of hydrogen in the cement paste reduces the bond strength of hot-dip galvanized reinforcement in normal-strength concrete. Image analysis after 28 days of cement paste aging indicates insignificant filling of hydrogen-generated pores by zinc corrosion products. Applying an additional surface treatment (topcoat) stable in an alkaline environment is necessary to avoid corrosion of the coating under hydrogen evolution and limit the risk of bond strength reduction.

1. Introduction

Corrosion protection of conventional steel concrete reinforcement with a hot-dip galvanized (HDG) coating can extend the service life of reinforced concrete structures. This coating increases the corrosion resistance to carbonation of the concrete cover and the action of chloride anions [1,2,3]. However, it is also known that in fresh concrete, the hot-dip galvanized coating corrodes not only by the formation of Zn²⁺ cations (anodic corrosion reaction—see Equation (1)) but also by the development of hydrogen, which is formed by a cathodic corrosion reaction—see Equation (2) [4,5].

$$\mathrm{Z}\mathrm{n}\to {\mathrm{Z}\mathrm{n}}^{2+}+{2\mathrm{e}}^{-}$$

(1)

$${2\mathrm{H}}_2\mathrm{O}+2{\mathrm{e}}^{-}\to {\mathrm{H}}_2+2{\mathrm{O}\mathrm{H}}^{-}$$

(2)

The generated hydrogen can increase the porosity of the cement paste at the interfacial transition zone (ITZ), thereby reducing the bond strength of the coated reinforcement in the concrete [6,7,8]. However, according to the results of some studies, the evolution of hydrogen bubbles and the formation of specific corrosion products (calcium hydoxyzincate dihydrate—CHZ) has an insignificant effect on bond strength [9,10], it is usually recommended to apply an additional protective coating on the surface of hot-dip galvanized concrete reinforcement [11]. This topcoating aims to prevent the corrosion of hot-dip galvanized steel in fresh concrete under hydrogen evolution. For this purpose, conversion coatings [12,13], organofunctional silane-based coatings [14], or, e.g., acrylate dispersions [15] can be used. The hot-dip galvanized coating is currently also considered as corrosion protection for conventional prestressing reinforcement steel of concrete [16], and a suitable method of corrosion protection needs to be developed.

Usually, the corrosion damage of prestressing reinforcement is influenced by the stimulating effect of chloride anions—see Figure 1. The critical level of contamination of the grout itself (0.2 wt. % converted to cement content) in the prestressing reinforcement protector by chloride anions is lower than that of conventional reinforced concrete structures (0.4 wt. % converted to cement content) [17,18]. The electrochemical corrosion damage of the prestressing reinforcement stimulated by chloride anions likely contributed to the collapse of the cable-supported Prague-Troja footbridge in 2017 (stress ribbon, Figure 2) [19] and also of Ponte Morandi in Genoa in 2018 [20].

Previous studies investigated using epoxy coatings as corrosion protection for prestressing reinforcement in concrete. Adding hard additive particles to the top epoxy layer significantly increased the surface roughness of the reinforcement, and thus increased the bond strength in concrete [21]. A 2020 patent recommended protecting the surface of the conventional prestressing reinforcement with a hot-dip galvanized coating (batch technology) [16]. However, the recommendation to additionally coat the outer surface of the hot-dip galvanized coating to avoid hydrogen-induced bond strength reduction needs to be questioned as sound experimental evidence about the bond strength of such coated prestressing reinforcement in normal-strength concrete (NSC) is missing.

However, it is clear that the geometry of the reinforcement surface and the composition of the hot-dip galvanized coating have a significant effect on the bond strength of concrete. It has been discussed in the literature that the composition of the hot-dip galvanized coating has a major influence on corrosion behavior in concrete environments [22,23,24]. It is recommended that the external layer should be composed of phase η (a solid solution of iron in zinc with an iron content of about 0.03 wt. %) to ensure the conditions for the transition of corrosion from the active to the passive state as sufficient zinc is necessary for the formation of a continuous layer of Ca[Zn(OH)₃]₂·2H₂O, the so-called CHZ—calcium hydroxyzincate dihydrate [22,23]. The hot-dip galvanized coating contains an inner layer of Fe-Zn intermetallic phases (see the composition of the HDG coating shown in Figure 3) under certain conditions (higher Si and P content in steel with a type of heat treatment playing a significant role), a coating can be formed when the outer layer is formed by the intermetallic phase ζ (FeZn₁₃; the iron content is 5.00–6.02 wt. %) [25,26,27,28,29]. It has been experimentally shown that this phase exhibits lower corrosion resistance in the alkaline environment of concrete (higher corrosion rate, higher kinetics of H₂ formation, formation of cracks across the coating) [30,31,32]. Thus, it is evident that the presence of an external layer of this phase can have a negative effect on the bond strength of the hot-dip galvanized steel in the concrete and on the extension of the service life of the structure since the Fe-Zn intermetallic phases exhibit lower resistance against chloride-induced corrosion.

This paper focuses on verifying the effect of hot-dip galvanized coating (batch technology) on the surface of currently used prestressing reinforcement (without applied prestressing) on corrosion behavior and bond strength under real conditions of the cement paste or normal strength concrete applications. In the case of corrosion, hydrogen-induced changes in the porosity of the cement paste at the interfacial transition zone (ITZ) are also tested.

This experimental verification aims to provide fundamental insights for the design of new and assessment of existing prestressed reinforced concrete structures with such coating (particularly for bridges in chloride-laden environments such as coastal areas or where de-icing salts are applied) concerning their service life (consideration of the effect of chloride anions) and possible changes in load capacity (effect of coating corrosion in fresh concrete on bond strength).

2. Materials and Methods

An experimental program was conducted to assess the corrosion properties and bond strength to concrete of hot-dip galvanized prestressing steel. The program included time-dependent measurements of the open-circuit potential (E_corr) in real cement paste conditions (individual wires were tested) and bond strength testing using pull-out tests (cubic specimens of normal-strength concrete). Scanning electron microscopy was also used to evaluate the microstructure of the hot-dip galvanized coating on the prestressing steel and the influence of coating corrosion under hydrogen evolution on the porosity of the cement paste at the interfacial transition zone (ITZ).

2.1. Materials Used

For the aforementioned experimental research, a high-carbon steel strand (conventional prestressing steel reinforcement according to prEN 10138-3 [33]) type Y1860 (1.1366) S7 15.7 (supplier: Freyssinet CS, a.s., Zápy-Brandýs nd Labem-Stará Boleslav, Czech Republic) was used. The composition of steel Y1860 S7 15.7 is presented in

Table 1. Composition of steel Y1860 (1.1366) S7 15.7 tested with prestressing strain (composition guaranteed by the producer).

Compound	C	Mn	Si	P	S	N	Fe
Content (wt. %)	0.81	0.72	0.26	0.008	0.009	0.006	balance

and the material characterization of the prestressing reinforcement steel is summarized in

Table 2. Material characterization of prestressing reinforcement steel (parameters guaranteed by the producer).

Property	Value and Unit
Ultimate tensile strength	1860 MPa
Nominal diameter	15.7 mm
Nominal density	~1172 g/m
Yield strength (0.1%)	246 kN
Young’s modulus	~195 GPa
Maximum elongation	3.5%

.

Samples of wires and strands of prestressing reinforcement were hot-dip galvanized in a commercial galvanizing plant (Apollo Metal, s.r.o., Čenkov, Czech Republic). Before coating, the samples were degreased at the Klokner Institute laboratory (CTU-Klokner Institute, Prague, Czech Republic) using mechanical cleaning with a CaO paste followed by conventional degreasing in 25 wt. % KOH for 20 min at 75 °C after rinsing in distilled water. The supplier of the surface treatment declared that the samples were also alkali-cleaned and pickled before the coating application [34]. Reference samples of wires (uncoated) were also degreased at the Klokner Institute laboratory before measuring the open-circuit potential. Reference samples of prestressing steel strands for bond strength testing were not degreased.

Portland cement CEM I 42.5 R (Českomoravský cement, a.s., Mokrá-Horákov, Czech Republic) was used for the preparation of the cement paste and concrete specimens, with its composition detailed in

Table 3. Cement composition guaranteed by the producer (CEM I 42.5 R).

Compound	CaO	SiO2	Al2O3	Fe2O3	MgO	SO3	K2O	Na2O
content (wt. %)	64.2	19.5	4.7	3.2	1.3	3.2	0.78	0.09

. For measuring the time-dependent E_corr of individual prestressing steel wires, a cement paste with a water-to-cement ratio (w/c) of 0.35 was used, prepared according to the mix design given in

Table 4. Content by m³ of cement paste for E_corr measurements.

Mixture	Portland Cement (kg)	Water (L)	Gravel (kg)	Sand (kg)
w/c: 0.35	14	4.9	0	0

. For bond strength tests, cubic concrete specimens with a w/c ratio of 0.55 were produced using normal-strength concrete (NSC) as per the mix design presented in

Table 5. Content by m3 of normal strength concrete for bond strength tests.

Admixture	Content (kg/m3)	Note
Cement (CEM I 42.5 R)	365	Pure Portland cement
Aggregate	900	Fraction 0/4—fine sand
	585	Fraction 4/8
	285	Fraction 8/16
Mixture (w/c)	0.55	-

[35].

2.2. Corrosion Investigations

A total of 14 cylindrical concrete specimens were created to measure the time-dependent E_corr (see Figure 4). These specimens contained prestressing steel wires as samples (7 uncoated wires and 7 hot dip galvanized wires). A cylindrical PE mold with a diameter of 50 mm and two centered PE caps were used. The mold and caps were 3D-printed (Prusa i3MK3S+ Prusa Research, Prague, Czech Republic). The total height of the mold was 150 mm. The bottom cap had no holes, and the gap between the cylindrical wall and the bottom cap was sealed with neutral elastic silicone-based sealant. The top cap had holes for inserting the prestressing steel wire specimen (uncoated or hot-dip galvanized). A hole was created at the top cap’s edge to place a reference electrode (foil) made of corrosion-resistant steel—FeCr18Ni9 (Goodfellow Metals, Huntingdon, UK). The distance between the reference electrode and the sample surface was 20 mm (see Figure 4) [6].

The cement paste was prepared with a higher water-to-cement ratio (w/c: 0.42) to ensure complete wetting of both the test wires and reference electrodes. To avoid potential interference with the corrosion process, plasticizers and aggregates were intentionally omitted. Therefore, the consistency of the cement paste could only be adjusted by adding water.

The evolution of the open-circuit potential (E_corr) between the reference stainless steel and the prestressing steel wire specimen was measured using an OWON B35T+ multimeter (OWON-Fujian Lilliput Optoelectronics Technology Co., Ltd., Zhangzhou, China). Measurements were taken on the first day of exposure after 30 min and subsequently every 2 h for 6 days (the molds were stored in a humid atmosphere above the water level: 65% RH; 20.5 ± 1 °C). After this period, the specimens were submerged in water for 7 days (after approximately 170 h). Subsequently, the E_corr of the samples was measured again (every 2 h), but the samples were removed from the water during measurements (and then returned to the water). Simultaneously, the potential difference between the reference electrode in the cement paste and a commercial saturated calomel electrode (SCE) was measured. Measurements were taken manually using an OWON B35T+ multimeter after opening the top cap of the mold (the average potential difference after 30 min of stabilization was determined to be −40 mV).

2.3. Bond Strength Investigation

To prepare the cubic normal-strength concrete (NSC) specimens, aggregates that absorbed a portion of the water were used [36]. Consequently, even though no plasticizer was added to the concrete mix, additional water was not required to ensure adequate wetting of the prestressing steel strand surface (actual w/c ratio: 0.55). The cement paste in the cylindrical specimens was manually compacted in four layers using a standard device (concrete mixer GIFOS MB 80; speed rotation: 30 rounds per minute) followed by vibration on a vibration table (GIFOS s.r.o., Brno-Horní Heršpice, Czech Republic). The NSC specimens in steel cubic molds were compacted using a hand-held device (Eibenstock EBR 125.1, 10,000 rpm-Elektrowerkzeuge GmbH Eibenstock, Eibenstock, Germany) for 1 min (the mold was filled in two separate layers and compacted). The prestressing steel strand specimen was anchored in the center of the steel cubic mold (for pull-out tests) using a wooden wedge (placed at the bottom of the mold) and a PE insert at the top (created by 3D printing)—see Figure 5.

All test specimens (cement paste, NSC) were cured in a humid atmosphere (95% RH; 20.5 ± 1 °C) for 28 days [37]. A total of 7 parallel specimens were prepared for bond tests.

The pull-out tests to determine bond strength were conducted on cubic specimens according to RILEM RC6 [38] and ASTM C234-91a [39] standards. Bond strength was determined based on the slip–bond stress relationship. The slip was measured using an LVDT sensor (Micro-Epsilon Messtechnik GmbH, Ortenburg, Germany) placed on the unloaded end of the reinforcement—see Figure 6. The pull-out tests were performed on an MTS 500 kN testing machine. The pull-out test was controlled by displacing the unloaded end of a reinforcing bar at a constant rate of 0.005 mm/s [40]. The test was terminated when a slip of 2.0 mm was reached.

2.4. Scanning Electron Microscopy

The coating microstructure was evaluated using scanning electron microscopy (SEM) (Jeol JSM-IT200, Tokyo, Japan) and optical microscopy (Weldinspect DIM-U) on cross-sections. Grinding was performed on an automatic grinder LaboPol-2 (Struers GmbH, Willich, Germany) using grinding papers (P60-P2400) and diamond polishing pastes.

Measurements of the changes in porosity of the cement paste at the interfacial transition zone (ITZ) between the prestressing reinforcement and the cement paste (with a composition similar to the samples for E_corr measurements) were conducted [6,14]. Steel strands were embedded to approximately 3/4 of their cross-sectional depth into 3 prismatic PE molds for each type of prestressing steel reinforcement (see Figure 7). After 28 days of curing at 60% RH, the reinforcements were removed from the specimens. After drying the mechanically divided cement paste samples (8 h, 85 °C), the porosity of the cement paste at the ITZ was measured. The total surface area occupied by pores (%) was evaluated and compared between the two types of interfaces (cement paste from uncoated prestressing reinforcement and hot-dip galvanized reinforcement). A scanning electron microscope (SEM) (Jeol JSM-IT200, Tokyo, Japan) and the free software ImageJ were used for evaluation. The total analyzed area from both samples was approximately 4 cm².

2.5. Roughness Investigations of Prestressing Reinforcement

The surface roughness R_a (measured in the longitudinal direction) was verified for individual wire specimens (uncoated steel and hot-dip galvanized steel) from the prestressing steel reinforcement. The surface roughness has an effect on the bond strength of reinforcement in concrete. A total of 8 wires were measured (6 measurements per wire). The mean value and standard deviation of each reinforcement were evaluated. The individual wires were clamped separately in a clamping vice, and the measurements were performed using a MITUTOYO SJ-210 instrument (Mitutoyo, Kawasaki, Japan).

2.6. X-ray Diffraction of Corrosion Products

An X-ray diffraction analysis of the composition of the corrosion products was performed on a sample of hot-dip galvanized prestressing steel reinforcement exposed in cement paste (28 days of aging, reinforcement taken from an experiment to verify the changes in the porosity of cement paste on ITZ). For the actual analysis, a PAN analytical X´Pert3 instrument (Malvern Panalytical V. B., Almelo, The Netherlands) using CuKα radiation over the angular range 5–110° was used.

3. Results and Discussion

The results of the presented test program are discussed in four subsections. In Section 3.1, the composition of the hot-dip galvanized (HDG) coating on the individual wires of prestressing steel strain (six outer and one inner wire in total) is evaluated by scanning electron microscopy (SEM) [41]. In Section 3.2, the open-circuit potential of uncoated and HDG wires separated from the strands is measured in the cement paste. In Section 3.3, the results of pull-out tests of the bond strength of the prestressing steel strand in normal strength concrete are presented. In that section, the porosity of the cement paste taken from the ITZ zone (from the surface of both uncoated and HDG prestressing strands) is evaluated by image analysis (ImageJ).

3.1. Structure of HDG Coating

A typical example of the thickness and continuity of the HDG coating on the surface of individual wires taken from strands of the prestressing reinforcement is shown in Figure 8 (cross-section taken by optical microscopy). The surface of the outer wires of the prestressing reinforcement exhibited uniform coating coverage. However, the thickness of the coating was locally very rough—a local increase of up to 400% compared to the mean thickness. This significantly increased the friction of the surface of the test reinforcement, which can affect the increase in the bond strength of the coated reinforcement with the concrete [6,14,42,43]. The differences in coating thickness were related to the locally different reactivity of the steel surface. The inhomogeneity of carbon and silicon representation in the steel may influence this. An increased manganese content in the steel should not play a role in this regard [44]. However, the non-uniformity of the distribution of crystal lattice defects in the steel, the local surface cleanliness, and possibly the distribution of internal stresses in the steel of prestressing strands can be discussed. A significant thickness of the HDG coating was often detected at contacts of the outer wire surfaces. Due to the higher thickness of the coating, cracking once a strand was prestressed occurred (Figure 9A). These cracks can cause the coating to spall off from the steel surface during the handling and storage of the reinforcement. A detailed study of the surface of the inner and outer wires revealed localized coating delaminations (see Figure 9B). Both locally high thickness and cracking of the coating can reduce the barrier protection properties of the coating on the surface of the prestressing reinforcement. The corrosion products can cause the debonding of other coating parts in a short exposure time due to stress cracking [45,46].

Imaging the cross-sections of the coating of an outer and inner (Figure 10) wire of the prestressing reinforcement revealed that a continuous HDG coating with a thickness of approximately 65 µm formed locally (typical locations shown) on the steel surface. The outer wires exhibited the usual coating composition of single Fe-Zn intermetallic phases (see Figure 3) and continuous outer η layers [25,26,27,28,29,47,48]. However, the thickness of that phase was locally less than 10 µm, which is the necessary thickness for the formation of a protective layer from the corrosion products on the dominating Ca[Zn(OH)₃]₂·2H₂O base [6,14,22,49,50]. However, the thickness of the coating on the surface of the inner wire was slightly higher—by about 10%. It appeared that the composition of the coating was atypical, i.e., the Γ phase (i.e., Γ + Γ₁), the δ phase (δ_1k + δ_1p), and the outer layer was a palisade structure of the ζ phase (FeZn₁₃). The ζ phase also showed a slightly higher thickness than usual. The outer layer from the η phase was absent (only local spots were observed). Thus, the HDG coating had a similar composition as if it were formed on silicon-killed steel (with Si content in the Sandelin region, i.e., 0.03–0.12 wt. %) [14,26,51,52]. The hydrogen overvoltage on the surface of the ζ phase was lower than on the surface of the η phase, accelerating the kinetics of hydrogen gas formation [30,31,32,53]. Earlier experimental works indicated that the hydrogen evolution could significantly compromise the integrity of the coating, thereby increasing the likelihood of compressive stresses causing transverse cracks, a significant hydrogen embrittlement of the coating, and susceptibility to brittle fracture [14,53,54]. The difficult accessibility of the steel surface for the zinc melt was a major contributor to the atypical coating composition on the inner wire of the prestressing reinforcement.

The results of surface roughness (Ra) measurements of uncoated and hot-dip galvanized prestressing steel wires are shown in Figure 11. The results indicate that the surface roughness (measured in the longitudinal direction) of hot-dip galvanized steel is higher than that of uncoated steel (approximately three times). Although the resulting coating is metallic sheen and the grain morphology is visually fine and uniform (produced after the samples are removed from the hot-dip galvanized bath). Similar values of Ra in the case of hot-dip galvanized and uncoated plain bars steel have been measured previously [6]. However, it can be considered that the surface roughness of single-strand hot-dip galvanized wires in strand (coating of the whole strand) will be higher (see Figure 8).

In general, it can be summarized that hot-dip galvanizing of whole strands was disadvantageous with regard to optimum protective properties. A metallographic analysis revealed a non-uniform coating thickness, the local absence of coating on the steel surface (between the outer and inner wires of reinforcement), and the formation of coating without a continuous η phase layer (on the surface of the inner wire of the prestressing reinforcement). To ensure the formation of a continuous coating of hot-dip galvanizing with an external η phase layer (solid solution of iron in zinc), it is necessary to hot-dip galvanize the individual wires of the prestressing reinforcement separately and then form the prestressing strain reinforcement.

3.2. Open-Circuit Potential Measurements

The time development of the open-circuit potential of uncoated steel (results for uncoated steel, US, in red color) and hot-dip galvanized steel, HDG (in blue color), are shown in Figure 12. The US values gradually increased over the time (up to 250 h here) interval, and the effect of keeping the samples under a water level (after ~170 h) was insignificant (a decrease in values of at most 50 mV/SCE). Thus, it is clear that relatively quickly after the experiment started, the surface of the uncoated steel was covered by a continuous natural passive layer, which was related to the high alkalinity of the cement paste pore solution (very likely pH > 13.0). This fact is described in detail in the literature [6,14,55,56,57,58]. The passive layer of usual thickness of around 15 nm consists of an inner protective layer probably composed of Fe₃O₄ and an outer significantly less protective layer of Fe₂O₃ [59,60,61,62]. Such a passive protective layer guarantees the corrosion of uncoated steel at a negligibly low corrosion rate, unaffecting the service life of concrete structures [14].

The time development of the open-circuit potential of HDG steel in the cement paste showed significant changes in the E_corr values with exposure time, and compared to uncoated steel, they were significantly dependent on the RH of the environment. At the beginning of exposure, all four samples corroded under hydrogen evolution (see Equation (2)) at an open-circuit potential of about −1200 mV/SCE. This was observed in both model concrete pore solutions (containing Ca²⁺) and real normal-strength concrete [6,13,14,22,49,50,63,64]. After about 20 h, there was a sudden increase in E_corr values up to about −500 mV/SCE and most likely a significant reduction in the corrosion rate of the HDG specimens.

Since the open-circuit potential of the HDG samples after this exposure time exceeded the threshold of −990 mV/SCE, the corrosion process proceeded with a cathodic oxygen reduction according to Equation (3).

$${\displaystyle \frac{1}{2}}{\mathrm{O}}_2+{2\mathrm{e}}^{-}+{\mathrm{H}}_2\mathrm{O}\to {2\mathrm{O}\mathrm{H}}^{-}$$

(3)

The limiting value of −990 mV/SCE for the corrosion of HDG steel in the alkaline environment of concrete is expressed from the equilibrium potential E_eq for Equation (2). The actual expression follows from the Nernst equation (see Equation (4) through Equation (6)) [6].

$${\mathrm{E}}_{\mathrm{e}\mathrm{q}}={\mathrm{E}}^0-{\displaystyle \frac{\mathrm{R}\mathrm{T}}{\mathrm{z}\mathrm{F}}}·\mathrm{l}\mathrm{n}\left({\displaystyle \frac{{\mathrm{a}}_{{\mathrm{H}}_2}·{\mathrm{a}}_{{\mathrm{O}\mathrm{H}}^{-}}^2}{{\mathrm{a}}_{{\mathrm{H}}_2\mathrm{O}}^2}}\right);{\mathrm{E}}^0=-0.828\mathrm{V}$$

(4)

$${\mathrm{E}}_{\mathrm{e}\mathrm{q}}=-0.828+0.059\left(14-\mathrm{p}\mathrm{H}\right)=-0.059·\mathrm{p}\mathrm{H}$$

(5)

$${\mathrm{E}}_{\mathrm{a}\mathrm{q}}({2\mathrm{H}}_2\mathrm{O}+2{\mathrm{e}}^{-}\to {\mathrm{H}}_2+2{\mathrm{O}\mathrm{H}}^{-})=-990\mathrm{m}\mathrm{V}/\mathrm{S}\mathrm{C}\mathrm{E}$$

(6)

According to some authors, this is caused by the formation of a passive layer of Ca[Zn(OH)₃]₂·2H₂O (CHZ), which effectively passivates the surface of HDG steel up to about pH 13.3 [13,14,22,49,63,64]. According to other authors, the passivity is not provided by the formation of CHZ but by the lack of water penetration at the surface of the HDG steel. These authors assume that sufficient protective passive drilling is formed not only by CHZ but also by ZnO and Zn(OH)₂ (the phase precipitating between the pores of CHZ) [65,66]. The formation of the protective passive layer, in this case, is related to kinetic parameters (mainly ion transport to the surface of the coated steel). The thermodynamic stability of the protective layer of corrosion products strongly depend on the SO₄²⁻, Cl⁻ ion content, and pH of the concrete pore solution [67,68,69,70].

However, after the samples are submerged under water at about 170 h, the open-circuit potential values decreased. For one of four HDG samples, the E_corr value even dropped below the threshold value of −990 mV/SCE. This again allowed for the progress of corrosion damage of HDG steel under hydrogen evolution at a very significant rate. The decrease in the measured values was related to facilitating water access to the corroding surface. Thus, it was evident that even after 170 h of exposure, a layer of corrosion products did not provide complete corrosion protection in a passive state. Moreover, this was observed even for specimens with wires coated separately, exhibiting an optimal composition with an outer layer formed from the η phase (Figure 10). However, locally, the thickness of the η phase may have been significantly lower than the specified critical value (10 µm) for passivation, and the transition to corrosion in the passive state on Fe-Zn intermetallic phases was very difficult [14,22,30,32,71,72].

The results presented in this section indicate that the transition of the surface of HDG steel in real concrete to a passive state may be somewhat imaginary and unrelated to the formation of a compact CHZ layer. Conversely, the lack of liquid water (consumed by cement hydration) at the surface of HDG steel may be related to the increase in E_corr, accompanied by a decrease in corrosion rate. A passive layer with a significant barrier protection effect may form significantly later.

However, the E_corr(t) values demonstrated that HDG conventional prestressing reinforcement corroded in real concrete under hydrogen evolution for a significant time (at least 20 h), which may be reflected in an increase in the porosity of the cement paste at the interface (ITZ). This increase in the porosity of the cement paste may be related to a decrease in the bond strength of the reinforcement in concrete [6,7,14,15,16].

3.3. Bond Strength

The effect of corrosion of HDG steel under hydrogen evolution on bond strength in concrete is often discussed in the literature. The evolution of gas penetrates the cement paste in the interfacial transition zone. This increases the porosity, which significantly reduces the proportion of the surface area of the reinforcement occupied by the cement paste. This leads to a reduction in the adhesion force f_ad that contributes to bond strength T_c,i according to Equation (7) [6,14]:

$${\mathrm{T}}_{\mathrm{c},\mathrm{i}}={\mathrm{i}}_{{\mathrm{f}}_{\mathrm{a}\mathrm{d}}}+{\mathrm{i}}_{{\mathrm{f}}_{\mathrm{f}}}+{\mathrm{i}}_{{\mathrm{f}}_{\mathsf{\sigma }}}$$

(7)

The friction force (f_f) differed for the US and HDG specimens as the hot-dip galvanizing process increases steel roughness [6,73,74]. This was confirmed in Figure 11. It can be concluded that an increased value of R_a (surface roughness) increased the bond strength of the HDG specimens in concrete. The effect of mechanical resistance of the concrete cover layer (f_σ) on bond strength is generally significant. In our case, it was assumed that the f_σ component was approximately the same for both specimens. Since it may apply that f_σ > f_ad for this type of reinforcement, the negative effect of hydrogen evolution on bond strength may be less critical in such cases [6,14].

The results of the bond strength (pull-out) tests for both types of specimens are displayed in Figure 13. Bond stress–slip curves for uncoated steel (in red) have a typical parabolic shape with increasing bond strength for prestressing strands [75,76,77]. For HDG specimens, the bond stress–slip curve attained a global maximum (~4.5 MPa at 0.04 mm slip). Afterwards, there was a gradual decrease in bond stress values as the slip value increased. However, at very low slip values (up to the global maximum of the curve), the increase in bond strength of the HDG strands was more significant than for uncoated steel. This was mainly attributed to the locally increased surface roughness (R_a) (Figure 11). Additionally, there may have been an increase in the surface roughness of the samples due to the formation of CHZ crystals [10,78,79].

Due to shear loading during the pull-out test (already at low slip values), longitudinal cracks likely formed between the corrosion products of the coating (or the top rough layer) and the cement paste, and consequently, the reinforcement was easier to be pulled out of the normal strength concrete cubic specimen. In general, the bond strength of HDG strands is lower than that of uncoated steel. The porous structure of the cement paste caused by hydrogen evolution at the interfacial transition zone (ITZ) was only manifested at higher slip values (bond strength of ~4.5 MPa) when the bond strength was significantly reduced. However, the influence of other factors on the bond strength of coated prestressing steel strands with concrete could not be excluded.

The failure pattern after pull-out tests for both types of strands was further investigated—see Figure 14. Uncoated steel and hot-dip galvanized steel showed a clear splitting failure, as the cubic specimens for the pull-out test were without transversal reinforcement [80,81]. A visual examination of failure surfaces revealed no significant differences between the two types of specimens regarding crack sizes and crack propagation. The reduction in bond strength in normal-strength concrete (NSC) was due to the changes caused by hydrogen formation or precipitation of the CHZ-based corrosion products. Note that the deceleration of cement paste hardening in the ITZ due to zinc corrosion products may partly reduce bond strength. Zinc corrosion products can significantly inhibit the hydration of silicate/aluminate-based cement, negatively affecting the mechanical properties of concrete [82,83,84,85].

Figure 15 shows the surface of hot-dip galvanized coating on surface of prestressing steels after bond strength test (pull-out test). The increased surface roughness (see Figure 11) of the hot-dip galvanized reinforcement conditions the formation of cracks/fracture patterns mainly on the highest surface (perpendicular to the reinforcement axis) in contact with normal-strength concrete. Locally, cracks/fracture patterns were detected in the η phase (solid solution of iron in zinc)-see Figure 15A. However, locations with cracks through zinc corrosion products were also detected, suggesting that the failure mode may partially pass in this area. Quite exceptionally, carcks/fracture pattern in ζ phase (FeZn₁₃) were detected. This phase is significantly more brittle than the η phase.

3.4. Cement Paste Porosity

Figure 16 shows examples of the ITZ of the cement paste from uncoated steel. The porosity of the phase interface from uncoated steel exhibited low porosity, typical for the cement paste with a water-to-cement ratio w/c = 0.35. The cracks across the ITZ were caused by the absence of reinforcement and aggregate in the cement paste and the tensile forces required to pull out the strands. The cement paste was marginally locally porous due to imperfect compaction.

Figure 17 displays the ITZ of HDG prestressing steel. The porosity of the cement paste was significantly larger. It is also evident that the pores formed were due to the corrosion of the coating under hydrogen evolution. Compared to the literature [9,53,71,86], it is evident that the corrosion products of zinc could not fill the pores in the cement paste formed by hydrogen evolution after 28 days of concrete aging. On the contrary, the results of the present work confirmed the findings of other works, which showed that zinc corrosion products were not able to fill the pores formed by hydrogen after 28 days of concrete (cement paste) aging [87,88], after 1 year of aging [6,14], even after 4 years of aging [89]. Hydrogen-formed pores (see Figure 17) had locally significant dimensions (hole diameter)—up to hundreds of micrometers. The total depth of the pores varied; corrosion products partly filled some. The total pore depth was expected to be less than the hole diameter (less than hundreds of micrometers). The reason for this is the partial possibility of gas leakage along the phase interface (steel/cement paste) and the partial filling of the bottom of the pores with zinc corrosion products.

Figure 18 presents the overall results of the comparative image analysis (surface occupied by pores) of the ITZ cement paste from both types of strands. Corrosion of the HDG strands under hydrogen evolution caused an increase in the porosity of the cement paste up to 20 times. Therefore, this may have been the leading cause of the observed reduction in bond strength of the HDG strands in normal-strength concrete (Section 3.3).

Figure 19 shows the result of X-ray diffraction (diffraction pattern) of corrosion products on the surface of hot-dip galvanized prestressing steel reinforcement after 28 days of curing in cement paste. Due to the circular surface of the reinforcement, the analysis was very difficult, and even a semi-quantitative comparison of the representation of the different phases would be burdened with a significant error. Notwithstanding this fact, it was clear that the detected surface had only a very thin layer of η (1-Zn) phase, as the presence of ζ (2-FeZn₁₃) phase was quite evident. Among the corrosion products, the expected ZnO and CHZ (Ca[Zn(OH)₃]₂·2H₂O) phases [6,22,49,50,63,64,67] were observed. According to the results of the pull-out test, it was clear that the CHZ phase was unlikely to contribute significantly to the bond strength [10,14]. The presence of ZnO in larger amounts indicated the higher alkalinity of the cement paste pore solution [14,50,63,64,65]. Also present were portlandite (Ca(OH)₂) and calcite (CaCO₃) phases, whose origin was from cement paste.

It followed that an additional topcoat of the surface of the HDG strands was necessary to limit the corrosion of the coating under hydrogen evolution in contact with fresh concrete. Previous work suggested that conversion coatings [11,13,14,90,91], organic coatings [92], and bitumen-based coatings [93] could be successfully used. Without topcoats, the bond strength of HDG strands in concrete (cement paste or cement grout) may be significantly reduced.

Corrosion of the hot-dip galvanized coating on the surface of the prestressing reinforcement in OPC led to the development of the hydrogen gas and a small reduction in coating thickness. Despite this, the coating commonly keeps its barrier protection properties. The observed reduction in bond strength due to a significant increase in the porosity of the cement paste in ITZ may be critical in cases where transfer and anchorage lengths provided by design are insufficient, e.g., when outdated design approaches are taken into account. It may then be required to increase the bond strength by extending anchor lengths.

In contrast, bond strength reduction may have an unimportant effect on the load-bearing capacity of the structures for which design provisions provide a large reliability margin and where bond strength is only partly utilized at ultimate limit states. In other cases, mild steel reinforcement may provide additional contribution to the load-bearing capacity when bond strength between tendons and the cement paste or grout is reached.

Corrosion of prestressing reinforcement is initiated when the concentration of chloride anions exceeds a threshold value. The level of local corrosion damage to the surface of the prestressing reinforcement is determined by chloride ion concentration, voids in or absence of grout, total wetting time, and the possible presence of other corrosion stimulants (e.g., SO₂, SO₄²⁻). A wide range of conditions that may occur in situ is difficult to capture by laboratory tests and consequently, the results presented in this study needs to be considered with caution. Regardless of this, hot-dip galvanized coatings exhibited a substantially higher resistance to chloride anions, and it can be concluded that they could significantly extend the time to activation of the underlying steel under in situ conditions, thereby extending service life of prestressing reinforcement. This positive effect was slightly reduced by the hydrogen-induced pores in the cement paste that allowed chloride anions to accumulate along the surface.

A challenge for further research is the investigation into the effect of prestressing—when bond strength is measured on hydrogen development-affected prestressed tendons, it is expected that a lower porosity of the cement paste would be reached at the same amount of developed hydrogen. However, this hypothesis needs to be verified by subsequent studies.

In combination with frequent stress cycles, the aforementioned pores and caverns in the cement paste along with the accumulated chloride anions may also contribute to the formation and propagation of fatigue cracks. This effect is expected to be relatively minor in comparison with the extension of service life provided by the coating; yet, this remains to be confirmed by further research.

Regarding implications for applications with other materials, continuous galvanizing is deemed to be slightly more efficient but excessively expensive for routine applications in civil engineering. Applications based on rapidly developing ultra-high-performance concretes (UHPCs) are foreseen—as the material is much denser compared to NSC, much lower porosities in the paste are expected, implying a lower reduction in bond strength.

Hot-dip galvanized coating can extend service life particularly of the structures susceptible to chloride-induced corrosion. In order to prevent hydrogen-induced corrosion of the coating, it is, however, necessary to apply an additional coating of the prestressing reinforcement—see above.

Subsequent life cycle-cost studies should help to identify situations in which hot-dip galvanization should be applied to prestressing reinforcement. It is expected that these will include structures in chloride-laden environment, in particular those with vertically curved tendons, with a significant probability of early chloride ingress (i.e., for the first five years approximately).

4. Conclusions

Focusing on hot-dip galvanized prestressing steel strands, this study investigated the effect of their corrosion in fresh normal-strength concrete under hydrogen evolution on bond strength. Initially, whole strands of conventional prestressing reinforcement were hot-dip galvanized. The corrosion behavior of individual coated wires was then studied by measuring the time development of the open-circuit potential in the cement paste. The effect of hydrogen evolution on the porosity of the cement paste at the interfacial transition zone (ITZ) was verified by image analysis.

Hot-dip galvanizing of whole strands is an inefficient method of corrosion protection, because this method does not ensure the formation of a uniform coating of similar composition for all wires in the strand. This coating exhibits lower corrosion resistance and stability in the concrete environment.

Corrosion of hot-dip galvanized prestressing steel in cement paste is accompanied by hydrogen evolution for a very long time. The formation of corrosion products based on Ca[Zn(OH)₃]₂·2H₂O may not provide sufficient surface passivation. Corrosion of hot-dip galvanized prestressing steel under hydrogen evolution may occur again after re-submergence of samples under water level.

It was shown that the hydrogen evolution increased the porosity of the cement paste at the interfacial transition zone (ITZ). This fact significantly contributed to the reduction in bond strength in normal-strength concrete, although hot-dip galvanizing prestressing steel reinforcement significantly increased surface roughness (R_a).

Based on these findings, it is concluded that for civil engineering applications, it is necessary to coat wires individually (batch hot-dip galvanizing process) and subsequently form the strand. To prevent corrosion of the coating with hydrogen evolution in cementitious materials, it is necessary to additionally protect HDG steel by other coatings that are sufficiently stable in an alkaline environment. It is recommended that wires be individually coated in this way (hot-dip galvanized coating as well as topcoating).