A Critical Review on Corrosion and Runoff from Zinc and Zinc-Based Alloys in Atmospheric Environments

Because of its anticorrosive properties, its bonding ability to other metals, its use in many production technologies, and its importance as an essential metal for humans, animals, and plants, zinc is number four among the most consumed metals worldwide (after iron, aluminum, and copper).¹  The most important use of zinc is in zinc-galvanizing to protect iron and steel from rusting. As such, with and without applied organic coatings, it finds much use in, e.g., architectural, infrastructural, and automotive applications. In second place comes zinc in zinc-containing alloys for electrical components, vehicles, and household fixtures, and in third place zinc in zinc oxide (ZnO) for protective skin ointment and for rubber manufacturing. But zinc is not only one of the most important construction materials but also an essential metal, necessary for the proper growth and health of humans, animals, and plants.¹ 

In view of its technical importance, atmospheric corrosion of zinc and galvanized steel has been studied extensively over several decades. Numerous field and laboratory studies of zinc in atmospheric environments have resulted in the ISO classification system,²  which is used for practical purposes in order to predict the corrosion rates of zinc in different corrosiveness classes. Zinc has also been extensively studied for a better fundamental understanding and is today probably the most investigated metal from an atmospheric corrosion perspective.

The last two to three decades have seen an increased environmental concern regarding the use and concomitant diffuse emission of metals into the surrounding environment. Since the beginning of the 1970s, international programs have been inaugurated (e.g., WHO’s Environmental Health Criteria [EHC] program and the International Programme on Chemical Safety [IPCS]) in order to increase the scientific-based knowledge on stocks and flows of chemicals, including metals, in the society and how they may induce adverse effects on humans and the environment. The first assessments were primarily focusing on human health aspects, whereas environmental effects were not fully addressed until 1989. The first EHC for zinc was published in 2001 followed by a full environmental risk assessment of zinc in 2006 and a chemical safety report in 2010 within the framework of REACH (Registration, Evaluation, Authorisation, and Restriction of Chemicals).³  A summary on flows of metals, including zinc, in society and in the environment was published in 2004.⁴ 

In this critical review the intent is to present important aspects on the corrosion and dispersion (runoff) of zinc from zinc and selected zinc-based alloys of relevance in outdoor constructions. Corrosion and runoff are processes of different physicochemical nature.⁵  The review seeks to identify the most important parameters which influence the corrosion and runoff processes, respectively. To place the data in a broader perspective, the review also briefly discusses how zinc runoff data can be used in environmental risk assessments and which additional data need to be considered.

The fact that the corrosion and runoff processes are of different nature and are governed by different parts of the corrosion product layer (herein often referred to as patina) is discussed first. Figure 1 (left) shows the corrosion rate and runoff rate of pure zinc during the first year of exposure in an urban site (Stockholm, Sweden).⁵  While the corrosion rate is slowly decaying, the runoff rate is fairly constant throughout the whole first year of exposure. These different time-dependences are caused by different physicochemical processes that govern corrosion and runoff processes, respectively. Figure 1 (right) displays schematically that the corrosion process, which is electrochemically driven, takes part at the interface between the metal and the corrosion product layer. The runoff process, on the other hand, is largely governed by chemical dissolution as the patina gradually develops in contact with the atmosphere, and takes place at the atmosphere/patina interface.⁵ 

Hence the role of the patina is two-fold, with different parts taking place during corrosion and runoff, respectively. To describe this dual role further the so-called GILDES-formulation of the interfacial regime between the zinc metal and the atmosphere is introduced. GILDES is a computer model which was developed to explore the interfacial regime between an oxide-covered metal and the atmosphere.⁶  It contains six parts (Gas, the Interface between gas and liquid, the Liquid, the Deposition layer, the Electrodic region near the solid surface, and the Solid), and has been developed for the exploration of the chemistry of metal or metal oxide surfaces covered by a layer of water and exposed to the atmosphere. The aim of GILDES is to use all chemical information available and then apply the model to predict the phases formed during atmospheric corrosion and their formation rate. It has been successfully applied to zinc in a few laboratory atmospheres representing both outdoor⁷  and indoor⁸  atmospheric environments. To exemplify the GILDES formulation, Figure 2 (left) shows the sequence of compounds detected during exposure of zinc in a typical marine environment.⁹ 

It starts with the instant formation of a very thin layer of zincite (ZnO), which is present already before the marine exposure, continues with a thin layer of hydrozincite (Zn₅(CO₃)₂(OH)₆) that evolves to simonkolleite (Zn₅Cl₂(OH)₈·H₂O), and ends with gordaite (NaZn₄Cl(OH)₆SO₄·6H₂O)¹⁰  as the final compound after extended exposure times. Figure 2 (right) shows schematically the main species and compounds that can be detected in each GILDES-region during marine exposure of zinc.

It is evident from Figure 2 that the patina consists of two regions with different functionalities, the inner electrodic and the outer deposition region. According to GILDES, the electrodic region is a very thin region at the interface between the patina and the solid that incorporates all electrochemistry governing the corrosion process. The deposition region, on the other hand, has been formed through a broad spectrum of chemical rather than electrochemical reactions, including equilibria reactions (gas-solid reactions, hydration chemistry reactions, acid-base reactions, surface complex formation, and solubility equilibria) and irreversible reactions (redox reactions involving oxygen and hydrogen, zinc oxidation, reduction reactions, and irreversible ligand- or proton-promoted dissolution).⁵ 

The important point to be made herein is that the corrosion process largely is influenced by the electrodic region, to some extent also by the deposition region, while the runoff process mainly occurs in the outermost part of the deposition region, where patina dissolution takes place in contact with the liquid region. Evidence for this dual or multi-role of the corrosion product layer can easily be observed on other metals, such as passivating metals, or on copper. On passivating metals, the observation of an inner barrier layer and an outer porous layer with different functionalities is well established, see, e.g., Macdonald, et al.,¹¹  and Chu, et al.¹²  Also, copper exhibits a patina which usually contains two different sublayers:^13-14  an inner cuprite (Cu₂O) layer, which largely governs the protection ability of the patina layer, and an outer layer, which commonly contains brochantite (Cu₄SO₄(OH)₆ in sulfur-dominating environments) or atacamite (Cu₂Cl(OH)₃ in chloride-dominating environments) and which determines the runoff rate of copper during its interaction with the surrounding atmosphere.^15-16 

As opposed to copper, the protective inner layer on zinc is more difficult to detect. However, detailed analysis of the corrosion products on zinc by transmission electron microscopy (TEM) revealed an inner layer of ZnO, having a thickness in the range 20 nm to 70 nm.¹⁷  Detailed analyses of self-repairing properties of the corrosion products formed on zinc show that this inner barrier layer is compact and dense, and that the barrier properties may improve with exposure time.¹⁸ 

Following the GILDES-formulation, it is concluded that the corrosion process is largely governed by the electrodic region at the metal/patina interface, to some extent also by the deposition region, while the runoff process mainly is governed by the deposition region at the patina/atmosphere interface. Differences between corrosion and runoff data of zinc will be further discussed in a later section.

The atmospheric corrosion of zinc is an electrochemical process, and the standard electrode reduction potential of zinc at 25°C is −0.76 V_SHE. The metal dissolves as Zn²⁺ in the aqueous adlayer, which acts as the medium for electrochemical reactions on zinc during atmospheric corrosion. The stability domain of the Zn²⁺ ion in the Zn-H₂O system is relatively large and ranges from around pH 8 to lower pH values. Zn²⁺ is an intermediate Lewis acid, and when following the Lewis hard and soft acid-base (HSAB) principle one expects the ion to coordinate with a broad range of Lewis bases, such as OH⁻, O²⁻, ⁠, ⁠, ⁠, ⁠, and ⁠.⁵  These and many other ions have shown to be reactants during the atmospheric corrosion process, which is evidenced by the relatively broad compositional range of corrosion products observed in both outdoor and indoor atmospheric environments.⁵ 

Zinc is one of the base metals on which the standardized ISO corrosiveness classification is based.²  In this classification there are a few atmospheric parameters that are more influential than others for predicting the corrosion behavior of zinc in different outdoor atmospheric environments. These include sulfur dioxide (SO₂) concentration or deposition, chloride ion (Cl⁻) deposition, temperature, and time of wetness (TOW, the time during which the temperature is >0°C, and the relative humidity [RH] > 80%).²  This standard was originally based on corrosion rate data mainly from the temperate region. Extended international exposure programs were therefore initiated. The first was ISOCORRAG,¹⁹  which included more than 50 test sites in other climatic regions than the temperate region, located in Europe, Argentina, Canada, Japan, New Zealand, and the United States. Later, the MICAT project was established to perform corrosiveness mapping in 12 Ibero-American countries and in Spain and Portugal.²⁰  An important outcome of these exposure programs was the establishment of corrosiveness categories of all environments, from C1 to CX, as explained in ISO 9223 and briefly described in Table 1.

Another important outcome of the international exposure programs was the exploration of the combined effect of temperature and RH. It turned out that in atmospheric environments dominated by SO₂, the corrosion rate of the investigated metals increases with temperature up to a certain maximum value and thereafter decreases, corresponding to the drying out of the aqueous adlayer and concomitant reduction in corrosion rate. In atmospheric environments dominated by Cl⁻, on the other hand, the decrease at higher temperature values is never observed. The hygroscopic salt particles can retain the aqueous adlayer even at higher temperatures, resulting in extremely high corrosion rates (Category CX in Table 1).

In the period between 1960 and 2000, the SO₂ levels decreased substantially in many parts of Europe, and it was soon realized that the relative contribution of SO₂ and other sulfur-containing pollutants were not any longer dominating. Instead, the focus was on other pollutants that could contribute, especially particulate matter and nitric acid (HNO₃). The overall result of all these international efforts is seen in the dose-response functions developed for zinc:⁵  

Here, R denotes the average corrosion effect in μm after 1 y of exposure. The corrosion effect has four major contributions. The first is a constant and corresponds to the background atmospheric condition. The second is the overall contribution of dry deposition effects and is expressed as a combined effect of SO₂, RH, and temperature (T), where f(T) = 0.062(T − 10) when T < 10°C, otherwise f(T) = −0.021(T − 10). These conditions describe the complex temperature behavior of the atmospheric corrosion of zinc, with a maximum in corrosion rate at 10°C. The third contribution is from wet deposition, and includes the amount (in mm/y) and concentration ([H⁺] = 10^−pH) of rain. The fourth contribution is from HNO₃, which turns out to have a very strong corrosion effect on zinc, as has been shown both in laboratory and field based exposures.^21-22  [SO₂] and [HNO₃], finally, represent annual average concentrations in μg/m³.

A consequence of the historical development in many parts of Europe with a steadily decreasing SO₂ concentration is the increasing relative importance of salt aerosols. Predictions have been made in the European region showing that the atmospheric corrosion of zinc and other metals will be dominated by Cl⁻ deposition in coastal and near-coastal areas, an effect which will increase further when the temperature is expected to increase because of climate changes.²³  The situation is, however, different in other parts of the world which still do not experience any SO₂ reduction or only limited reduction of this and other acidifying gaseous pollutants.

Given the importance of Cl⁻ deposition at high humidity conditions in combination with or without SO₂, an exposure program was undertaken to explore the atmospheric corrosion of zinc in 18 tropical sites in Asian countries and Australia.²⁴  Regression analysis of corrosion and environmental data indicated that the most important factors to control zinc corrosion were the climatic parameters temperature and rainfall, and also time of wetness. However, no significant influence of any gaseous pollutants on zinc corrosion rate could be discerned, despite large variations in SO₂ concentration between the sites. The result shows that dose-response functions obtained from one or several programs may not necessarily be valid in other programs, because of differences in environmental conditions between the programs.

Of particular concern is the natural and anthropogenic emission of aerosol particles, originating from, e.g., sea spray (from breaking ocean waves), mineral dust (from windblown surface soil), biomass burning, industrial processes, and vehicle traffic. The interaction of these particles and zinc or zinc-alloy surfaces is quite complex. Upon adsorption most particles sorb water whereby they transform from agglomerated solid particles to droplets. This deliquescence process follows the variations in RH and generally results in highly complex chemical species, which may undergo further changes when emitted gases in the atmosphere interact with the wetted aerosol particles. Local emission variations, the complex aerosol chemistry, and lack of predicted emission data makes it very difficult to give some general prediction of the corrosion effect of aerosol particles on zinc and zinc-alloys. However, when limiting the effect to NaCl particles alone, one can expect a significant increase in corrosion effect as a result of the combined effect of temperature, RH, and Cl⁻ concentration on zinc.²⁵ 

Zinc-based coatings on steel in general corrode two orders of magnitude slower than conventional cold rolled steel in many types of environments. By adding magnesium, aluminum, or a combination of both, the corrosion rate may decrease another order of magnitude.^26-27  Considering the technical importance for, e.g., the automotive industry, the last decade has seen a large number of investigations aiming at exploring the role of these two alloying elements for the corrosion resistance of zinc. In this section, an effort is made to summarize major findings. The authors start by presenting important investigations on pure zinc. It should be emphasized that the zinc-containing layer on steel in, e.g., hot-dip galvanized steel in reality consists of a series of intermetallic Fe-Zn alloy layers, each characterized by a decrease in Fe content when going from the steel substrate toward the outer zinc-rich phase.^28-29  Each layer has a unique electrochemical nobility which makes the corroding system quite complex when deducing mechanistic information from hot-dip galvanized or other zinc-coatings on steel. For simplifying reasons, many fundamental studies have therefore been performed on pure zinc rather than on galvanized steel.

During the decades between 1970 and 2000, the fundamental and practical understanding of the atmospheric corrosion of pure zinc and other metals was largely based on the exposure of metal plates in both laboratory and field exposure conditions. Based on data generated, the present authors were able to deduce main sequences for corrosion product formation on zinc sheet or galvanized sheet in atmospheric environments.⁹  Despite large variations in gaseous and aerosol pollutants, RH, time of wetness, and solar radiation, two main sequences were identified: one in more sulfur-dominated environments and one in more chloride-dominated environments. These sequences are shown schematically in Figure 3 for sheltered conditions. Both sequences start with the instant formation of zinc oxide (ZnO) and zinc hydroxide Zn(OH)₂ and subsequent formation of hydrozincite, Zn(CO₃)₂(OH)₆, during time scales indicated in the figure. This compound, together with Zn(OH)₂, is the whitish corrosion product formed on zinc during wet storage or exposure in low polluted environments.

In environments dominated by SO₂ or other sulfur-containing pollutants, the upper sequence of compounds is formed with zinc hydroxysulfate containing different amounts of water (Zn₄SO₄(OH)₆·nH₂O). The phases are commonly seen in rural, urban, and industrial environments, and can remain as end product in these environments. If chloride aerosols are present in the atmosphere, such as in certain industrial or road environments, the Cl⁻ ions can become incorporated into the lattice and eventually form Zn₄Cl₂(OH)₄SO₄·5H₂O. This is the end product in many highly corrosive atmospheres characterized by high deposition rates of SO₂, sulfates, and chlorides.^5,30 

In marine environments, or other environments dominated by aerosols containing NaCl or other chlorides, the corrosion products formed instead include Zn₅Cl₂(OH)₈·H₂O and later NaZn₄Cl(OH)₆SO₄·6H₂O, a result of the incorporation of S-containing aerosols from the ocean. Characteristic for the formation of these compounds is their gradual formation through a number of principal steps, facilitated by the structural resemblance of several of the compounds involved.^5,10 

The interaction of NaCl aerosols on zinc have been described in quite some detail in a series of papers. Such investigations are of particular relevance in view of the increasing importance of aerosols relative to gaseous pollutants, as explained above. Significant progress has been made in the understanding of corrosion phenomena under aqueous droplets produced by wetting of salts.^31-36  The droplet is usually formed around a single salt crystal which is exposed to increasing RH until the point of deliquescence is reached and the crystal transforms into a droplet. Although the droplets formed under these experimental conditions are much larger than those formed as natural marine aerosols, it is nevertheless believed that fundamental information has been gained which can help to understand the atmospheric corrosion in marine environments.

Figure 4 helps to depict some of the major processes found for pure zinc with a single NaCl droplet. A characteristic feature is the well-separated anodic and cathodic regions. The anode in the central part has been formed as a pit through chloride- or acid-promoted dissolution of the ZnO covered surface. If the conditions for maintaining the pit are fulfilled, driven by high enough zinc dissolution and local acidification in the pit, a pH gradient is formed across the surface from the anode (acid) to the peripheral cathodic region (basic) through the cathodic reaction with dissolved oxygen forming hydroxyl ions. Zn²⁺ and Zn(OH)⁺ ions formed in the pit will migrate out of the pit, react with (from dissolved CO₂ into the droplet) and Cl⁻. This creates conditions to form both Zn(CO₃)₂(OH)₆ and Zn₅Cl₂(OH)₈·H₂O at various parts, depending on pH and Cl⁻ ion concentration.

It should be emphasized that this depicted model is simplified and becomes more complicated when adding SO₂ or sulfate-containing aerosols, for more details see, e.g., Cole and coworkers,^24,37  and Chen, et al.³⁴  Another important phenomenon not seen in the figure is the so-called secondary spreading effect, which is observed during single droplet experiments when the moisture film spreads out from the central droplet after certain stages of exposure and alters the local electrochemical and chemical conditions for the corrosion processes.³⁸ 

Nevertheless the schematic figure can act as a platform when discussing the role of the alloying elements magnesium and aluminum on galvanized steel sheet. Such efforts have been based mostly on laboratory exposures, less often on field exposures. The improvement of hot-dip galvanized steel sheets by aluminum has a history which goes back to the early 1970s with the development of Galvalume^† (Zn + 55 wt% Al + 1.6 wt% Si, from here on denoted Zn55Al)³⁹  and Galfan^† (Zn + 5 wt% Al; UNS Z38510⁽¹⁾, from here on denoted Zn5Al).⁴⁰  Later came the development of even more corrosion resistant coatings through addition of magnesium. It resulted in several commercial Mg-Al coatings by Nippon Steel Corporation around the year 2000, such as ZAM^† (Zn + 6.0 wt% Al + 3.0 wt% Mg) and Super Dyma^† (Zn + 11.0 wt% Al + 3.0 wt% Mg + 0.2 wt% Si).⁴¹  Driven by mainly European industries, numerous investigations have been implemented during the last decade to further explore the beneficial role of the elements aluminum and magnesium. The way to produce the alloy coatings has been through hot-dipping in Zn-Al-Mg baths or by physical vapor deposition of Zn-Mg alloy coatings.⁴¹  Magnesium has usually been added in the range from 0.2 wt% to 11 wt% and aluminum in the range from 0.1 wt% to 3 wt%, and both elements result in a highly complex microstructure, which includes different phases besides the main phase Zn: MgZn₂, binary eutectic MgZn₂-Zn, binary eutectic Zn-Al, and ternary eutectic phases. These additions have resulted in layers on steel which clearly represent alternatives to hot-dipped zinc and electrogalvanized zinc coatings, and with significantly improved corrosion protection ability in the neutral salt spray test or in cyclic automotive corrosion tests.^27,41-44  However, the improvements reported very much depend on the actual laboratory exposure conditions. They are expected to vary even more when considering actual differences in natural field exposure conditions. Critical parameters in this context are Cl⁻-load, pH of electrolyte introduced, RH, and temperature.

Because of the variations of these critical parameters, it has so far been difficult to develop a universal mechanism for the beneficial role of the alloying elements magnesium and aluminum alone or in combination on the corrosion resistance of zinc coatings, despite a substantial amount of studies on this topic. The new mechanistic insight that has been generated during the last decade has been based on a few novel fundamental approaches: the influence of the complex microstructure on atmospheric corrosion initiation, the possibility for in situ measurements of the dissolution of alloying elements, and the detailed compositional analysis of corrosion products formed. Within this context it is hard to give full justice to all investigations made, and this survey therefore will be limited to some key investigations. The present authors begin by explaining the role of magnesium alone, which is largely based on the comprehensive study of Hosking, et al.⁴²  In doing so, refer to Figure 5, which is a modification of the figure shown for bare zinc sheet (Figure 4).

As for pure zinc, an important prerequisite is the creation of anodic (center) and cathodic (periphery) areas. Mg²⁺ and Zn²⁺ ions anodically dissolve from the MgZn₂ phase in the Zn-Mg matrix and enter into the electrolyte. In the initial stage, ZnO covers the substrate outside the anodic area where the cathodic reaction can proceed. When the magnesium ions reach close to the cathodic area they form stable precipitates of Mg(OH)₂ on the cathodic area. This suppresses the cathodic reaction rate and facilitates the formation of Zn₅Cl₂(OH)₈·H₂O, either directly from chloride complexes at the anodic region or via the transformation of zinc oxide. Zn₅Cl₂(OH)₈·H₂O is a relatively compact compound and improves the corrosion protective ability of the system. It will continue to form as long as the oxygen reduction process at the cathode is reduced through formation of Mg(OH)₂ on the cathodic area. Hence, a main beneficial role of alloying zinc coatings with magnesium is the reduction of the cathodic reaction through precipitation of Mg(OH)₂ and the formation of protective Zn₅Cl₂(OH)₈·H₂O.^45-46  Studies have shown that the beneficial effect of magnesium has an optimum Mg-concentration range of 4 wt% to 8 wt%, when exposed to pre-deposited NaCl and humidified air.⁴⁵  By adding aluminum to the Zn-Mg coating a few additional beneficial effects can be seen (Figure 5), primarily the formation of a layered double hydroxide (shortened LDH in the figure) of type Zn₆Al₂(OH)₁₆(CO₃)·4H₂O,⁴³  and also the formation of a thin film of Al₂O₃ with barrier properties (see below for Zn-Al coatings). Several studies provide evidence for the combined effect of Mg and Al, based on both field and laboratory conditions.^27,44,47 

For Zn-Al coatings on steel, there exists today a substantial amount of information on corrosion rates and long-term corrosion product formation, both for Zn55Al^48-51  and Zn5Al.^52-53  Zn55Al possesses a relatively simple microstructure with an aluminum-rich dentritic phase (often denoted the α-phase) and a zinc-rich interdentritic phase (the β-phase). This is seen in Figure 6⁵⁰  displaying an atomic-force microscope (AFM) in image (a) and a corresponding Volta potential image in (b). After slight diamond polishing of the as-received Zn55Al surface, the initial Al₂O₃-film covering all parts of the surface was removed revealing the two-phase structure underneath. The Volta potential on dendritic parts (Al-rich) of the surface turn out to be higher than on interdendritic parts (Zn-rich), from which is expected that interdendritic parts are more susceptible to corrosion initiation than dendritic parts, at least initially.⁵⁰ 

Figure 7 is a compilation of all compounds found in corrosion products formed on zinc,⁹  Zn5Al,⁵³  and Al^54-56  upon unsheltered exposure in marine environments.

On all three materials, oxides and/or hydroxides of zinc or aluminum are instantaneously formed whenever they are exposed to moist-containing atmospheres. In the presence of CO₂, Zn(OH)₂ reacts to form hydrozincite (Zn₅(CO₃)₂(OH)₆) and Zn₆Al₂CO₃(OH)₁₆·4H₂O. Sufficiently high Cl⁻-deposition rates also result in the formation of hydroxychlorides, such as simonkolleite (Zn₅Cl₂(OH)₈·H₂O) on bare zinc and Zn₅Cl₂(OH)₈·H₂O and/or Zn₂AlCl(OH)₆·H₂O on the Zn-Al coating. Because of biological activity in the oceans, marine environments are also characterized by significant sulfate emission rates. This is a main reason for the frequent detection of gordaite (NaZn₄Cl(OH)₆SO₄·6H₂O) on both bare zinc and on Zn5Al in marine environments.

As originally proposed for zinc,⁹  several compounds in corrosion products bear structural resemblance. They consist of layered structures with octahedral or tetrahedral building blocks in which the layers are held together by, e.g., ⁠, ⁠, or Cl⁻ ions.⁵  Such compounds, which are characterized by layered structures, include Zn₅(CO₃)₂(OH)₆, Zn₅Cl₂(OH)₈·H₂O, NaZn₄Cl(OH)₆SO₄·6H₂O, Zn₂AlCl(OH)₆·H₂O, and Zn₆Al₂CO₃(OH)₁₆·4H₂O, whereby the former three form on bare zinc and the latter two on Zn-Al alloys. During the cycling weathering conditions the slow transformation from one compound into another can proceed through replacement of, e.g., or OH⁻ ions with, e.g., or Cl⁻ ions. This sequential process of the gradual buildup of corrosion products can successively result in more protective corrosion products in many atmospheric environments, see Leygraf, et al.,⁵  for more details.

Several of the compounds seen in the corrosion product sequences displayed in Figure 7 have been detected upon exposure of zinc and Zn-Al alloys in different Cl⁻-containing atmospheres, including laboratory exposures (the new revised VDA Corrosion Test Method, Standard SEP 1850 VDA 621-415 B, for automotive applications, and the single drop-test discussed above) as well as marine field exposures.^41,53  The similarity in detected compounds for both laboratory and field tests suggests that laboratory exposures can reproduce the same or similar compounds as in marine environments, although in different proportions.⁵³ 

It should emphasized that the different mechanisms presented above for the role of magnesium and aluminum in Zn-Mg-Al coatings are highly dependent on actual physicochemical conditions, and can only be expected to operate to a varying extent depending on actual exposure. One obvious example of an influential parameter is the Cl⁻-deposition rate which determines to what extent the Cl⁻-containing compounds in Figure 7 are formed at the expense of -containing compounds. Another important parameter is the concentration of CO₂, which may be dramatically reduced in geometrically confined volumes. The reduction or absence of this gaseous constituent has been shown to significantly increase the corrosion rate of hot-dip ZnMgAl coatings compared with ambient CO₂ concentrations. At low levels of CO₂, the main corrosion products detected were the layered double hydroxide Zn₆Al₂CO₃(OH)₁₆·4H₂O and Zn₅Cl₂(OH)₈·H₂O, whereas Zn₅(CO₃)₂(OH)₆ and Zn₅Cl₂(OH)₈·H₂O were detected at ambient CO₂ levels.⁵⁷  In addition to all mechanisms presented above, which largely are based on the detection and distribution of different compounds in the corrosion products, the corrosion protective ability may depend on compounds which have actually not been detected but, nevertheless, may play a decisive role. One example could be the very thin (less than 100 nm) layer of ZnO that has been shown to form a protective inner barrier layer on bare zinc and only could be detected with highly sensitive TEM.¹⁸ 

As discussed in a previous section, corrosion is primarily an electrochemical process in which the metal is oxidized in the anodic reaction, whereas metal runoff with time predominantly is governed by chemical processes taking place at the interface between the patina and the environment (Figure 1). There are several reasons why atmospheric corrosion rates and runoff rates are not equal. The total amount of corroded metal contains two parts: one part which retains in the corrosion product layer and one part which dissolves from the patina by the action of precipitation (metal runoff) or possibly via flaking or wear processes (e.g., Leygraf, et al.,⁵  and Oesch and Heimgartner⁵⁸ ). From this follows that metal corrosion rate data cannot directly be used to assess metal release rate data; such an assumption would highly overestimate the extent of metal release. Available corrosion rate data or dose-rate predictions^5,59  are often based on 1 y measurements or on corrosion rates averaged over a few years. During this time, the corrosion rates are significantly higher compared with long-term conditions, when the patina is established and acts as an efficient barrier that hampers the corrosion process. As a result of substantial reductions in corrosive pollutant levels (e.g., SO₂) seen in several parts of the world (e.g., Europe, North America), available literature data on corrosion rates are furthermore often considerably higher than more contemporary rates.⁶⁰  Differences in short-term corrosion rates and runoff rates of zinc and galvanized steel are therefore expected, and directly measured annual average runoff rates are typically 50% to 90% of the corresponding annual average corrosion rates up to 5 y of exposure^61-74  at sites characterized by different corrosiveness and annual rainfall quantities (in the range 500 mm to 1,000 mm). Similar results have been observed for zinc sheet exposed up to 4 y at seven different sites in Switzerland^58,62  for which the zinc runoff rate was calculated, rather than measured, by subtracting the amount of zinc in adhering corrosion products (assuming Zn₅(CO₃)₂(OH)₆ as the main corrosion product) from the total measured mass loss. These calculated release rates of zinc were on an average 49% and 69% of the corrosion rates after 1 y and 4 y of exposure. Comparative runoff rate measurements at one of the sites (Dübendorf) showed that such calculations highly overestimate the runoff rate, see Figure 8(a), as the patina also contained other corrosion products not considered in the calculations. At more polluted sites and for more complex patina compositions, this effect will be even more pronounced. Nevertheless, this approach can provide conservative and more realistic estimates of runoff rates, rather than using corrosion rates directly. Lower runoff rates compared with corrosion rates (84% lower) during short-term exposures are further illustrated in Figure 8(b) for zinc sheet exposed during 1 y at a marine site (Brest) in France⁷²  and for galvanized steel (99% lower) exposed for 15 month in Valparaiso, Chile.⁶³  As discussed below, the differences depend on factors such as the corrosiveness of the test site, the formation rate and composition of corrosion products, and on climatic/environmental conditions. At sites of low or minor annual rainfall quantities (<250 mm/y), such as the site in Valparaiso, Figure 1(c), the difference can be even larger as the partial dissolution of zinc from the patina during daily repeated cyclic wet/dry periods is not washed from the surface to any larger extent but will rather re-precipitate and gradually build up the patina.

Even though different mechanisms govern atmospheric corrosion and metal runoff, both phenomena are highly intertwined and influenced by the same environmental and exposure conditions. Whereas the corrosion rates often rapidly decrease with time, resulting from the formation of a protective patina, long-term runoff rates seem to be relatively constant or even reduced with time as the barrier properties of the patina improve (Figure 8[a] and discussion below).

The metal release process of single rain events is time dependent and strongly dependent on actual pollutant levels and environmental conditions, both prior to and during a rain episode. The predominating corrosion products that form on zinc at most atmospheric conditions are usually poorly soluble and adhere strongly to the surface where they act as efficient barriers that reduce the corrosion rate. These corrosion products should not be mistaken for soluble, poorly adhering zinc salt particles that rapidly dissolve in the presence of water/humidity. Some zinc in the corrosion products will be dissolved into the thin aqueous surface layer (in which pollutants and deposits dissolve) and be released (washed) from the surface by the action of impinging rainwater. This fraction is denoted the zinc runoff or zinc release and influences only marginally the adherent patina. In the absence of rainwater, dissolved zinc will re-precipitate as corrosion products. The amount of released zinc that can be dissolved and released from the corrosion products depends not only on the patina composition and action of rain water, but also on parameters such as exposure site and environmental/pollutant conditions, environmental conditions prior to a rain event, surface orientation, degree of sheltering, and surface inclination, rain characteristics/intensity, and sampling period (e.g., see literature^63,70,75-80 ). Several investigations have concluded that most metal is released in the first portion (first flush) of the runoff volume (e.g., see literature^{63,70,75,79,81-83} ), where after it is reduced to lower, relatively constant, levels upon extended rain duration. The part of metal runoff in the first flush portion of a rain event is strongly related to the flow rate (or rain intensity) passing over a surface. At a given total rainfall volume, more zinc is released per surface area during a longer low intensity rain episode than during a shorter high intensity rain episode.

As a consequence, large variations in released concentrations of zinc are reported both during and between different rain episodes. For example, zinc concentrations between 0.3 mg/L and 30 mg/L, with a volume weighted mean concentration of 4.9 mg/L, have been reported for 38 different rain events sampled during 1 y in Munich, Germany.⁷⁵  Data from continuous measurements of released zinc from bare zinc sheet during a 15 y urban exposure in Stockholm, Sweden showed a mean concentration (225 sampling periods of several rain events) of 4.1 mg/L Zn (min: 1.3 mg/L, max: 11.9 mg/L) for the whole time period.^84-85  Corresponding data generated during 10 y at marine conditions ranged between 1.6 mg/L and 6.9 mg/L (mean value 3.7 mg/L).^71,85  Similar results (1.1 mg/L Zn to 12.2 mg/L Zn) have been observed,^86-90  whereas other studies reported significantly lower concentrations, ranging from 0.03 mg/L to 0.5 mg/L,⁹¹  or higher concentrations (8.0 mg/L Zn to 38.1 mg/L Zn).^87,92  However, any direct use or comparison between runoff concentrations of different sites or events may lead to erroneous conclusions as the measured concentrations of released metals in, e.g., roof and façade runoff strongly depend on the surface area exposed to precipitation, the duration and intensity of the rainfall event, the actual rain volume impinging the surface of interest, dilution effects, etc.^{63,75,78,93-95}  Implications on the environmental risk assessment are discussed below.

Normalization of observed released concentrations of zinc to the nominal surface area and inclination, the rainfall quantity (in mm—different from the collected runoff volume because of surface inclination and wind direction) and time period must be made to enable comparison of results from, e.g., sites of different rainfall quantities or between different rain events.^2,61  In order to quantify the extent of zinc release from corroded surfaces it is also essential to correct for background levels of deposited metals. Relatively high background deposition rates of zinc from the atmosphere⁹⁶  or from industrial plants have been observed in the literature.^64,76,90,97  The methodology on how to determine and report metal runoff rates at atmospheric conditions is described in ISO Standard 17752:2012.⁹⁸ 

As a consequence of the successive buildup of more insoluble corrosion products, the runoff rate of zinc from zinc sheet and galvanized steel is gradually reduced with time on a long-term perspective. As long as the prevailing environmental and meteorological conditions (e.g., SO₂, pH, rainfall quantity) do not show any significant variations on an annual perspective, the long-term runoff rate reaches a relatively constant (or slightly diminishing) level.^62,79  This effect is illustrated in Figure 9 for bare zinc sheet exposed for 15 y at urban conditions during which a patina, predominantly consisting of a basic zinc hydroxide (Zn₅(CO₃)₂(OH)₆), gradually is built up.^5,61  Similar findings and trends (based on shorter exposure periods) have been reported also for other exposure conditions, e.g., Leuenberger-Minger, et al.,⁶²  and Veleva, et al.⁶⁹ 

Relatively few studies exist that have performed long-term zinc runoff rate measurements from bare zinc sheet at outdoor field conditions. These investigations are usually performed using model surfaces of different size, orientation, and surface inclination.^65,99  Reported annual zinc runoff rates from bare zinc and galvanized steel in the literature are compiled in Figure 10(a) whereby the runoff rates have been normalized to a surface inclination of 45°. The data in Figure 10(a) include 117 different annual rates and originate from 33 investigated sites with different rainfall quantities (220 mm/y to 2,200 mm/y), environmental characteristics, exposure length, and exposure conditions. The runoff rates range from 0.6 g/m²·y to 25.3 g/m²·y (mean value 4.97 g/m²·y; 50% of all rates are between 2 g/m²·y and 6.4 g/m²·y). The highest zinc runoff rates are observed in industrial sites characterized by high pollutant levels. Because observed differences in runoff rates reflect differences in both exposure and environmental conditions, reported runoff rates have been normalized to the corresponding annual rainfall quantity (when available), Figure 10(b). More than 85% of the reported data show runoff rates between 0.002 g/m²·mm and 0.010 g/m²·mm (amount of metal runoff [g] per surface unit [m²] and annual rain amount [mm]) without any differences between marine, rural, or urban sites with low levels of SO₂. The amount of released zinc per given rainfall unit increases as the SO₂ concentration exceeds 10 μg/m³. The results clearly show that the amount precipitation is a major factor that determines the zinc runoff rate and that the effect of chlorides seems less important compared with SO₂. This is consistent with both field and laboratory investigations that show the importance of SO₂ deposition on the runoff process (e.g., see literature^{64,80,97,99-102} ).

Models to predict zinc runoff resulting from atmospheric corrosion were elaborated during the 1990s in which the importance of rainfall quantities (surface wetness), rain pH, and wet and dry deposition of weak and strong acids (including gaseous concentrations and deposition velocities of SO₂, HNO₃, and HCl) and of particles on the solubility of corrosion products were explored.^80,101  Later on, an equation could be deduced:

in which observed zinc runoff rates (R, in g/m²·y) at sites of similar annual rainfall quantities were correlated to SO₂ concentration (μg/m³).⁹⁹  Because of scarce availability of literature data, the correlation was only based on limited data. Nevertheless, the model illustrated the importance of SO₂ on the zinc runoff process. Based on detailed field and laboratory investigations on the importance and influence of surface inclination, rainfall quantity, and rain pH on the zinc runoff rate,^66,70  a more elaborate model was established. However, this model did not include the importance of SO₂ deposition or concentration, previously known to influence the runoff rate. As a result of the lack of sufficient data highlighting the effect of SO₂, runoff rates were instead calculated based on mass balances (zinc runoff = zinc mass loss – zinc in adhering corrosion products) from gravimetric data on mass loss and on the assumption that Zn₅(CO₃)₂(OH)₆ is the predominating corrosion product. As previously discussed and shown in Figure 8,⁶²  such a calculation may overestimate the importance of SO₂ on the runoff rate and provide a conservative measure of its effect. Based on the same approach as previously used for copper^103-104  and on data from bare zinc sheet and galvanized steel after 1, 2, and 4 y of unsheltered exposure within the ICP Materials research program,¹⁰⁵  the following model was established (Equation [3]):

Here R is the predicted annual runoff quantity of zinc per surface unit area (g/m²·y), [SO₂] the annual average value of the SO₂ concentration (μg/m³), and [Rain] the total amount of rain per year (mm/y). Predicted versus observed zinc runoff rates from zinc sheet or galvanized steel are presented in Figure 11. The model predicts 70% of all observed rates (panels of different age, zinc sheet, hot-dipped galvanized sheet, electrogalvanized sheet) within 40% from their measured value. However, it should be noted that quite a few environmental datasets added into the model are uncertain because of lack of information (in particular for rain pH).

Calculated zinc runoff maps of Europe are presented in Figure 11(b) based on annual average values of rainfall quantities and SO₂ taken from the cooperative program for monitoring and evaluation of the long-range transmission of air pollutants in Europe¹⁰⁶  and rain pH data compiled as annual average values for the period 1980–2000. Maps of rain pH, SO₂, and rainfall and details on how they were generated are given in Odnevall Wallinder, et al.¹⁰⁴  The maps clearly show that the amount of zinc runoff has decreased significantly over the time span investigated, a consequence of reduced SO₂ concentrations and rain acidity over most parts of Europe.

Depending on the area of application, the given exposure condition, the intended service life, and the requirement to hinder cut-edge corrosion, zinc-based coatings on steel sheet are commonly used in outdoor constructions and outdoor applications. The Zn-Al alloys Zn5Al and Zn55Al are two frequently used coatings that, at similar exposure conditions, possess improved corrosion resistance (lower corrosion rate) compared with galvanized steel and zinc sheet.^{50-51,71,107-108}  The amount of zinc runoff from these (and other) coatings is strongly related to the presence and surface coverage of zinc-rich corrosion products (similar as formed on zinc sheet and galvanized surfaces) and of aluminum-rich, or aluminum-zinc-rich corrosion products of low solubility and/or with efficient barrier properties. The formation of corrosion products is also strongly related to the microstructure, as discussed above. From an increased formation of poorly soluble corrosion products and the presence of aluminum-rich corrosion products follows a reduced zinc runoff rate from Zn-Al coatings compared with zinc sheet and galvanized steel. As illustrated in Figure 12 for Zn55Al (containing 43.3 wt% Zn) and Zn5Al (containing 95 wt% Zn), the amount of zinc runoff from alloy coatings cannot be calculated based on their bulk composition and knowledge on zinc runoff rates from bare zinc sheet. Long-term field data shows that the annual runoff rate of zinc from Zn55Al is 64% lower (after 1 y) and 95% lower (after 10 y) than the corresponding data for bare zinc sheet exposed in urban conditions, Figure 12(a). In marine conditions the zinc runoff from Zn55Al is 80% (after 1 y) and 90% (after 5 y) lower⁵⁰  than for bare zinc sheet. For Zn5Al, observed annual zinc runoff rates were 40% to 50% lower than for zinc sheet in urban conditions and remained at relatively stable levels up to 10 y, Figure 12(b).⁵¹  Also in marine exposures the zinc runoff rate was 40% to 60% lower than for bare zinc sheet up to 5 y of exposure.⁵³  These results are consistent with literature findings at an industrial/marine site with approximately 40% and 20% lower runoff rates of zinc from Galvalloy^† (a similar alloy as Zn5Al) and Zalutite^† (a similar alloy as Zn55Al), respectively, after 1 y of exposure.⁷⁸  The efficiency of aluminum-rich corrosion products to hinder zinc runoff has also been shown for Zn-Al-Mg alloys.¹⁰⁹  The extent of aluminum runoff from Zn-Al alloys or magnesium from Zn-Al-Mg alloys has not been reported in the literature, with a few exceptions. Long-term studies of Zn55Al exposed up to 10 y at urban and up to 5 y at marine unsheltered conditions show 20 to 30 times lower release rates of aluminum compared with zinc.⁵⁰  Even lower levels of released aluminum compared with zinc have been determined for Zn5Al exposed at similar conditions.^53,85 

Some runoff data also exist for other coatings such as spray-coated zinc and zinc-aluminum alloys, e.g., Bertling, et al.,⁸⁴  and for pre-weathered zinc sheet.^{64,71,94,97,110}  Runoff rates from these surfaces are in the same order of magnitude, or lower, compared with bare zinc sheet. The release of zinc from cut-edges of Zn-Al and Zn-Al-Mg alloys has been shown to be of minor importance and, furthermore, to be reduced with time.¹⁰⁹ 

The presence of different additional surface treatments and organic coatings (of different composition and thickness) will reduce the released amount of zinc further as long as they remain intact, but will gradually (often locally) lose their barrier properties with time upon environmental exposure (e.g., resulting from UV radiation, chloride deposition, gaseous pollutants, etc.).^{61,71-72,78,84,94,109}  Thin (10 nm) chromate-based surface treatments on galvanized steel have as an example shown barrier effects up to 5 y of exposure at urban conditions and thin organic coatings (100 nm) on the same material to maintain their barrier properties also after 10 y.⁶¹  Similar observations, though more rapid coating degradation, have been made at marine conditions up to 5 y.⁷²  More efficient and long-term barrier effects of top-coatings with time are illustrated in Figure 13 for hot-dipped galvanized steel with and without an organic coating (duplex, 60 μm thick) exposed up to 9 y at marine conditions. The amount of zinc runoff from the coated surface was for most years less or significantly less than 5% compared with the bare surface except for the fourth year (less than 9%). Similar results are reported up to 10 y at urban conditions.^61,85  Very low levels of zinc have also been shown to be intermittently released from coatings due to the presence of zinc-rich pigments.^61,78 

Reported runoff rates reflect total amounts of released zinc without considering the chemical speciation of zinc or how it changes upon environmental interaction. This kind of data can be used to assess the total amount of dispersed zinc from an outdoor application but cannot be used directly to assess any potential environmental risks.⁴  Such an assessment requires additional knowledge on the chemical speciation (chemical form) of released zinc, which depends on the chemistry of actual dispersion conditions, and which governs the bioavailability. Further aspects that need to be considered and that influence both the total amount of dispersed zinc and its speciation include dilution effects of zinc runoff with other storm water sources and the retention and interaction with different solid surfaces and with dissolved organic matter upon environmental entry via different aquatic settings. Reported total released zinc concentrations are measured at the immediate release situation (often directly at the rooftop) and vary significantly with time and between different rain episodes, with levels related to differences in rain characteristics, rain amount, and prevailing environmental and pollutant conditions. At the immediate release situation at the rooftop, zinc is predominantly present as the free hydrated Zn²⁺ ion and some weakly bonded (labile) zinc complexes, all in a bioavailable form.^{71,78,84,111-113}  These findings are supported by chemical speciation modeling considering the pH of the zinc runoff water, which typically varies between 5.8 and 7.1 for galvanized steel⁷⁸  and between 4.5 and 7.2 for different bare and coated zinc-based materials and with low levels of organic matter at the rooftop.⁸⁴  However, when dispersed via downspouts, through urban storm water systems, or directly into an aquatic setting, the free Zn²⁺ ion will rapidly react with organic and inorganic species in the water forming different labile or strongly bonded complexes with OH⁻, ⁠, or dissolved organic matter of low bioavailability and reduced toxicity,¹¹⁴  or be retained to different extent on surrounding solid surfaces. Surfaces in the near vicinity of a building, such as concrete in pavement and in storm water systems, and sewage piping and soil have shown to act as efficient sinks for released zinc and to reduce the bioavailable fraction, e.g., see literature.^{61,84,94,115-119}  Laboratory investigations at realistic rain and runoff water conditions using slabs of pavement concrete revealed a retention capacity of zinc which varied between 7% and 25% per contact meter depending on flow rate. Such findings suggest that most zinc in the runoff water is retained within 20 m from the building, Figure 14.⁶¹ 

These observations are supported by literature findings on porous pavement cement, e.g., Dierkes, et al.,¹²⁰  and Legret and Colandini.¹²¹  Efficient (>90%) long-term retention of zinc in storm water runoff has recently been reported for porous Portland cement concrete.¹²²  Long-term zinc retention investigations on soil of different characteristics (pH, cation exchange capacity, composition) clearly show a high capacity of all soils to retain all zinc in runoff water (95% to 100%) and to strongly bind released zinc to organic matter. Zinc in the pore water was present in a non-bioavailable form (non-toxic toward algae) and strongly bonded to complexes with organic matter, Figure 14.⁸⁴  Similar findings have been observed for zinc retained in constructed wetlands.¹²³ 

These results and literature findings suggest that zinc to a large extent is retained by different solid surfaces already in the near vicinity of a building and that the concentration (and bioavailability) of dispersed zinc becomes substantially reduced upon environmental entry.

• In this review some critical issues have been addressed that concern the corrosion and dispersion (runoff) from zinc and zinc alloyed with aluminum and magnesium when exposed to atmospheric environments.

• Corrosion and runoff are intertwined processes influenced by the same environmental and exposure conditions. Yet, the underlying physicochemical mechanisms that influence the processes are different. Corrosion is largely controlled by electrochemical reactions at the metal/patina interface, whereas runoff predominantly is governed by chemical reactions at the patina/atmosphere interface.

• For both processes, the evolution of the patina layer is crucial. The corrosion rate is often observed to slowly decrease with time, resulting from the formation of a successively more protective patina. The runoff rate, on the other hand, is more constant on a long-term perspective (not during individual rain events) but may also exhibit a slow long-term reduction, as the barrier properties of the patina eventually may improve.

• Despite large environmental variations between exposure sites, there exist two main sequences for the gradual evolution of compounds in zinc patina: one in more sulfur-dominated and one in more chloride-dominated environments. Because of climatic changes and the reduction of sulfur-containing atmospheric species in many parts of the world, the chloride sequence is expected to gradually dominate over the sulfur sequence.

• The atmospheric corrosion of bare zinc sheet in chloride-containing atmosphere is relatively well understood. Characteristic features are separated anodic and cathodic areas, the instant formation of a thin layer of zinc oxide, and the subsequent formation of hydrozincite (Zn₅(CO₃)₂(OH)₆), simonkolleite (Zn₅Cl₂(OH)₈·H₂O), and gordaite (NaZn₄Cl(OH)₆SO₄·6H₂O) at various parts of the zinc surface with different corrosion protective functions. Even though the deposition of chlorides increases the rate of corrosion of zinc, its effect on the extent of zinc runoff is minor.

• Successful efforts to improve the corrosion properties of zinc by alloying with magnesium and aluminum have resulted in zinc layers with more complex microstructure and also a broader range of mechanisms responsible for the improved protective ability. The role of the alloying elements has been attributed to several mechanisms, such as the precipitation of magnesium hydroxide at cathodic areas to reduce the oxygen reduction rate, the increased formation of protective simonkolleite and of aluminum oxide, and a layered double hydroxide, the latter two with significant barrier properties. No universal mechanism exists, as the proposed mechanisms most likely are dependent on actual exposure conditions.

• By compiling a large body of reported zinc runoff data, it is concluded that the amount of released zinc is highly dependent on the amount of rainfall. Per rainfall unit, the zinc release increases when the SO₂ concentration exceeds 10 μg/m³, while the effect of chloride deposition is less important compared with SO₂.

• Based on all runoff data a model is presented for bare zinc sheet that predicts 70% of all observed runoff rates within 40% from their measured value. Corresponding predictions of Zn alloys cannot be based on the bulk alloy content, but need to consider actual properties of the corrosion product layer.

• In order to perform an assessment of potential environmental risks of zinc-containing outdoor surfaces it is necessary to consider not only the total amount of dispersed zinc in a given exposure situation, but also the chemical speciation (chemical form) of zinc in different environmental settings of varying chemistry, the dilution effects, and the capacity of different solid surfaces to retain dispersed zinc.

The authors are most grateful to Docent Johan Tidblad, Swerea Kimab, Sweden for invaluable contributions in the elaboration of the predictive zinc runoff model and for producing zinc runoff maps. We also wish to acknowledge Dr. Clara Anghel for her initial efforts to establish a zinc runoff model and M.Sc. Tingru Chang for technical assistance with the manuscript preparation.