CA1104961A - Process for coal liquefaction - Google Patents

Abstract

ABSTRACT

In two-stage solvent refining of coal by mixing commi-nuted coal under hydrogen pressure with a solvent produced in the process for initial solubilization and reacting the mixture under hydrogen pressure in a second stage, the overall process is improved by enriching the recycled solvent in hydroaromatic hydrocarbons of 14 or less carbon atoms.

Description

This invention relates to an improvement in solvent refining of coal in which components of coal suitable for fuel are extracted from comminuted coal by a solvent and recovered as a low melting point mixture of reduced sulfur and mineral matter content adapted to use as fuel in conventional furances.
The present emphasis on the conversion of coal to sub-stitute solid and liquid fuels has led to several alternative processes which are now being considered. The end use of the resultant converted coal will primarily determine the degree of conversion that must be accomplished and the quality of the desired product. The optimal use of the coal will depend on the sp0cific application.
Among the many processes present,ly being considered is the solvent refinlng o~ coal (SRC) in which coal is treated at an ele~ated temperature ln the presence of a hydrogen donor solvent and hydrogen gas in order to remove the mineral makter, lower the sulfur content of the coal and to oonvert it into a low melting solid which can be solubllized in simple organi¢
solvents. This SRC can also be upgraded through ca~alytic hydro-,~ 20 genation to produce a liquid of higher quality.
Little is known at present as to the exact mechanisms bywhich the coal is transformed into soluble ~orm or of the detailed chemi~al structure o~ the soluble product or even the parent coal.
It is known that many coals are easily solubilized and for others solubillzation is more difficult. Some correlations have been made batween the rank of the coal and ease of so1ubllization and product yield. A somewhat better correlation has bèen found with the petrography o~ the coal. Little is known about the relation-ships to product quality.

3~ ' ~ -2-,,. ,., : , ': ' , .
' ': ' ' , : . . , . , :
.. ... . . .

The initially dissolved coal (SRC) may ha~e utility as a substitute clean fuel or boiler fuel; however, for substitute fuels of higher quality, specifications on viscosity, melting point, ash, hydrogen and sulfur contents are much more stringent~
Attempts to meet these speci~ications by operating the S~C pro-cess more severely have met with many difficulties such as low liquid yields, high hydrogen consumption, difficulty of separat-ing unreacted residue and excessive char formation, which often completely plugs process transfer lines and reactors.
Alternative methods of improving speclfications through catalytic hydrogenation are also difficult. The problems which arise are threefold: (1) SRC components are susceptible to fur-ther condensa~ion and may deposit as coke on catalysts used ~or their converslon, (2) they can also foul the catalysts by physl-cal blockage as their size approaches the pore size o~ conven-tlonal catalysts and (3) they may contain metal ~ontaminants and thelr highly polar nature (particularly nitrogenous and sulfur compounds) can lead to selective chemisorption and thus poison the catalysts.
The precise chemical nature o~ the SRC is still unknown, generally its composition is discussed in terms o~ solubility.
Several classifications are commonly used. These include oils which are hexane or pentane soluble, asphaltenes which are ben-zene soluble and pyridine soluble-benzene insoluble materials.
Of these the asphaltenes and pyridine soluble-benzene insoluble materials are believed to be responsible for high viscosity, sol-vent incompatabllity and processing di~ficulties. Little is known about the pyridine soluble-ben~ene insoluble materials. These have been re~erred to as 'tpre-asphaltenes" which implies that asphaltenes are derl~ed from them, however, this has yet to be established.

~3-.

" " ''''' ~

6~

More information is available on the nature of asphaltenes.
It is common experience that coal liquids contain large quantities of materials known as asphaltenes. In ~act~ it has even been suggested that the formation of asphaltenes is a necessary step in the liquefaction of coal.
The term asphaltene is a rather nebulous and all-inclusi~e classification of organic materials for which a detailed chemical and physical identification is quite difficult, and has not yet been accomplished.
This classificati~n generally refers to high molecular weight compounds, boiling above 650F, which are soluble in ben-zene and insoluble in a light para~finic hydrocarbon (e.g.~ pentane).
Usually no distinction is made regarding polarity, as the term has been used customarily in the characterization of heavy'petroleum fractions (resids, etc.) where the amount of highly polar materials ls small. Howe~er, in coal liquids this may not necessarily be the case due to the hig~ degree of functionality of coal ltself. Thus~
coal liquids of low molecular weight may s~ill be "asphaltenes."
There ls considerable variation in the molecular weight of solu-bil~ize~ coals which arises from differences in the parent coals or different solvent or solvent-reactant systems at the same tempera-ture o~ reaction. This could well be related to colloidal proper-ties of coal liqulds. It is well documented that asphaltenes ~ound in heavy petroleum fractions are colloidal ln nature.
Some comments on'the chemical nature of coal asphaltenes ; 4~ have recently been made. Asphaltenes f'rom Synthoil Process liquids ,.~ , ~ .
were separated into a basic fraction (containing oxygen only as ether or ring oxygen and basic'nitrogen as in pyridine) and an acidic fraction (containing phenolic OH and nitrogen as in pyrrole).

~ 30 The two fractions were found to have very di~ferent properties.
' :

.

.
. : : . .. . .
-. ,~ ,: . ' ' , ~ ' : ' The basic fraction could be hydrotreated only with difficulty~
while the acid fraction underwent facile hydrotreating. This is consistent with reported data on the influence of nitrogen hetero-cycles on conventional hydroprocessing.
Based on these results an acid-base pair structure for asphaltenes was proposed and this structure was extrapolated to that of coal itself. This structure is quite different from the more amphoteric nature of coal which has been proposed previously.
Mechanisms have been proposed for the noncatalyzed forma-tion of asphaltenes from coal. In this work it was concluded thatasphaltenes were a necessary product of coal liquefaction and that oils were derived from asphaltenes. The more polar pyridine soluble materials were not investigated and were assumed to be equivalent to unreacted coal. The maxlmum yield of asphaltenes was ~ound, however, to be a ~unction of the condltions of coal conversion; hydrogen donor solvents greatly reduced the propensity for formation of asphaltenes at low conversion. In addition, it was not determined whether the asphaltene fractions resulting from different conditions were of the same chamical and/or physical nature. Thus, asphaltenes may be inherent constituents of coal products or they could well be the result o~ either thermal or catalytic transformations of more polar materials.
In considerinæ what may be involved in the ~ormation of asphaltenes during coal solubilization or conversion, it may be instructive to consider what is known of coal structure. Coal i5 a rather ~ompllcated network of polymeric organic species the bulk of which is porous in the natural form; the pore system varies from coal to coal. ~ependlng upon the specific nature of the porous structure of each coal~ its chemical constituents and the reaction conditions, the rate of diffusion and mass transport of organic .

molecules through the pores could have a strong effect on the rates of dissolution, hydrogen transfer and hydrogenation and hydrocracking reactions and thus on the ultimate yield of soluble product.
As the rank of coal becomes higher, an increasing number of colloidal size aggregates (20-50A) can be observed by X-ray scattering and diffraction.
If, in the early stages of the dissolution of coal these colloidal aggregates dissociate to some degree and go into solu-tion~ the molecular weight of the lowest unit appears to be con-sistent with the lowest molecular weights observed in solubilized coals (~ 500 MW). This compar~son may be coincidental, however~
Unfortunately, in order to dissolve coal it ls generally ~ound that temperatures in excess o~ 300C are necessary. It is also known that coal begins to pyrolize and evolve volatile matter at temperatures as low as 250C (depending on rank), and by 350C
considerable material has evolved. This strongly suggests that exkensive internal rearrangement of the coal occurs durlng the dissolution process. Rearrangement can include hydrogen migration to produce highly condensed aromatic rings as well as further association of small colloidal aggregates or condensation of reac-tive species. Ma~or physical changes in the pore system of the solid coal have also been reported.
This rearrangement could possibly be responsible ~or some of the very high molecular weights (~ 3000 M~) observed with some solvents. No detailed relationships of solvent type and/or reac-tion condition to the molecular weight distribution of solubilized coal has yet been established. Similarly, the possibility of rever sible molecular weight changes, due to recondensation causing in-creased molecular welghts at various temperatures, has not beenlnvestigated thoroughly~

An alternative route to high molecular weight is through the catalytic in~luence of inorganic coal minerals which are present in the processing of coal. It is known that some coals are more reactive than o~hers, producing higher yields of liquid products at shorter residence times. It is belleved that this is due to the fact that the initial coal products are reactive and condense to char unless proper reaction conditions are established. This further condensation could well be a catalytic phenomenon induced by intrinsic coal minerals.
Another more subtle consequence of certain inorganlc constituents is their influence on the physical properties of pyrolytic coal chars~ and thus on the diffl1sional properties imposed on reactive intermediates. The volume of char has been observed to vary by a factor of ~our or more~ with little change in weight, by varying the type of inorganic contaminants in a given bituminous coking coal. The pore system of the resultant chars must be vastly different and changes of this type magnltude in the physical structure of the coal or char could greatly influence ma~s transport of intermediates produced within the pore system. Mass trans~er limitation during the pyrolysis and hydrogasification of some coals at high temperatures has recently been established.
This study showed that for some coals, reactive primary products are formed which can recombine to produce char i~ the conditions are not properly ad~usted. The criticality was ~ound to be the rate of' di~f`usion of the reactive species out of the coal relative to its rate of conversion to char.
At lower temperatures 5 the rates of reaction are~ of - course, slower and thus less susceptible to mass t:ransport , limitations. However, the imposition of a liquid phase~
commonly used in liquefaction processes, may greatly enhance diffusional restrictions. Recent model studies conducted in aqueous systems, have shown that restriction of di~fusion through porous structures with pore radii ranging from 45A to 300A for even relatively small solute molecules is very signlficant.
At t.he present stage of the art, the accumulated infor-mation is largely empirical, with little basis ~or sound extra-polation to predict detailed nature of solvent and processing condltlons ~or optlmum yield and quality of solvent refined coal.
It is recognized that the poorly understood asphaltenes are probable sources of many of the problems encountered, e.g.
formation o~ char at processing conditions conducive to efficient separation of mineral matter (ash) and sulfur from desired product at hlgh yield.
In the process of converting coal to a low sulfur, low melting solid by use of recycled product fractions as solvent, several reaction steps occur. Generally coal is admixed with a suitable solvent recycle stream and hydrogen and the slurry is passed through a preheater to raise the reactants to a desired reaction temperature. For bituminous coal, the coal is sub-stantially di~solved by the time it exits the preheater. Sub~
bituminous coals can be dissolved but care must be exercised not to raise the temperature too high and thus promote charring.
The products exiting from the preheater are then trans-ferred to a larger backmixed reactor where further conversion takes place to lower the heteroatom content o~ the dissolved coal to speci~ication sulfur content and melting point. The geometry of this reactor 1s such that the linear flow rate through it is not sufficient to discharge a substantial quantity of particulate matter of a desired size. Thus the reactor volume becomes filled (at steady state) up to about 40 vol % by solids which are produced from the coal. These solids have been shown to be catalytic for the removal of heteroatoms and the introduction of hydrogen into the coal products and solvent. The products exiting the reactor are initially separated by flash distillation, which depressurizes the stream and remove gases and light organic liquids. The products are further separated (filtratlon, centrifugation, solvent precipitat.~on, etc.) and the filtrate is distille~ to recover solvent range materlal (for recycle) and the final product SRC.
The present invention provldes:
1. A process for solvent refining coal by heating a mixture of comminated coal and a steady state recycle solvent derived as hereinafter recited, maintaining the mixture under hydrogen pressure at reaction conditions for solubilization of coai components, separating undissolved solids from the resultant reaction products and separating solvent refined coal product from a recylce solvent ~raction for mixture with coal as aroresaid, characterized by adding light hydrogen donor components having 14 or less carbon atoms to the recycle solvent whèreby the solvent mlxed with coal has a hi~her proportion of such light hydrogen donor compounds than the original recycle solvent fraction.
Under typical conditions presently practised, we have found that ~he recycle solvent contains a mixture of hydroaromatic compounds thydrogen donors) and condensed aromatic compounds in whlch the hydroaromatics are below the thermodynamic ratios allswed under the cond1tions of the reactions (about 2000 psi H2 _ g _ ~&~

and 400 - 450C.). Phenols are also present in the recycle solvent, with certain consequences.
These hydroaromatics achieve a steady state concentration whlch is dictated by the rate of hydrogen consumption (by hydrogen donation to coal) and the rate of rehydrogenation oP condensed aromatics (catalyzed by reactor solids). In the preheater only hydrogen consumption occurs. In the reactor both hydrogen consumption and rehydrogenation occur but apparently in existing processes the rehydrogenation step is slow; thus, thermodynamic equilibrium is not established. We have ~ound that the ma~or contribution to hydrogen donation are partially hydrogenated aromatic hydrocarbons (e.g. tetralin, dihydrophenanthrene) and partially hydrogenated aromatic phenols (ethers can also be present).
The concentration o~ hydrogen donors in the solvent thus varies in the solvent depending on its partlcular location in the proce~s.
We have also found that the ma~or contributors to the hydrogen donor capacity of typlcal recycle solvents are hydroaromatlc h~drocarbons having fewer than 14 carbon atoms~ in particular tetralin and methyltetralins and hydrophenanthrenes.
Based on these observations we prepose an improved coal lique~actlon process in which the hydrogen donor capacity of a given solvent is maintained at a higher steady state level than is presently practiced~ This i9 done by adding a light hydroaromatlc feed to the recycle s~ream aPter a standard steady state condition has been achieved (including buildup of catalytic reactor solids).
This hlgher level of hydrogen donor capacity will be sustained ~s any hydrogen consumed will be replaced through the catalytic action oP the reactor solids.

,. . ' - . ~
.

6~

The regenerated li~ht hydroaromatics are recoverable hy proper malntenance o~ the ~lash separator temperature and by a second condenser (cold separator). These recovered light hydro-aromatics are then recombined with heavier recycle solvent streams and admixed with fresh coal feed.
The only perturbation of this capacity whlch might occur would be due to the production of the same specific components (hydroaromatics and aromatics) ~rom the coal itsel~ or the conversion o~ hydroaromatics to lower molecular lQ welght speci.es. Our experience indicated that these two processes are relatively slow, but if the hydrogen donor capacity does app-roach a lower value occasionally the light hydroaromatic stream can be catalytically upgraded in a separate reactor.
One advantage to this process is that catalytic regenera-tion need only be done occasionally on small streams. The overall solvent properties (polycondensed aromatics and phenols) wil-l~ not be changed by ~his process. Thus, the solvent will be capable cf dissolving even the most polar SRC's. The SRC product quality will be improved under set reactlon Gondltions as the solvent will be more reactive. Alternatively~ the process ~low rate can be increased.
Another advantage of this process is that the tendency of the SRC
product to form char through competitive regressive reactions will be dramatically reduced. The process can ~urther be improved by removal of phenols ~rom the solvent.
The fact that reactor solids catalytically regenerate the solvent is known. There has already been proposed a process which continuously regenerates a hydrogen rich solvent through catalytic hydrogenation externally. That process, however, because o~ the requirement o~ severe solvent rehydrogenation, produces a solvent which has poor solubility properties for SRC and excessive amounts .

of hydrogen are consumed merely to dissolve the coal. A further proposed process ~H-Coal) employs a catalyst present in the reac-tor, which catalyst, however, is an expenslve commercial catalyst that loses activity due to poisoning by coal inorganic constituents and must continuously be replaced by fresh catalyst.
By contrast, the present invention requires only occasional rehydrogenation o~ a small portion of the solvent and little, i~
any, catalyst deactivation occurs as the stream being processed is ~ree of inorganic contaminants. T~e solvent properties of the recycle solvent of this inventlon are much superior to those of the prior proposals and even highly polar SRC's can be di~solved. In sum~ the present invention allows much more flexibility in the SRC
process with less tendency for char formatlon and a more reactive solvent which can lead to smaller reactors and lower capital costs.
The overall hydrogen consumption for a given SRC product speciflca-tion may also be reduced.
The process of the invention applies se~eral findings derived from study of solvents in solvent refining of coal. The term "solvent" is used here because of custom in the art despite the fact that the liquid applied in solvent refining of coal performs important functions in the chemical reactions involved in addltlon to the physical function of dissolving components of the coal charged to the system and maintaining a~ solutes the pro-ducts o~ chemical conversion arising in the process. In fact, high solvent power for the ultimate SRC product is not an adequate indi-cator of suitability for use as solvent applied to the coal charged to the process. As the process proceeds, the chemical character of the solute changes by removal of polar groups and the product takes on a greater degree of hydrocarbon nature.

In order to be commercially attractiveg the process must provide solvent as a recycle product o~ the solvent refining pro-cess. We have found that these considerations are satisfied to better effect by enhancing the proportion of hydrogen donors con-stituted by partially hydrogenated polycyclic aromatic hydrocarbons of 14 or less carbon atoms such as tetralin, methyl tetrallne and hydrophenanthrene. To accomplish this result, this invention pro-- vides for enriching the recycle solvent in such compounds. One method contemplated is to add these desirable compounds from a source external to the process. An alternative technique is sepa-ration of these compounds from the reaction products and blending the same with heavier recycle solvent removed at a later stage.
The light fraction may be ~ub~ected to catalytic hydrogenation before blending with the heavy fraction and may be blended in desired proportions to suit needs of the process dictated by nature of the coal under treatment and the pro~ected end use o~ the product.
The novel technique permits control of the degree of h~drogen donor components. If too much hydrogen is added back to the solvent, it becomes deficient in aromatic content and the coal products become insoluble.
The present invention will be more fully understood by consideration of speciflc embodiments described below with refer-ence to the drawings.
Figure 1 is a graphical representation of hydrogen donor oapacity of the solvent in SRC prooessing; and Figure 2 is a diagrammatic flow sheet illustrative o~ best modes contemplated by us of carrying out our invention.
The nature of changes in hydrogen donor content of the solvent is set out in the graph constituting Figure 1. The graph contemplates a process in which coal and recycle solvent are pre-heated under hydrogen pressure and passed to a back mixed reactor contalning mineral solids derived from coal which is also under hydrogen pressure.
It is typical of such processes that the ratio of hydrogen donors to corresponding aromatics (e.g. 7 tetralin/naphthalenej is significantly below the khermodynamic limit imposed by conditions at thermodynamic equilibrium of hydrogen, hydrogen donor and aro-matic hydrocarbon. As shown at the lefthand side of the graph, solvent enters the preheater at thesteady state limit on hydrogen donors. Hydrogen donor content drops rapidly through the preheater and on into the reactor as the donors are stripped oP hydrogen to satisfy demands by coal fragments for the hydrogen which inhlbits polymerization and formation of insoluble char. In the reactor, polycycllc aromatics are hydrogenated to regenerate donors under the catalytic effect of the accumulated solids. During an initial period in the reactor, donor content continues to decline as the demand for trans~erred hydrogen exceeds-the rate o~ rehydrogenation of polycyclic aromatics. As that demand drops, rehydrogenation becomes the dominank reaction with rise of hydrogen donor content to thesteady state limit at time of discharge from the reactor.
The inventlon as illustrated by Figure 2 is pre~erably applied a~ter 8teady state has been achieved by a system of the type described, although it may be applied continuously beginning on start-up. A typical operation may consist in solvent reflning o~ Monterey Mine Illinois ~6 coal on which inspection data are shown in Table 1.

Name of Coal Illinois #6 c State Illinois County Maco~pin Seam 6 c Name of Mine Monterey % Moisture ~as rec) 12.81 u ~ Ash (as rec) 9.43 u % Volatile Matter 41.73 Fixed Carbon 47.45 o ~ BTU (as rec) 10930.
BTU 12536.
Free Swelllng Index ---C 69.72 H 4.98 a % 0** 8.20 % N 1 o8 % S (total) 5 14 % S (pyritic) 2.26 ~ ~ % S (organic) 2.70 P ~ % S (sul~ate) 0.18 % Cl o.o6 % Ash 10.82 : * All analyses are given on a dry weight basls unless okherwlse sta~ed.
.

** By dlf~erence.

'~0 ..
~h~

a) a~ a o ~

89 3 1 l 1 2 2 l loo Mean Maximum ReI'lectance in Oil ( 564 nm) . o . 47%

... .. . . ..
. . .: -:
- . . ... . . . . ..

Thatcoals,crushed to pass 100-200 mash standard sieve having a maximum particle dimension of about .15 .07 mm is admitted by line 10 to mixer 11 where it is mixed with 1 to 6 parts by weight of recycle solvent from line 12 and hydrogen from line 13. Alternately hydrogen can be added only to the backmixed reactor 15. The mixture passes to and through a pre-heater 14 where it is brought to a temperature of 350-480C during a transit time of 1-10 minutes. Components of the coal are largely ~aken into solution in preheater 14 and the reac~ions characteristlc of the process are initiated, with resultant depletion of donor hydrogen. The reaction mixture is transferred to back mixed rea~-tor 15 operated to retain undissolved coal solids to the extent of up to 40% of the reactor volume. Residence time of the reaction mixture in reactor 15 is about 20~120 minutes average while the reactlon of dissolved coal proceeds in known manner concurrently with hydrogenation of polycyclic aromatic compounds to regenerate hydrogen donor capacity at temperature of 300-460C and 500-3000 psig. During initial operation the flash separatar 21 does not have to be used to full capacity and a portion of the e~fluent of reactor 15 is conducted through by~pass line 16 to solids sepera-tor 17 ~or removal o~ ash, unreacted coal, iron sulfides, coke and the like by filters, centrifuges, precipitation or other appropri-ate means. The clari~ied liquid passes to distillation ~acility 18 for recovery of solvent refined coal (SRC) by line 19 free of recycle solvent which is returned by line 12 to mixer 11 as des~
cribed above. A portlon or all of the recycle solvent may be ~ diverted through phenoI extractor 20 for separate recovery of ; phen~ls, e.g., by caustic wash.
The described operation conforms generally to known practice 3 and is conducted for a period adequate to achieve steady state (say 6~

20 to 200 hours) as shvwn by constant composition of recycle solvent and SRC. Upon reaching steady state, the full effluent of reactor 15 is diverted ko flash separator 21 where reduction of pressure ko about 15-150 psig causes evaporation of compounds having 14 carbon atoms or less. That vapor phase fraction is cooled to about 180-350C at 15-150 psig and passed to cold sepa-rator 22 from which normally gaseous compounds, boiling below about 20C are removed by vent 23 for use as fuel or other purpose.
Lighk liquids withdrawn by line 23 from separator 2~ are recycled ko mixer 11 in a ratio to heavy liquid recycle solvenl; from llne 12 such that total recycle solvent to mixer 11 contains a proportlon of hydrogen donors havlng 14 or less carbon atoms ~reater than the prlor steady state operatlon with return o~ a slngle recycle solvent stream.
A portion or all of the light recycle ln line 23 may be diverted through phenol removal facility 24 for recovery o~ pro-duct phenols, to ad~ust solvent properkies and the like. Ad~ust-ment of hydrogen donor content may be achleved by dlversion of khe li6ht solvent recycle in whole or part, continuouslg or inter-mittently through catalytic hydrogenation reackor 25.
If monophenols are to be preserved ln external hydrogena-tlon, then a second distillation may have to be done to remove and separately recycle them. This could be done by an interme-diate distillatlon ak ~210C. (This saves oresol but would bypass tetralin/naphthalene which would still be hydro~enaked.) It will be apparent that increase of the content o~ hydro-gen donors containing 14 or less carbon atoms may be achieved by adding such compounds from an external source inskead of or as a supplement to the technique descrlbed above, and such operation is contemplated within the scope of the invention.
', :

. ~

- ,. : -, , . , .. . , - - . ~ . .. .

.

5a6~

As well known in the art, coals vary in composition and require varied conditions for optimal production o~ SRC. In general, the invention contemplates use o~ solvent in the range of l to 6 parts of solvent per part of coal by weight. In the mixer, coal and solvent will optionally be mixed with 5 to 50 standard cubic feet of hydrogen per pound of coal then heated in the preheater for l to lO minutes to a temperature in the range of 350C to 460C. Alternatively the hydrogen can be added directly to the back mixed reactor. In reactor 15 khe mixture is held ~or a period of 20 to 120 minutes at 350C to 460C and 500 to 3000 psig. The recycle solvent is separated as a ~raction of the reaction products having a boiling range o~ 190C to 500C
and a quantity which will satis~y the needs o~ the reaction stage when admixed wlth an amount o~ light recycle solvent adequate to give the descrlbed ratio of 14 or less carbon atom donors.

.
. .

Claims (5)

1. A process for solvent refining coal by heating a mix-ture of comminuted coal and a steady state recycle solvent derived as hereinafter recited, maintaining the mixture under hydrogen pressure at reaction conditions for solubilization of coal com-ponents, separating undissolved solids from the resultant reaction products and separating solvent refined coal product Prom a recycle solvent fraction for mixture with coal as aforesaid, characterized by adding light hydrogen donor components having 14 or less carbon atoms to the recycle solvent whereby the solvent mixed with coal has a higher proportion of such light hydrogen donor compounds than the original recycle solvent fraction.

2. A process according to Claim 1 wherein hydrocarbons having 14 or less carbon atoms are separated from the reaction products and added to the recycle solvent as the light hydrogen donor components.

3. A process according to Claim 1 wherein the light hydrogen donor components are constituted by a fraction consist-ing essentially of compounds of 14 or less carbon atoms separated from the reaction products and subjected to partial catalytic hydrogenation.

4. A process according to Claim 1 wherein phenols are separated from the recycle solvent fraction before mixture with the coal.

5. A process according to Claim 3 wherein the fraction consisting essentially of compounds of 14 or less carbon atoms is process for separation of phenols therefrom before mixture with the coal.