Carbon stored in the subsurface as wood could be released to the atmosphere by either upward transport of the wood, or in-situ reactions that degrade the wood to gas, which is then transported upward to the atmosphere.  This would affect the permanence of the carbon storage.

 Particles injected at depths of more than 5m are expected to be physically immobile over time scales of millennia in many locations.  Layers of sand can be fluidized to create injectites13 but it is unlikely that angular interlocking particles of wood could be mobilized this way.   The maximum average soil erosion rate estimated by the National Resource Conservation Service is approximately 1 m/1000 yrs33.  We conclude that particles of wood injected at a depth of 10m or more would be exposed by erosion on the millennia time scale only under extreme conditions.

Wood and other biomass is readily degraded by fungi and other organisms at the ground surface, but the degradation rates are much slower under saturated anoxic conditions 11,34.  Enzymes produced by erosion bacteria degrade wood under saturated, anoxic conditions, leading to a reduction in strength and alteration of the cell structure35,36.  This process is slow, however, to our knowledge no studies have been conducted evaluating degradation rates of layers of wood particles under saturated, anoxic conditions at depths of tens of meters.  As a result, we currently have no direct evidence on the degradation rates of wood under the shallow geologic storage conditions, so we turn to analog studies for insights into degradation rates that might occur until a dedicated study can be performed.     

One source of evidence for degradation rate is the occurrence of wood in archeological sites11,34,36.  For example, intact wooden boats up to several millennia old11,36–38 have been found in anoxic, water-saturated sites at a variety of locations37–41. The study of old wooden piles used for foundations indicates that wood degradation rate decreases with depth, presumably due to a decrease in oxygen concentrations with depth 10,41,42.  This is meaningful because our applications would be deeper than the several meters of burial depth of most archeological sites, suggesting that degradation rates may be even slower. 

Intact wood many millennia or tens of millennia old has been documented in the geologic record 43,44.  One example is the occurrence of baldcypress trees between 40ka and 60ka years old with their  primary cell structure intact, which were buried and recently exposed in the Gulf of Mexico 43.  Wood fragments were common in cores of sediments more than 1 My old from the Bengal delta, where its appearance ranged from essentially unaltered wood to altered brown and black44.      

Results from the archeological and geological record demonstrate that some carbon can be stored in wood in the subsurface for millennia or longer, but those results are unable to quantify the release of carbon that might have occurred.  To estimate the fraction of released carbon, we turn to laboratory studies of the permanence of storage of carbon in landfills.  A review by O’Dwyer and others45 shows the fraction of carbon in wood lost under anoxic conditions ranges from essentially 0 up to 0.08 in published laboratory studies, with a few outliers where the fraction was greater.  A laboratory reactor study9 found fractions of carbon released ranged from 0.005 to 0.05, and they concluded that 0.014 was an appropriate value to use for the fraction of carbon in wood released under anoxic conditions.  They also found that the release rate was greatest at the outset of the study and it decreased sharply with time over the first two years of the study.  This result is similar to the finding using a reactors with different wood species where the fraction of degradation products ranged from 0 to 0.078 and the degradation rate decreased significantly over a few years46. These data suggest that the fraction of carbon released by degradation of wood particles will be less than 0.08, and it could be as little as 0.01. 

Lab studies and field observations outlined above provide useful benchmarks, but degradation rates of wood particles in shallow geologic storage could be affected by grain size, oxygen transport, microbial community, nutrient concentration and other factors.  The grain size of injected wood particles would likely be on the order of 1 mm, whereas laboratory studies typically use coarser particles of 10mm or larger.  Diameters of intact trees are one to two orders of magnitude larger (0.1 to 1m).  The greater surface area to volume ratio, and the smaller grain size could cause the degradation rates during shallow geologic storage to be faster than during landfill studies and faster than rates that could occur in-situ.  However, wood particles would be packed together in layers during shallow geologic storage, which would reduce their available surface area for reaction compared to individual particles.   

Landfill studies typically circulate leachate containing an established microbial community, and they monitor the nutrient concentrations and supplement them as needed9,46–48.  This would likely cause the microbe population to be larger and more active in the landfill studies than during shallow geologic storage.  Wood particles would be injected into deep, anoxic conditions in a few minutes, whereas burial of wood at geologic and archeologic sites would likely have occurred over longer time scales, resulting in a longer exposure to oxygen.  These factors would cause the degradation rate under analog conditions to be faster than during shallow geologic storage.     

We conclude that a small fraction (less than 0.1) of the carbon injected into shallow geologic storage would be returned to the atmosphere over time scales of centuries to millenia, according to available analog studies.  

The upward propagation of a hydraulic fracture during injection will increase the risk of compromising storage permanence, however, this risk can be mitigated as a consequence of the multiple injection scenario described during the field experiment.  Upward propagation of a hydraulic fracture occurs where the minimum compression direction is horizontal.  This can be a result of the ambient stress state, or from stress changes due to injection.  In either case, the creation of a sub-vertical pressurized fracture will increase the horizontal compressive stress.  Creating multiple vertical pressurized fractures will further increase the horizontal compression until it exceeds the vertical compressive stress.  Subsequent hydraulic fractures will be sub-horizontal in response to this altered state of stress.  Similar techniques are known to reorient vertical hydraulic fractures49–51.  This means that multiple small injections could be used to suppress upward propagation.