Agate: a mineral that develops with age, water and moganite

Agate is banded or variegated chalcedony and this distinctive appearance allows a ready identification from any source. Many agate thick sections from basic igneous hosts are reminiscent of a series of distorted onion-like rings with the initial bands closely replicating the shape of the supporting gas cavity. However, the banding is frequently distorted and this general pattern is known under various names, for example, fortification or wall lining. A second type is less common and demonstrates apparently gravity-controlled horizontal bands. Agate host rocks are varied but the most abundant agate sources are the gas cavities of basic igneous rocks. This article limits discussion to agates from these basic hosts. However, agates can also be found in some igneous acidic hosts (for example, rhyolite), sedimentary rocks (for example, limestone) and in some fossils.

Agate is greater than 97% silica (SiO2) and shows little variation between different samples. Under normal earth surface conditions, silica occurs in a number of forms. It is most commonly found as alpha-quartz and this is the major component in agate. A second silica constituent is moganite with a concentration at 2 to 25%. Moganite is found in agate that has not been heated by metamorphism or in the laboratory. Together with alpha-quartz, they are usually the only forms of silica identified in agate. However, other forms of silica such as cristobalite and tridymite have been occasionally identified in agate. In agates from basic igneous hosts, calcite is a rarity, as demonstrated by an examination of 180 worldwide agates using powder X ray diffraction. Trace calcite was identified in just 7 agates and 4 came from one source – Las Choyas, Mexico (Moxon unpublished data). The major non-volatile impurities include iron, aluminium, magnesium and sodium but concentrations are very small and measured in parts per million; collectively the total is much less than 1% (Götze, 2001). Surprisingly, water (up to 2%) is the major agate impurity. However, only a fraction of the water is due to actual water molecules (H2O). The bulk of detected water is obtained through water loss from two neighbouring silanol groups (≡Si—OH).

Agate/chalcedony has been found in host rocks as young as 13 millions years (Ma) and as old as 3,720 Ma (Moxon et al., 2006). However, agate has never been made in the laboratory. It develops during ageing and these changes allow valid speculation about agate in its early years. This article considers the essential role that water and moganite play in these age-related changes. Mineral age-related development is rare but it is not unique to agate. One other example showing ageing changes is bone apatite. The change offers a means of approximately dating ancient bones (Bartsiokas and Middleton, 1992).

Relationship between agate and host rock age

Agates could form around the same geological time as the host or hundreds of millions years later. Hence, the relative timing of host and agate formation is a prerequisite before any attempt can be made to link agate and host rock age. Many agate host rocks have been dated using radioactive isotopes. Unfortunately, the agates themselves have not been dated due to either a limited radioactive isotope content and/or costs. Agate/chalcedony from the Yucca Mt., Nevada, USA is an exception and has been radiometrically dated together with the host. Yucca Mt., aged 13 Ma, was under consideration as a potential nuclear waste repository and investigated for more than 30 years. The intended nuclear storage has resulted in many studies examining water flow including its effect on the Yucca Mt. host rock minerals. One of these studies dated the chalcedony that had formed in Yucca Mt. tuffs. Initial chalcedony coatings appeared around 4 Ma after the host formation (Neymark et al, 2002). However, radioactive isotopes are generally limited in agate and published agate/chalcedony dating, as far as I am aware, is limited to this one example.

Alternative methods are required and agate properties do vary from source to source offering the opportunity to test potential links with the host rock age. Once a property has been identified, it needs to be investigated with respect to the known host rock age. A plot of any quantitative data against host rock age will produce either random values and is therefore of no use for dating purposes, or show trends with respect to the age of the host rock. If trends can be identified, then that particular property potentially provides a method of roughly dating the approximate agate age. Over the years, I have looked at a variety of property changes and found some that produce a host age link. Quartz crystallite size, density, moganite and total water content do show either partial or total connections with the host rock age. A number of agates need to be examined from a particular area but at best, the sample mean values show a variation range of 5 to 15% and none match the precision of radiometric dating.

Quartz crystallite size. Most minerals are crystalline: they have a definite form due to the long-range order of component atoms or ions. Even complete crystals in macrocrystalline quartz are composed of many smaller crystal units known as crystallites. Crystallites are measured in nanometres (1 nm = 10-9 m) and the crystallite size can be determined using powder X-ray diffraction (XRD). X-rays are part of the electromagnetic spectrum that has a decreasing wavelength when passing from infrared à visible lightà ultraviolet à X-rays. For XRD, a diffractometer generates and directs X-rays at finely ground crystalline powders and the resulting diffraction and interference effects are recorded electronically. The X-ray angle of incidence is measured in degrees theta but the collected data adds the angle of reflection and is recorded as 2 theta. Each pattern of peak positions is as diagnostic of a particular mineral as are human fingerprints. An amorphous (non-crystalline) substance such as glass or plastic is without peaks and shows as a single broad hump. The collection of peaks (Fig.1) are unique to α-quartz and the agate samples are from (a) Mt. Warning, Australia (23 Ma, host age); (b) Chihuahua, Mexico (38 Ma); (c) Lake Superior, USA (1,100 Ma). All show the same peaks that are given by Brazilian macrocrystalline quartz (d). The largest signal intensity is around 26o 2 theta. This is approximately five times greater than the second largest signal at around 20o 2 theta and 10 to 20 times greater than the remainder. A plot using full intensity would result in these two main peaks minimising the rest. Square root of signal intensity reduces this effect. Points of interest are:

1) Agate shows the same signal positions as the Brazilian macrocrystalline quartz.

2) The signals become narrower from (a) to (b) to (c) to (d). The narrower signals demonstrate an increasing crystallite size.

3) These XRD scans have used a fairly rapid time scan. A slower scan would reveal moganite in all the agate samples. However, the moganite signal (m) still shows its presence in the younger agates.

Moganite is the most recently discovered form of silica having been identified in 1984 and finally accepted as a new mineral in 1999. Detection of moganite in agate is usually done using powder XRD. However, moganite quantification can be difficult at low concentrations as it is widely distributed and makes a limited contribution to the moganite XRD signals. Raman spectroscopy is more sensitive and readily identifies trace moganite that has been detected in agate from hosts as old as 1,100 Ma. Visual evidence of this identified development will be considered next.

Scanning electron microscope (SEM)

Optical microscopes are limited to a magnification of about 1,000 x and a thin section examination of agates from hosts aged between 30 Ma and 1,100 Ma does not produce any observable age-related changes. Electron microscopes use accelerated electrons in place of light and the SEM has an increased magnification up to around 500,000 times. The SEM can be used to examine a fractured agate surface that has been coated in a conducting material such as gold or carbon. Early SEM work on fractured agate identified globules in the non-white areas (Lange et al., 1984). Further work distinguished differences between the white bands and the non-white area. The white bands in the older agates were described as showing a stacked plate-edge like structure while the globular structure showed a general age-related increase in size (Moxon, 2002). There is a common trend with an increasing age-related globular growth but differences are not readily quantifiable because there are variations in globular size within the same agate.


Deposits Fig 1ed_1
Fig. 1. X-ray diffraction signals produced by agates from: a) Mt. Warning, Australia (23 Ma host rock age); (b) Chihuahua, Mexico (38 Ma); (c) Lake Superior, USA (1,100 Ma). All show the same signals that are given by the Brazilian macrocrystalline quartz d). A weak moganite signal (m) is shown in two of the agate samples (a) and (b). The square root of reflection intensity has been used in all cases; silicon (Si) is added as an internal standard.

Relationships between different features are lost at high SEM magnifications and one examination routine is shown by the link between increasing size and detail in the Lake Superior agate (Fig. 2, I à IV). Here, the globular nature of the clear area (a) contrasts with a fine white band (b). This band is very unusual, the normal plate-like structure is not observed and it appears to be a collective infill of broken fragments. Further structural detail is demonstrated by taking enlargements around the centre producing the Fig. 2 micrographs. Much of the observed surface debris is caused by the preparation. However, the higher magnification shown in Fig. 2 (IV) does allow an observation of genuine globular growths that have also developed on the white band at (e). The micrograph in Fig. 3 shows a white band that is well formed and typical of that found in agates older than about 60 Ma. The bottom edge of the micrograph (c) shows the repeated vertical stacking of the plate-like edges. Occasional twists of the “plates” have produced flat surfaces (d).

The white band differences between young and old agates can often be demonstrated without the use of the SEM. In my experience, agates from younger hosts (for example, Mt Warning Australia (23 Ma); Isle of Rum, Scotland (60 Ma); Woolshed Creek, New Zealand (89 Ma); Rio Grande do Sul, Brazil (135 Ma) show white bands that are less intense and often more diffuse than those from older agates such as those from Botswana (180 Ma) or Lake Superior (1,100 Ma). If a lapidary diamond saw is available, then an age-related judgement can be made. The agate slabs in Fig. 4 I and II, both 2mm thick, are respectively from Botswana (180 Ma host age) and New Zealand (89 Ma). When viewed against strong white light, the white bands in the older agates reveal a distinct orange/brown colour (white band “a” in Fig. 4 I and III). Younger agates either transmit the full visible light spectrum or produce a weaker brown colour (Fig. 4, IV). The difference is caused by the loss of the blue end of the visible light spectrum. Blue has a smaller wavelength and the transmission of white light through the well-developed plate edge-like structures results in the smaller wavelengths being bounced and scattered. Hence, the final transmitted light in the older agates is mainly from the larger wavelengths: orange/red giving this brownish colour. The poorly formed white bands in younger agates transmit more, or all, of the smaller visible wavelengths.

Deposits Fig.2ed_1
Fig. 2. Scanning electron micrographs showing a freshly fractured Lake Superior agate. The micrographs show an increasing magnification from (I) to (IV). Each higher magnification has focussed on the approximate centre of the previous image. In Fig. 2 (I), (a) is a globular region and (b) shows a most unusual weak white band that was not apparent in the hand specimen.

High temperature dehydration of agate

Agate enthusiasts will be aware of the Brazilian enhydro(s) agates where bulk water has become trapped at the centre of the agate. Less well known is the fact that practically all agate contains water molecules that are free to enter and leave through structural pathways (Moxon, 2017). The ease of movement depends on the age of the agate and the surrounding water vapour pressure. In addition to free H2O, there is the less well-known silanol water (≡Si—OH). These silanol groups are found on the surface and within the agate structure. Strictly, silanol water is not water but the hydroxyl group. Agate is rich in these groups and, over the geological time scale, two neighbouring Si—OH groups combine to release water and form the Si—O—Si bond; this silanol water is often referred to as structural defect water. The total water content includes neighbouring silanol group water loss as well as free water and can be found by heating agate powders of less than 50 micron at temperatures greater than or equal to 1,000oC.

Evidence for the development of agate

My prime reason for examining property variations is to seek possible links between a particular property and the host rock age. Data plots of quartz crystallite size, density, moganite and total water content all show some valid regional links with respect to host rock age. Quartz crystallite size demonstrates the best accuracy and when plotted against host rock age, it demonstrates interesting development (Fig. 5a). The plot exhibits a four-stage agate development pattern for the first 450 Ma. Initially, there is 60 Ma of linear growth followed by growth cessation for the next approximate 200 Ma. Growth restarts for around 30 Ma followed by little change for the next 150 Ma. There is no further change with agates from hosts aged 1,100 Ma. Agates from Brazil (135 Ma host age) #10 and New Zealand (89 Ma) # 9 are clearly off trend: suggesting agate formation ages of 25 and 30 Ma respectively (data from Moxon and Carpenter, 2009).


Deposits Fig.3ed_1
Fig. 3. A well developed white band in an agate from Ethiebeaton quarry, Scotland (412 Ma). The sections at (c) show the typical plate-edge like structure with flat surfaces (d) where the plates have twisted. The micrograph is a montage assembly made from four individual micrographs.

The total water loss in agate with respect to host age has recently been investigated (Moxon, 2017). The study was able to demonstrate that the mean values of the total water show a linear decrease over the first 60 Ma (Fig. 5b). Other than the problematic New Zealand (89 Ma) agates, samples with a host age between 60 and 120 Ma were not available. Suitable agates of this age would have established the post 60 Ma trend. Extrapolation implies that agates between 60 and 120 Ma would lie somewhere between A and B (Fig. 5b). Lake Superior agates (approximately 1,100 Ma) are not shown but the total water is similar to the 180 Ma Botswana agates suggesting the total water content is independent of age for agates greater than 180 Ma. The outlier agates from Brazil # 10 and New Zealand # 9 indicate a formation age of 27 and 41 Ma respectively.

There is an increase in density over the first 55 Ma with stability over next 350 Ma (Fig. 5c). An upper point of density reference is shown by Madagascan macrocrystalline quartz at 2.647 gcm-3; although a more realistic density maximum for agate is provided by the smaller grain size in quartzite pebbles (d = 2.635 gcm-3; Fig. 5c). The density increases with age for the first 60 Ma and the outlier agates from Brazil # 10 and New Zealand # 9 suggest a formation age of 39 and 20 Ma respectively (Fig. 5c). Data from Moxon et al. (2006)

The relationship between moganite and host rock age is the final property (Fig. 5d). Once again, there is a cessation after 60 Ma with little change in the moganite content over the next 1,000 Ma. Unlike the total water plot, there is less room for doubt regarding moganite content in agates from hosts aged between 60 and 180 Ma. Extrapolation would suggest that agate from hosts greater than 60 Ma would have a moganite content closer to the approximately 4% found in agates from hosts older than 180 Ma. The outlier agates from Brazil #10 and New Zealand # 9 suggest formation ages of 25 and 40 Ma respectively (data from Moxon and Carpenter, 2009).

Deposits Fig 4 ed_1
Fig. 4. (I) and (II) show 2mm thick slabs of Botswana agate (180 Ma host age) and New Zealand agate (89 Ma host age) respectively. The slabs are photographed in front of a reading lamp and the central white bands in the Botswana agate are dark brown (a) in the transmitted light (III). The diffuse white banding in the New Zealand agate shows as a pale brown colour (IV). Scale bars = 1cm.

Summary of the evidence

1) All four plots in Fig. 5 indicate a cessation of change occurring after about 60 Ma. Given the moganite data plot, it is likely that total water in agates from hosts aged between 60 and 180 Ma would show a similar percentage water content to that found in hosts older than 180 Ma.

2) All four properties have demonstrated similar host age links. These properties show that agates generally form around the same geological age as the host. There are exceptions and the data shows that Brazilian and New Zealand agates have formed long after the host formation.

3) Agates from Brazil and New Zealand are outliers in all plots. The mean predicted formation age range of Brazilian and New Zealand agate using crystallite size, total water, and moganite content is within 25 to 27 and 31 to 43 Ma respectively. The density determinations have been excluded from these mean values; the predicted ages of 39 and 21 Ma for the Brazilian and New Zealand agates are respectively so much higher and lower than extrapolated data from the other three plots. The poor density values are due to the small density spread of only around 0.05 gcm-3 between the highest and lowest values: any minor errors have a large effect on the outcome.

4) The density and crystallite size show age-related increases for the first 60 Ma; the water and moganite content show age-related decreases. An important question, is there any relationship between these increasing and decreasing changes?

Age-related development of agate

A case can be made for the decreasing water and moganite content as the cause of the increasing crystallite size and density. Moganite is more soluble than quartz so water movement would dissolve the moganite and recrystallisation from a low concentration silica solution within the agate would produce new quartz growth, giving an increase in crystallite size. However, this is conjecture and evidence is required.

High temperature and pressure investigations have been carried out on agate rods that were sealed in gold and heated to temperatures up to 550oC at 1000 atm. pressure (Moxon and Carpenter, 2009). Water was not added and any detected changes in moganite content and crystallite size relied on the water released from the agate during the heating process. The decreasing moganite and increasing crystallite size produced plots that were approximate mirror images. Furthermore, minimal quartz development was obtained when similar sized agate rods were heated in an open furnace for 122 days at 550oC. Hence, agate development occurs when agate and released agate water is heated in a sealed environment. Increasing the temperature produces a reduction of reaction time, but the high temperature reaction mechanism was different from the mechanism occurring under normal earth surface temperature conditions.

Agate has a complex water arrangement, some water molecules are trapped while others are free to enter and leave the agate. The free water molecule concentration varies but in the younger agates it is generally around 30% of the total water (Moxon, 2017). It is the structural silanol groups (Si-OH) that are the important sources of water for the age-related development. Using a variety of silica sources, the number of silanol groups has been calculated at around 5/nm2 (Zhuravlev, 1987). The close occurrence of these silanol groups allows two neighbours to combine and eliminate a water molecule leaving the newly formed Si—O—Si under strain. Moganite is more soluble than quartz and over the geological time scale the water is sufficiently mobile to dissolve moganite leading to a re-crystallization as quartz. Some of the quartz is formed in water pathways and the remainder produces larger quartz crystallites. The better packing in the larger crystallites causes the age-related increase in density.

Further support for a change in the age-related silanol groups has been demonstrated using cathodoluminescence. Cathodoluminescence is used in conjunction with the scanning electron microscope when electrons are directed onto a polished agate surface. Light is emitted and the intensity of the various emitted wavelengths can be examined. Different structural defects produce different wavelength signals. Silanol groups and the strained Si—O—Si bond in quartz cause respective red and orange emissions (Stevens-Kalceff and Phillips, 1995). An investigation was carried out on 40 agate samples aged between 13 and 3720 Ma (Moxon and Reed 2006). The data demonstrated an age-related decreasing red emission that was matched by an increasing orange emission. Hence, an age link between silanol groups and the strained Si—O—Si bonds. Over the geological time scale, this released water is able to dissolve the more soluble moganite for a later re-crystallization as quartz. This change is shown in the structural chemical equation.

Deposits Fig 5 ed_edited-1
Fig. 5. The relationship between four determined properties and host rock age are shown by: (a) crystallite size; (b) total water content; (c) density; (d) moganite content. The dashed lines are for eye guidance. (1) Yucca Mt, USA; (2) Mt Warning, Australia; (3) Chihuahua, Mexico; (4) Cottonwood Springs, USA; (5) Washington, USA; (6) Las Choyas, Mexico; 7) Khur, Iran; (8) BTVP, Scotland; (9) Mt Somers, New Zealand; (10) Rio Grande do Sul, Brazil; (11) Semolale, Botswana; (12) Nova Scotia, Canada; (13) Agate Creek, Australia; (14) Thuringia, Germany; (15) Derbyshire, England; (16) East Midland Valley, Scotland; (17) West Midland Valley, Scotland. Error bars show one standard deviation.

Further work

Comments in this article apply to agates from basic igneous hosts and other agatised material might start from different sources. Nevertheless, it is possible that all agate development could follow a similar development pathway. If correct, then approximate dating of fossilised material younger than 60 Ma might be achieved. However, the XRD data and thin sections in Fig. 6 show that collecting evidence will not be straightforward. The first sample shows fossil wood from the Morrison Formation, Utah, USA (Fig. 6, Ia) produces an X-ray diffractogram that is comparable with agate from basic igneous hosts. The second sample (Ib) from the same location shows a sample that has a calcite content much greater than is found in any of the 180 examined agates from basic igneous hosts. High temperature studies are limited by calcite as it decomposes at around 840oC and the mass loss of carbon dioxide adds to any water loss. The high calcite content shows different formation conditions have applied even in these two neighbouring samples.

Agate thin sections are shown in Fig. 6, II, III and IV. Around 1cm of initial growth produces an unusual band of weak fibrous growth in this New Zealand agate (Fig.6, IIa). The central region shows the expected well-developed agate fibrosity shown by agates from basic igneous hosts. All the fibres grow towards the agate centre. An example of Morrison Formation fossil wood is shown in Fig. 6, III; although the fibrous quartz growth is towards the centre, the actual fibres are clearly fragmentary.

The final thin section (Fig.6 IV) is agate from the Chert Beds at Lyme Regis, UK. The host (a) leads to the first microcrystalline quartz deposit (b) before finalising with typical agate deposits (c), (d) and (e). I have never seen an initial formation of microcrystalline granular quartz in any agate from a basic igneous host. The limited comments on these three samples suggest that agates in sedimentary hosts and fossil wood have more complex histories than their basic igneous-hosted counterparts.

Deposits Fig 6 ed_1
Fig. 6. (I) X-ray diffractograms using two samples of fossil wood from the Morrison Formation, Utah, USA. Sample I (a) produces a similar signal collection to those shown by agates from basic igneous host rocks (Fig. 1). However, sample I (b) demonstrates a high proportion of calcite. (C-calcite, Q-quartz, Si- silicon that is added as an internal standard). (II) A thin section of New Zealand agate showing weak fibrosity (a); the central area demonstrates the more typical agate fibrosity (b). The black markings are added reference points for cathodoluminescence experiments. (III) A thin section of a third sample from the Morrison Formation, Utah. Here, the fibrosity is distinctly fragmented. (IV) Sample from the Lyme Regis Chert Beds with host at (a). Granular microcrystalline quartz (b) is the first quartz deposit Typical agate fibrosity is shown at (c, d and e). The thin sections were photographed with the polars crossed. All scale bars = 2cm.


Agate development poses interesting questions regarding the conditions of agate at birth and the changes that occur over the geological time scale. Around 1,100 Ma ago, the Lake Superior agates started their development journey. I speculate that an examination of these agates after a few million years would have only shown weak white bands set against the red iron oxyhydroxide background. It is the ageing process that is responsible for the present appearance of these ancient agates. I have yet to obtain agates from hosts aged less than 8 Ma but an investigation of younger agates might allow an extrapolation of data back to zero time. This would produce a further insight into the initial state of agate before its later development.


The described studies could not have been carried out without the generous donation of samples from David Anderson, Glenn Archer (Outback Mining, Australia), Jeannette Carrillo (Gem Center U.S.A., Inc.), Roger Clark, Nick Crawford, Brad Cross, Dick Dayvault, Robin Field, Gerhard Holzhey, Brian Isfield, Herbert Knuettel (Agate Botswana), Reg Lacon, Brian Leith, Ian Lennon, Maziar Nazari, Dave Nelson, Leonid Neymark, John Raeburn, John Richmond and Vanessa Tappenden, Bill Wilson and Johann Zenz. I am indebted to the Dept. of Earth Sciences, Cambridge University for laboratory and equipment access. Fig. 5b (Moxon, 2017) is reproduced with the permission of the Mineralogical Society of Great Britain and Ireland.

About the author

Terry Moxon graduated in chemistry from Manchester University and carried out research into agate genesis in the Geology Dept. Sheffield University. For the past 17 years he has been a part-time research worker into the properties of agate in the Earth Science Department at Cambridge University.


Bartsiokas, A. and Middleton, A.P. (1992) Characterization and Dating of Recent and fossil Bone by X-Ray Diffraction. Journal of Archaeological Sciences, 19, 63-72.

Götze, J., Tichomirowa, M., Fuchs, H., Pilot, J. and Sharp, Z.D. (2001) Geochemistry of agates: a trace element and stable isotope study. Chemical Geology, 175, 523-541.

Lange, P.,Blankenburg, H-J. and Schron W. (1984) Rasterelectronmikroskopische Untersuchgen an Vulkanachaten. Zeitschrift für Geologische Wissenschaften, 12, 669-683.

Moxon, T. (2002) Agate: a study of ageing. European Journal of Mineralogy, 14, 1109-1118.

Moxon, T. (2017) A re-examination of water in agate and its bearing on the agate genesis enigma. Mineralogical Magazine. Mineralogical Magazine, 81, 1223-1244

Moxon, T and Reed, S.J.B. (2006) Agate and chalcedony from igneous and sedimentary hosts aged 13 to 3480 Ma: a cathodoluminescence study. Mineralogical Magazine, 70, 485-498.

Moxon, T. and Carpenter, M.A. (2009) Crystallite growth kinetics in nanocrystalline quartz (agate and chalcedony). Mineralogical Magazine, 73, 551-568.

Moxon, T., Nelson, D.R. and Zhang, M. (2006) Agate recrystallisation: evidence from samples found in Archaean and Proterozoic host rocks, Western Australia. Australian Journal of Earth. Sciences, 53, 235-248.

Neymark, L.A., Amelin, Y., Paces, J.B. and Peterman, Z.E. (2002) U-Pb ages of secondary silica at Yucca Mountain, Nevada: implications for the paleohydrology of the unsaturated zone. Applied Geochemistry, 17, 709-734.

Stevens-Kalceff, M.A. and Phillips, M.R. (1995) Cathodoluminescence microcharacterization of the defect structure of quartz. Physical Review B, 52, 3122-3134.

Zhuravlev, L.T., (1987) Langmuir, 3, 316-318.

Terry Moxon (UK)

Buy Fossils, Crystals, Tools
Subscribe to Deposits
Join Fossil Hunts
UK Fossil Locations